128 62 18MB
English Pages 306 [304] Year 2015
Deterrence and Defense
BOOKS FROM THE CENTER OF INTERNATIONAL STUDIES WOODROW WILSON SCHOOL OF PUBLIC AND INTERNATIONAL AFFAIRS PRINCETON UNIVERSITY Gabriel A. Almond, The Appeals of Communism, G. A. Almond and J. S. Coleman, editors, The Politics of the Developing Areas Robert J. C. Butow, Tojo and the Coming of the War Bernard C. Cohen, The Political Process and Foreign Policy: The Making of the Japanese Peace Settlement Percy E. Corbett, Law in Diplomacy Herman Kahn, On Thermonuclear War W. W. Kaufmann, editor, Military Policy and National Security Klaus Knorr, The War Potential of Nations Klaus Knorr, editor, NATO and American Security Lucian W. Pye, Guerrilla Communism in Malaya Rolf Sannwald and Jacques Stohler, Economic Integration: Theoretical Assumptions and Consequences of European Unification Translated by Herman F. Karreman Glenn H. Snyder, Deterrence and Defense Sidney Verba, Small Groups and Political Behavior: A Study of Leadership Charles De Visscher, Theory and Reality in Public International Law. Translated by P. E. Corbett Myron Weiner, Party Politics in India
Deterrence and Defense TOWARD A THEORY OF NATIONAL SECURITY
BY GLENN H. SNYDER
PRINCETON UNIVERSITY PRESS PRINCETON, NEW JERSEY
1961
COPYRIGHT © 1961 BY PRINCETON UNIVERSITY PRESS ALL RIGHTS RESERVED L.C. CARD: 61-12102
PRINTED IN THE UNITED STATES OF AMERICA BY PRINCETON UNIVERSITY PRESS, PRINCETON, NEW JERSEY
Preface THIS BOOK was motivated largely by the apparent lack of any definite agreed criteria for making rational decisions in national security policy. Many of the "great debates" in this field seem to be characterized, even partially caused, by the lack of such criteria. We have no scarcity of prescriptive arguments for or against some policy alternative, and many of these are very well reasoned and powerfully stated. Such prescrip tive analyses should certainly continue, both in the scholarly world and in the government. But there is an urgent need for something else—a systematic method, or theory, which can be used by the governmental generalist, or the interested layman, to compare and weigh the argu ments for particular policies, which often sharply conflict. This method must attempt to clarify the kinds of values we get from having and using military forces, the ways in which these values are related, how they are produced by different types of forces, and how choices can be made which maximize total value. This does not purport to be a general theory of national security. It deals only with what I perceive to be the central issue in national se curity policy—namely, the choice between deterrence of, and defense against, military attacks, and the allocation of resources and the man agement of diplomatic declarations in the service of these two ends. Thus, except for incidental treatment, it leaves out such matters as economic aid, military aid, psychological warfare, disarmament, and most aspects of diplomacy. A theory which attempted to embrace all these matters would certainly be over-ambitious, if not impossible to formulate. But I believe there is an important and manageable core of logic in the military segment of the field, a logic which centers on the twin goals of deterrence and defense. Rather than "taking a position" for or against some policy or set of policy alternatives, I have attempted only to elaborate and illustrate the nature of, and relationship between, these two central values, and to suggest criteria for combining them in the most efficient way. Chap ter 1, the most theoretical chapter, attempts to analyze the nature and constituent elements of deterrence and defense and the relation of these concepts to the notion of balance of power. Chapters 2, 3, 4, and 5, which are more policy-oriented, deal with problems of deterrence and defense in the context of all-out war, NATO strategy, the "Grey Areas," and declaratory policy and force demonstration. Chapter 6, returning to a more theoretical orientation, presents an analytical framework for choosing between deterrent and defense values and for making choices among various types of military capabilities. I wish to express my deep indebtedness to the Center of International
vi
Preface
Studies, Princeton University, for providing me with the financial means to write this book, over a period of two years, and for making possible the very fruitful advice and stimulation which I received from other members of the Center during that time. In this connection, I particu larly wish to thank Professor Frederick S. Dunn, Director of the Center, and Professor Klaus Knorr, Associate Director. Professor Knorr's care ful criticism of the manuscript, at various stages of its production, was especially helpful. Among the others who contributed, either in con versations or by their comments on portions of the manuscript, are Edgar S. Fumiss, Jr., and S. T. Beza of Princeton University; Arthur L. Burns of the Australian National University; Daniel Ellsberg, Herman Kahn, and William W. Kaufmann, all of The RAND Corporation; Thornton Read of Bell Telephone Laboratories; Robert E. Osgood of the University of Chicago; and Morton Halperin and Thomas C. Schelling of Harvard University. In my case, an acknowledgment of support and assistance from my wife is due for reasons much more compelling than custom and the usual wear and tear on the domestic relationship which accompanies the production of a book. Her sensible comments on various difficult points did much to clarify them for me; also she contributed to the typing of portions of an early draft. GLENN H. SNYDER December 1,1960
Contents PREFACE
ν
1. DETERRENCE AND DEFENSE; A THEORETICAL INTRODUCTION
3
A. Examples of Choices and Conflicts Between Deterrence and Defense
5
B. The Technological Revolution
8
C. Deterrence 1. The Logic of Deterrence 2. Denial vs. Punishment 3. A Mathematical Illustration 4. Uncertainty
9 12 14 16 27
D. Defense 1. The Components of Defense Value 2. Strategic Value and Deterrent Value
30 31 33
E. The New Balance of Power 1. Differences Between the Balance of Terror and the Balance of Power 2. Interaction Between the Balance of Terror and the Balance of Power
41 42 46
2. DETERRENCE AND DEFENSE WITH STRATEGIC NUCLEAR POWER
52
A. A Note on Terminology
53
B. Deterrence of All-out Surprise Attack
54
C. Defense by Offense in All-out War 1. All-out Countercity Retaliation 2. All-out Counterforce Retaliation 3. Limited Countercity Retaliation as a Bargaining Tactic 4. Limited Counterforce Retaliation 5. Strikes Against the Enemy's Tactical Forces 6. No Retaliation
63 68 69 69 73 74 75
D. Deterrence of Soviet Aggression in Eurasia
79
E. Qualitative Requirements
85
• ··
Contents
VlU
F. Civil Defense
95
G. The Conditions of Stability
97
H. Pre-emptive Attack
104
I. Accidental War
110
J. Summary and Conclusions
115
3. DETERRENCE AND DEFENSE IN WESTERN EUROPE
120
A. The Sword and the Shield in NATO Strategy
120
B. The Role of Tactical Nuclear Weapons 1. Extended Limited Nuclear War
137 144
C. Future Military Alternatives for NATO 1. A U.S. First-strike Counterforce Capability 2. A Conventional Defense Capability 3. Independent Strategic Nuclear Forces for NATO Countries 4. An Integrated NATO or European Nuclear Deterrent 5. Limited Retaliation 6. Combination of Alternatives
146 148 148
4. DETERRENCE AND DEFENSE IN THE "GREY AREAS"
157 174 193 219 225
A. Overt Attack
225
B. Indirect Aggression
231
5. DECLARATORY POLICY AND FORCE DEMONSTRATIONS
239
A. Declaratory Policy 1. Clarity vs. Ambiguity 2. Reiteration of Threats 3. Who Should Make the Threat? 4. Publicity vs. Secrecy 5. The Context
240 246 249 249 250 251
B. Force Demonstrations 1. The Effects of Force Demonstrations
252 254
Contents
ix
6. THE RECONCILIATION OF DETERRENCE AND DEFENSE
259
A. The Requirements Approach vs. the Limited Economy Approach
260
B. The Balancing of Deterrence, Defense, and Resource Cost 1. A Mathematical Example
265 269
C. The Marginal Utility Criterion 275 1. The Independent Utility of Strategic Nuclear Forces 276 2. The Independent Utility of Tactical Conventional Forces 280 3. Interdependent Utilities 282 INDEX
291
Deterrence and Defense
1 Deterrence and Defense: A Theoretical Introduction NATIONAL SECURITY still remains an "ambiguous symbol," as one scholar described it almost a decade ago.1 Certainly it has grown more ambiguous as a result of the startling advances since then in nuclear and weapons technology, and the advent of nuclear parity between the United States and the Soviet Union. Besides such technological complications, doctrine and thought about the role of force in interna tional politics have introduced additional complexities. We now have, at least in embryonic form, theories of limited war, of deterrence, of "tactical" vs. "strategic" uses of nuclear weapons, of "retaliatory" vs. "counterforce" strategies in all-out war, of "limited retaliation," of the mechanics of threat and commitment-making, of "internal war," "pro tracted conflict," and the like. Above all, the idea of the "balance of terror" has begun to mature, but its relation to the older concept of the "balance of power" is still not clear. We have had a great intellec tual ferment in the strategic realm, which of course is all to the good. What urgently remains to be done is to tie together all of these concepts into a coherent framework of theory so that the end-goal of national security may become less ambiguous, and so that the military means available for pursuance of this goal may be accumulated, organized, and used more efficiently. This book can claim to make only a start in this direction. The central theoretical problem in the field of national security policy is to clarify and distinguish between the two central concepts of deter rence and defense. Essentially, deterrence means discouraging the enemy from taking military action by posing for him a prospect of cost and risk outweighing his prospective gain. Defense means reducing our own prospective costs and risks in the event that deterrence fails. Deterrence works on the enemy's intentions; the deterrent value of military forces is their effect in reducing the likelihood of enemy mili tary moves. Defense reduces the enemy's capability to damage or de prive us; the defense value of military forces is their effect in mitigating 1 Arnold Wolfers, " 'National Security' as an Ambiguous Symbol," Political Science Quarterly, Vol. LXVII, NO. 4 (December 1952), pp. 481ff.
4
Deterrence and Defense:
the adverse consequences for us of possible enemy moves, whether such consequences are counted as losses of territory or war damage. The concept of "defense value," therefore, is broader than the mere capacity to hold territory, which might be called "denial capability." Defense value is denial capability plus capacity to alleviate war damage. It is a commonplace, of course, to say that the primary objectives of national security policy are to deter enemy attacks and to defend successfully, at minimum cost, against those attacks which occur. It is less widely recognized that different types of military force con tribute in differing proportions to these two objectives. Deterrence does not vary directly with our capacity for fighting wars effectively and cheaply; a particular set of forces might produce strong deterrent effects and not provide a very effective denial and damage-alleviating capability. Conversely, forces effective for defense might be less potent deterrents than other forces which were less efficient for holding terri tory and which might involve extremely high war costs if used. One reason why the periodic "great debates" about national security policy have been so inconclusive is that the participants often argue from different premises—one side from the point of view of deterrence, and the other side from the point of view of defense. For instance, in the famous "massive retaliation" debate of 1954, the late Secretary of State Dulles and his supporters argued mainly that a capacity for mas sive retaliation would deter potential Communist mischief, but they tended to ignore the consequences should deterrence fail. The critics, on the other hand, stressed the dire consequences should the threat of massive retaliation fail to deter and tended to ignore the possibility that it might work. The opposing arguments never really made contact be cause no one explicitly recognized that considerations of reducing the probability of war and mitigating its consequences must be evaluated simultaneously, that the possible consequences of a failure of deterrence are more or less important depending on the presumed likelihood of deterrence. Many other examples could be cited. Perhaps the crucial difference between deterrence and defense is that deterrence is primarily a peacetime objective, while defense is a wartime value. Deterrent value and defense value are directly enjoyed in different time periods. We enjoy the deterrent value of our military forces prior to the enemy's aggressive move; we enjoy defense value after the enemy move has already been made, although we indirectly profit from defense capabilities in advance of war through our knowl edge that if the enemy attack occurs we have the means of mitigating
A Theoretical Introduction
5
its consequences. The crucial point is that after the enemy's attack takes place, our military forces perform different functions and yield wholly different values than they did as deterrents prior to the attack. As deterrents they engaged in a psychological battle—dissuading the enemy from attacking by attempting to confront him with a prospect of costs greater than his prospective gain. After the enemy begins his at tack, while the psychological or deterrent aspect does not entirely disappear, it is partly supplanted by another purpose: to resist the enemy's onslaught in order to minimize our losses or perhaps maximize our gains, not only with regard to the future balance of power, but also in terms of intrinsic or non-power values. That combination of forces which appeared to be the optimum one from the point of view of de terrence might turn out to be far inferior to some other combination from the point of view of defense should deterrence fail. In short, maximizing the enemy's cost expectancy may not always be consistent with minimizing our own. Thus we must measure the value of our military forces on two yardsticks, and we must find some way of com bining their value on both yardsticks, in order accurately to gauge their aggregate worth or "utility" and to make intelligent choices among the various types of forces available. Before launching into a theoretical analysis of the concepts of deter rence and defense, it may be useful to present a sampling of policy is sues involving a need to choose between deterrence and defense; the examples will be treated in more detail in subsequent chapters.
A. Examples of Choices and Conflicts Between Deterrence and Defense A strategic retaliatory air force sufficient only to wreak minimum "unacceptable" damage on Soviet cities—to destroy, say, 20 cities— after this force had been decimated by a surprise Soviet nuclear attack, would have great value for deterring such a surprise attack and might be an adequate deterrent against that contingency. But if deterrence were to fail and the Soviet attack took place, it would then not be ra tional to use such a minimum force in massive retaliation against Soviet cities, since this would only stimulate the Soviets to inflict further dam age upon us and would contribute nothing to our "winning the war." If we are interested in defense—i.e., in winning the war and in mini mizing the damage to us—as well as in deterrence, we may wish to have (if technically feasible) a much larger force and probably one of dif-
6
Deterrence and Defense:
ferent composition—a force which can strike effectively at the enemy's remaining forces (thus reducing our own costs) and, further, either by actual attacks or the threat of attacks, force the enemy to surrender or at least to give up his territorial gains. The threat of massive nuclear retaliation against a Soviet major ground attack in Western Europe may continue to provide considerable deterrence against such an attack, even if actually to carry out the threat would be irrational because of the enormous costs we would suffer from Soviet counterretaliation. Strategic nuclear weapons do not provide a rational means of defense in Western Europe unless they not only can stop the Russian ground advance but also, by "counterforce" strikes, can reduce to an acceptable level the damage we would suffer in return. We may not have this capability now and it may become altogether infeasible as the Soviets develop their missile technology. For a means of rational defense, therefore, NATO may need enough ground forces to hold Europe against a full-scale attack by Soviet ground forces. This does not mean, however, that we necessarily must maintain ground forces of this size. If we think the probability of attack is low enough, we may decide to continue relying on nuclear deterrence primarily, even though it does not provide a rational means of defense. In other words, we might count on the Soviet uncertainties about whether or not nuclear retaliation is rational for us, and about how rational we are, to inhibit the Soviets from attacking in the face of the terrible damage they know they would suffer if they guessed wrong. An attempt to build an effective counterforce capability, in order to have both a rational nuclear defense and a more credible nuclear de terrent against ground attack in Europe, might work against the deter rence of direct nuclear attack on the United States. Since such a force, by definition, would be able to eliminate all but a small fraction of the Soviet strategic nuclear forces if it struck first, the Soviets might, in some circumstances, fear a surprise attack and be led to strike first themselves in order to forestall it. Tactical nuclear weapons in the hands of NATO forces in Europe have considerable deterrent value because they increase the enemy's cost expectation beyond what it would be if these forces were equipped only with conventional weapons. This is true not only because the tactical weapons themselves can inflict high costs on the enemy's forces, but also because their use (or an enemy "pre-emptive" strike against them) would sharply raise the probability that the war would spiral to all-out dimensions. But the defense value of tactical nuclear
A Theoretical Introduction
7
weapons against conventional attack is comparatively low against an enemy who also possesses them, because their use presumably would be offset by the enemy's use of them against our forces, and because in using such weapons we would be incurring much greater costs and risks than if we had responded conventionally. For deterrence, it might be desirable to render automatic a response which the enemy recognizes as being costly for us, and communicate the fact of such automation to the enemy, thus reducing his doubts that we would actually choose to make this response when the occasion for it arose. For example, a tactical nuclear response to conventional aggres sion in Europe may be made semi-automatic by thoroughly orienting NATO plans, organization, and strategy around this response, thus in creasing the difficulty of following a non-nuclear strategy in case of a Soviet challenge. But such automation would not be desirable for de fense, which would require flexibility and freedom to choose the least costly action in the light of circumstances at the time of the attack. The Continental European attitude toward NATO strategy is gener ally ambivalent on the question of deterrence vs. defense; there is fear that with the Soviet acquisition of a substantial nuclear and missile capability, the willingness of the United States to invoke massive re taliation is declining, and that therefore the deterrent to aggression has weakened. Yet the Europeans do not embrace the logical consequence of this fear: the need to build up an adequate capacity to defend Eu rope on the ground. A more favored alternative, at least in France, is the acquisition of an independent strategic nuclear capability. But when Exiropean governments project their imaginations forward to the day when the enemy's divisions cross their borders, do they really envisage themselves shooting off their few missiles against an enemy who would surely obliterate them in return? One doubts that they do, but this is not to say that it is irrational for them to acquire such weap ons; they might be successful as a deterrent because of Soviet uncer tainty as to whether they would be used, and Soviet unwillingness to incur the risk of their being used. Further examples easily come to mind. For the sake of deterrence in Europe, we might wish to deploy the forces there as if they intended to respond to an attack with nuclear weapons; but this might not be the optimum deployment for defense once the attack has occurred, if the least-cost defense is a conventional one. For deterrence of limited ag gressions in Asia, it might be best to deploy troops on the spot as a "plate-glass window." But for the most efficient and flexible defense
8
Deterrence and Defense:
against such contingencies, troops might better be concentrated in a central reserve, with transport facilities for moving them quickly to a threatened area. As Bernard Brodie has written,2 if the object of our strategic air forces is only deterrence, there is little point in developing "clean" bombs; since deterrence is to be effected by the threat of dire punishment, the dirtier the better. But if we also wish to minimize our own costs once the war has begun, we might wish to use bombs producing minimum fall-out, to encourage similar restraint in the enemy. For deterrence, it might be desirable to disperse elements of the Strategic Air Command to civilian airfields, thus increasing the num ber of targets which the enemy must hit if he is to achieve the neces sary attrition of our retaliatory power by his first strike. However, this expedient might greatly increase the population damage we would suffer in the enemy's first strike, since most civilian airfields are located near large cities, assuming that the enemy would otherwise avoid hit ting cities.3
B. The Technological Revolution The need to choose between deterrence and defense is largely the result of the development of nuclear and thermonuclear weapons and long-range airpower. Prior to these developments, the three primary functions of military force—to punish the enemy, to deny him territory (or to take it from him), and to mitigate damage to oneself—were em bodied, more or less, in the same weapons. Deterrence was accom plished (to the extent that military capabilities were the instruments of deterrence) either by convincing the prospective aggressor that his territorial aim was likely to be frustrated, or by posing for him a pros pect of intolerable cost, or both, but both of these deterrent functions were performed by the same forces. Moreover, these same forces were also the instruments of defense if deterrence failed. Long-range airpower partially separated the function of punishment from the function of contesting the control of territory, by making possible the assault of targets far to the rear whose relation to the land battle might be quite tenuous. Nuclear weapons vastly increased the 2 Bernard Brodie, Strategy in the Missile Age, Princeton: Princeton University Press, 1959, p. 295. 3 This particular choice between deterrence and war costs has been analyzed by Thomas C. Schelling in an unpublished paper which I have been privileged to read.
A Theoretical Introduction
9
relative importance of prospective cost in deterring the enemy and reduced (relatively) the importance of frustrating his aggressive en terprise. It is still true, of course, that a capacity to deny territory to the enemy, or otherwise to block his aims, may be a very efficient deterrent. And such denial may be accomplished by strategic nuclear means, though at high cost to the defender. But it is now conceivable that a prospective aggressor may be deterred, in some circumstances at least, solely or primarily by threatening and possessing the capability to inflict extreme punishment on his homeland assets and population, even though he may be superior in capabilities for contesting the control of territory. Nuclear powers must, therefore, exercise a conscious choice between the objectives of deterrence and defense, since the relative proportion of "punishment capacity" to "denial capacity" in their mili tary establishments has become a matter of choice. This is the most striking difference between nuclear and pre-nuclear strategy: the partial separation of the functions of pre-attack deter rence and post-attack defense, and the possibility that deterrence may now be accomplished by weapons which might have no rational use for defense should deterrence fail.
C. Deterrence4 Deterrence, in one sense, is simply the negative aspect of political power; it is the power to dissuade as opposed to the power to coerce or compel. One deters another party from doing something by the implicit or explicit threat of applying some sanction if the forbidden act is performed, or by the promise of a reward if the act is not performed. Thus conceived, deterrence does not have to depend on military force. We might speak of deterrence by the threat of trade re strictions, for example. The promise of economic aid might deter a country from military action (or any action) contrary to one's own in4 Other treatments of the theory of deterrence include Bernard Brodie, "The Anatomy of Deterrence," World Politics, Vol. xx, No. 2 (January 1959), pp. 17392; Morton A. Kaplan, "The Calculus of Deterrence," World Politics, Vol. xi, No. 1 (October 1958), pp. 20-44; William W. Kaufmann, "The Requirements of Deter rence," in W. W. Kaufmann (ed.), Military Policy and National Security, Prince ton: Princeton University Press, 1956; Thomas W. Milburn, "What Constitutes Effective Deterrence?" Conflict Resolution, Vol. m, No. 2 (June 1959), pp. 13846; Glenn H. Snyder, "Deterrence by Denial and Punishment," Research Mono graph No. 1, Center of International Studies, Princeton University, January 2, 1959; and Glenn H. Snyder, "Deterrence and Power," Conflict Resolution, Vol. iv, No. 2 (June 1960), pp. 163-79. Robert E. Osgood has allowed me to read several of his manuscripts on the subject which were unpublished at this writing.
10
Deterrence and Defenses
terests. Or we might speak of the deterrence of allies and neutrals as well as potential enemies—as Italy, for example, was deterred from fighting on the side of the Dual Alliance in World War I by the promise of substantial territorial gains. In short, deterrence may follow, first, from any form of control which one has over an opponent's present and prospective "value inventory"; secondly, from the communication of a credible threat or promise to decrease or increase that inventory; and, thirdly, from the opponent's degree of confidence that one intends to fulfill the threat or promise. In an even broader sense, however, deterrence is a function of the total cost-gain expectations of the party to be deterred, and these may be affected by factors other than the apparent capability and intention of the deterrer to apply punishments or confer rewards. For example, an incipient aggressor may be inhibited by his own conscience, or, more likely, by the prospect of losing moral standing, and hence politi cal standing, with uncommitted countries. Or, in the specific case of the Soviet Union, he may fear that war will encourage unrest in, and pos sibly dissolution of, his satellite empire, and perhaps disaffection among his own population. He may anticipate that his aggression would bring about a tighter welding of the Western alliance or stimulate a degree of mobilization in the West which would either reduce his own security or greatly increase the cost of maintaining his position in the arms race. It is also worth noting that the benchmark or starting point for the po tential aggressor's calculation of costs and gains from military action is not his existing value inventory, but the extent to which he expects that inventory to be changed if he refrains from initiating military action. Hence, the common observation that the Russians are unlikely to under take overt military aggression because their chances are so good for making gains by "indirect" peaceful means. Conceivably the Soviets might attack the United States, even though they foresaw greater costs than gains, if the alternative of not attacking seemed to carry within it a strong possibility that the United States would strike them first and, in doing so, inflict greater costs on the Soviet Union than it could by means of retaliation after the Soviets had struck first. In a (very ab stract) nutshell, the potential aggressor presumably is deterred from a military move not simply when his expected cost exceeds his expected gain, but when the net gain is less or the net loss is more than he can ex pect if he refrains from the move. But this formulation must be qualified by the simple fact of inertia: deliberately to shift from a condition of peace to a condition of war is an extremely momentous decision, in-
A Theoretical Introduction
11
volving incalculable consequences, and a government is not likely to make this decision unless it foresees a very large advantage in doing so. The great importance of uncertainty in this context will be discussed below. In a broad sense, deterrence operates during war as well as prior to war. It could be defined as a process of influencing the enemy's inten tion(S, whatever the circumstances, violent or non-violent. Typically, the outcome of wars has not depended simply on the clash of physical capabilities. The losing side usually accepts defeat somewhat before it has lost its physical ability to continue fighting. It is deterred from con tinuing the war by a realization that continued fighting can only gen erate additional costs without hope of compensating gains, this expecta tion being largely the consequence of the previous application of force by the dominant side.5 In past wars, such deterrence usually has been characteristic of the terminal stages. However, in the modern concept of limited war, the intentions factor is more prominent and pervasive; force may be threatened and used partly, or even primarily, as a bar gaining instrument to persuade the opponent to accept terms of settle ment or to observe certain limitations.6 Deterrence in war is most sharply illustrated in proposals for a strategy of limited retaliation,7 in which initial strikes, in effect, would be threats of further strikes to come, designed to deter the enemy from further fighting. In warfare limited to conventional weapons or tactical nuclear weapons, the stra tegic nuclear forces held in reserve by either side may constitute a deterrent against the other sides expanding the intensity of its war effort. Also, limited wars may be fought in part with an eye to deterring future enemy attacks by convincing the enemy of one's general willing ness to fight. The above observations were intended to suggest the broad scope of the concept of deterrence, its non-limitation to military factors, and its fundamental affinity to the idea of political power. In the discussion following, we shall use the term in a narrower sense, to mean the dis couragement of the initiation of military aggression by the threat (im plicit or explicit) of applying military force in response to the aggres sion. We shall assume that when deterrence fails and war begins, the 5 For an excellent extended discussion of this point, with case studies, see Paul Kecskemeti, Strategic Surrender, Stanford: Stanford University Press, 1959. 6 See Thomas C. Schelling, The Strategy of Conflict, Cambridge: Harvard Uni versity Press, 1960, chap. 3. 7 This strategy will be discussed in Chapter 3.
12
Deterrence and Defense:
attacked party is no longer "deterring" but rather "defending." Deter rence in war and deterrence, by military action, of subsequent aggressions will be considered as aspects of defense and will be treated later in this chapter. 1. THE LOGIC OF DETERRENCE The object of military deterrence is to reduce the probability of enemy military attacks, by posing for the enemy a sufficiently likely prospect that he will suffer a net loss as a result of the attack, or at least a higher net loss or lower net gain than would follow from his not attacking. If we postulate two contending states, an "aggressor" (mean ing potential aggressor) and a "deterrer," with other states which are objects of conflict between these two, the probability of any particular attack by the aggressor is the resultant of essentially four factors which exist in his "mind." All four taken together might be termed the aggres sor's "risk calculus." They are (1) his valuation of his war objectives; (2) the cost which he expects to suffer as a result of various possible responses by the deterrer; (3) the probability of various responses, including "no response"; and (4) the probability of winning the ob jectives with each possible response. We shall assume, for simplicity's sake, that the deterrer's "response" refers to the deterrer's entire strategy of action throughout the war precipitated by the aggressor's move— i.e., not only the response to the initial aggressive move, but also to all subsequent moves by the aggressor. Thus the aggressor's estimate of costs and gains is a "whole war" estimate, depending on his image of the deterrer's entire sequence of moves up to the termination of the war, as well as on his own strategic plans for conducting the war, plans which may be contingent on what moves are made by the deterrer during the war.8 8 By way of example, NATO capabilities might raise the prospect of the follow ing possible reactions to a Soviet attack on West Germany: massive retaliation, limited retaliation with nuclear weapons on the Soviet homeland, a tactical nuclear response confined to the local theater of battle, a conventional response, or no response at all. Theoretically, the Soviets would assign a probability and a net cost or gain for themselves to each possible response (the net cost or gain repre senting the summation of their territorial gains, other gains, and war costs as a consequence of the entire war following the initial response), calculate an ex pected value for each response by multiplying the probability times the assumed net cost or gain, and determine an expected value for the aggression by summing the expected values for all possible responses. If the expected value were negative, or positive but less than the positive expected value of non-military alternatives, the Soviets would be deterred.
A Theoretical Introduction
13
Obviously, we are dealing here with factors which are highly sub jective and uncertain, not subject to exact measurement, and not com mensurate except in an intuitive way. Nevertheless, these are the basic factors which the potential aggressor must weigh in determining the probable costs and gains of his contemplated venture. Certain generalizations can be made about the relationship among these factors. Factor 3 in the aggressor's calculus represents the "credi bility" of various possible responses by the deterrer. But credibility is only one factor: it should not be equated with the deterrent effective ness of a possible or threatened response, which is a function of all four factors—i.e., the net cost or gain which a response promises, discounted by the probability (credibility) of its being applied. An available re sponse which is very low in credibility might be sufficient to deter if it poses a very severe sanction (e.g., massive retaliation) or if the aggres sor's prospective gain carries very little value for him. Or a threatened response that carries a rather high credibility but poses only moderate costs for the aggressor—e.g., a conventional response, or nuclear retalia tion after the aggressor has had the advantage of the first strategic strike—may not deter if the aggressor places a high value on his ob jective and anticipates a good chance of attaining it. The credibility factor deserves special attention because it is in terms of this component that the risk calculus of the aggressor "interlocks" with that of the deterrer. The deterrer's risk calculus is similar to that of the aggressor. If the deterrer is rational, his response to aggression will be determined (within the limits, of course, of the military forces he disposes) largely by four factors: (1) his valuation of the territorial objective and of the other intangible gains (e.g., moral satisfaction) which he associates with a given response; (2) the estimated costs of fighting; (3) the probability of successfully holding the territorial ob jective and other values at stake; and (4) the change in the probability of future enemy attacks on other objectives which would follow from various responses. Variations on, and marginal additions to, these fac tors may be imagined, but these four are the essential ones. The deterrer will select the response which minimizes his expectation of cost or maximizes his expectation of gain. (As in the case of the aggressor's calculus, we assume that the deterrer's estimates of cost and gain are "whole war" estimates—i.e., the aggregate effects not only of the de terrer's initial response, but also of all the aggressor's countermoves, combined with the deterrer's counter-countermoves, over the entire progress of the war.) The credibility of various possible responses by
14
Deterrence and Defense:
the deterrer depends on the aggressor's image of the deterrer's risk calculus—i.e., of the latter's net costs and gains from each response—as well as on the aggressor's assessment of the deterrer's capacity to act rationally. The aggressor, of course, is not omniscient with respect to the de terrer's estimates of cost and gain. Even the deterrer will be unable to predict in advance of the attack how he will visualize his cost-gain pros pects and, hence, exactly what response he will choose once the aggres sion is under way. (Witness the United States response to the North Korean attack in 1950, which was motivated by values which apparent ly did not become clear to the decision-makers until the actual crisis was upon them.) Nor can the aggressor be sure the deterrer will act rationally according to his own cost-gain predictions. Because of these uncertainties, the aggressor's estimate of credibility cannot be precise. More than one response will be possible, and the best the aggressor can do is attempt to guess how the deterrer will visualize his gains and losses consequent upon each response, and from this guess arrive at a judgment about the likelihood or probability of each possible response. The deterrer evaluates the effectiveness of his deterrent posture by attempting to guess the values of the four factors in the aggressor's risk calculus. In estimating the credibility factor, he attempts to guess how the aggressor is estimating the factors in his (the deterrer's) calcu lus. He arrives at some judgment as to whether the aggressor is likely to expect a net cost or net gain from the aggressive move and, using this judgment and his degree of confidence in it as a basis, he determines the probability of aggression. Happily, the spiral of "guesses about the other's guesses" seems to stop here. In other words, the aggressor's de cision whether or not to attack is not in turn affected by his image of the deterrer's estimate of the likelihood of attack. He knows that once the attack is launched the deterrer will select the response which promises him the least cost or greatest gain—at that point, the deter rer's previous calculations about "deterrence" of that attack become irrelevant. 2. DENIAL VS. PUNISHMENT It is useful to distinguish between deterrence which results from capacity to deny territorial gains to the enemy, and deterrence by the threat and capacity to inflict nuclear punishment.9 Denial capabilities— 9 This
distinction is discussed by Robert E. Osgood in "A Theory of Deterrence,"
A Theoretical Introduction
15
typically, conventional ground, sea, and tactical air forces—deter chief ly by their effect on the fourth factor in the aggressor's calculus: his estimate of the probability of gaming his objective. Punishment capa bilities—typically, strategic nuclear power for either massive or limited retaliation—act primarily on the second factor, the aggressor's estimate of possible costs, and may have little effect on his chances for territorial gain. Of course, this distinction is not sharp or absolute: a "denial" re sponse, especially if it involves the use of nuclear weapons tactically, can mean high direct costs, plus the risk that the war may get out of hand and ultimately involve severe nuclear punishment for both sides. This prospect of cost and risk may exert a significant deterring effect. A "punishment" response, if powerful enough, may foreclose territorial gains, and limited reprisals may be able to force a settlement short of complete conquest of the territorial objective. However, there are some differences worth noting between these two types or strategies of deterrence. Apart from their differential impact on the cost and gain elements of the aggressor's calculations, the two types of response are likely to dif fer also in their credibility or probability of application. As a response to all-out nuclear attack on the deterrer, the application of punishment will be highly credible. But for lesser challenges, such as a conventional attack on an ally, a threat to inflict nuclear punishment normally will be less credible than a threat to fight a "denial" action—assuming, of course, that denial capabilities are available. While the making of a threat of nuclear punishment may be desirable and rational, its fulfill ment is likely to seem irrational after the aggressor has committed his forces, since punishment alone may not be able to hold the territorial objective and will stimulate the aggressor to make counterreprisals. The deterrer therefore has a strong incentive to renege on his threat. Realizing this in advance, the aggressor may not think the threat a very credible one. A threat of denial action will seem more credible on two counts: it is less costly for the deterrer and it may be effective in frustrating the aggressor's aims, or at least in reducing his gains. A denial response is more likely than reprisal action to promise a rational means of defense in case deterrence fails; this consideration supports its credibility as a deterrent. A related difference is that the threat of denial action is likely to be appraised by the aggressor in terms of the deterrer's capabilities; threats mimeographed, 1960, and in my own "Deterrence by Denial and Punishment," op.cit.
16
Deterrence and Defense:
of nuclear punishment require primarily a judgment of intent. It is fairly certain that the deterrer will fight a threatened denial action if he has appropriate forces;10 the essential question for the aggressor, therefore, is whether these forces are strong enough to prevent him from making gains. In the case of nuclear reprisals, however, the capa bility to inflict unacceptable punishment is likely to be unquestioned, at least for large nuclear powers; here the aggressor must attempt to look into the mind of the deterrer and guess whether the will to apply punishment exists. Thus a denial threat is much more calculable for the aggressor than a reprisal threat—assuming that a comparison of military capabilities is easier than mind-reading. This may make a de nial strategy the more powerful deterrent of the two if the deterrer has strong denial forces; but if he obviously does not have enough ground and tactical forces to block conquest, the threat may be weaker than a nuclear reprisal threat. Even if there is doubt in the aggressor's mind that the reprisals will be carried out, these doubts may be offset by the possible severity of his punishment if he miscalculates and the threat is fulfilled. 3. A MATHEMATICAL ILLUSTRATION It is possible to express some of these relationships mathematically. The illustrations which follow are based on the assumptions that each side is able to translate and combine all of its own relevant values into a single numerical utility, that each can and does estimate probabilities for the other's moves, and that each acts rationally according to the principle of "mathematical expectation."11 This principle states that the 10
It is possible that the aggressor may be able to deter "denial" resistance by threatening to take punitive action if resistance occurs. This is perhaps most feasible with respect to allies of the country attacked whose troops are not de ployed on the territory of the victim. 11 The numerical illustrations are intended simply to set out as starkly as possible the essential logic of deterrence; there is no intent to light a torch for the "quantifiability" of the factors involved, which are, of course, highly intangible, unpre dictable, unmeasurable, and incommensurable except in an intuitive way. It is worth keeping in mind, however, that decision-makers do have to predict, to measure, and, in some sense, to make incommensurable factors commensurate if they are to reach wise decisions. Although, in practice, the factors cannot be given precise numbers, it is legitimate, for theoretical purposes, to pretend that they can be in order to clarify the logic or method by which they should be weighed and compared. The logic is just as applicable to imprecise quantities as to precise ones; to express it in mathematical terms can provide a useful check on intuitive judgment and may bring to light factors and relationships which judgment would miss.
17
A Theoretical Introduction
"expected value" of any decision or act is the sum of the expected values of all possible outcomes, the expected value of each outcome being determined by multiplying its value to the decision-making unit times the probability that it will occur. To act "rationally," according to this criterion, means simply to choose from among the available courses of action the one which promises to maximize expected value (or mini mize expected cost) over the long run. Imagine a world of four states: A, B, C, and D. By a happy alpha betical coincidence, we can say that A is the "aggressor" and D the "deterrer." (Those who prefer not to think in abstractions may think of A as the Soviet Union, D as the United States, B as Western Europe, and C as non-Communist Asia.) Both A and D are thermonuclear pow ers. They are in a condition of "nuclear stalemate"—that is, neither, by striking first at the other, can prevent the other from striking back with an effect outweighing any possible gains. We assume, therefore, that surprise attack on D is not a rational move for A, since D is practically certain to retaliate. The model deals with "secondary deterrence"— i.e., deterrence of enemy attack not against oneself, but against a third party.12 In the first illustration (Figures 1 and 2), which we might call the "massive retaliation" model, A has substantial conventional ground A's CALCULUS S O V I E T
DS CALCULUS
U N I T E D
Attack .60
Not attack .40
0
Retaliate
-100
0
0
Not retaliate
- 20
0
_ I2
ο
Not attack
Retahate .10
-100
Not retaliate .90
+ 20
Expected value
+
8
Fig. 1
U N I T E D
U N I O N
S T A T E S
Attack
S T A T E S
S O V I E T
U N I O N
0
Fig. 2
12 For the preparation of this section, I am indebted to Daniel Ellsberg for per mitting me to read his unpublished manuscript on "The Theory and Practice of Blackmail." This paper, which contains a mathematical model similar to the one I am presenting here, was delivered as part of the Lowell Lectures, "The Art of Coercion: A Study of Threats in Economic Conflict and War," Boston, March 1959.
18
Deterrence and Defense:
forces, but D has none. B and C are the objects of contention between A and D, and they are allied to D. B and C have neither nuclear nor conventional forces. We assume that both A and D are rational but that they are not certain of each other's rationality. To simplify, we postulate that A has only two possible moves: to attack B with full conventional strength or not to attack at all. Similarly, D has only two available responses: to "retaliate massively" or to do nothing—i.e., to acquiesce in the loss of its ally. The cost of all-out war for both sides is 100 and the value of B to A and D is 20. We assume, for simplicity, that an all-out response will preserve the independence of B. Figure 1 represents A's cost-gain estimates ("payoffs") and the proba bilities he associates with each of Ds possible responses to an attack. Figure 2 represents D's payoffs and his estimate of the probability of attack. For convenience, we have made their valuations and losses symmetrical. A estimates the probability of retaliation by D by attempting to guess D's payoffs and D's capacity to act rationally in accordance with them. This probability, in theory, is determined by the size and direction of the "gap" between D's two payoffs, as estimated by A.13 If A were to estimate D's payoffs to be roughly equal, and if he assumed that D would act rationally, logically he would assign a probability of about .5 to "retaliation." It would be a "toss-up," in A's calculations, whether D would retaliate or not. In the example, however, A suspects (cor rectly) that the gap is large and that retaliation is irrational for D. But he is not sure, because he is not privy to D's estimate of the conse quences of war and of D's valuation of those consequences, and of D's valuation of the continued independence of B. Moreover, he is not sure that D will act rationally. Therefore, he believes the chances of retalia tion are small but not zero—in the diagram, one chance out of 10. This risk is not sufficient to deter him because his expected value from at tacking is greater than the expected value from not attacking. A calcu lates the expected value for attack as follows: .10 (-100) plus .90(20) equals 8. If D were able to divine A's calculations, he would expect an attack with certainty. But on the best evidence available to him about A's payoffs and probability estimates (the latter depending on how he 13 Strictly speaking, not only the absolute size of the gap, but also the ratio between the size of the gap and the size of the payoffs, would have to be con sidered.
A Theoretical Introduction
19
thinks A visualizes his own [D's] payoffs), the most he can say is that As "expected value" is probably positive and small. This leads him to consider an attack more likely than not, but not much more. Therefore he assigns a probability of .60 to the attack. This is D's measure of the effectiveness of his deterrent posture. D's "expected value" of minus 12 (probability of A's attack times D's losses with his best response) is a rough measure of D's degree of insecurity. It is useful to note that the probability of retaliation at which A will be indifferent between attacking or not (at which his expected value for "attack" is 0) is approximately .17.14 From D's point of view, this is the "required credibility" of his threat—i.e., the minimum credi bility necessary to make it an effective deterrent.15 Required credibility depends on A's payoffs; credibility depends on A's image of D's pay offs and his appraisal of D's rationality. Obviously, D wishes at all times to keep "credibility" higher than "required credibility." If it is lower, as in Figure 1, his opponent is likely not to be deterred. D can attempt to bring credibility over the required threshold either by lowering the threshold, which involves changing A's payoffs, or by increasing credi bility, which requires shifting his own payoffs or, more accurately, changing A's image of them. He can also try to increase A's doubts about his (D's) capacity to act rationally. Let us consider first how D may affect A's payoffs and thus influence required credibility. Obviously, he can increase his nuclear striking power and thus increase the costs which he can inflict on A in retalia tion. If his capabilities are already at the level where they can complete ly destroy A's society and productive economy, such a move would have little effect on A's calculations. But let us assume they are not. Then an increase might conceivably raise A's all-out war costs to 200, as shown in Figure 3. This change (assuming that A's capabilities do not change) makes D's nuclear deterrent effective by reducing required credibility to .09. Or, to put it another way, it makes A's expected value negative. Of course, if A increases his own retaliatory capability, too, the actual credibility of D's threat of retaliation may also decline. But there is no inherent reason why, with similar increases in armament on both sides, 14 Since the probabilities must sum to 1.00, the probability of "no retaliation" is .83 when the probability of retaliation is .17. 15 Daniel Ellsberg has shown that when a deterrent threat is expressed in a 2 χ 2 matrix, a "critical risk" for the potential aggressor can be calculated. The aggres sor's critical risk is identical with the deterrer's "required credibility." See Ellsberg, op.cit.
20
Deterrence and Defense: AS CALCULUS
D
A
Attack
Not attack
Retaliate .10
-200
0
Not retaliate .90
+ 20
0
Expected value
—
2
0
FIG. 3
these effects should exactly offset each other. They are more likely to do so if both parties are symmetrical in their estimates of the consequences of all-out war or, more exactly, if the aggressor believes such sym metry exists. But if A thinks D is less concerned than himself about in creases in potential damage in all-out war—perhaps because D has a much better civil defense program—a rise in the nuclear striking pow er on both sides would be likely, on balance, to increase Ds deterrent power. Required credibility, in other words, would fall more than credibility. Such a relationship is illustrated in Figure 4. The intersec tion of the lines represents the point at which D's deterrent power just Probability of retaliation
Credibility Required credibility
Severity of retaliation
Fig. 4 begins to be effective; reciprocal increases in striking power beyond this point will increase D's deterrent effectiveness, since credibility exceeds the credibility requirement by a greater and greater margin. Of course, if the position of the lines is reversed, so that the solid line intersects the dotted line from above, D will reduce his deterrent influence by an increase in capability which A matches.
21
A Theoretical Introduction
If the forces that D adds are suitable for "counterforce" action— e.g., very accurate missiles, piloted bombers, and facilities for accurate reconnaissance of enemy airbases and missile launching sites—the ef fect would be to reduce his own costs in all-out war and thus to increase the credibility of his massive retaliation threat, and perhaps also to in crease total war costs for A and hence reduce required credibility. D would also increase the credibility of his threat if he were to increase substantially his civil defense and air defense capabilities. Suppose, as an alternative, that D were to build up a sizable army— though not one strong enough to defeat A's ground forces—and deploy it on B's boundary with A. Also, suppose that both D's and A's ground forces have tactical atomic weapons. The situation might then change, as shown in Figures 5 and 6. A's CALCULUS
D'S CALCULUS A
A Attack
Attack .40
Not attack
Not attack .60
Massive retaliation .05
-100
0
-100
0
Tactical nuclear retaliation .30
- 40
0
- 70
0
Conventional response .65
12
0
- 24
0
Expected value
- 9.2
Fig. 5
-
9.6
0
Fig. 6
This change in D's capabilities forces A to consider three possible initial responses: massive retaliation, tactical nuclear warfare, and a conventional surface response. "No response" is ruled out because D's ground forces are practically certain to fight if they are attacked. A expects net costs of 40 in the event of a tactical nuclear response, which he calculates by considering two possible outcomes and their probabilities: a tactical nuclear war which stays at that level, and an initial tactical nuclear response which spirals to all-out war.16 The fact 16 If A's cost of all-out war is 100, if his cost of fighting a tactical nuclear war is 40, if he expects to conquer B in a tactical nuclear war which stays at that level but not in one that spirals to all-out war, if the destruction of B's assets in a tactical nuclear war is of no consequence to A, and if he thinks there is a .25 chance that
22
Deterrence and Defense:
that D has created an alternative to all-out war leads A to reduce the probability of an initial all-out response to .05. A ascribes a much higher probability to a conventional response than to a tactical nuclear one because he believes D fears the consequences of tactical nuclear war fare. Figure 6, which shows D's payoffs, indicates that A is correct in this assumption. A estimates a net gain of 12 if D responds convention ally. This figure also is calculated from a "cluster" of "sub-outcomes" and their probabilities, including the possibility that the conventional response will stay limited, that it will grow to tactical nuclear war and stay at that level, or that it will eventually spiral to all-out war. A's en tire calculation yields him an expected value of about minus 9—which, of course, deters him. The chief point to be noted about this last example is that over-all deterrent effectiveness can be increased by the provision of ground forces which, although they cannot hold the objective and they depreci ate the credibility of an immediate massive retaliation, do insure some violent response and pose risks of still greater eventual violence, risks which the aggressor is unable to accept. By holding constant the probability of any one of the three initial responses, a "required credibility" can be calculated for either one of the other two. For example, if the probability of massive retaliation is held at .05, the probability of a tactical nuclear defense must be at least as high as .13 for effective deterrence. If we introduce the assumption that A is uncertain whether he can "win"—i.e., conquer B—in a tactical nuclear war, he must discount the value of his possible territorial gain (20) by the probability that he will be unsuccessful. This raises the over-all expected cost of the tactical nuclear response for A because he has less potential gain to set off against his estimate of potential loss and risk. We might also note that A's valuation of the objective affects required credibility, as A's estimate of D's valuation affects credibility. For ex ample, if the object of contention is C rather than B, and C is valued at only 10 by both sides, required credibility in the massive retaliation model (Figures 1 and 2) is only .09, rather than .17 as it is when B is the object. But since A recognizes that C carries also a lower value for D, he will attach less credibility to the massive retaliation threat. If the comparative valuations of B and C are symmetrical for both sides, there is a prima facie case for believing that the relation between required a war which starts at the tactical nuclear level will eventually become all-out, A's calculation is .25 (—100) plus .75 (20—40) equals —40.
A Theoretical Introduction
23
and actual credibility—and hence the deterrence effectiveness of the threat—is roughly the same for both objects. This offends common sense, which seems to indicate that very violent threats—such as that of massive retaliation—have lower effectiveness as well as lower credibility when the prize is small than when it is large. This intuitive appreciation rests on the assumption—a plausible one—that the potential aggressor ignores very small probabilities; that probabilities of retaliation as low as, say, one chance out of 50, are considered equivalent to zero in his calculus. If the ratio of the value of the objective to the anticipated cost of the war is higher for the deterrer than for the aggressor—if, for example, C is valued at 30 by D and at 10 by A when both anticipate war costs of 100—then (assuming that both sides are aware of this asymmetry in valuations) required credibility is lower and actual credibility is higher than when the ratios are symmetrical.17 An entire later chapter will be devoted to the deterrent usages of diplomatic declarations and threats.18 However, some preliminary ob servations are appropriate here. Threats by the deterrer may serve to impress upon the aggressor that a certain objective is valued highly by the deterrer. Threats may also change the deterrer's payoffs and conse quently enhance the credibility of the threatened action. They do so by increasing the cost of not responding in the threatened way, by impli cating additional values beyond the bare value of the territorial objec tive, values which would be lost if the threat were not carried out. For example, in our massive retaliation model (Figures 1 and 2), if D were to pledge his honor and prestige in an unambiguous threat to retaliate, the cost of failure to retaliate might be raised from 20 to 30 or 40. Retaliation would still be irrational for D, but A would be less certain of this than he was before. Fearing that in making the threat D might have increased his cost of not retaliating above his cost of retaliating, A would assign a somewhat greater probability than .10 to the retalia tion outcome. In addition to the intrinsic losses in failing to carry out a threat, 17 Note that this generalization does not depend on "interpersonal comparability of utilities," but rather on the comparability of different forms of utility in the value scale of each side. Thus the appropriate intercountry comparison is that between the ratios of value-of-the-prize to anticipated war costs on each side. To say that one side "values" an objective higher than the other has meaning only when "higher" means "in relation to the expected costs of fighting for the objec tive." 18 For a lucid discussion of threats, commitments, and bargaining, see Thomas C. Schelling, op.cit., esp. chaps. 2 and 5. See also Daniel Ellsberg, op.cit.
24
Deterrence and Defense:
there may also be deterrent and political losses in the form of reduced credibility of other threats and a reduced capacity to attract allies. In making a threat, in other words, the deterrer places in hostage not only his own honor and moral self-respect but also certain aspects of his future deterrent power. In doing so, he makes it rationally more difficult to fail to make the response, and the aggressor, recognizing this, expects a response with greater probability. The credibility of an irrational response may be increased if the de terrer can appear to commit himself to this response by some device which removes or reduces his freedom of choice. Such "automation" is itself rational—even though, paradoxically, the response is not—if the aggressor can be expected to believe with high confidence that the commitment is irrevocable and is thereby deterred. Automation differs from threat-making in that it does not change the underlying payoffs, but rather inhibits the choice of all responses except the one which is being relied on for deterrence. Complete mechanical automation is probably impossible in military affairs, but it can be approached in various ways. The use of ground troops as a "trip-wire" for nuclear retaliation is a kind of automating device. A military commander may be given conspicuous advance authority to order retaliation as soon as he sees that an attack is under way. Or military organization and plan ning may become oriented around a particular response to such a degree that the inertia in its favor is very difficult to overcome in a sudden crisis. The integration of tactical atomic weapons into NATO forces, and the plans to use them in case of major aggression against Western Europe, are a clear case in point. The commitment to massive retalia tion may also be subject in some degree to such administrative automation. The enemy does not have to believe that automation is complete— that freedom of choice has been entirely relinquished—in order for deterrent effects to accrue. Even if the deterrer can only commit himself partially to an irrational response—i.e., restrict his freedom of choice but not eliminate it—doubts may arise in the aggressor's mind as to whether the deterrer has retained enough flexibility to be able to over come the built-in bias in favor of this response. One might say that the commitment of the United States to resist aggression in the Middle East—contained in the "Eisenhower Doctrine"—is at least partially automatic by virtue of the Doctrine's provision that U.S. intervention will take place at the request of the country attacked. The deterrer may increase the credibility of a seemingly irrational
A Theoretical Introduction
25
response by creating the general impression that he is prone to act ir rationally. It is certainly paradoxical, even bizarre, to say that a certain amount of "irrationality" can be an aspect of "rationality." Yet this follows logically if deterrence is to depend on a threat which it would be madness to carry out. "Rationality" may be defined as choosing to act in the manner which gives best promise of maximizing one's value position, on the basis of a sober calculation of potential gains and losses, and probabilities of enemy actions. This definition is broad enough to allow the inclusion of such "emotional" values as honor, prestige, and revenge as legitimate ends of policy. It may be perfectly rational, in other words, to be willing to accept some costs solely to satisfy such emotions, but of course if the emotions inhibit a clear-eyed view of the consequences of an act, they may lead to irrational behavior. It would be irrational, however, to satisfy a momentary passion if sober judgment revealed that the satis faction obtained would not in the long run be worth the cost. Irrationality may take the form either of failing to act in accordance with one's best estimate of costs, gains, and probabilities, or of faulty calculation of these factors in the light of the evidence available. Irra tionality of the latter kind may stem from such sources as commitment to a dogma or theory which is inapplicable to the situation or which shuts out relevant data; education, training, and experience which pre vents attainment of the "whole view"; or limited or distorted perspec tives resulting from bureaucratic parochialism. Irrationality may work in the direction either of recklessness or of timidity; in most cases, for deterrence, one wishes to project an image of excessive boldness rather than of excessive caution. But is it possible for a government to appear irrational and be rational? This would mean appearing to be willing to act contrary to one's payoffs, or appearing to have miscalculated them, when in fact one has calculated, and intends to act, rationally. A democratic government may not be able to practice credibly this kind of sophisticated deceit. In the United States it is difficult enough to achieve consensus about what is rational; to attempt to go further and practice "calculated lunacy," while secretly intending to act sanely, may be infeasible, if only because of the risk of being found out in the pretense. Perhaps the only way a democratic govern ment can appear irrational is actually to be irrational. If this is true, it leads to some interesting thoughts. For example, even if I, as an individual, believe it would be irrational for the United States deliberately to initiate nuclear war in any circum-
26
Deterrence and Defense:
stances, I may not be inconsistent if I also believe that it is rational for the President not only to threaten, but actually to intend, to undertake such a war in defense of West Berlin. I may not wish to argue against this intent if I think the deterrent effect is great enough. Or I may think it would be irrational for the United States to retaliate all-out against Russian cities after a surprise attack which reduced SAC by threefourths and left the Soviet Union with an enormous preponderance in striking power, but for the sake of deterrence I might not want the government to think so, if I believed the Soviets were sensitive to our government's apparent intentions. On the other hand, the United States government has demonstrated a considerable ability to keep secrets when it is not under serious do mestic political pressure to reveal them. Even if the enemy suspects that our proclaimed intentions are not our real ones, it may be possible to prevent him from getting enough information to confirm this sus picion. It may be possible deliberately to create enough doubts about our capacity to calculate and act rationally to deter a moderately con servative enemy by a threat which we know and he knows it would be irrational to fulfill. One should not underestimate the capacity of democratic govern ments to practice self-deceit or uncalculated irrationality. For exam ple, it would not be too much of an exaggeration to say that U.S. military budgets between 1954 and and 1961 were based on the premise that the resources available for military use were limited, that the limited resources should be allocated so as to produce the maximum firepower, that maximum firepower could be obtained by concentration on nuclear weapons, and that nuclear weapons were therefore the most efficient means of fighting wars of any size larger than a mere incident or "brushfire." This syllogism, and the fervor with which the basic premise of limited resources was believed, forced the government to stress the efficacy of nuclear deterrence in a wide range of contingencies, in order to appease domestic and allied opinion, as well as to justify its own predilections. There is impressive evidence that the government com mitted itself so thoroughly to the nuclear deterrent thesis that by a certain process of psychological repression it failed to consider thor oughly the consequences of having to carry out the deterrent threat. If this habit of thinking continues, the government may not be able in time to disengage itself from its own preconceptions and act rationally when the casus retaliati arises. This kind of self-deceit may not, on
A Theoretical Introduction
27
balance, be desirable, especially if the enemy fails to recognize it. But if the Soviets do recognize it, deterrence should be enhanced. Of course, the deterrer must consider not only the usages of apparent irrationality in his own policy, but also the degree of irrationality of the potential aggressor. A certain minimum degree of rationality in the latter is required for the communication of threats, and to enable the aggressor to understand their import and risk for himself. Ideology and doctrine which distort the aggressor's image of the deterrer's values and emotional propensities may work against deterrence. Before World War II, the judgment of the Japanese and German leaders that the Western democracies were weak-willed and soft, unwilling to continue fighting after early defeats, or unable to stand up for long under the horrors of warfare, were examples of the kind of irrationality which tends to inhibit deterrence. Certain aspects of Soviet ideology and doc trine may also reduce the efficacy of deterrent threats and capabilities. For example, the Soviets' obsessive fears about the intention of "capi talist" nations to "encircle" them and eventually to attack them probably contribute to the current prominence of "pre-emptive war" in Soviet military doctrine and may reduce our capacity to deter them from an all-out surprise attack.19 However, there may be other aspects, such as the injunctions against "adventurism," which tend to promote caution and careful weighing of risks and which therefore support deterrence. The potential aggressor's irrationality may affect his appreciation of the deterrer's possible irrationality. For example, if the Soviets really believe their ideological tenet that the "capitalist" elites are cool, calcu lating, and highly rational,20 any image of irrationality which we pro ject, whether contrived or not, may be fruitless. 4. UNCERTAINTY I have assumed that the aggressor will have sufficient evidence bear ing on the deterrer's payoffs to be able to assign precise probabilities to the responses open to the deterrer. He does have some evidence, of course—chiefly, the record of the deterrer's past reactions to aggression, the existing size, composition, and deployment of his military establish ment, and his policy declarations, including expressions of articulate public opinion in the deterring country and its allies. Supplementing 19 See
Herbert S. Dinerstein, War and the Soviet Union, New York: Frederick A. Praeger, 1959. 20 See Nathan Leites, The Operational Code of the Politburo, New York: Mc Graw-Hill, 1951, pp. 18, 23.
28
Deterrence and Defense:
these sources, the aggressor could turn to what evidence is available concerning the "national character" or "psychology" of the deterrer as it pertains to foreign and military policy. For example, in judging the probability of American nuclear retaliation after an attack in Western Europe, the Soviets would be wise to take note of certain American attitudes: our deep emotional involvement with Western Europe, and our strong sense of honor, which might lead us to fulfill a commitment at whatever cost. After making use of all these sources of evidence, the potential ag gressor would still have only a set of very general inferences about the deterrer's probable behavior—not certain predictions, or even state ments of precise probability. The hypothetical stimulus—i.e., the con templated aggression—is likely to be unique in many important re spects; therefore there is not likely to be any similar situation in the deterrer's action record. Observance of the deterrer's preparedness of course will indicate fairly clearly what the deterrer can or cannot do. But within the range of what is possible, preparedness is not a reliable indicator of intent, chiefly because of the various capabilities that will be available and because some forces may be designed to deter rather than to provide a capability for action. Nor are threats a reliable indi cation of intent, for they may be uttered for deterrent purposes only— that is, they may be bluffs and, moreover, they are likely to be vague. The aggressor faces uncertainty at two levels concerning the de terrer's payoffs: first, in his estimate of the opponent's estimate of the consequences of his response; and, secondly, in his estimate of the opponent's valuation of the consequences. The Soviets, for example, cannot know what consequences the U.S. leaders see arising from the initiation of tactical nuclear war in Europe. Do we believe that the first bombs dropped would influence the Soviets to call off the war in fear of suffering further costs? Or do we believe that tactical nuclear war in Europe would inevitably spiral into all-out war? Many other pos sible expectations lying between these two extremes might be imagined. But beyond that, even if the Soviets knew how the United States esti mated the consequences, they could not know how we valued them. They might well doubt, even if we did foresee a high probability that a limited nuclear war would not stay limited, that we had fully realized ourselves how much we valued the things that would be lost to us in an all-out war. All of these considerations tend to raise the question whether it is valid at all to introduce "probabilities" into the analysis of deterrence.
A Theoretical Introduction
29
Everything depends, of course, on whether the enemy thinks in prob ability terms. If the Soviets simply come to a flat decision as to whether massive retaliation (or whatever response they are appraising) will occur or will not occur, then the analysis of deterrence would be con siderably different and rather simpler than we have presented it. The problem for the deterrer then would be to make sure that his military posture and threats posed greater costs than gains for the aggressor, and make sure that his threat was believed. There would be no need then to speak either of "levels of credibility" or of "expected value." Threats would be either credible or not credible. Of course, the de terrer would still face the problem of estimating how much evidence supporting the threatened intent (or absence of contrary evidence) would be necessary to achieve credibility. I am inclined to believe, however, that decision-makers, Russians in cluded, do implicitly think in rough probabilities. The phrase "calcu lated risk" connotes more precision than is probably actually practiced, but it does suggest at least an awareness of the relevance of probabili ties and even some sort of rough-and-ready expected value calculation. Of course, the probabilities cannot be "objective" ones, such as one calculates in estimating the chances that a certain face will turn up when a die is thrown, or in projecting a "frequency distribution" from a large number of identical cases. But potential aggressor and deterrer do have some information bearing on each other's intent, enough to establish at least rough orders of subjective "likelihood." Such subjective "feelings" of likelihood may become more precise in the actual crucible of decision than their verbal statement would suggest. If the Soviets consider a massive response to ground attack in Europe to be "unlikely" in the abstract, they will be driven to make some judgment of "how unlikely" when they actually sit down around a table to decide whether to order such an attack. If the judgment then becomes "possible, but highly improbable," this will carry a meaning for the participants which is close enough to a mathematical statement of probability to allow them to make some rough judgment concerning the "expected cost" of the venture—i.e., some combination of "likelihood" and "value" (or cost) of different possible outcomes. If pressed, the decision-makers probably would be able to choose the probability figure which seemed to them the most plausible, even if they were careful also to assign a wide margin of error to their choice. At least they would be able to specify a range of probabilities which seemed to be more plausible
30
Deterrence and Defense:
than any others and equally plausible. Such a statement would take the form: "Less than 10 per cent, but not more than 50 per cent." If we assume that the aggressor will see enough evidence in the deterrer's behavior to justify the assignment of a range of "most plau sible" probabilities to possible responses, or if his subjective feelings of likelihood or uncertainty (whether or not based on evidence) in effect take the form of such a range, the next question to ask is: what use does the aggressor make of this estimate? If he is to calculate an "expected cost" for his venture, he needs, for working purposes at least, a precise probability, not a range of plausible ones. Does he "split the difference" between the extremes of the range? Or does he, more conservatively, choose to base his decision on the most unfavorable probability within the range of the plausible? Or, even more conservatively, does he base it on the most unfavorable probability which is at least conceivable, though not as plausible as some others? The answer would seem to de pend on the risk-taking propensities of the aggressor. Since it is the aggressor who takes the initiative, it is he who must bear most of the burden of weighing the uncertainties. In the nuclear age, when the eventual outcome of even the smallest border skirmish might be utter devastation, the aggressor's uncertainty is an important deterring factor. This does not mean, however, that it is uniformly de sirable to "keep the enemy guessing." To increase the aggressor's uncer tainty is desirable with respect to deterrent threats—such as massive retaliation—which he probably is inclined to disbelieve in the first place. To increase his uncertainty forces him to attach a somewhat higher working probability to the outcome "all-out war" in order to cover himself against miscalculation. But of course it would be better to in crease his certainty that the threatened response would be carried out. And if the aggressor is thought to be already fairly certain that aggres sion will produce costs for him greater than his gains, it obviously does not pay to reduce his certainty.
D. Defense21 The deterrer, in choosing his optimum military and threat posture in advance of war, must estimate not only the effectiveness of that 21 The reader is reminded that I am using the word "defense" in a rather special sense, which is narrower than one ordinary usage of the term and broader than another. Obviously it is narrower than the usage which makes "defense" synonymous with all military preparedness. It is broader, however, than "capacity to hold terri tory in case of attack," which I would prefer to call "denial capability."
A Theoretical Introduction
31
posture for deterrence, but also the consequences for himself should deterrence fail. In short, he is interested in defense as well as in deter rence; his security is a function of both of these elements. Capabilities and threats which produce a high level of deterrence may not yield a high degree of security because they promise very high costs and losses for the deterrer should war occur. We turn now to a discussion of the factors which go into the deterrer's estimate of the consequences of the failure of deterrence and into his evaluation of the defense effectiveness of his military forces. In other words, we will now talk about how military forces influence the "numbers in the boxes" in the deterrerdefender's matrices given in the preceding section. 1. THE COMPONENTS OF DEFENSE VALUE As already pointed out, military forces affect not only the probabili ties of various enemy moves, but also the potential costs which the defender would suffer should the enemy undertake aggressive moves. We will assume, for simplicity, that the consequences of aggression are always net costs to the defender. This does not mean, of course, that aggression should not be resisted, for much greater losses might be entailed in not resisting. The extent to which given military forces poten tially can mitigate the defender's costs and losses of all kinds is the measure of the "defense value" of those forces in various contingencies. It is assumed that, to all possible enemy moves, the most rational re sponse will be made—i.e., the response which the defender thinks will minimize his aggregate loss during, and as a result of, the ensuing war. The potential costs of enemy moves—followed by one's own optimum responses—are counted in two categories of value: intrinsic value and power value. Intrinsic values are "end values"; they are valued for their own sake rather than for what they contribute to the power relations between the protagonists. They include such things as the value we place upon our own independence (including all the subsidiary values which flow from this independence), the value we attach to the independence (or nonCommunization) of other countries with which we feel a cultural or psychic affinity (apart from what their independence contributes to our own security), the economic values which we find in trading with other free countries (to the extent that these values would be lost should these countries fall under Communist control), and moral values such as self-respect, honor, and prestige. Some of these intrinsic values obviously attach to the continued independence either of our own coun-
32
Deterrence and Defense:
try or of other countries, others are inherent in the response or lack of response rather than in the political entity or territory attacked. Other intrinsic values are the material assets and lives (again as valued for their own sake rather than for their contribution to the power equa tion ) which would be lost in the act of resisting aggression. Power values are "instrumental values," not end values. That is, they are valued not for their own sake but for what they contribute to the security of intrinsic values. It might be more precise to say that the "assets" at stake in international conflict are valued on two scales—a power scale and an intrinsic scale—and that the aggregate value of any given asset is the sum of its valuation on both scales. "Assets" of course may be either tangible, like raw materials and productive resources, or intangible, like prestige or a set of enemy expectations concerning one's willingness to fight in future contingencies. The "defense value" of given military forces includes both the power values and the intrinsic values which can be preserved by using the forces in various contin gencies. Power values are of three major kinds: strategic value, deterrent value, and political value. Strategic value is the potential contribution of the territorial prize to the military capabilities of either side: it includes such familiar elements of power as population, industrial capital, natural resources, and strategic location, valued as war poten tial. More precisely, the strategic value of any country or territory is the probable effect of its loss in increasing the chances that the aggressor would be able to take other areas, or in increasing the costs to the de fender of holding other areas. While strategic value is entirely a function of the war-making poten tial inherent in the contested territory, deterrent value is an attribute primarily of the act of responding to aggression. Deterrent value may be described as the effect of a response in reducing the probability of enemy attacks against other areas in the future—i.e., reducing it below what it would be if no resistance were offered to the immediate aggres sion. The enemy may be discouraged from making future attacks either if (1) his territorial gains from his present move are limited or denied entirely, or if (2) the costs we inflict upon him by our response are greater than he had expected. In the main, deterrent value stems from the evidence which our response provides to the enemy concerning our future intentions. Political value is the effect of a response, and of its direct conse quences, on the alignment or attitudes of third countries. Although we
A Theoretical Introduction
33
have classified political value under the rubric of "power values," it is really a mixture of intrinsic and power effects. In its power dimension, political value may be subsumed under either strategic value or deter rent value. That is, the political loss of an ally to the other side may increase the enemy's capability to make future conquests, or reduce our own capability to prevent such conquests. It may also increase the enemy's inclination to attempt future aggression and reduce our own capacity to deter it. It may reduce the cohesion of our own alliances. All of these are power effects. However, we also place an intrinsic value on 'liaving friends" abroad, on keeping democratic countries out of the Communist sphere of influence, on economic relations which might be disrupted by a transfer of political allegiance or the adoption of a "neu tralist" stance. Political value is the sum of all such considerations. A net loss in power value means an increase in the probability, and/or the potential adverse consequences, of future aggressive enemy moves, after completion of the "first" enemy move, and the ensuing war (if the move is resisted). A strategic loss is an increase in the potential esti mated loss as a result of such future attacks. A deterrent loss is an in crease in the probability of future enemy moves. A political loss may affect either the probabilities or the costs of future attacks (or both). Power values, especially strategic value, may be lost as a result of losses of lives and attrition of military capabilities and future war poten tial in the process of fighting. More subtle strategic forms of war cost might include such things as a weakening of the population's willingness to sacrifice in future wars or willingness to stand up to enemy threats in future political crises. This, in other words, along with the intrinsic value of the lives and economic assets sacrificed in the war, is an offset to the power and intrinsic values which may be gained or saved by fighting. In summary, the total value of a response (and the defense value of the military forces available for making the response) is the sum of the power values and intrinsic values which can be saved or gained by the response, minus the power values and intrinsic values lost as the result of war casualties and damage. 2. STRATEGIC VALUE AND DETERRENT VALUE Much of the inconclusiveness of the recurring "great debates" about military policy might be avoided if the concept of "strategic value" could be clarified and clearly separated from the deterrent effects of military action. The strategic value of a particular piece of territory is
34
Deterrence and Defense:
the effect which its loss would have on increasing the enemy's capability to make various future moves, and on decreasing our own capacity to resist further attacks. The deterrent value of defending or attempting to defend that piece of territory is the effect of the defense on the enemy's intention to make future moves. The failure to recognize this distinction contributed to the apparent about-face in United States policy toward South Korea, when we decided to intervene after the North Korean attack in June 1950. Earlier, the Joint Chiefs of Staff had declared that South Korea had no strategic value—apparently meaning that its loss would have no significant effect on the U.S. capacity to fight a general war with the Soviet Union. This determination was thought to justify—or at least was used as a rationalization for—the withdrawal of U.S. combat forces from the Korean peninsula in 1948 and 1949. Secretary of State Dean Acheson strengthened the impression that "no strategic value" meant "no value" when, in a speech early in 1950, he outlined a U.S. "defense perimeter" in the Far East which ex cluded Korea. Then when the North Koreans, perhaps encouraged by these high-level U.S. statements, attacked in June 1950, the United States government suddenly discovered that it had a deterrent interest, as well as strong political and intrinsic interests, in coming to the rescue of South Korea. The dominant theme in the discussions leading up to the decision to intervene was that if the Communists were "appeased" this time, they would be encouraged to make further attacks on other areas.22 The chief motive behind the intervention was to prevent such 22As former President Truman has stated: "Our allies and friends abroad were informed through our diplomatic representatives that it was our feeling that it was essential to the maintenance of peace that this armed aggression against a free nation be met firmly. We let it be known that we considered the Korean situation vital as a symbol of the strength and determination of the West. Firm ness now would be the only way to deter new actions in other portions of the world. Not only in Asia but in Europe, the Middle East, and elsewhere the con fidence of peoples in countries adjacent to the Soviet Union would be very ad versely affected, in our judgment, if we failed to take action to protect a country established under our auspices and confirmed in its freedom by action of the United Nations. If, however, the threat to South Korea was met firmly and suc cessfully, it would add to our successes in Iran, Berhn and Greece a fourth success in opposition to the aggressive moves of the Communists. And each success, we suggested to our allies, was likely to add to the caution of the Soviets in under taking new efforts of this kind. Thus the safety and prospects for peace of the free world would be increased." Harry S. Truman, Years of Trial and Hope, New York: Doubleday and Co., 1956, pp. 339-40. The primary political value of the intervention, as U.S. decision-makers saw it, was that it would give other free nations confidence that they could count on U.S. aid in resisting aggression. The most salient intrinsic values were moral value in
A Theoretical Introduction
35
encouragement from taking place, and positively to deter similar at tempts in the future. Another case in point was the debate about the desirability of a United States commitment to defend the Chinese offshore islands of Quemoy and Matsu. Those who took the negative in this debate stressed that these two small islands held no "strategic value" for the United States, that they were not "vital" to the defense of Formosa, etc. Former Secretary of State Dean Acheson declared that the islands were not worth a single American life.23 Administration spokesmen, on the other hand, emphasized the political and deterrent value of defending Quemoy and Matsu. President Eisenhower, for example, said that this coun try's allies "would be appalled if the United States were spinelessly to retreat before the threat of Sino-Soviet armed aggression."24 Secretary of State Dulles asserted that the stakes were not "just some square miles of real estate," but the preservation of confidence in other countries— both allies and enemies—that the United States would resist aggression. It was better to meet the challenge at the beginning, Mr. Dulles said, than after "our friends become disheartened and our enemies over-confident and miscalculating."25 Power values are sometimes discussed in terms of the "falling domino" theory. According to this reasoning, if one objective is lost to the enemy, other areas contiguous to the first one inevitably will be lost as well, then still additional areas contiguous to these, etc., as a whole row of dominoes will fall when the first one is knocked over.26 In its extreme form, the domino thesis would value any objective, no matter how small, opposing the aggressive use of force, support for the "rule of law" in interna tional affairs, support for the collective security system embodied in the United Nations Charter, and the special responsibility the United States felt for the Republic of Korea, whose government it had played a major role in establishing. "Support for the collective security system" of course had deterrent and political as well as moral overtones. 23 New York Times, October 3, 1958, p. 3. 2i Ibid., October 5, 1958, p. 1. 25 Ibid., September 26, 1958, p. 1. 26 Apparently the domino theory was first given public expression by President Eisenhower on April 7, 1954, when he said, in reply to a request that he explain the strategic value of Indo-China to the United States: "You had a row of dominoes set up, and you knocked over the first one, and what would happen to the last one was the certainty that it would go over very quickly. So you could have a beginning of a disintegration that would have the most profound influences." The President then referred to "the possible sequence of events, the loss of IndoChina, of Burma, of Thailand, of the peninsula, and Indonesia following." Ibid., April 8, 1954, p. 18.
36
Deterrence and Defense:
as dearly as the value which the United States placed on the continued independence of all other non-Communist countries. Thus we should be as willing to fight for one place as another, since a failure to resist once inevitably means future losses. The important thing is to "draw a line" and resist violations of the line, whatever their dimensions and wherever and whenever they may occur. The domino theory tends to overstate power values: since the enemy may have limited aims and may be satisfied with a small gain, his in crease in capability from a single small conquest may not significantly shift the balance of capabilities in his favor, and the loss of single small areas may not have adverse political effects among neutrals and allies.27 Nevertheless, the domino image does highlight an important truth: the strategic and intrinsic value of the immediate territorial prize is not a sufficient criterion for evaluating the wisdom of resisting aggression, or for estimating the forces necessary for successful resistance. The enemy's possible ultimate objective must also be considered, as well as the effect of resistance in discouraging him from attempting further progress toward that objective, and in forestalling political changes among other countries which would tend to further that ultimate objective. There is a relationship between the strategic, political, and intrinsic value which the enemy believes one attaches to a given objective, and the deterrent value which can be realized by responding to an attack on that objective. For example, a failure to resist effectively a Com munist attack on the offshore islands of Quemoy and Matsu might not increase perceptibly the chances of Chinese Communist attacks on other non-Communist countries in Asia, if the Communists did not be lieve we placed a high intrinsic and strategic value on these islands. On the other hand, it could be argued that a determined and costly re sponse to an attack on an objective which the enemy thinks means little to us in strategic and intrinsic terms is likely to give him greater pause with respect to his future aggressive intentions. Thus, if the objective is to "draw a line" to deter future aggression, perhaps the best place to draw it is precisely at places like Quemoy and Matsu. The enemy would reason that if the United States were willing to fight for a place of such trivial intrinsic and strategic value to itself, it must surely be willing to 27 It is hard to believe, for example, that a Communist Chinese conquest of Quemoy and Matsu would have reduced the confidence of the European allies in the willingness of the United States to defend Europe. The solidarity of NATO might have been weakened by a U.S. attempt to defend the islands.
A Theoretical Introduction
37
fight for other places of greater value. Thus, the deterrent value of de fending any objective varies inversely with the enemy's perception of its value to us on other accounts. There is a further consideration: if it is thought necessary to fight a certain amount of war, or risk a certain amount of war, to convince the other side of our willingness to fight generally, what better place to do it than at places like Quemoy and Matsu, where it is least likely that the war will spiral to all-out di mensions? Mutually shared expectations are extremely important in determining the deterrent value of military actions. The United States did not lose much in deterrent utility by failing to intervene in Hungary in 1956, because both sides regarded Hungary as part of the Communist camp. But a failure to defend Berlin would severely undermine the U.S. capa bility to deter future Communist incursions in Europe or elsewhere. The consequences of enemy moves, and the defense value of forces for resisting them, are subject to modification by policy declarations. Threats and commitments may involve one's honor and prestige in a particular area or objective, and this involvement increases the deter rent, political, and intrinsic value of defending such places and the value of forces which are able to defend. Thus the adverse conse quences of an unresisted Communist attack on Quemoy and Matsu were increased by the various official statements, including the Formosa Resolution passed by Congress, to the effect that these offshore islands were "related" to the defense of Formosa. But these consequences were not increased as much as they might have been, had the United States made an unequivocal commitment to defend the islands. Of course, losses of power values through the loss of an ally or neu tral to the enemy may be offset by increased mobilization of domestic resources. The cost of the additional mobilization required might be taken as a measure of the power value of the territory in question. Thus the defending power might ask itself: "If I let this piece of territory or this ally be taken over by the enemy, how many additional resources will I have to spend for military weapons to have the same degree of security I have been enjoying?" Once war is entered into, consideration of deterrent possibilities may call for a different strategy than would be the case if we were interested only in the strategic and intrinsic values of the particular area attacked. If the latter were our only interest, our war aims might be limited to restoration of the status quo ante; deterrence of future aggressions,
38
Deterrence and Defense:
however, might dictate more ambitious aims. In the Korean War, for example, it is possible that if closer consideration had been given to de terrent benefits, the U.N. armies might have pushed on farther than they did—if not to the Yalu, then perhaps at least to the "narrow neck" of the Korean peninsula. The opportunity was not taken to show the Communists that their aggressions were likely to result in losses not only of manpower but also of territory; that in future limited wars they could not hope to end up at least where they started. In general, we will be willing to suffer higher costs in fighting a limited war if deterrence is an objective than if it is not. In other words, it may be desirable to fight on longer and in the face of a higher cost expectancy if an important objective is to assure the enemy of our will ingness to suffer costs in future contingencies. The objective of deterrence may call for the use of different weapons than would the simple objective of blocking enemy conquest of an area at least cost. Our use of nuclear weapons probably would support the Communist estimate of our willingness to use them in the future; and, conversely, to refrain from using them when such use would be militar ily advantageous would weaken that estimate. However, as in the deci sion whether or not to fight at all, the strategic and intrinsic value of the immediate objective is relevant to the deterrent effects: the use of nuclear weapons to defend highly valued objectives might support but little the probability that they would be used to meet lesser challenges;28 the failure to use them when the prize was small would not necessarily signal a reluctance to do so when the object of the conflict was vital. Finally, for deterrent reasons it might be desirable to attempt re sistance against a particular limited enemy attack even though we knew in advance that our resistance would fail. The purpose would be to in form the enemy, for future reference, that although he could expect to make gains from limited aggression in the future, these gains could be had only at a price which (we hoped) the enemy would not want to pay. Proposals for limited nuclear retaliation against one or a few enemy cities in response to limited ground aggression may draw on this kind of reasoning. Of course, the concept of "deterrence by action" has no relevance in 28 On the other hand, any use of nuclear weapons would set a precedent. The symbolic or psychological barrier to their use which had rested on their previous non-use would be eroded. The Russians might believe, after they had been used once, that the probability of their use in any future conflict had increased.
A Theoretical Introduction
39
determining the appropriate response to a direct thermonuclear attack on the United States, or in valuing the forces for the response. In that event there would be no future contingencies which would seem worth deterring or worrying about at all, compared with the magnitude of the catastrophe which had already taken place. The primary values would be intrinsic values associated with reducing war damage, per haps limiting the enemy's territorial gains in Eurasia, and preserving the independence of the United States itself.29 Power values lost by the defender represent power values gained by the attacker, although the values may not be equally important to each side. For example, the Middle East has strategic value for the United States because its geographic location and resources add significantly to the West's capacity to fight limited war in Europe and elsewhere, and because the area, in the hands of the Soviets, would increase the Soviets' capacity to fight such wars—because of its position athwart vital trans portation routes if not because of its oil resources. Our strategic loss if the Middle East should fall under Communist control would be the sum of the deprivation to the West's future military capabilities and the increment to the Soviet capabilities. Similarly, the strategic gain to the Soviets would be the sum of their own direct gain in military re sources plus the losses for the West. It is less obvious that deterrent values also have this reciprocal char acter. When, by fighting in Korea, we demonstrated our willingness to defend free institutions in Asia, not only did we gain "deterrent value" with respect to other possible Communist moves in Asia; the Commu nists lost something analogous to it in their own value system. Presum ably they became less confident that overt aggression could be attempt ed again without U.S. intervention. Their "expected value" from future aggressive moves declined perhaps below what it was before Korea, and certainly below what it would have been if the North Korean aggression had been unopposed by the United States. When an aggressor state successfully completes a conquest, or has its demands satisfied short of war, its willingness in the future to make war, or to make demands at the risk of war, presumably is strengthened 29 We might, of course, attempt to "deter" the enemy from continuing his attacks, thus reducing our war costs and perhaps preserving our independence and the essential fabric of our society, by a discriminating use of the weapons we had left after absorbing a surprise attack, accompanied by appropriate bargaining tactics. Possibilities along this line will be explored in the next chapter.
40
Deterrence and Defense:
by the reduction of expected cost or risk which it perceives in such future moves. This reduction in the perceived chances of being opposed in the future we might label "expectational value," to differentiate it from "deterrent value," which is peculiarly associated with status quo powers. Deterrent and expectational values are in obverse relationship —i.e., when the defender loses deterrent value by failing to fight or to carry out a threat, the aggressor gains expectational value, and vice versa—although again the gain or loss may have a stronger psychologi cal impact on one side than on the other, since the value in question is highly subjective. This distinction is similar to Thomas Schelling's distinction between "compellent" and "deterrent" threats.30 A compellent threat is used in an aggressive way; it is designed to persuade the opponent to give up some value. A deterrent threat, on the other hand, is intended to dis suade the opponent from initiating some positive action. A successful conquest would increase an aggressor's compellent power with respect to other possible victims, especially if the fighting had included the use of nuclear weapons; other countries would lose deterrent power, since their psychological capacity to resist demands would be weakened by the aggressor's demonstration of willingness to risk or to undertake nuclear war. Strategic gains by the Soviets might appear in their risk calculus as an increased probability that future attacks on other areas would be successful, or perhaps as a decreased expectation of cost in making future conquests. Gains in expectational value would appear as a de creased probability of resistance to future attacks, or perhaps as a reduced probability of a high-cost response by the defender or its allies. The aggregate of strategic gains and expectational gains produces an increase in "expected value" to be gained from future moves (or a re duction in "expected cost"). This might not always be the case if the consequence of a successful aggression were to stimulate an increased level of military mobilization by the United States and its allies and/or an increased determination to resist future attacks. Thus a successful limited attack might backfire and reduce the Soviets' strategic position as well as their expectational value, although of course they would retain whatever intrinsic values they had gained by their conquest. 30 Thomas
C. Schelling, op.cit., pp. 195-96.
A Theoretical Introduction
41
E. The New Balance of Power31 The separation of the functions of punishment and denial, and espe cially the enormous magnification of the capability to inflict punishment, have profoundly affected the traditional concept of "balance of power." In effect, nuclear weapons, long-range aircraft, and missiles have super imposed a new balance upon the old. These weapons have not simply added higher levels of potential destructiveness to the traditional bal ancing process; they have changed the very nature and meaning of "balance." Two balancing systems—the strategic balance of terror and a truncated tactical balance of power—now operate simultaneously, each according to different criteria, but interacting in various ways which are not yet thoroughly understood. The "power" that was balanced in the pre-nuclear balancing process was essentially the military power to take or hold territory. Moreover, territory, and the human and material resources on it, were the pre dominant source of power. The motive for engaging in the balancing process was to prevent any single state or bloc from becoming so pow erful that it could make territorial conquests with impunity and eventu ally achieve hegemony over the other states in the system. The objec tives were, first, to deter the potential disturber from initiating war by forming alliances and building up armaments sufficient to defeat him; and, second, if deterrence failed, to defend or restore the balance by engaging in war. Whether a balance existed or not depended essentially on whether the states interested in preserving the status quo were able to deny territorial gains to the expansion-minded state or states. The balance of terror centers on a different form of power—not the power to contest the control of territory directly, but the power to inflict severe punishment, to prevent the enemy from inflicting punishment on oneself, and to deter by the threat of punishment. In its pure deterrent form, a balance of terror exists between two nuclear powers when neither can strike first at the other without receiving a completely in tolerable retaliatory blow in return. A balance does not exist when one power, in striking first, can eliminate all but a tolerable portion of the opponent's capacity to strike back. At present, three states participate in the balance of terror. (France 31 This section essentially is a condensation of my article, "Balance of Power in the Missile Age," Journal of International Affairs, Vol. xiv, No. 1 (1960), pp. 21-35.
42
Deterrence and Defense:
can hardly be considered a full-fledged participant as yet.) Of the three, two—the United States and Great Britain—are so closely allied, politically and culturally, that for all intents and purposes they are one unit. Thus the balance of terror is essentially a bipolar system, tending to heighten, sharpen, and preserve the bipolarity which has existed in the non-nuclear balance since World War II and which is now beginning to break down with the emergence of China as a great military power and the increase in the economic strength of the Western European countries. There is good reason to believe that other states eventually will become nuclear or thermonuclear powers, thus possibly creating a multipolar balance of terror system. 1. DIFFERENCES BETWEEN THE BALANCE OF TERROR AND THE BALANCE OF POWER The traditional balancing process continues to operate as a balance between conventional forces (and the potential for building such forces) in all situations in which there is no significant possibility that nuclear weapons will be used. Hereafter, we shall refer to this balance as the "tactical" balance of power, differentiating it both from the strategic balance of terror and from an over-all balance of power in volving interactions between the strategic balance and the tactical balance. Some significant differences between these two balancing sys tems in their "pure" form are worth noting.32 One difference is that, in the strategic balance, quantitatively match ing the enemy's capabilities is virtually irrelevant as a Criteriojn for balance. A balance of terror exists when neither side can eliminate enough of the other's forces in striking first to avoid an unacceptable retaliatory blow. Depending chiefly on technological conditions, especially the degree of vulnerability of the opposing forces, a potential attacker may be balanced with a force only a fraction of the size of the attacker's forces; or balance may require having more forces than the potential attacker. The proper criterion is to be able to inflict unac ceptable retaliatory damage. By contrast, in the modern tactical balance centering on conven32 The term "tactical balance of power" refers chiefly to the balance of con ventional capabilities; "strategic balance of terror" to the balance of long-range nuclear air and missiles forces. Tactical nuclear weapons fall somewhere in be tween, but I am inclined to consider them principally as components of the bal ance of terror. Presently, I shall discuss the "mixed" balance—i.e., the over-all balance of power when the strategic and tactical balances interact.
A Theoretical Introduction
43
tional ground forces, as in the traditional balancing process, simply equaling the strength of the enemy's forces is still the most plausible balancing criterion, although of course a sophisticated calculation would require that it be modified to take account of factors such as a possible advantage of the defense over the offense, possibilities for post-attack mobilization, geography, asymmetries in supply capabili ties, etc. The balance of terror is primarily a deterrent balance rather than a defensive balance. That is, a "balance" is said to exist when a po tential aggressor faces the prospect of retaliatory damage sufficient to deter him, not when he faces the prospect of defeat or frustra tion of his aims. Conceivably, a balance of terror could exist in the defensive sense, if the forces on both sides were so invulnerable that the side which absorbed the first blow could still retaliate with suf ficient force to destroy or prostrate the attacker. But the forces re quired for winning the war after being attacked would be consider ably larger in number and probably different in kind than the forces required to deter the attack. The tactical balance of power, on the other hand, centers primar ily on the function of defense. A balance of power exists lWhen the defending side has enough forces to defeat the attacker or at least to prevent him from making territorial conquests. Deterrence is the consequence of this defensive capability, not of a capacity to inflict unacceptable costs. In the tactical balance, the requirements for de terrence and for effectively fighting a war more or less coincide; this is not the case in the balance of terror. Another difference concerns the strategic value of territory and of territorial boundaries. In the tactical balance, the strategic value of territory and of the human and material assets associated with territory continues to be high. The traditional elements of national power, such as manpower, natural resources, industrial strength, space, geographic separation, command of the seas, and so on, remain the primary sources of power and they are important criteria for determining the existence or non-existence of a tactical balance. These territorially based elements are also a source of power in the balance of terror, but their significance is less and considerably differ ent than in the tactical balance. Strategic nuclear weapons have reduced the importance of geographical separation of the opponents in the balance of terror, since ICBM's can reach from continent to
44
Deterrence and Defense:
continent. However, distance still retains some significance in the strategic balance of terror. An aggressor can reduce the required range and hence increase the accuracy and possible payload of his missiles by obtaining control of territory between himself and his prospective nuclear opponent. He may also increase the points of the compass from which he can attack, thus complicating the opponent's warning and air defense problem. He may increase the space available for dispersal of his striking forces, and he may obtain useful staging bases and post-attack landing points for his long-range aircraft. The acquisition of industrial and resource assets by conquest may increase a nuclear power's capability to produce additional strategic weapons. While "raw" manpower is not a significant source of power in the balance of terror, an aggressor may turn to his own uses the scientific brainpower of a conquered nation. On balance, however, the strategic value of territory and its associated assets is probably smaller in the balance of terror than in the tactical balance. Overconcentration on the strategic balance and the contingency of all-out war has caused us, in recent years, to downgrade excessively the importance of industrial potential for war. War potential continues to be a source of power in the tactical balance not only prior to war but also after the war has begun. Stockpiles of raw materials, stand-by war production plants, and the like can be translated into actual military power during the progress of a limited war, provided of course that the forces ready in advance of the attack can hold off the enemy until the additional power can be mobilized. However, in the balance of terror, industrial potential provides only pre-attack power, not postattack power. Once the war has started, if such potential were not destroyed, its usefulness probably would be limited to survival and reconstruction. Even in a very restrained war, involving only counterforce attacks on military installations, with minimum damage to economic assets, a decision probably would be reached before indus trial potential could be brought into play. In the tactical balance, alliances are useful for both deterrence and defense, in roughly equal proportions; the costs of war are low enough and the incentives to prevent the conquest of an ally are high enough that allies are likely to see a net advantage in coming to each other's aid. The conquest of an ally means a very serious erosion of one's own power and security position, and such erosion may be prevented at bearable intrinsic costs, if the necessary forces are available. Since
A Theoretical Introduction
45
the potential aggressor is aware of this, the credibility of alliance ob 1 ligations tends to be high. In a world of many nuclear powers—i.e., in a "multipolar" balance of terror—alliances are likely to have less utility and credibility for protection against nuclear attack. Obviously, a country which could mount a completely unacceptable retaliation to a nuclear attack on itself would not need allies for security against this contingency. (It might, of course, enter into an alliance for security against non-nu clear attack.) Countries which doubted their individual capacity to deter a nuclear attack might feel they could gain security by com bining. In combination, they might be able to muster enough retal iatory power to deter either an attack on the whole alliance simul taneously or an attack on a single member. The alliance's capacity to deter attack on a single member would depend critically on the amount of his forces which the Aggressor would have to use up in attacking the first victim. It is conceivable that the attacker would so deplete his own forces that the other mem bers of the alliance could strike without fear of serious (retaliation; at least the prospect of this would limit the amount of force which the attacker could use against the initial victim and might deter the attack. But if the aggressor could retain substantial and invulner able forces while successfully attacking a single member, the sup porting allies would feel powerful incentives to renege. Fulfilling the alliance obligation would mean accepting severe destruction. These costs might be suffered in vain, for there would be little chance of saving the attacked ally by nuclear retaliation. And, in retaliating, the supporting allies would be using up forces which they would need for their own future protection. Thus the alliance pledge may not seem very credible to a prospective nuclear aggressor. Nevertheless, alliances might have some deterrent value in a multi polar balance of terror, because of the aggressor's uncertainties, be cause an alliance would limit the amount of force which an aggressor would be free to apply against a single victim, and because deter rence does not depend on absolute credibility. A nuclear attack on a single country would be a very momentous act which might stimu late enough emotional reaction and irrationality among the victim's allies to trigger retaliation on their part. The aggressor would have to realize that the possible damage he might suffer at the hands of the whole alliance would be very much higher than the value he placed on conquest of a single member. The magnitude of the pos-
46
Deterrence and Defense:
sible retaliatory damage might very well offset in his mind the low credibility of an alliance response. 2. INTERACTION BETWEEN THE BALANCE OF TERROR AND THE BALANCE OF POWER The strategic and tactical balancing processes do not function inde pendently, but impinge on each other in various ways. In its pure form, the balance of terror operates only to deter an all-out nuclear attack by the Soviet Union against the United States and Great Britain, or by the latter countries against the Soviet Union. The traditional or tactical balancing process operates independently only in conflict situations which involve no possibility of the use of nuclear weapons. Thus it operates between non-nuclear countries and blocs with respect to issues in which the nuclear powers are not significantly interested, and in minor conflicts between nuclear powers. But between the two bal ancing processes in their pure forms is a wide range of situations in which they are interdependent. For example, since the end of World War II, the United States has used its dominant position in the balance of terror to deter a consider ably wider range of contingencies than a direct nuclear attack on itself. The means of the balance of terror—the threat and capability to inflict punishment—have been applied to the furtherance of certain ends in the tactical balance of power, notably the deterrence of a large-scale Soviet ground attack in Western Europe. Consequently, in U.S. and Western policy, the scope of the tactical balancing process has shrunk to the deterrence and defense of limited aggression, primarily outside Western Europe. The validity of this concept became increasingly questionable after 1953 and 1954, when the Russians exploded a hydro gen bomb and then demonstrated that they had a modern, long-range delivery capability. These and further Russian advances in missilry since 1957 have tended to reduce the plausible scope of our threat of a "first strike" and to increase the scope of the tactical balancing process. However, it is not likely that the scope of the balance of terror will narrow to its pure form or that the tactical balance will widen to its pre-nuclear dimensions. In other words, some form of nuclear response remains a possibility whenever the interests of one of the superpowers are challenged militarily—especially if challenged directly by the other superpower. In some circumstances, and with appropriate capa bilities, the threat of massive retaliation may retain some significant
A Theoretical Introduction
47
credibility. Even if a nuclear response takes only a limited form, there is always the chance—and a conservative aggressor must consider it a good chance—that the war will escalate to severe levels of destruc tion, perhaps to all-out war. Thus the modern balance of power takes a "mixed" form: any conventional military attack by one nuclear power (or its adherent) against the interests of another nuclear power creates a risk of nuclear reprisal of some kind; whether a balance exists with respect to that act depends not only on the defender's "denial" forces but also on the attacker's appreciation of the risk that punishment will be imposed and the possible severity of that punishment. The greater the severity of the provocation, the greater the relative importance of the punishment component. The idea of a mixed balance can be extended and elaborated in terms of a "spectrum of violence." The pole of least violence would be "peaceful competition"—the use of economic aid, propaganda, infiltra tion of subversive agents, and the like—to affect the internal political complexion of a country. Such competition shades into lower-keyed forms of violence like sabotage and the fomenting of rebellion. The next higher stage would be military (materiel) aid to rebellious or potentially rebellious groups or to the established government for use against such groups. Successively higher stages would be covert as sistance in manpower to either side in a civil war—e.g., by military missions, advisers, and "volunteers"—then limited interstate conven tional conflict, then limited or tactical nuclear warfare, then limited strategic warfare involving the territories of the superpowers, and finally all-out nuclear warfare. The word "spectrum" is appropriate because it suggests a gradual shading or progression from one intensity to another and gets away from a common tendency to think of the dimensions of warfare as sharply defined categories, such as "limited war" and "all-out war." One suspects that such sharp delineation does not correspond to the realities of human behavior. At each shading of the spectrum—i.e., for each possible aggressive act—an over-all balance of power either does or does not exist, de pending on the potential aggressor's image of the defender's capa bilities and intentions. A balance of capabilities would exist at any level if the defender had sufficient forces to deny gains to the aggres sor or could impose such high costs as to offset any possible gains. But true balance requires an additional condition: that the aggres sor recognizes a will or intent on the part of the defender to use his
48
Deterrence and Defense:
capabilities. This might be recognized only as a "likelihood" or "prob ability," rather than with certainty, and still present a prospect of defeat, or high costs, sufficiently serious to deter attack. A potential aggressor is "balanced" at each level of violence if his objectives can be denied him at that level, if his costs of fighting at that level would be higher than his expectation of gain, or if, by attacking at that level, he would exceed a certain critical threshold of risk that the defender would "up the ante" to a higher level at which either the aggressor's aims would be frustrated or he would suffer unacceptable costs. A balance exists at any level in both a deterrent and a defensive sense when the aggressor knows he would not be able to make gains in attacking at that level. A balance exists only in a deterrent sense when the aggressor can make territorial gains but foresees too great a risk of suffering war costs incommensurate with the gains. For example, should the Russians attack conventionally in Europe, they would be balanced or deterred either if the NATO "shield" could contain their attack, or if, in attacking with sufficient force to break through the shield, they in curred too serious a risk that NATO would respond with tactical nu clear weapons or make some other nuclear response which would either frustrate the Soviets' objectives or produce excessive costs for them. An over-all balance of power depends on the existence of a balance at all possible levels of violence, in the sense described above. The potential disturber would not be balanced in an over-all sense if there were one or more weak links—i.e., levels of conflict at which he could achieve his goal at tolerable cost and risk. The defender or status quo power can achieve a balance at some levels by having forces which can frustrate the enemy's aims; at other levels, by threatening to impose unacceptable costs on the aggressor. There are feedbacks between the various levels. For example, if we were to use nuclear weapons in response to a North Vietnamese attack on South Vietnam, we might alienate other Asian countries to the extent that our position in the balance of "peaceful competition" would be grievously weakened. Consideration of this might deter us from using nuclear weapons; also, if the North Vietnamese were to dis cover or suspect strongly that we would not use nuclear weapons for this reason, they would be less likely to be deterred. For some years, the Soviets have been making gains in the balance of peaceful com-
A Theoretical Introduction
49
petition more or less as a by-product of their strengthening position in the balance of terror. Also, the Russians have attempted to use their new missile strength to strengthen their position in the tactical balance by making threats of rocket attacks on various NATO mem bers that are clearly intended to weaken the cohesion of the alliance. As the above discussion has implied, nuclear technology has in creased the importance of intentions, relative to capabilities, in the balancing process. Intentions have always been important, of course. In the pre-nuclear balance, the balancing process was set in motion by the perception of the disturber's aggressive intent, as well as by his military capabilities and war potential. And the adequacy of the bal ance as a deterrent rested in part on the aggressor's being clear about the intentions of the states which would eventually oppose him. But both sides could be fairly sure that, once the conflict was joined, all states which did participate would do so to the full extent of their military power. An important calculation for each side, therefore, con cerned the balance of total capabilities. The relation between total capabilities is still important at the level of the balance of terror—i.e., in the deterrence and fighting of all-out war. But for conflicts beginning at lower levels, the balance between over-all capabilities is less important, and a new dimension has been added to the factor of intentions—namely, each side's assessment of the other's intent regarding what portion of its destructive power will be used. Each knows that the other can inflict costs far outweighing the value of any political objective if it cares to do so. Total capabilities establish the bounds of what is possible, but what is probable depends on a reciprocal assessment of wills, which in turn depends on each side's appraisal of the opponent's values at stake in each particular issue, his gambling propensities, his tendencies toward irrationality, his ideological or organizational commitments to certain responses, and his image of one's own characteristics in these respects. Such estimates are of course highly subjective and uncertain, and the pervasive uncertainty adds an important element of stability to the over-all balance of power. Each side is driven to think in terms of probabilities, and when even the smallest military action may eventuate in nuclear war and totally unacceptable costs, small probabilities are likely to be important. Consequently, there is considerable deterrent
50
Deterrence and Defense:
value in making threats which the threatener knows, and the threat ened party suspects, it would be irrational to carry out; if the threat increases the probability of unacceptable costs to the other side by only a few percentage points, it may be sufficient to deter. This is to say that the existence of a balance of power, or the capabili ties requirements for balancing, can hardly be determined without attempting to look into the "mind" of the enemy. One might say that a subjective "balance of intentions" has become at least as important as the more objectively calculable "balance of capabilities." A corollary of the increased relative importance of intentions is that methods of communicating intent have become more important means in the balancing process than they have been in the past. First, nations are becoming more sensitive to what they say to each other about their intentions; the psychological importance of threats and other declara tions is on the increase. Secondly, the function of military forces them selves may be shifting in the direction of a demonstrative role: the signaling of future intentions to use force in order to influence the enemy's intentions, as opposed to being ready to use, or using, force simply as a physical means of conquest or denial. Hence the enhanced importance of deterrence in the modern balance of power as compared with defense. We are likely to see more imaginative and subtle uses of "force demonstration" in time of peace—the Russians have given the lead with such acts as the U-2 incident, test-firing of missiles in the Pacific near U.S. territory, and the shooting-down of the U.S. RB-47 in the summer of 1960. Warfare itself may in the future become less a raw physical collision of military forces and more a contest of wills, or a bargaining process, with military force being used largely to demon strate one's willingness to raise the intensity of fighting, with the object of inducing the enemy to accept one's terms of settlement. While direct conflict or competition is going on at a low level of the spectrum of violence, selective force demonstrations using means appropriate to higher levels may take place as threats to "up the ante." The three chapters which follow will deal with various segments of the spectrum of violence from the point of view of the national security of the United States. Chapter 2 will treat the problem of deterring a nuclear strategic attack on the United States, of acting to mitigate the consequences of such an attack should it occur, and of deterring attacks elsewhere by the threat of all-out nuclear retaliation. Chapter 3 deals with the interaction of conventional, tactical nuclear, and
A Theoretical Introduction
51
strategic nuclear weapons in the deterrence of and defense against Soviet moves in Western Europe. Chapter 4 will discuss the more ambiguous forms of conflict which seem most likely to occur in the Middle East and Asia. Declaratory policy and force demonstration will be the subjects of Chapter 5.
2 Deterrence and Defense with Strategic Nuclear Power MUCH of the discussion about national defense in recent years has centered on the question of how much strategic nuclear power the United States needs.1 Any estimate of "needs" is necessarily subjective, depending on one's valuation of security, as against the values which must be given up to obtain security. But even allowing for differences in personal value preferences, the opposing positions in the debate about strategic power have been remarkably far apart. On the one hand, some believe that the United States itself is in considerable dan ger of being attacked by surprise with nuclear weapons—especially in the years immediately ahead, when the Soviets may have a numerical preponderance in long-range missiles. In this view, the United States long-range nuclear power may not be sufficient, unless drastic steps are taken, even for the minimal objective of deterring such an attack. Some of those who take this position—notably Air Force officers— argue further that this minimal objective is not an adequate criterion for strategic force requirements; that additional forces are needed to "win" a nuclear war should deterrence fail, and to deter enemy moves short of all-out attack. Others apparently believe that nuclear airpower already is far in excess of requirements; that we have the nuclear1 strength to "destroy Russia many times over" and that it should be reduced to release resources for other military uses, such as ground forces and other forces for fighting limited wars. The great disparity between these views indicates a need for clarification of the criteria for deciding how much strategic nuclear power is enough. The requirements for strategic capabilities depend on the purposes which these forces are expected to serve. Three distinct purposes are conceivable: (1) deterrence of an all-out Soviet surprise nuclear at tack on the United States; (2) defense—i.e., the efficient waging of 1 "Strategic" capabilities refer to nuclear and thermonuclear weapons designed chiefly for employment directly against the homeland of the opponent. "Tactical" capabilities are weapons, either nuclear or conventional, intended primarily for use on the local battlefield or against military targets in the immediate rear of the battlefield.
with Strategic Nuclear Power
53
war—after such an attack; and (3) deterrence of major Soviet aggres sions (conventional or nuclear) elsewhere. This chapter will chiefly explore the requirements for the first two purposes—deterrence and defense in the context of all-out attack on the United States—but the third one also will be treated briefly.
A. A Note on Terminology As a preliminary to the following discussion, certain definitions are necessary. A clear distinction should be made between a "first-strike" capability and a "strike-back" capability.2 A first-strike capability re fers to the amount of damage we could do to targets in the Soviet Union if we struck first. A strike-back capability refers to the degree of damage which could be inflicted on Soviet targets after the Soviets had struck first and destroyed a portion of our own force. It is quite obvious, of course, that any given prewar force will have both a firststrike and a strike-back capability, depending on whether it is used before or after an enemy attack. A further distinction is that between a "punitive"3 capability and a "counterforce" capability. A punitive capability is the amount of de struction which our forces can inflict on the Soviet economy and popu lation. A counterforce capability is the degree of attrition which our forces can inflict on the Soviet nuclear forces. Here again, the same prewar inventory of forces would have both a punitive and a counterforce capability, depending on the targets at which the forces were fired. Various factors, to be discussed below, determine the utility which different types of forces have for each of these missions, but practically all strategic forces have some utility for either purpose. And of course the amount of punitive or counterforce destruction which can be real ized with a given set of forces depends on whether the forces strike first, or whether they strike back after having been partially destroyed by an enemy first strike. Obviously, four different combinations are derivable from these two distinctions: a first-strike punitive capability, a first-strike counterforce 2 I use the term "strike-back capability" rather than "second-strike capability" because the latter phrase can be construed as meaning the residual forces which either side may have left after having struck once. 3 A punitive capability is sometimes referred to as a "retaliatory" capability. I have rejected this usage because a first-strike counterforce attack may also be carried out in retaliation for some serious provocation.
54
Deterrence and Defense
capability, a strike-back punitive capability, and a strike-back counterforce capability. Again, all four of these types of capability may be present in the same strategic force. The two types which are mosit relevant to strategic deterrence are the first-strike counterforce capa bility and the strike-back punitive capability, and the discussion which immediately follows will concentrate on these categories. The strikeback counterforce capacity may be useful for defense in all-out war, depending on whether the side which strikes first has any strategic forces left which can be effectively attacked by the side striking back; this capability will be considered in a later section of this chapter. The first-strike punitive capability is irrelevant to deterrence between the United States and the Soviet Union because it would be irrational for either of these countries to strike first massively at the other's cities, leaving the other's nuclear strategic forces undamaged and free to retaliate massively against the attacker's cities. However, this type of capability may be useful for other countries as a deterrent against ground attacks in Europe; it will be considered in the next chapter.
B. Deterrence of All-out Surprise Attack The strategic force required to deter a Soviet all-out nuclear attack on the United States may be roughly defined as a force which, after suffering attrition from the enemy's first strike, could still strike back at the enemy's cities with an effect serious enough to outweigh what ever values the Soviets hoped to gain by attacking. Such a force will be referred to hereafter as a "minimum strike-back punitive capability," or occasionally as a "minimum deterrent" force. The minimum deterrent calculation is based on the assumption that the enemy's first strike would be directed primarily or entirely against our strategic nuclear forces.4 The Soviet force which is necessary to overcome the U.S. minimum deterrent and create a Soviet willingness to strike first is described as a "sufficient first-strike counterforce capa bility." It is a force which, in striking first, could eliminate enough of the U.S. strategic forces to make their retaliation acceptable—i.e., not damaging enough to negate the gains the Soviets expected to make by striking first.5 4 The
plausibility of this assumption is examined below. If both sides had perfect information about relative capabilities and the other side's expected costs and gains, each side's requirement for a "sufficient first-strike counterforce capability" would be simply the obverse of the other's "minimum 5
with Strategic Nuclear Power
55
The major variables which enter into the determination of our mini mum deterrent requirements (and the Soviets' first-strike require ments) may be listed as follows: (1) The number and type of the enemy's attack vehicles likely to be used in the surprise attack. ( 2 ) T h e reliability, accuracy, a n d "yield" of t h e e n e m y ' s vehicles. ( 3 ) T h e e n e m y ' s firing efficiency—i.e., h i s ability t o c o o r d i n a t e firing of his delivery vehicles so as to maximize the effect of the initial sur prise salvo. ( 4 ) T h e a c c u r a c y of t h e e n e m y ' s intelligence c o n c e r n i n g t h e loca tion of our own striking forces. ( 5 ) T h e timeliness a n d reliability of o u r w a r n i n g system. ( 6 ) T h e efficiency of o u r a i r defenses. ( 7 ) T h e effectiveness of o u r passive f o r c e defenses—i.e., dispersal, hardening, mobility, concealment, etc. ( 8 ) T h e efficiency of t h e e n e m y ' s a i r defenses. ( 9 ) T h e efficiency of t h e e n e m y ' s civil defenses. (10) The enemy's threshold of unacceptable damage. Factors 1 through 7 determine how much of the United States forces would survive the surprise attack, assuming the enemy sent all of his first-strike vehicles against our forces. Factors 8 and 9 determine how much damage we would be able to inflict on the Soviet Union in retalia tion. The last factor determines how much prospective damage is nec essary to deter. For deterrence, of course, the relevant consideration is how these variables are estimated by the potential attacker. The United States, therefore, in determining its minimum deterrent requirement, must attempt to guess how these factors are visualized by the Soviet Union. All of them, in the United States calculus, are subject to more or less uncertainty, and most of them will be uncertain in the enemy's calcula tions as well. The enemy presumably will be certain about factors 1 strike-back punitive capability." If one side had first-strike forces sufficient to make it willing to strike first, obviously the other would not have a minimum capability to deter attack. However, because of the great uncertainty affecting each side's calculations, the two concepts are not quite reciprocal. In order to hedge against uncertainties, each side will set itself a minimum strike-back re quirement larger than that which is just adequate to reduce the other's first-strike counterforce capability below the level of "sufficiency." And neither will be willing to strike first unless it has a first-strike force considerably greater than that "objec tively" required to reduce the other's retaliation to acceptable dimensions.
56
Deterrence and Defense
and 10—his first-strike force and the amount of retaliatory damage he is willing to accept. He can also be sure that some retaliation will follow his attack. There may be some question about the targets of the retaliatory blow—i.e., whether it would be directed against cities or force installations—and about whether the United States would release its entire residual force immediately after the attack. But a minimum degree of caution would enjoin the Soviets to expect that retaliation would be against cities and would be all-out, even though (as we shall presently discuss) this kind of indiscriminate reprisal might not be the most rational action for the United States. It would seem that this near-certainty of all-out retaliation would be a very strong deterrent in itself. The Soviets' uncertainties concerning the other factors, which would determine how much of the U.S. forces they would be able to eliminate in their first strike, and how much of the surviving U.S. forces would be able to penetrate Soviet air defenses, probably further strengthen deterrence. Even if, on paper, the Soviets could calculate a near-certainty that all but a trivial fraction of the U.S. forces would be destroyed in the surprise attack, many things could go wrong. For example, some of the attacking missiles might be fired prematurely, thus giving us an extra degree of warning time and allowing almost all our forces to be launched before the bulk of the attacking vehicles arrived.6 The United States may have such con fidence in its warning system that it is willing to fire all its missiles on warning alone; in a few years, after the Ballistic Missile Early Warn ing System is completed and missile count-down times have been significantly shortened, the Soviet Union would have to assume, to be on the safe side, that we did have such confidence. The Soviets could not be entirely sure of the accuracy and reliability of their missiles,7 8 Some authorities believe that the risk of an uncoordinated salvo resulting from personal unreliability or mechanical failure may be a sufficient deterrent, even should the Soviets, on paper, calculate that they have the quantitative capability for a successful first strike. Of course, it would be wise not to bank on this to the extent of failing to provide ourselves with a force large enough to deter even a coordinated attack. 7Here it is worth noting that even if missile tests over pre-surveyed firing ranges had indicated high accuracy, a reasonably conservative attacker would not expect similar accuracy against enemy targets in an actual war—first, because enemy targets will not have been located with the same precision as test targets; and, secondly, because prewar computations of accuracy from tests would neces sarily be based on a small sample, which would be subject to ordinary statistical error.
with Strategic Nuclear Power
57
and would be still less sure of the strength of our anti-aircraft and anti-missile defenses, the vulnerability of our missile sites, the number of SAC planes which could become airborne before their own attacking vehicles arrived at SAC bases, the effectiveness of their own air de fenses, etc. When the penalty of miscalculation is so enormous, a rational Soviet leadership would naturally estimate all these factors on the conservative side—that is, assume the worst, from the Soviet point of view. However, the United States also faces uncertainties about how the enemy calculates all these factors. Considering the huge stakes in volved, it would be prudent for the United States to assume the worst about Soviet capabilities and appraisals of costs and gains—i.e., to place its minimum deterrent requirement at a level which would be sure to convince the Soviets that their damage would exceed their most hopeful estimate of possible gains. Perhaps the most uncertain factor, for the United States, is the degree of prospective damage which would be sufficient to deter the Soviet Union from attacking. Would the Soviets be deterred by the prospect of losing ten cities? Or two cities? Or fifty cities? No one knows, although one might intuitively guess that the threshold is closer to ten than to either two or fifty. The only available benchmark is that the Soviet Union did lose about 20 million lives (and an enor mous amount of material wealth) during World War II and yet managed to rebuild its society and economy within a very few years. However, this is probably an unreliable measure because the Soviets did not deliberately and willingly accept these costs in World War II but suffered them as a consequence of resisting invasion of their own homeland. The only alternative to not accepting them was the com plete loss of national independence. One can only guess, but it does seem likely that the Soviets would be deterred from deliberately start ing World War III by an expectation of loss considerably below their World War II losses. Of course, the Soviets' threshold of acceptable damage is a function not only of the value they place on the lives of their people and their material assets, but also of the strength of their incentives to attack. What is the nature of these incentives? Why would the Soviets want to attack the United States, anyway? Curiously enough, this question, which seems so basic, is often ignored in our preoccupation with technology and military capabilities. Possible Soviet motives for a first
58
Deterrence and Defense
strike would seem to fall into four categories: expansion of the Com munist orbit in Europe and Asia; pre-emption of an expected United States first strike; elimination of the United States as a future competi tor and threat to the security of the Soviet Union; and conquest of the United States. If the only incentive of the Soviets were the first—to make conquests in Europe and Asia—and if they had tactical superiority in the Eastern Hemisphere, their degree of acceptable damage would be a function not only of their valuation of these gains, but also of their degree of fear that the United States would be provoked into making an all-out first strike (or perhaps some less than all-out use of nuclear weapons) by a Soviet tactical move. (Presumably the Soviets would not measure the value of their expected territorial gains from the prewar status quo, but rather from the gains they could hope to make by non-military methods. If they expected Western Europe and Asia eventually to fall into their hands as a result of political evolution, or of political-psychological-economic methods of expansion, they would have no incentive to act militarily at either the tactical or the strategic level, unless they placed a high value on hurrying up the process. Let us assume, how ever, that non-military alternatives are ruled out.) The Soviets would have three basic alternatives: (1) not to attack at all; (2) to attack with tactical (ground) forces only; or (3) to strike first at the United States. They would choose (3) if the expected cost of the U.S. retaliation were less than the value of the expected territorial gains, and if this net gain were greater than the expected value from attacking with tactical forces only. The latter calculation would include a consideration of the risk and consequences of a U.S. first strike. If there were no chance of a U.S. first strike, the Soviets obviously would choose (2) rather than (3); they would prefer to make their conquest in the cheapest manner. However, if the chances and the probable cost of a U.S. first strike were higher than some critical threshold, they would choose to strike first themselves, as long as this move promised a net gain. It follows, under the assumption of Soviet tactical superiority, that the strike-back capability which is required to deter Soviet attack on the United States tends to be larger, the greater the U.S. first-strike capa bility. At the extreme, assuming Soviet war aims are limited to Europe and Asia, if the United States decided not to have even the semblance of a first-strike capability and not to threaten even limited forms of nuclear retaliation, we would need no strike-back capability at all,
with Strategic Nuclear Power
59
since the Soviets would gain nothing by hitting us first. The larger our first-strike or limited retaliation capability, the more strike-back capa bility we would need, but the upper limit for the latter would be those forces which, in striking back, could inflict more damage on the Soviets' homeland than their prospective territorial gains were worth. Still assuming the Soviets are interested only in making gains in Europe and Asia, we must consider also situations in which "not attack ing" either tactically or strategically involves a loss for them. This would be the case when the Soviets had engaged their prestige in a threat or demand, or when they had gotten themselves involved in a limited war in support of one of their satellites. Then the costs of backing down would enter into the Soviets' calculations. In such tense situations, the enemy faces essentially three alternatives: backing down, attacking or continuing to fight at the tactical level of warfare, or striking first strategically. Degrees of tension may be located on a spectrum ranging from situations in which the protagonists are merely making threats and demands, to threats plus demonstrative military action of some kind, to conventional tactical warfare, and upward through various intensities of nuclear warfare below the all-out level. The costs to the enemy of backing down depend upon the degree to which his prestige has been involved by prior action and threats and the amount of prior gain he would have to give up in capitulating. The costs and risks to him of starting or continuing tactical warfare depend on the expected direct costs of such warfare, the expected gain, and the estimated probability and severity of a U.S. first strike. When the opponents are merely in non-violent conflict involving threats and counterthreats, the Soviets would suffer some intangible costs in back ing down, and perhaps run some risk of a U.S. first strike (depending on U.S. first-strike capabilities) if they were to make a tactical move. Either alternative, however, might well seem better than initiating all-out war. By contrast, if the United States and the Soviet Union were to become locked in a tactical nuclear war in Europe, the cost to the Soviets of capitulating would be very high, and continuing to fight at the tactical level would carry much greater risks either of a U.S. first strike or of a gradual escalation of the war to all-out intensity, especially if the United States were losing the limited war. In this situation the Soviets would feel a much greater temptation to strike first—this alter native would have improved relative to the other two. Thus the required strike-back capability varies not only with our first-strike capability
60
Deterrence and Defense
and the value which the enemy places on territorial gains, but also with the degree of tension which may occur. Our strike-back capability should be strong enough so that in all reasonably conceivable political crises, and in all degrees of limited conflict, striking first at the United States will not appear to be the best of all the available Soviet alterna tives. The second possible reason for a Soviet first strike—a pre-emptive attack—is closely related to the reason just discussed, since Soviet fears of a U.S. first strike (or perhaps limited nuclear action) after a Soviet tactical move provide the only incentive for the Soviets to initiate all-out war simply to make gains in Europe and Asia when they have tactical superiority. However, the notion of pre-emptive attack usually is taken to mean an attack undertaken to forestall an enemy attack which is expected "out of the blue"—i.e., without having been provoked by one's own tactical actions. Admittedly the distinction is a blurred one. However, because of certain complicating factors, we have chosen to discuss pre-emptive attack in a later section of this chapter. The third and fourth possible incentives—eliminating the United States as a future competitor and conquering the United States—may be considered direct gains, as opposed to the indirect gain from fore stalling a U.S. first strike. Even though these incentives were not im portant enough or feasible enough in themselves to create a motive for a Soviet first strike, they might be important supplements to the primary motive of Communist expansion in Eurasia. For example, even if our strategic capabilities were not large enough to create a serious risk of an immediate U.S. first strike following tactical provoca tion, the Soviets might fear that we would be provoked into a crash program to build a sufficient first-strike capability and that, once we had it, we would use it to smash the Soviet Union and liberate the conquered areas. From this consideration, the Soviets might develop a "preventive" motivation to strike first. Conceivably, the Soviets might even hope to conquer the United States; they might believe that a surprise nuclear attack would !so disorganize U.S. society and reduce its military forces that invasion would be possible, or that the United States would be forced to sur render under the threat of further nuclear punishment. It does not seem likely that the Soviets could accomplish this objective concur rently with consolidating their conquests in their own hemisphere.
with Strategic Nuclear Power
61
Their own tactical forces would be busy enough in Europe and Asia, and if the United States were able to deliver a substantial retaliatory blow, these forces might not be able to complete even these conquests until minimum reconstruction had taken place. And the United States very probably would not be willing to surrender simply on pain of further nuclear destruction. However, if the Russians believed they would be able to recover from the thermonuclear exchange before we did, they might anticipate at least eventually being able to take over the United States, perhaps after inflicting further nuclear damage. Of course, to be sure of deterrence of a Soviet first strike, we should want to estimate all these Soviet incentives on the high side. We should very probably want at least a modest first-strike capability in order to help deter the most serious challenges to our interests overseas. Thus the Soviets are likely never to be completely free of some risk of either an all-out U.S. first strike or some form of U.S. nuclear action should they provoke us severely. Even if the Soviet incentives were limited to the indirect one of removing this risk, they might very well decide to remove it, however small, if they believed their total gains in Europe and Asia would be greater than the retaliatory damage they would suffer. Adding the direct incentives, we conclude that our minimum punitive force should be at least sufficient to strike back with a blow severe enough to offset all the gains the Soviets might expect to make in all theaters, including the conquest of the United States itself, plus an extra margin to offset the Soviet costs of backing down in periods of high political tension or limited violence. Even if we were able to discover the Soviets' critical threshold of unacceptable damage, we would not have ascertained the proper size of our strategic forces. We would still face many uncertainties bearing on whether or not we would have the necessary forces left after the enemy's attack to inflict this degree of damage, and whether the enemy recognized that we would have. These uncertainties would be con nected with relative capabilities—how many enemy attacking vehicles would penetrate our air defenses, how accurate they would be, how good our passive defenses were, etc. We would want to add additional forces to cover these uncertainties. We might also want to have some extra forces to deal with the possibility of an irrational enemy. Although we thought the prospect of losing 30 cities, for example, would deter any rational Soviet leadership, we might want to double this prospect, or even triple it, to make sure it was sufficient to deter a
62
Deterrence and Defense
leader who was subject to certain moderate forms of irrationality such as fits of temper, overoptimism about intelligence estimates, or just plain inability to comprehend the consequences of war. We might also want to have some extra forces to compensate for the possibility that the Soviets would not believe our threat to retaliate all-out against their cities with our residual forces, or that they believed they could blackmail us out of this intent. We might want more to hedge against possible Soviet technological breakthroughs—in the field of anti-missile defense, for example, or in techniques for keeping mobile missiles and missile-firing submarines under surveillance. Other reasons for providing more insurance could easily be en visaged. Just how much insurance is enough is not something that can be decided with precision. Yet there would seem to be certain intuitive benchmarks which serve to narrow the problem to manageable limits. For example, it seems clear that if we were certain of having enough forces left after the enemy's first strike to "destroy RuSsia"8 or to de stroy all major Russian cities, this would be excessive for deterrent purposes. At the other extreme, a force which might be completely destroyed by the enemy's surprise attack, or which the enemy might think he could destroy, obviously would be too small. Somewhere in between lies the proper "minimum deterrent." Perhaps a rational way to calculate it would be to make the most conservative, yet reasonable, estimate of what would be unacceptable damage for a rational enemy in a situation of severe international tension, then determine the size of the force which would certainly be able to inflict at least this much damage under a conservative estimate of the enemy's strike-first capa bilities, air defenses, and civil defenses. This might be called the "low minimum" deterrent. Above this level, additional forces would yield diminishing returns for deterrent purposes, but we might still wish to add on further increments to cover some of the more remote uncer tainties mentioned above. The amount added would depend on intui tive judgment, reflecting the degree of subjective security we would feel with various force levels, the costs of building, protecting, and maintaining the forces, and the urgency of competing military and nonmilitary demands on the national resources. At some point, however, additional increments to strategic nuclear power would become re dundant for deterrence of attack on the United States. 8 This is the term often used by government spokesmen to indicate the require ments for deterrence.
with Strategic Nuclear Power
63
Of course, there are certain other deterrents, besides our own strikeback capabilities, which would work against a Soviet surprise attack. Presumably the Soviet leaders are not entirely devoid of conscience; the initiation of all-out war would be morally costly. It might also be costly politically—the Soviets might fear an adverse reaction among their own people and among the satellite populations, perhaps leading to rebellion. They would have to consider whether they would be able to maintain their own authority in conditions of great social and polit ical disruption. There would occur considerable revulsion against the Soviets in countries outside the Communist empire. The Soviet leader ship might doubt its ability to consolidate the political gains which the elimination of the U.S. strategic nuclear forces might make possible. Above all, a decision as momentous as starting an all-out war would be an extremely difficult one to make, even for a ruthless calculating leadership which was certain or almost certain of making a net gain. In the over-all deterrent, all such non-military factors and costs would supplement the costs which the Soviets would expect from U.S. re taliation; to put it another way, they might reduce the military require ment for a U.S. strike-back force. However, since these inhibitions are highly uncertain in their effect, it would be prudent not to give them much weight in determining the proper size of the minimum deterrent force.
C. Defense by Offense in All-out War9 Of course, the primary objective in dealing with the threat of all-out attack is to deter it. But we can never have absolute confidence in deterrence; it makes some sense, therefore, to consider the strategy and weapons which would be most useful for fighting an all-out war should deterrence fail. Until recently, this subject has been given little attention, apparently because of the widespread view that an all-out war would simply be an unmitigated catastrophe; there seemed little reason, therefore, to give any serious thought to how we would go about fighting such a war, what our war aims would be, how much force and what kind of weapons would be needed, and how we would attempt to terminate the war. However, a general nuclear war might 9 In
writing this section I profited considerably from hearing Herman Kahn's "Three Lectures on Thermonuclear War," Princeton University, March 1959. (A revised version of these lectures was published in 1960 by Princeton University Press, under the title, On Thermonuclear War.)
64
Deterrence and Defense
not automatically be an unlimited disaster; whether it became so might depend critically on how we had planned and prepared in advance to fight it. The concept of "minimum deterrence" rests on the assumption that the Soviet leadership will believe we intend to retaliate all-out against Soviet cities after a Soviet first strike, or that they will not wish to take the risk that we might do so. The assumption is a strong one, because a nation which has been subjected to thermonuclear attack cannot confidently be expected to refrain from doing as much damage as it can in retaliation. Such indiscriminate retaliation might very well be irrational. But a thermonuclear attack based on an expectation that the victim would behave rationally would be a very dangerous gamble for the attacker. Yet, if there is any chance of deterrence failing, we certainly are interested in behaving rationally (even though, before the war, the threat of an irrational response might be a very potent threat for the purpose of deterrence). The most rational strategy for fighting the war must be determined in the light of what our objectives would be in all-out war. It seems clear that our primary objective would be to minimize our own losses, counted in terms of physical damage to the United States, losses of allied territory, and possible compromise of our own political independence. A secondary objective (which may, in some circumstances, be in conflict with the primary one) would be to win the war, or at least to reduce as much as possible the postwar power position of the Soviet Union relative to that of the United States. The word "win" in connection with all-out war is usually put inside quotation marks because of the general feeling that the question of "who won the war" would simply be overwhelmed in the minds of the survivors by the immense destruction and the problems of reconstruc tion and survival. However, the concepts of "winning" and "losing" have to do with the military or power outcome of the war, as they always have in wars throughout history. They have nothing to do with the in trinsic costs of damage suffered in fighting the war. The winner is the side that comes out of the war with a distinctly improved power posi tion vis-a-vis the opponent, compared with what its power position was before the war. There are many degrees of "winning," ranging all the way from complete conquest to minor rectification of a boundary. In all-out war, an important component in determining the victor and the
with Strategic Nuclear Power
65
loser would be the degree of damage suffered by each side, measured according to its power dimension or its "strategic value." The other primary component would be changes in territorial control. We might not be able to prevent the Russians from improving their relative power position in each of these components; yet we would still be interested in limiting the degree of their victory—i.e., minimizing their territorial conquest and maximizing their damage relative to our own. These aims would have to be weighed, however, against the intrinsic costs which would be suffered in pursuing them. The Soviets, if they fought the war rationally, would act according to a similar calculus of costs and gains, measured in both intrinsic and power terms. Before exploring the various alternative strategies which might be feasible, we shall make two major assumptions concerning the situa tion which is likely to exist after a Soviet surprise attack. First, it is assumed that the Soviet surprise attack would be directed almost entirely against our strategic nuclear forces. This does not mean, of course, that all cities would be spared; some would inevitably be damaged in the initial strike because of their proximity to SAC airbases and missile launching sites, and because of stray bombs and missiles. If some of the SAC aircraft had been dispersed to civilian airports, or if the enemy thought it necessary to hit such airports to prevent our airborne aircraft from having landing places upon return ing from their retaliatory sorties, the city and population damage would be greater. Yet the enemy would have strong incentives to refrain from hitting our cities, beyond such unavoidable damage as was incident to his counterforce attack. His first aim would be to reduce as much as possible our ability to retaliate against him. It would be foolhardy for the Soviets to expend any of their weapons on cities if this meant delaying or forgoing the destruction of a signifi cant portion of SAC. Besides using up forces which might have been used to reduce our power of retaliation, deliberate attacks on cities also would provide maximum stimulation for an all-out U.S. response against Russian cities, which the Soviet leaders might hope to avoid. Finally, avoiding cities would preserve them as hostages that the Russians could later threaten to destroy, either to deter the United States from unlimited retaliation against Russian cities, or to persuade the United States to accept Soviet terms for ending the war.10 10 We might strengthen our deterrence somewhat by deliberately placing our missile sites near large cities. The enemy's first strike would then be maximally
66
Deterrence and Defense
The second assumption is that the Soviet Union would have some strategic nuclear forces left after making a surprise attack. This assumption seems plausible on technical, military, and political grounds. Even if the Soviets wanted to mount the most powerful first strike possible, it would be technically impossible for them to use up all their capability in a single strike. Some aircraft which escaped being shot down might return to their bases. Some submarines, after firing their missiles off the U.S. coasts, would be able to return to port for reloading and refueling. Some missiles undoubtedly would mal function on the first firing order and would be subject to repair or re placement. The Soviets probably would have stockpiles of missiles and bombs for reloading submarines and bombers, and perhaps missile launchers. There might be a certain number of missiles and bombs on production lines at the time of attack. Aside from these technical factors which would provide a postsurprise attack capability, the enemy might have good reason deliber ately to hold back some of his forces from the first strike. This might be the case, for example, if the Soviets thought there was a good chance that some U.S. forces would escape destruction in the initial attack. They might then wish to reserve some forces to go after the surviving U.S. installations after reconnaissance determined their location. Al ternatively, the Soviets might wish to retain some forces to deter the United States from using all of its residual forces against Russian cities, by threatening counterretaliation against U.S. cities. If Soviet aims included the conquest of the United States itself, they would want to have enough striking power left after their first attack to persuade the United States to surrender. The Soviets could not hope to avoid U.S. retaliation altogether by threatening to use their residual force against surviving U.S. cities. If the United States had determined in advance to retaliate immediately with all its alert forces, there would be no time, of course, for the Soviets to employ blackmail threats to prevent them from being launched. The missiles in these alert forces could not be recalled. However, it is conprovocative, we would have less to lose in retaliating immediately against Russian cities, and the Soviets' post-attack bargaining position would be weakened because the additional city destruction which the Russians could threaten might seem trivial in comparison to the destruction we had already suffered. But the deterrent benefits of such deployment would have to be weighed against the defense lia bility—the certainty that if the enemy did strike, we would suffer maximum dam age in the first blow.
with Strategic Nuclear Power
67
ceivable that the United States leaders could be convinced—after two or three hours of reflection, helped along by Soviet communications and threats—that the Soviets had achieved such preponderance in striking power that further retaliation against the Soviet Union would not only be futile in terms of winning the war, but also bring about further heavy destruction to U.S. cities and population which could be avoided. The manned bombers which had already taken off might then be re called. In addition, all the non-alert bomber and missile forces which had escaped destruction, as well as naval surface vessels and missilefiring submarines which were not within firing range of the Soviet Union at the time of the surprise attack, would be subject to neutraliza tion by blackmail.11 Actually, the Soviets might face a very ticklish choice between doing the maximum damage to the U.S. striking force in the first blow, and preserving a maximum capacity for bargaining and blackmail after that blow. Just where the balance was struck presumably would depend on the weights which the Soviets gave to two considerations: the marginal counterforce destruction to be realized from throwing more forces against U.S. installations, and Soviet expectations concerning what we would do with our residual forces—whether we would fire them off indiscriminately or hold them back for bargaining purposes. If these assumptions are correct—and they would seem to be at least 11
Herman Kahn has argued that the residual U.S. forces which could retaliate immediately or "reflexively" after the Soviet surprise attack would be a relatively small proportion of the whole. This may well be true in the years immediately ahead, but seems rather doubtful when the United States has come to rely heavily on quick-firing solid-fueled missiles, such as the Minuteman. When the Minuteman era arrives, perhaps by 1964 or 1965, the portion of our forces which can respond automatically or reflexively is likely to be a large percentage of the total surviving force. Then the Soviets would be able to avoid maximum retaliatory costs, through blackmail tactics, only if the U.S. government had deliberately decided, in advance of the attack, to hold back a significant portion of the quickreacting missiles for bargaining purposes, or perhaps to await assessment of dam age, reconnaissance of available counterforce targets in the Soviet Union, and retargeting. Whether or not the United States should plan to delay the firing of a portion of these weapons depends critically on how the balance is struck between the objectives of deterrence and defense. If our objective is only to deter, as large a proportion as possible of our forces should be set to respond reflexively, to guarantee maximum costs to the enemy. But if we wished to limit our own further losses after the attack takes place, we might decide to hold a maximum of forces back as bargaining counters or for discriminating attacks against those counterforce targets which seemed most fruitful—at the risk that deterrence might be weakened by the enemy's discovery of this intention.
68
Deterrence and Defense
plausible—what use of our remaining forces would contribute most to the twin objectives of mitigating damage and achieving the best possi ble postwar power position vis-a-vis the Soviet Union? The answer would seem to depend on the relative size, nature, and vulnerability of the surviving U.S. forces as compared with the residual Soviet forces, and the presumed Soviet willingness to negotiate an end to the war. Seven possible strategies are worth considering: (1) (2) (3) (4) (5)
All-out retaliation against Russian cities. All-out retaliation against the Soviet residual nuclear forces. Limited retaliation against Soviet cities as a bargaining tactic. Limited retaliation against Soviet nuclear forces. Strikes against Soviet tactical forces a n d their supporting facilities. ( 6 ) Some combination of t w o or more of t h e above strategies. ( 7 ) N o retaliation.
1. ALL-OUT COUNTERCITY RETALIATION This strategy at least has the merit of coinciding with the threat which is likely to have most deterrent potency in advance of war. In addition, it would satisfy revenge motives and would contribute to weakening the postwar power position of the Soviet Union. However, if there are counterforce targets in the Soviet Union which can be hit, or if there is some chance of terminating the war quickly by bargaining, and espe cially if the number of Soviet cities which can be destroyed is relatively small—say, 20 or 30—such a strategy probably would be irrational to carry out. It could not defeat the enemy or destroy his capacity or will ingness to make war; it probably would provoke him to destroy addi tional American cities which might otherwise have escaped punishment; and it would use up whatever bargaining power we still had by virtue of our residual nuclear forces. Missiles already fired, bombs already dropped, and Soviet cities already destroyed are subtractions from bar gaining power because they represent losses of capabilities to destroy assets as well as reduction of targets capable of being destroyed. However, there may be two sets of conditions which would make such a response rational: (1) when the residual Soviet nuclear forces are so well protected and concealed that they cannot be hit (or when their location is not known), and when the Soviets are believed certain to use all their remaining force against our cities whether or not we retaliate against their cities; or (2) when the undamaged U.S. forces
with Strategic Nuclear Power
69
can destroy enough Soviet cities to eliminate the Soviet Union as a functioning political entity, and when the residual Soviet nuclear forces are so small that the additional damage they would produce is consid ered a reasonable price to pay for the elimination of the future Soviet threat. If we could do so at small additional cost to ourselves, we would have an incentive to inflict as much damage on the Soviet Union as we could, in order to create a severe reconstruction problem for them and reduce the chances that they would recover first and then attempt to disrupt our own reconstruction. 2. ALL-OUT COUNTERFORCE RETALIATION If the residual Soviet forces were not all invulnerable, and if we be lieved the Soviets were completely impervious to bargaining possibili ties—that, once having started the war, they would proceed to inflict the maximum possible damage on us, first on our forces, and then on our cities—our primary objective would be to reduce as much as possible their ability to damage us. We would strike back with all our power against the enemy's nuclear forces, starting with the most vulnerable ones, which would include airbases and missile-submarine ports. Strik ing at missile launchers and silos might yield some dividends if our warning system were good enough to tell us which ones were still unfired and if we had accurate intelligence concerning their precise loca tion or a good reconnaissance capability. If our surviving forces were large enough, we might even want to hit missile launchers which had already fired their weapons if we believed the enemy had a reload capability. Nuclear bomb and missile construction plants also would be fruitful targets, as would stockpiles of missiles and bombs if we knew their location, and if they were not completely invulnerable to attack. If our surviving forces were larger than necessary to hit all vul nerable counterforce targets, we might use the remainder against Soviet cities and industrial centers in order to maximize Soviet postwar re construction problems and minimize postwar Soviet power. Or we could use them against the Red Army and its supporting facilities in the hope of blocking or retarding its movement into Europe or Asia. 3. LIMITED COUNTERCITY RETALIATION AS A BARGAINING TACTIC But the enemy may be willing to bargain. He will wish to limit his own war damage as much as possible; to this end he may offer some
70
Deterrence and Defense
sort of terms, or an ultimatum of some sort, immediately after his first blow. If the first strike has reduced our nuclear forces to an insignificant level, the terms are likely to be harsh: at best, "Do not retaliate beyond certain limits or you will suffer severe destruction of your cities"; at worst, "Capitulate or suffer extreme devastation." If our forces really have been almost eliminated and the enemy has a considerable force left to carry out his threat, then of course he holds all the bargaining cards and we would not be able to make any credible counterthreats. However, it is conceivable that we might have a sizable force left, perhaps because the enemy had run into bad luck or ineffi ciency in his first strike, or because the war had started by accident, or because the attack had been pre-emptively motivated—to forestall a U.S. first strike which the Soviets had considered imminent. In such a case we might be able not only to deter the Soviets from continuing their strategic attacks on the United States, but also to persuade them to give up or limit their territorial gains in Europe and Asia. We would support such demands by threatening to retaliate against Soviet cities and inflict costs more than offsetting whatever gains the Soviets could make by continuing the war. Our determination to inflict such costs if necessary could be underlined by limited strikes against a few Russian cities, accompanied by a threat to continue such strikes until the Soviets gave in to our demands. Such a strategy would be costly, of course, since the Soviets no doubt would retaliate in kind. If they were willing to call off the strategic exchange, provided they were given carte blanche in Europe and Asia, we might prefer to settle on this basis. However, if we placed a high enough value on forestalling Russian conquests overseas, we might be willing to accept some additional costs in limited countercity exchanges if we thought a few such exchanges would convince the Russians of our willingness eventually to unleash all of our residual forces against Rus sian cities. Of course, whatever forces we used up in such limited at tacks would be subtractions from our ability to cause further damage; we would be trading capability to cause damage to create an impression of will to continue causing it. If we assumed the Soviets would give in at any point only if they expected further damage at least equivalent to their expectation of gain, we must at all times have in reserve the capacity to inflict such unacceptable damage.12 12 It
could be maintained that we would always have to keep in reserve at least twice the striking power needed to produce unacceptable damage, since, if
with Strategic Nuclear Power
71
If the Soviets had a residual force more than sufficient to inflict dam age on U.S. cities offsetting the U.S. valuation of the territories which were objects of the bargaining, the Soviets might expect the United States to give up its demands after a few exchanges. The outcome of the bargaining process would seem to depend on three variables: the ratio between the residual forces on each side, each side's valuation of the objects of the bargaining, and each side's valuation of the assets and lives that would be lost in continuing the strategic blows. (Perhaps more important than the "objective" force ratio and valuations would be each side's image of the other's forces and valuations.) If the U.S. forces were large enough to pose the threat of unacceptable damage but the residual Soviet forces were very much larger, the bargaining advantage might shift to the Soviets according to their degree of superiority, although not necessarily in direct ratio to this superiority. However, the Soviet advantage in the force ratio might be offset if the United States could convince the Soviets that it valued the continued independence of countries in Europe and Asia much higher than the Soviets valued their conquest, or that it valued its assets and lives at risk lower than the Soviets valued theirs. If the Soviets had preponderance in ground and tactical forces, they would have the advantage of being able to offset losses of cities by ter ritorial gains. They also would have the advantage of momentum— once the Eed Army was on the move and facing ineffective opposition, to stop it would be difficult, let alone recalling it. The Soviet leaders could rather credibly say that they did not have sufficient control over the Army commanders to persuade them to give up gains already made. Restoring the status quo ante would mean a tremendous loss of prestige for the Soviets. The Soviets could also point out that since it was they who had started the war, they obviously had considered and resolved to accept the costs of U.S. retaliation—hence the threat of U.S. retalia tion could have no effect in reversing this decision. To the extent that the U.S. leaders recognized these Soviet advantages, their will to carry out their threat would be weakened, and if the Soviets expected the we had just this amount, our threat to use it in extremis would not be credible: we would know and the Russians would know that after we had expended our entire residual forces, we would have no more power to regulate the enemy's behavior. Thus we would gain nothing by carrying out the threat, and we would have no power to deter further nuclear attacks on ourselves. Having twice the capability to produce unacceptable damage would enhance the credibility of the threat to do so at least once.
72
Deterrence and Defense
U.S. will to be weak for this reason, their own will to carry out their threat would be strengthened. However, the position of the United States would not be hopeless. The revenge motive would be strong; even if the government were able to control its own and the people's natural inclination to retaliate im mediately and with full force, this control might be very precarious, and this precariousness would be recognized by the Soviet leadership. It would be all the more precarious because of the disruption of com mand and communications channels which would inevitably have re sulted from the Soviet surprise attack. It is true that if the Soviets had made their surprise attack with the intention of following it with an attempt to negotiate, they might very well have spared such vital con trol centers as the White House, the Pentagon, and the Strategic Air Command Headquarters, in order to insure that communications could be carried on with the American decision-makers who had the maxi mum authority and power to restrain the U.S. response. Even so, some disruption would have occurred at lower levels. The American govern ment could very credibly say that it could not be confident of maintain ing absolute control over the actions of field commanders, either because communications had been cut, or because these commanders might not accept restraints imposed from Washington. Once the Soviets had begun deliberately to destroy American cities, restraints would become progressively more difficult to maintain; there would be an in creasing probability that all the remaining American forces would be released in one final orgasm of destruction. This appears to be a situation in which the de-centralization of a democratic society, as well as certain traits in the American character—such as the tendency for emotion to prevail over reason in war—may work to our advantage, provided of course that these characteristics and traits are fully recognized by the enemy. It might be unrealistic for the United States to believe that by bar gaining and limited retaliation it could reimpose the territorial status quo ante in the Eastern Hemisphere. However, it might be reasonable for the United States to expect to limit the Soviet gains—meaning that there might be some settlement, short of the Communization of all Eurasia, which the Soviets would accept, and which would be worth the cost to the United States in imposing it. We may assume that after certain vital objectives were obtained, additional conquests would have diminishing utility for the Soviets. Presumably, at some point the
with Strategic Nuclear Power
73
Soviets would be willing to give up further gains rather than risk addi tional severe retaliation against their cities. For bargaining purposes, invulnerable forces would of course be more desirable than vulnerable ones. If they were vulnerable, the enemy might prefer to take them out as quickly as possible, rather than make bargaining concessions, and there would be great pressure upon our selves to fire weapons which were momentarily subject to destruction by new enemy strikes, rather than hold them in reserve. The enemy would face some difficulty in locating our weapons which were still unfired, but this difficulty might be overcome in part by post-attack reconnaissance if he had been successful in eliminating a good portion of our air defenses in his first blow. For vulnerability reasons, such weapons as the Polaris submarine and mobile ICBM's would seem to be most useful as bargaining cards. To speak of the possibility of bargaining after the United States has been hit by a surprise nuclear attack may seem rather strange. We have taken an apocalyptic view of all-out war for so long that it requires a strong act of will even to entertain the idea of terminating such a war by negotiation. The degree of reasoned self-restraint required for a bargaining strategy may not be within our psychological capabilities. Yet it must be admitted that such a strategy may be preferable to in discriminate countercity retaliation as a means of "defense" after a surprise attack, even though the threat of an indiscriminate response probably is the better "deterrent" to such an attack. Bargaining sup ported by limited countercity retaliation also is preferable to an all-out counterforce response when counterforce targets are few, or when their location is unknown, or when they are highly invulnerable. Certainly it would be in the interest of the United States to terminate the war as quickly as possible, with minimum additional damage to the United States population and economy, and minimum losses to our interests overseas—or some minimal combination of both these objectives. This interest surely overbalances the emotional satisfaction to be gained from maximum revenge. And, as noted above, there are easily imagin able circumstances in which the enemy would also wish to end the war quickly, even at the sacrifice of some potential tactical gains. 4. LIMITED COUNTERFORCE RETALIATION If, after the surprise attack, there were some counterforce targets in the Soviet Union which were vulnerable, and whose location was known,
74
Deterrence and Defense
a bargaining strategy might be supported by limited, selective attacks on these targets, rather than against cities. But if such attacks are to be part of a bargaining strategy, they must not be carried out all at once, by the commitment of our entire residual force, but one or a few at a time, so as to preserve at all times a maximum capacity to threaten future damage. One advantage of such a limited counterforce strategy is that it would simultaneously serve the objectives of mitigating potential war costs and bargaining. If the Soviets prove unamenable to bargaining, we at least reduce their capacity to inflict further costs on us. If they do want to bargain, counterforce strikes may be almost as good as countercity attacks to indicate our determination to continue the fight if our terms are not met. There is the possibility, of course, that the progres sive attrition would work against the United States—either because we used up more of our forces in attacking the enemy's forces than we were able to destroy, or because he could attack our residual forces more effectively than we could attack his. If this were the case, limited countercity retaliation might be preferable, especially if we had civil de fense facilities which allowed evacuation of the population of major cities to fall-out shelters in rural areas. The response might conceivably mix some attrition of the enemy's forces with some strikes against cities. The countercity strikes would emphasize our willingness to engage in a countercity exchange; the counterforce blows would reduce his capability for such an exchange. This poses a problem of balancing; its solution would depend largely on the degree of resolution which we thought the enemy perceived in us, and in the degree of attrition of the enemy's forces which we could cause by counterforce strikes. If the enemy's forces were substantially invulnerable, there would be little point of course in hitting back at them; there might be some point in hitting a few of his cities. If his forces were vulnerable, more advantage might be gained by reducing his ability to damage us than by supporting our determination to dam age him. 5. STRIKES AGAINST THE ENEMY'S TACTICAL FORCES Still another type of response would be retaliation against the enemy's ground forces and the supply and transportation facilities supporting them. This would be a form of "counterforce" response, but intended not to reduce the enemy's capacity to hurt us directly with nuclear
with Strategic Nuclear Power
75
blows, but rather to weaken his ability to make territorial gains in Eu rope and Asia. One of the enemy's objectives in launching a surprise nuclear attack would certainly be to make such gains. Presumably he would delay mobilization and concentration for his ground movement until after his initial nuclear attack, in order not to raise suspicion of his intent. Therefore, following the surprise attack, his troop concentra tions could be attacked with significant effect while they were still on Soviet or satellite territory. Supply dumps and other supporting installa tions would also make attractive targets. Such facilities, and troop units, may be more attractive targets than missile launching pads and airbases simply because we are likely to have more reliable information about their location. Attacks on ground forces probably would cause rather high incidental damage to the civilian population; yet they would be less provocative than deliberate attacks on cities—that is, less likely to provoke the Russians to a massive indiscriminate response. Such a strategy would indicate a desire to fight a restrained war, pri marily against military targets, and therefore would be conducive to the preservation of a more or less rational atmosphere which in turn would promote negotiations leading to an early termination of the war. This might be a better bargaining strategy than attacks on the remain ing Soviet nuclear forces, since it might raise doubts in the minds of the Soviet leaders that they would be able to consolidate territorial gains to offset their war costs. For bargaining purposes, some combina tion of attacks on ground forces, the most vulnerable strategic forces, and perhaps a few cities would seem the most reasonable. 6. NO RETALIATION Finally, the alternative of no retaliation at all should be considered. Emotionally, it would be bitter medicine, of course, to have to accept a nuclear attack without inflicting any punishment on the enemy in re turn. Emotional values are legitimate values, and we may "rationally" be willing to accept additional costs in order to satisfy them. But, even allowing for this, in some circumstances no retaliation may turn out to be the most rational course. This would be the case if the enemy had succeeded in eliminating virtually all of our forces in his first strike, or if he had established a great preponderance in his favor. Then any at tempt at bargaining probably would be futile, there would be no possi bility of causing a significant attrition to the enemy's forces by counterforce action, and to throw all our remaining forces at his cities probably
76
Deterrence and Defense
would result in further damage to the United States which might have been avoided. It might be the course of wisdom to retain our undam aged forces as a deterrent against invasion of the United States itself and as a nucleus for the eventual rebuilding of our strategic capability. Of course, if our response to a surprise attack is to be something other than an all-out retaliation against Soviet cities, it is rather important that we plan such restraint in advance. It is difficult to imagine the United States government planning for all-out punitive retaliation and then changing its mind in the confusion following an attack. Such be havior would mean being irrational when conditions are optimum for rational thinking (before the attack) and suddenly becoming rational when conditions are most conducive to irrationality (after the attack has occurred). Of course, the Soviet government might well try to "educate" the United States government, after the attack, as to what its rational response would be, but the conditions would not be propi tious for such education. The conclusion is inescapable that if we are to act rationally after the war starts, the war plans must reflect this intention. As pointed out earlier, rational action might very well involve some thing other than all-out retaliation against Russian cities. But it is the prospect of an all-out punitive response which is the most potent deter rent to a Soviet surprise attack. If war plans calling for a restrained, discriminating response were to become public knowledge, deterrence would be weakened. Therefore, it may be desirable to declare and ap parently intend to retaliate all-out against cities, but actually to plan to do otherwise. Here we arrive at a crucial question. Is it possible for a democratic government to practice such deceit? Must we place all our bets on either deterrence or defense—i.e., either actually plan to do the thing that promises the greatest cost to the enemy even though it means maximum cost to ourselves if deterrence fails, or plan to minimize our own costs in war even though this causes a weakening of our deterrent threat? We may be able to have some of both worlds, if not the best of both. That is, it may be possible to threaten immediate all-out countercity retaliation for deterrence, and yet have a set of contingency plans call ing for more discriminating responses. The existence of such plans probably could be kept secret if sufficient precautions were taken. The appropriate response would be selected after the Soviet first strike, ac-
with Strategic Nuclear Power
77
cording to the apparent possibilities for bargaining and for counterforce action, and weapons would be assigned targets accordingly. Aircraft and missile crews would have to be assigned first-priority targets in advance of the attack for training purposes and to hedge against failure of the retargeting mechanism as a result of the attack. These first-pri ority targets should probably be counterforce targets, attacks upon which would be sure to have some utility. The most important of these would be assigned to those of our own forces which were most likely to survive the enemy's surprise attack. Large home airbases, smaller staging airbases on the fringes of Soviet territory, submarine pens, and important installations supporting the enemy's ground forces would be among such targets. But there should also be provision for shifting some forces to subsidiary targets—cities, in case limited countercity retalia tion, with bargaining, appeared the most fruitful strategy after the first strike; and missile sites, in case some Soviet missiles appeared not to have been fired. It may be doubtful whether our reconnaissance, dataprocessing, and control systems would be good enough in the post-attack environment to do more than very crude retargeting. Yet, for efficient defensive action, provision should be made for the tightest possible control and selection of retaliatory actions, to insure that military action could be kept consistent with the diplomatic post-attack negotiations, should such negotiations prove possible. The administration's war plans in the spring of 1960 apparently called for an all-out combined counterforce-countercity reaction to a surprise attack—that is, a combination of alternatives (1) and (2) dis cussed above. Secretary of Defense Thomas S. Gates said that "we are adjusting our power to a counterforce theory; or a mixture of a counterforce theory plus attacks on industrial centers and things of that sort." Secretary Gates also said, however, that "what we actually do depends on circumstances," and General Nathan F. Twining, the Chairman of the Joint Chiefs of Staff, stated that various alternative war plans were available for the contingency of surprise attack.13 However, there was 18 House Committee on Appropriations, "Department of Defense Appropriations for 1961," Hearings, Part 1, 86th Congress, 2nd Session, pp. 26-27. These state ments seemed a good deal more tentative than the objective stated by General Thomas D. White, the Chief of Staff of the Air Force, a year earlier: ". . . to do the greatest possible damage to the Soviet Union as a whole with attention to applying that destruction in such a way as to do as much damage as possible to their residual military striking force." House Committee on Appropriations, "De partment of Defense Appropriations for 1960," Hearings, Part 1, 86th Congress, 1st Session, p. 929.
78
Deterrence and Defense
no indication that any of the alternative war plans contemplated a limited-retaliation bargaining strategy. The dominant concept appar ently is one of all-out retaliation, with first priority going to the enemy's residual nuclear forces and second priority to economic strong points. If we wish to go beyond the mere objective of deterrence, and to be prepared to fight an all-out war most effectively should the enemy initi ate such a war, the military requirements are larger and somewhat different in kind than for deterrence alone. For "minimum deterrence," all that is required is to be able to strike back at a relatively small num ber of large-city targets. In "defense" or the waging of all-out war, it may be desirable and possible to hit some counterforce targets in order to minimize the enemy's capability to damage us. Since these targets are likely to be small and well-protected, this may pose a requirement for multiple missile firings at each target—and hence for large numbers of missiles—or for very accurate weapons such as the manned bomber which can also conduct reconnaissance. At the extreme, besides the counterforce requirement, the defense objective also would create a utility for enough weapons to infliot paralyzing destruction on the Russian economy—enough destruction, at least, to eliminate the Soviet regime and reduce Soviet power potential to a level that would no longer constitute a threat to Europe and the United States. Short of this, "defense" would require at least enough strike-back capability to bar gain the enemy into terminating the war on terms more or less favorable to ourselves. Of course, if we had pre-attack forces of this size, the Soviets would be extremely unlikely to start the war at all. We would have a "maxi mum deterrent" rather than a "minimum deterrent." The enemy would face the prospect not only of much greater damage than if we had only a minimum deterrent, but also of being balked in his war aims and possibly even of losing the war or of suffering the complete collapse of his society. Whether it would be desirable to supplement the minimum deterrent force with such additional forces would depend on whether their cost was thought to be justified by the combined effect of the re duced probability of attack, the diminished war losses, and the possi bility of winning the war, or at least of destroying Soviet future power potential to a degree commensurate with our own destruction. It is quite conceivable, of course, that the chances of war would be considered so low with only a minimum deterrent force that the "defense" value pro vided by additional forces would not be thought sufficient to justify
with Strategic Nuclear Power
79
their additional cost in peacetime resources. However, there is another possible value in having a larger-than-minimum capability—its contri bution to the deterrence of major Soviet aggressions outside the United States.
D. Deterrence of Soviet Aggression in Eurasia To deter a direct Soviet thermonuclear attack on the United States, the essential requirement is that we have the capability to inflict intoler able "strike-back" damage on the Soviet Union. The factor of intent or will is not critical, because the Soviets can hardly gamble on the chance that we will not retaliate with all the forces we have left after the Soviet attack. But for the deterrence of major Communist attacks in Europe and Asia by strategic nuclear means, the relative importance of these two factors is reversed. Here the capability to retaliate so severely as to negate any conceivable enemy gains cannot be questioned, but the enemy may very well doubt our will to do so, in the face of his capacity for counterreprisal. In short, the dimension of credibility here becomes crucially important. The credibility of the threat of nuclear retaliation to, say, a largescale Soviet ground attack on Western Europe depends on convincing the enemy that we think we would gain more by carrying out this threat than we would lose. For maximum credibility, the Soviets would have to believe that we possessed a sufficient first-strike counterforce capabil ity—i.e., one that we could rationally use in response to a Soviet tactical aggression. If we assume that the United States and its allies are greatly inferior to the enemy in ground or tactical forces, such a capability must meet two primary requirements. First, it must be able, in striking first, to eliminate enough of the Soviet strategic nuclear forces so that their retaliatory blow (assuming that all the residual Soviet forces are de livered in retaliation) would be less costly to us than the value we placed on the continued independence of the countries attacked by the Soviet Union, and on other more positive gains we might expect to make. Secondly, it must be able to prevent the Soviets' intended con quests. This latter requirement could be met by strategic forces in either of two ways. Conceivably, a United States first strike could, by hitting cities, decision-making centers, and the ground forces themselves and their supporting installations, destroy the Soviet will or capability to continue fighting and force a withdrawal or collapse of the Red Army.
80
Deterrence and Defense
But if this result could not be confidently expected, the United States must be able to hold back sufficient strategic forces from the first strike to threaten the Soviets with enough countercity damage on later strikes to persuade them to restore the territorial status quo. Either way, the first-strike requirement is greater than simply reducing the enemy's retaliatory blow to a level commensurate with our own value of the objective. There would be little comfort in knowing the objective would be "worth" the costs if the objective were expected to be lost!14 The objective for the United States might of course include values beyond merely forcing the Soviets to disgorge whatever territorial gains they had made. There would be a certain amount of moral value sim ply in honoring our commitment to our allies, even if retaliation could not prevent their conquest. Moreover, it is quite likely that if the United States ever did have the capability and will to launch the first blow in all-out war, the war aim would include the toppling of the Soviet regime itself, and the establishment of democratic institutions in Russia and the Eastern European satellites. The contemplation of the possibility of this positive gain, as well as the reduction of the Russian threat to the future security of the United States, might make the United States willing to suffer much greater losses in all-out war than if the only object of the fighting were to save Europe and Asia from Communist control. In other words, it would reduce the U.S. requirement for a sufficient firststrike capability. These larger objectives might actually be inseparable from the more immediate aim of preventing Soviet conquests; that is, perhaps a necessary condition for stopping and rolling back the Red Army would be the collapse of the Soviet government and the advent of new leaders whose lack of responsibility for initiating the war would allow them to repudiate it and terminate the war on Western terms. The United States calculation of a "sufficient" first-strike require14 There may be a certain amount of "trade-off" between our expected cost from the enemy's retaliation, and our expected territorial losses. If the requirement of restoring the territorial status quo is relaxed, so that the Soviets may be able to make limited tactical gains, the damage-reducing requirement becomes more stringent: the Soviet strike-back capability must be reduced further by our first strike to compensate for our expected tactical losses. In the spring of 1959, General Thomas D. White, Chief of Staff of the Air Force, described SAC's first-strike objectives as follows: "One, to destroy the enemy's capability to destroy us—that would be the first priority; next would be to blunt the enemy attack on our deployed military forces and other forces in Europe and in Asia; and third, systematically destroy the Soviet Union's ability to wage war." Significantly, the alternative of terminating the war by bargaining supported by threats of direct strikes against cities was not mentioned. See ibid.
with Strategic Nuclear Power
81
ment is roughly parallel to the Soviet calculation of the first-strike cap ability it needs to undertake safely a nuclear attack on the United States. However, there are three important asymmetries which tend to make the U.S. requirement larger. One is related to a point just mentioned: a first strike by the United States is irrational unless Soviet territorial gains can be prevented or at least restricted. The Soviets, in contem plating their first strike, do not face the problem of having to defeat the Western allies on the ground with strategic weapons. Their tactical superiority provides them with this capability. A related point is that the United States must be able to threaten unacceptable damage with its second salvo unless the Red Army can be physically stopped by tac tical interdiction strikes. What is "unacceptable" damage to the Soviets after they have already committed themselves to tactical aggression— i.e., the degree of threatened destruction necessary to persuade them to restore the status quo—is likely to be larger than the amount required to deter them when they are only contemplating aggression. Secondly, a Soviet nuclear attack on the United States would have the benefit of maximum surprise. A U.S. first strike after Soviet tactical provocation would not have this advantage. Simultaneously with their tactical move, the Soviets certainly would alert their air defense and striking forces and put in motion measures for civil defense. A portion of their bomber force would get into the air and major cities might be evacuated. Such measures would greatly reduce the counterforce effectiveness of our first strike and would also reduce its by-product population damage. A third asymmetry concerns military intelligence: the Soviets prob ably will always have a much more precise knowledge of the number and location of U.S. missile sites and airbases, as well as their degree of protection, than the United States will have of the Soviet strategic forces. This will make the United States much less confident of the chances of a successful first strike than the Soviets can be, and will require the United States to fire more weapons at a single target to compensate for inexact information about its location. As against these considerations there are two asymmetries which probably favor the United States. One is that since the Soviets will al ready have attacked at the tactical level, thus engaging all the intangible values associated with the U.S. commitment to its allies and taking upon themselves the moral burden of initiating warfare, the U.S. decision to strike first at the strategic level may be psychologically easier to reach than a Soviet decision to make a surprise attack "out of the blue." The
82
Deterrence and Defense
second—perhaps more questionable—is that the United States is likely to value the protection of its allies higher than the Soviets value their conquest. It seems to be generally true of human nature, although not without exception, that the party which already has or controls some thing is willing to spend more resources and energy to prevent its loss than the party which merely covets it is willing to sacrifice to take it. Intuitively, this rule would seem to hold in most situations of interna tional conflict, barring extreme emotional, ideological, or irrational pro pensities in the aggressor. The United States probably is willing to accept more damage in the defense of Western Europe than the Soviets would be willing to accept as the price of taking it. This asymmetry in valuations tends to reduce the U.S. first-strike requirements as com pared with the Soviets'. It must be admitted that the requirements for a sufficient first-strike counterforce capability—one which would make massive retaliation a rational act, and the threat of it maximally credible—are extremely stringent; much more stringent in terms of both quantity and quality of weapons than a minimum strike-back capability. The targets of a first strike would not be a limited number of large cities but a very large number of small military installations, many of them concealed, mobile, or otherwise well-protected. In fact, a sufficient first-strike capability is probably unattainable for the United States, barring a very significant improvement in air defense which is not matched by the Soviet Union. Even in the short run, while the Soviets still rely chiefly on aircraft for their strike-back capability, some of these aircraft undoubtedly would escape our first strike and would reach their targets—enough of them to inflict retaliatory costs so grievous that not even the prospect of sav ing Western Europe would justify them. A sufficient strike-first capa bility will become still more difficult to obtain when the counterforce targets are multiplied by Soviet emplacement, hardening, dispersal, and concealment of large numbers of long-range missiles. This does not mean, however, that the threat of a first-strike has lost, or will lose, all its credibility. It can stand repeating that what counts in deterrence is not what we are willing to do, but what the enemy thinks we may do. It may be possible for the United States to have a first-strike capability which at least creates an expectation in the enemy's mind that retaliation may take place. The high credibility which would be provided by a "sufficient first-strike counterforce capability," as de fined above, may not be necessary for effective deterrence. Even a very
"with Strategic Nuclear Power
83
small chance, in Soviet estimations, that the United States might re taliate strategically to a massive ground attack in Europe might deter such an attack, in view of the very severe damage the Soviets would suffer if they miscalculated our intent. There are several reasons why the enemy must consider a first strike by us a possibility, even if he doubts that we have a sufficient first-strike capability. One reason is that he does not know how we estimate the consequences of his retaliation. He has to consider that even if he is sure of being able to retaliate unacceptably after our first strike, we may have underestimated his capability to do so. Secondly, the enemy does not know the values and utilities by which we judge the conse quences we expect. He will surely be in doubt, for example, about what level of retaliatory damage we would consider unacceptable. He can not be sure but what our prestige and moral investment in our commit ment to Western Europe may be high enough to make us willing to accept very high damage in fulfilling that commitment. Thirdly, the enemy can never be sure we will act rationally. In the fourth place, he must consider that even if we do not initiate all-out war immediately following his first tactical move, our initially limited response might eventually develop into something approaching all-out war—because of a spiraling situation growing out of a limited nuclear war, because of accidents, or because our expectations, values, emotions, or propensities to irrationality may change during the course of the conflict. Finally, the threat of a first strike might have some credibility, even if the Soviets knew with certainty that we did not have enough counterforce power to eliminate a hard core of very well-protected and wellconcealed forces which could inflict damage well beyond our level of tolerance if they were all used against our cities. The thought would occur to them that we might strike first in the belief that we could establish a significant preponderance in residual forces by doing so, and that we could then bargain the Soviets out of retaliating all-out with their undamaged forces. Of course, this would require that we hold back enough forces from our first strike to threaten completely unac ceptable subsequent damage to Soviet cities. The Russians would know, and know that we would know, that the blind discharge of all their remaining forces against our cities would yield little gain for them and would allow us to threaten or attack their cities without further fear of retaliation. Therefore, they might wish to hold back most of their residual forces to deter us from striking again. Then the stage would be
84
Deterrence and Defense
set for a war of limited selective blows, against either cities or tactical targets, with concurrent diplomatic negotiations—a war from which we might expect to emerge with less total value losses (in terms of damage plus territorial losses in Europe) than if we had not initiated strategic nuclear warfare. Or we might believe that the Soviets would retaliate against our residual strategic forces (in which case we would suffer considerably less damage than if they struck a full-force blow against our cities), and that after the Russian counterforce strike we would still retain enough forces to threaten enough damage against Soviet cities to persuade them to restore the territorial status quo or at least restrict their conquests. Admittedly, striking first under either of these assumptions would be very risky. It would assume a degree of rationality and sheer physical and administrative control in the Soviet leadership which probably is quite unrealistic. The point is, however, that the Soviets do not know but what we might not act on such assumptions. As it becomes more widely realized that it is irrational to expend all one's forces against the cities of an opponent who still possesses substantial residual forces, such a strategy will become more plausible, even to the Soviets. For all these reasons, the threat of a first strike by the United States is likely to retain some credibility even when our capability for such a strike falls well short of the standard of "sufficiency," and when we do not actually intend to strike first under any circumstances. Presumably the degree of credibility will vary according to the size of our first-strike forces. The required size of a sufficient first-strike capability is sensitive, of course, to the value of the objective, an attack upon which is to be de terred. In other words, a first-strike capability is "sufficient" with respect to any particular enemy provocation or move when acquiescence in that move would be more costly than acceptance of the enemy's retaliatory blow after our first strike. For example, to be willing to carry out a first strike after a Soviet incursion into Iran, or to be able to threaten such a strike, before the Soviet move, with enough credibility to deter, we would need a much larger first-strike capability than in the case of a massive Soviet ground assault on Western Europe. The more serious the provocation, the less demanding is our need to limit our own dam age by counterforce strikes.
with Strategic Nuclear Power
85
E. Qualitative Requirements Of course, practically all strategic weapons can be used alternatively either to threaten retaliation against the enemy's cities as a deterrent to his first strike, or to strike back at his cities or forces in "defense" after the enemy has struck first, or to strike first at the enemy's forces. Puni tive and counterforce capabilities are not mutually exclusive. However, there are differences between the requirements for the punitive and counterforce missions, and among types of weapons and carriers, which tend to make some of them more suitable for one of these purposes than for the other.15 These differences center on questions of accuracy, pene tration capability, destructive yield, range, reaction time, and vulner ability. For deterrence of direct attack on the United States there is a pre mium on reducing vulnerability of weapons installations, by such methods as hardening, dispersal, mobility and concealment, and, in the case of aircraft, on a capacity for quick take-off after receipt of warn ing. Such protective measures are also desirable, of course, to maximize capabilities for counterforce action and other forms of defense if deter rence fails. Protection is unnecessary for first-strike deterrence except for those forces which are held back from the first strike for bargaining purposes. A decision to have a substantial first-strike capability would reduce the invulnerability requirements for a minimum strike-back capability, since a first-strike capability would require large numbers of weapons, and numbers can substitute to some extent for protective measures, provided the weapons are well dispersed. Accuracy is not essential to the strike-back punitive mission, since cities are large and "soft" targets. However, it is crucial for the counterforce mission, whether on a first strike or in striking back. The reason, of course, is that counterforce targets are small and likely to be well protected, perhaps to the extent that a direct hit would be necessary to destroy them. Numbers and, to a degree, yield are substitutable for ac curacy. To compensate for imperfect accuracy, several very powerful weapons may have to be used against each counterforce target, a con sideration which may multiply the quantitative requirement for coun terforce attacks, a requirement which is large anyway because of the large number of targets which it may be desirable to hit. 15 George
W. Rathjens, Jr., discusses differences between counterforce and punitive requirements in Klaus Knorr (ed.), NATO and American Security, Princeton: Princeton University Press, 1959, pp. 73ff.
86
Deterrence and Defense
Of the various weapon systems now available, or in sight, which are most suitable for each of the three strategic objectives we have been discussing: deterrence of surprise attack, defense after surprise at tack, and deterrence of major Soviet aggression against other countries? The outside scholar without access to classified information, especially cost information, cannot give a definitive answer to this question, but at least it is possible to discuss certain apparent characteristics of the weapons and compare them with the requirements for each of the ob jectives. Five weapon systems will be briefly considered: the fixed landbased ICBM, the mobile land-based ICBM, the land-based IRBM (fixed or mobile), the submarine-based IRBM, and the manned bomber. For the deterrence of surprise attack, assuming this is the only ob jective, the essential problem can be simply stated: which weapon sys tem, or combination of systems, will provide an assured minimum capa bility to strike back at Soviet cities at least peacetime cost to ourselves? This statement of the problem subsumes a variety of complexities, but these can be reduced somewhat if we make two simplifying assump tions: that the enemy's first strike will be made by ballistic missiles fired from the Soviet Union or possibly from submarines off our coasts, and that we have a warning system which will provide at least tentative warning 15 or 20 minutes before impact of the missiles fired from the Soviet Union. The first requirement, of course, is survivability of our retaliatory forces. Here the fixed, dispersed, hardened, land-based ICBM gets fairly high marks because of its small size, distance from the enemy's firing points, and ability to survive anything but a direct hit or a very near miss. It seems unlikely that the accuracy and reliability of the enemy's missiles and of his target intelligence will ever be good enough so that he can be confident of destroying a well-hardened launching site with only one attacking missile. A reasonably conservative attacker will be lieve it necessary to fire several at each site, although advances in tech nology may reduce the required number to as low as two per site. Recent tests of the U.S. Atlas missile and Soviet missiles have indicated that the accuracy of ICBM's is likely to be much better than had previously been expected.16 But against this it must be pointed out that accuracy 16 President Eisenhower said in his State of the Union Message in January 1960 that the Atlas ICBM had been hitting, on the average, within two miles of the target in recent tests. Earlier, Moscow had announced that a Soviet long-range shot into the Pacific had struck within 1.24 miles of the aiming point (New York
with Strategic Nuclear Power
87
against a pre-surveyed test target may be a totally different thing than accuracy against an enemy target the location of which is not known with similar precision. Also, accuracy predictions developed from testfirings are subject to the statistical liability of any small sampling—the smaller the sample, the greater the margin of possible error. Under foreseeable air defense technology, the penetration capability of the ICBM will be high—much higher at least than the manned bomb er. Missiles can carry penetration aids of various kinds to confuse the enemy's air defenses, although the amount of such equipment may be limited in later-generation missiles, such as Polaris and Minuteman, because of their limited payload capacity. ICBM's may never be as ac curate as manned bombers, but they will be accurate enough to hit large cities, which is all the accuracy they will need as a punitive deterrent force. Mobile land-based ICBM's—for example, Minutemen mounted on railroad cars or trucks and shuttling back and forth in random fashion between pre-selected firing points—may turn out to be a surer and cheaper means of assuring a strike-back capability than hardened fixed sites if the enemy is believed to have developed great firing accuracy against fixed targets. There is some danger, since mobile ICBM's will be "soft," that the enemy could determine their general location at any particular time and eliminate them all by area attacks—i.e., by blanket ing whole areas with enough blast pressure to destroy all aboveground structures. However, such attacks would also ensure very heavy civilian casualties and economic damage, which might weaken U.S. inhibitions against striking back indiscriminately with surviving forces against Russian cities. This consideration might strengthen deterrence by weak ening the Russians' hopes of limiting their own damage by post-attack Times, January 23, 1960). Aimed at a very hard missile-launching site (e.g., able to withstand 100 pounds per square inch of "overpressure"), the Soviet shot would have destroyed the site if it had carried a 5-megaton warhead. A shot hitting two miles away from such a target would destroy it with a 30-megaton warhead. If both accuracies are interpreted as "circular error probable" (CEP)— meaning that half the shots flred would fall within the stated distances from the target—the Soviets would have to fire four 5-megaton warheads to be 90 per cent confident of destroying the target, while the United States, with a two-mile CEP, would have to fire ten missiles of 5-megaton yield to achieve more than a 90 per cent probability of a kill. Thus small differences in accuracy make a con siderable difference either in the size of warhead or number of shots required to destroy a target, and several shots are required to be sure of eliminating a very hard target even with these quite good accuracies and very heavy warheads.
88
Deterrence and Defense
bargaining and blackmail. On the other hand, a mobile system which the enemy could counter successfully only by area attacks would work against our "defense" objective of limiting our own damage should war occur. Other means of beating a mobile system are conceivable—for ex ample, reconnaissance satellites transmitting continuous television pic tures of missiles on the move, supplemented by terminal guidance of attacking missiles.17 Fixed land-based IRBM's in Europe, or at other points around the Soviet periphery, appear less desirable than ICBM's for the deterrence of surprise attack because of their proximity to the Soviet Union, which makes them much more vulnerable to being destroyed by the Soviet first strike. However, mobile IRBM's in Europe would have a much better chance of surviving, and could be a valuable part of the minimum de terrent against surprise attack, if political obstacles could be overcome.18 The submarine-based Polaris IRBM undoubtedly has strong elements of invulnerability, although its actual degree of invulnerability is a mat ter of controversy, chiefly between the Air Force and the Navy. Its invulnerability is said to inhere in its combination of three elements: mobility, concealment, and the protection provided by the sea itself. However, there are conceivable ways in which it could become quite vulnerable to enemy attack, especially considering the huge submarine fleet of the Soviet Union. If the Soviets were to find a way to locate, identify, and trail all our Polaris submarines "on station" around the Soviet periphery, they might be able to destroy them all nearly simul taneously. The Soviets can be expected to put large resources into solving this problem because the counterforce benefits from solving it 17 Major General G. I. Pokrovsky, a Soviet officer, has written of "self-seeking, pilotless missiles" and observation of mobile targets by television. See his Science and Technology in Contemporary War, translated and annotated by Raymond L. Garthoff, New York: Frederick A. Praeger, 1959, p. 115. 18 One possible political obstacle would be the fear of allied governments that the spectacle of missiles constantly moving about on their territories would create serious anxieties among their populations. On the other hand, General Maxwell Taylor, the former Army Chief of Staff, has stated that a number of allies would prefer mobile installations to fixed ones because they believe that fixed-based missiles would be more likely to draw fire to their countries in case of general war. Preparedness Investigating Subcommittee, Senate Committee on Armed Serv ices, in conjunction with Senate Committee on Aeronautical and Space Sciences, "Missiles, Space and Other Defense Matters," Hearings, 86th Congress, 2nd Session, p. 206.
with Strategic Nuclear Power
89
would be very large. Every Polaris submarine successfully torpedoed would mean the destruction of 16 missiles. It is conceivable that the Soviets might try to eliminate Polaris sub marines one by one, on the assumption that an attack on a single sub marine would not trigger all-out retaliation. They would be much less likely to think they would avoid retaliation if they made piecemeal attacks on fixed or mobile installations in the United States, not only because of the unavoidable by-product damage, but also because to at tack the territory of an enemy is a much more provocative act than to attack his ships at sea. An argument sometimes used in support of the Polaris system is that it would "draw fire" away from the United States.19 This is certainly an advantage from the point of view of defense or minimizing losses should deterrence fail. However, it might detract from the deterrent value of Polaris, if this system were relied upon exclusively. The Soviets might believe that a counterforce attack which did not damage the United States itself would not produce sufficient emotional stimulation to trigger an all-out indiscriminate response by the United States. Since the United States economy and population would be undamaged, we would have much more to lose in retaliating and accepting Soviet counterretaliation than if the Soviet first strike had been directed at land installations, in which case we would already have suffered severe population damage. This is not an argument against having a Polaris system at all, but only against relying exclusively upon it. In a mixed force including both Polaris and land-based installations, this particular liability of the Polaris would not apply. There are other questions concerning the Polaris submarine. Because of the difficulties of communicating with submarines, the order to re taliate may not be received: it can be assumed that one of the enemy's prime targets on his first strike will be the Polaris communication sys tem. If the submariners are cut off from contact with the outside world, they will not know that a war has begun, although of course they may suspect it. But they would hardly fire their missiles on suspicion alone. During the delay in re-establishing communications, the Russians might expect to wipe them out, or to persuade the U.S. authorities not to order them to fire, by threatening additional destruction to U.S. cities in re10 For a statement of this argument, see Oskar Morgenstem, The Question of National Defense, New York: Random House, 1959.
90
Deterrence and Defense
taliation. On the other hand, it might be argued that if the Russians were uncertain about their capability to destroy the Polaris communi cation system, or the submarines themselves nearly simultaneously, they would be deterred from taking the risk. Despite these possible liabilities, the Polaris system undoubtedly will be an extremely valuable component in a mixed retaliatory force. As compared with missiles, the manned bomber has both disadvan tages and advantages for the deterrence of surprise attack. Airbases are larger targets than missile launching sites and even a near miss with a megaton weapon would probably destroy all aircraft parked aboveground at the base. Aircraft can be sheltered, but a direct hit on their runways could prevent them from taking off, unless methods were de vised for making take-offs directly from shelters. Aircraft not on alert therefore get low marks for vulnerability. However, aircraft on airborne alert or on 15-minute ground alert are probably less vulnerable than almost any type of missile, assuming that 15-minute warning will be available. Aircraft can take off on tentative, unconfirmed warning, sub ject to a system of "positive control" or "fail safe," under which they turn back at a certain point unless they receive definite coded orders to proceed to their targets. Since missiles cannot be called back once they are launched, they can be launched only upon unequivocal warning or after impact of enough enemy bombs and missiles to leave no doubt that an attack is under way. If they must ride out an attack, they will of course suffer losses which aircraft on alert might avoid.20 The bomber is much more vulnerable to enemy air defenses than is the missile, but the bomber's penetration capability should not be underrated. The U.S. bomber attrition rate in World War II was 1.8 per 20 It is technically possible to build a "destruct" mechanism into a missile war head so that it can be destroyed before it reaches its target. However, the mech anism might not be completely reliable at long ranges; a missile's flight time gives little time to disconfirm or confirm a warning; to destroy a full-scale salvo of missiles would destroy a significant portion of our capability; and there would be danger that the enemy would learn the destruct signal, in which case we would have handed him gratis a very good anti-missile defense system. See House Committee on Appro priations, "Department of Defense Appropriations for 1961," op.cit., Part 2, pp. 386-87. However, the later missiles have a partial equivalent of the fail-safe system in their capacity to be "counted down" on warning and held just short of firing. There is probably some risk of accident in this procedure, but it shortens reaction time and allows the firing of a maximum number of missiles immediately after the impact of the enemy's first missiles confirms unambiguously that the war is on.
with Strategic Nuclear Power
91
cent,21 a fact which does not seem to support estimates sometimes heard that 70 or 80 per cent might be lost in World War III. Of course, air defenses have improved greatly with the addition of anti-aircraft missiles and sophisticated detection and tracking devices. But these devices, relying heavily on electronics, are subject to jamming and con fusion by electronic countermeasures. Bombers may also pursue evasive tactics, such as low-level sneak attacks, and they can "saturate" defenses by a well-coordinated attack. In addition, the enemy's air defenses will be subject to destruction; in fact, they are likely to be first-priority targets in our retaliatory blow. Air-launched missiles, such as the Hounddog and the later Skybolt, capable of being fired from 500 to 1,000 miles away from the target, may give aircraft a capacity to blast a path through air defenses at little risk to themselves. ICBM's and IRBM's might also be used for this purpose. Air-launched missiles, especially the projected 1,000-mile Skybolt, may also enable aircraft to stay outside the enemy's defense screen alto gether and fire at targets other than air defense installations. Thus the missile-carrying bomber combines the missile's superior penetration capability with the bomber's ability to protect itself by taking off on warning, subject to recall; it provides more time in which to determine whether an attack is under way before making the irrevocable missile launch, and thus lessens the danger of accident or avoids missile losses consequent upon riding out an attack. The primary utility of the manned bomber as part of the "minimum deterrent" lies in its ability (if on alert status) to become airborne be fore the enemy's first missile wave arrives. Very few of the bombers which are not on alert can be expected to survive the enemy's first blow, unless they are well sheltered. Unalert and unsheltered bombers con tribute very little to deterrence of surprise attack by missiles, or to de fense in all-out war, although of course some of them must always be undergoing repair, and some on unalert status will be necessary for training. But beyond this, unalert and unprotected bombers would con tribute only to a first-strike capability, and their existence in the face of a missile threat can be justified only in terms of this objective. If the strategy for fighting an all-out war, in which the enemy strikes first, calls for some counterforce action, the bomber has certain advan tages over missiles for this mission. It is more accurate than the 21 From a speech by General Thomas S. Power, Commander-in-Chief, Strategic Air Command, printed in ibid., p. 277.
92
Deterrence and Defense
missile, under present and foreseeable technology, and therefore is more suitable for attacking small targets such as airbases and missilelaunching sites. In addition, its bombs can be much more powerful than missile warheads, thus making it the ideal weapon for destroying very hard installations. Moreover, because of its visual reconnaissance capa bility, the bomber may be able to attack mobile missiles, well-concealed installations, and targets whose location is not known with precision. Missiles, on the other hand, can efficiently attack only fixed targets of known location. A bomber can carry several bombs, and may therefore be capable of attacking several targets on one sortie. Depending on re fueling facilities, the bomber's capacity to re-run the air defense gaunt let on the return trip, and the availability of landing bases at home or on friendly territory, the bomber may be able to return and fly another mission. Finally, bombers would be useful for assessment of damage and for making sure of the destruction of highly important targets which might have been missed in an earlier missile barrage. There would be no need for such post-attack reconnaissance in a strictly punitive mission because presumably we know where the largest Russian cities are with consid erable precision, and could be fairly confident of hitting them with missiles. IRBM's based in Europe, and perhaps in the Far East and Alaska, would have more utility for counterforce purposes than ICBM's or Polaris submarine-based missiles because of their greater accuracy and perhaps heavier payloads—provided they survived the enemy's first strike. Since most fixed-base IRBM's probably would be lost in the enemy's first strike, however, only mobile land-based IRBM's should be counted as part of a strike-back capability. For the counterforce mission, the heavy liquid-fueled missiles such as the Atlas and Titan might have an important function which could not be performed effectively by the much less powerful solid-fueled Polaris and Minuteman missiles.22 The enemy may have hardened cer tain installations to such a degree that even a direct hit with the latter missiles could not destroy them. With their more powerful warheads, 22 Public estimates of the payload of Minuteman and Polaris usually are in the range of 100-500 kilotons. Warheads of the Atlas and Titan may yield several megatons. A kiloton is equivalent to the explosive power of 1,000 tons of TNT; a megaton, to one million tons. The bomb dropped on Hiroshima in World War II had a 20-kiloton yield.
with Strategic Nuclear Power
93
the Atlas and Titan might be able to do so. Their relative value in this role, as against manned bombers, would be sensitive to the estimated efficiency of the enemy's air defenses against bombers. If national policy is toinclude the objective of deterring Soviet attacks in Europe and Asia by the threat of a first strike, counterforce capabili ties are all-important. In relation to this objective, manned bombers, land-based IRBM's, and high-payload liquid-fueled missiles take on greater utility. The relatively high vulnerability of the fixed, land-based IRBM and the liquid-fueled ICBM is of no consequence, of course, when they strike first. The bomber does have an important disadvantage in a first strike, as compared with the missile: its lesser capacity to achieve surprise. As suming the enemy has a good warning system against bombers, most of the enemy's strategic forces would have been launched before the attacking bombers arrived. If missiles were used first, to gain maximum surprise, and to clear a path through enemy air defenses for a later wave of bombers, the enemy would have had unequivocal warning, and, again, the number of fruitful counterforce targets for the bombers might be small. This liability would be reduced if very high-speed bombers, such as the B-70, were developed. It may be, however, that the manned bomber's utility for a first strike will be limited to the "clean-up" mis sion—attacking delivery vehicles which the enemy had been unable to get off quickly, and installations such as airbases and submarine pens, the destruction of which would degrade the enemy's capacity for re using aircraft and submarines returning from the first retaliatory wave. The solid-fueled Minuteman is superior to the liquid-fueled Atlas and Titan for the deterrence of surprise attack not only because of its great er capability for protection and its cheapness and ease of handling, but, more importantly perhaps, because of its fast reaction time. Fast re action is important, first of all, because an attacking ICBM takes only about 30 minutes to arrive at its "address" and because warning of the attack may be even shorter—15 or 20 minutes. To have a capability to react in this space of time poses the possibility for the enemy that most of our retaliating missiles would get into the air before his arrived, in which case his counterforce attack might turn out to be futile. Even if the enemy doubted that we would react on warning alone (and it might be unwise for us to plan to do so), the possibility that we might would introduce an important element of uncertainty into his planning and the
94
Deterrence and Defense
possibility of much greater damage to himself than he would suffer if he could make a successful counterforce attack. By similar reasoning, a fast reaction capability places very stiff re quirements on the enemy's firing efficiency. If some of his missiles arrive only slightly prematurely, this might provide enough warning for us to get the bulk of our missile force launched before the arrival of the main body of the enemy's missiles. Quick reaction capability is valuable for carrying out a counterforce strike-back strategy; any delay in striking back at the enemy's forces increases the enemy's opportunities to do further damage to ours, and also may reduce the counterforce targets which remain available for us to hit.23 Finally, the shorter our reaction time on missiles, the less confident the enemy would be that we could be blackmailed out of using our residual missiles against his cities after the initial surprise attack. The faster our reaction, the less time there would be for the Soviets to take civil defense measures, and therefore the greater damage they would suffer in our retaliatory blow if we did retaliate immediately. These considerations obviously favor the deterrence of surprise attack. From the point of view of defense, we might not wish to react quickly if our best defensive strategy were thought to be a bargaining one; then we would want to delay to allow the communication of threats and offers. However, a quick reaction capability need not prevent such a delay. A fast reaction capability might increase the risk of accidental war, if combined with a policy of firing missiles on warning alone. This point will be taken up in more detail below. In summary, while the minimum objective of deterring direct attack on the United States perhaps would be served most efficiently by fixed and mobile solid-fueled ICBM's, missile-firing submarines, and perhaps some alert aircraft, the objective of waging all-out war most efficiently, and deterring lesser Soviet moves by the threat of a first strike, adds utility to fixed-base liquid-fueled ICBM's, fixed and mobile IRBM's, and manned bombers. It is in some ways desirable, however, to have something of almost everything, even if the objective is limited to "minimum deterrence." A mixed force greatly complicates the enemy's 23 Some delay may be desirable, however, to make sure that the attack upon us was not accidental, and to retarget our surviving forces in the light of whatever post-attack information we were able to get about which of our weapons had been eliminated and which of the enemy's missiles remained unfired, if any.
with Strategic Nuclear Power
95
counterforce attack problem. If, besides hitting fixed missile sites on land, he must also blanket entire areas to be sure of eliminating mobile missiles, keep a fleet of submarines under close surveillance, and cope with a fleet of manned bombers, some of which will be firing missiles from outside his air defense screen, his chances of avoiding terrible retribution from at least one of these sources are likely to seem very low. When he must diversify his counterforce attack to take care of various retaliatory threats, his uncertainty of success will be compound ed, and the cost to him of building a sufficient first-strike capability will be higher than when he is able to concentrate his counterforce eiforts on one or two types of defending capabilities. Also, a diversified force tends to complicate the enemy's air defense problem, and it insures against a major enemy technological breakthrough in the means of at tacking a single system. Of course, there are miscellaneous other forces available which would add something to the deterrence of surprise attack. These include airbreathing ICBM's such as the Snark, air-breathing shorter-range missiles such as the Matador, tactical aircraft in Europe, many of which have a nuclear capability, aircraft carriers with nuclear bomb-carrying aircraft of medium range, and missile-firing surface ships.
F. Civil Defense The most obvious function of civil defense measures is to save lives in an all-out war which the enemy starts. In the popular image, this is the only utility which civil defense measures can provide. The general public apathy toward civil defense can be explained in terms of this rather restricted image of its role in national strategy, combined with two expectations: that if thermonuclear war does occur, it will be so destructive that any prewar measures for protecting the population will turn out to be futile, and that it is very unlikely to occur anyway. We have already suggested that an all-out nuclear war is not likely to involve massive attacks on cities from the outset. The first blow is more likely to be directed at the defender's nuclear forces and civilian casualties from the blast effect of this blow would be limited to areas around nuclear force installations, which might of course include some large cities. Aside from this unavoidable by-product effect of the coun terforce first strike, the enemy might wish to avoid cities in order to maximize his chances of saving his own. If the Soviets do decide to start
96
Deterrence and Defense
the war, their primary objective will be to disarm the United States at the strategic nuclear level, not to kill as many people as they can. If this reasoning is correct, there may well be time, after the initial strike, for people in undamaged areas to get into shelters before fall-out reaches them, or before the war enters a city-busting phase. There may even be time to evacuate some portion of the urban population to safer rural areas. There is reason to believe, then, that millions of lives could be saved by a relatively cheap program of civil defense limited to construction and stocking of fall-out shelters, supplemented by plans for evacuating cities, and instruction in such things as how to get to shelters, how to decontaminate, the amount of time required for outside radiation to fall to a safe level, etc.24 Civil defense preparations also might contribute to the establishment of a bargaining position after the enemy's first blow. If we had evacu ated all the population of our largest cities to fall-out shelters, we would be able to threaten much more credibly that we would hit the enemy's cities if he failed to terminate the war on our terms. The enemy's ad vance recognition that we had this capability to put our urban popula tions quickly into a place of relative safety might help to deter his initial strike. However, it might have just the opposite effect—by sug gesting that we were seriously interested in limiting our own damage in war, civil defenses might indicate that we expected to retaliate with discrimination and caution, thus reducing the enemy's expectation of damage. The value of civil defense programs in simply reducing civilian casualties in an all-out war naturally carries operational urgency ac cording to how probable all-out attack is believed to be. In other words, the urgency of civil defense—and of all other capabilities, including offensive counterforce capabilities, which might be thought desirable for mitigating our costs in all-out war—depends on the presumed effec tiveness of our deterrence. If war is believed to be very, very unlikely, then even moderate measures of civil defense may be a waste of re sources. From this it follows that a choice must be made between civil defense and the accumulation and protection of a strike-back capability. If we are sure of having a strike-back force sufficient to deter attack, the case for civil defense, from the purely defensive point of view, must 24 See Herman Kahn, et al., "Report on a Study of Nonmilitary Defense," Report R-322-RC, The RAND Corporation, July 1, 1958.
with Strategic Nuclear Power
97
rest on the possibility that we may miscalculate the offensive require ments for deterrence, or that the Soviet leadership may act irrationally, or that war may break out by accident. But civil defense has another utility—its contribution to the deter rence of Soviet and satellite moves in Europe and Asia. The credibility of our threat to retaliate against such moves depends partly on the Russian image of the amount of damage we would suffer as a result of Soviet counterretaliation. Civil defense, in this role, is a complement to the offensive counterforce capability as well as to the active air defense of cities. An ability to move the bulk of our population into places of shelter in advance of our first strike would surely increase the chances, in Soviet calculations, that we would carry out such a strike under sufficient provocation. But, here again, the utility of civil defenses is interdependent with the amount of offensive capability we decide to have. If we should decide to abandon the whole idea of deterrence in Europe and Asia by the threat of massive retaliation and not try to have even the facade of a sufficient first-strike capability, then the utility of civil defense for this form of deterrence would be zero. In sum, the value of civil defense varies inversely with the presumed deterrent potency of our strike-back offensive capability, and varies directly with the size of our first-strike capability, so long as the latter is not so large as to provide by itself an absolutely effective deterrent.
G. The Conditions of Stability We may say that a "balance of terror" exists when each side has some what more than the minimum strike-back requirement—i.e., when neither side, in striking first, can destroy enough of the opponent's forces to make the latter's retaliation bearable. But the existence of a deterrent balance is something different from the stability of the bal ance. Stability in one sense refers to the degree of change in the military, technological, or political situation which is necessary to give one side a sufficient first-strike capability or sufficient incentive to strike first. For example, the balance would be unstable if either side required only a small additional expenditure of resources to achieve a first-strike capa bility which could reduce the opponent's retaliation to acceptable pro portions. It would also be unstable if some moderately conceivable technological or scientific breakthrough—say, in air defense or in de vices for tracking mobile missile launchers—suddenly gave one side a
98
Deterrence and Defense
first-strike capability. Or it might be unstable if the commitment of one side's honor and prestige in a policy declaration made it willing to accept significantly greater retaliatory damage, so that a first-strike capability which previously had been insufficient became sufficient to justify a first strike. A second form of instability would be a force relationship which produced strong fears on one or both sides that the other was about to strike first, thus creating an incentive to strike first to pre-empt the other's attack. Stability, at any given time, is a function of two factors: the extent to which either side's strike-back capability exceeds the necessary mini mum, and the "attacker-to-target ratio." The latter refers to the amount of attacking forces required to eliminate a given amount of the de fender's forces; it may also be called the degree of "structural stability." Both factors together determine the degree of effort required of the attacker to achieve a sufficient first-strike capability. Stability is highest when both sides have much more than the mini mum strike-back capability and when the attacker-to-target ratio is high. Then a very large increase in either side's forces would be neces sary to obtain a sufficient first-strike capability; only a very major and rather fantastic scientific discovery would overcome the deterrent bal ance; and no political commitment or threat would be likely to in crease either side's values at stake to a degree necessary to offset the costs of all-out war. And with such a configuration of forces, neither side would be likely to fear a first strike by the other, so there could be no incentive for pre-emptive attack. At the other extreme, the balance would be very unstable if both sides just barely had a minimum strike-back capability and if the attacker-to-target ratio were low, so that only a slight increase in one side's forces, or a minor technological breakthrough, or a small rise in political tension, would create a situation in which one side felt it could rationally strike first. Incentives to pre-empt would be high be cause of mutual fears that the opponent had, or would soon develop, a sufficient first-strike capability. Of course, there could be many intermediate degrees of stability between these extremes. For example, if one side could muster a wide margin of forces over its minimum deterrent, while the other side just barely met this criterion, stability would be fairly high if the stronger country were a status quo power with little incentive to strike first. But
with Strategic Nuclear Power
99
stability would depend on the continuance of benevolent intentions on the part of the stronger side. Even with a high degree of structural and over-all stability in the sense just discussed, there may be countervailing factors which make for instability. For example, even though a very major unilateral scien tific breakthrough would be required to give one side a first-strike capability, if such breakthroughs are possible or likely, the situation may be unstable. The sudden perfection of a good anti-missile defense system by one side would drastically shift the balance of terror in its favor, whatever the previous degree of structural and over-all stability. Or if one or both sides entertain a policy of firing missiles on warning alone, there may be serious dangers of accidental war. However, as will be discussed below, the danger of accidental war is closely related to the degree of stability on other counts; when neither side has any thing approaching a sufficient first-strike capability, there is little likeli hood of their being "trigger-happy." It is instructive to consider the degree of structural stability which would seem likely to exist with various types and combinations of weapons. If both sides were to rely entirely on fixed land-based missiles, rea sonably well-hardened and dispersed, the attacker-to-target ratio would depend largely on the accuracy of the attacking missiles and their lethal radius, the latter defined as distance from the point of impact within which the target would be destroyed.25 For example, if accuracy were such that there was a .50 chance of a single missile landing within the lethal radius, four missiles would have to be fired to achieve upward of a .90 chance of killing the target and the situation would be, struc turally, very stable. The attacker would need many more missiles than the defender to be sure of getting off with acceptable retaliatory dam age. But if there were a .50 chance of an attacking missile landing within a circle of only half the lethal radius, the attacker would have a .94 chance of destroying the target with a single missile. Then the balance of terror would be structurally unstable; one side might be willing to strike, for example, if its missile stocks only equaled the defender's and if it were willing to accept the retaliatory destruction which approxi mately 6 per cent of the defender's missiles could inflict. 25 Lethal radius is a function of the "hardness" of the target—the amount of blast pressure it can withstand—and the yield of the warhead. This relation be tween accuracy and lethal radius is discussed by Thornton Read in "Strategy for Active Defense," unpublished at this writing.
100
Deterrence and Defense
The degree of preponderance, if any, which the attacker needs, is sensitive not only to the attacker-to-target ratio, but also to the absolute number of missiles which the defender has. This follows from the fact that the damage which the attacker will accept is constant, while the number of defending missiles which escape the first strike, with given accuracy and lethal radius, varies with the total number of defending missiles. For example, suppose retaliation against cities by 25 missiles is just unacceptable damage for the attacker, the defender has 50 missiles, and the attacker has a .5 chance of killing any particular missile site with any one missile. Then the attacker needs only 50 missiles for a sufficient first-strike capability. If both sides are symmetrical in their accuracies and willingness to accept damage, each has a sufficient firststrike capability and the situation is unstable. But if one side increases to 100, the other must fire 200 (two at each of the defending sites) to be confident of retaliation by only 25 missiles. If the defender increases to 200, the attacker needs 600. If the defender increases to 300, the at tacker needs 1,200, and so on.26 The less accurate the attacker's missiles, the faster his first-strike requirement multiplies as the defender in creases his stocks. Thus, when the number of missiles on both sides is reasonably large, the requirement for a sufficient first-strike capability will be some multiple of the requirement for a minimum strike-back capability, and the multiple itself will be greater, the larger the number of the defender's missile sites.27 It is thus characteristic of the missile arms race, in seeming contrast to traditional arms races with conven tional weapons, that it becomes more stable as it proceeds. The incen tives to continue the arms race diminish rapidly as the numbers of 26 The mathematical reason for this "multiplier eifect" is that increasing numbers of attacking missiles are wasted in "overkills" as the defender's launching sites increase in number. When both have 50 missiles, the attacker reduces the re taliating force to 25 by firing one missile at each defending site. When the defender has 100, the attacker must reduce the probability of missing each target to .25, to have a statistical likelihood of reducing the number of retaliating missiles to 25. The attacker must fire two missiles per target: the probability that both will hit the target is .25, the probability that one or the other will hit is .50, and the probability of a miss by both is .25. When more than one missile is fired at a target, the probability that the target is missed is the product of the probabilities of a miss with each missile. When the defender has 200, the attacker must fire three per target to reduce the probability of a miss per target to .125 (.50 χ .50 χ .50). With 300 defending missiles, the probability of a miss must be reduced to .0625 or (.50)4, requiring four missiles per target. 27 This relationship is discussed by Thomas C. Schelling, "Surprise Attack and Disarmament," in Klaus Knorr (ed.), op.cit., p. 184.
with Strategic Nuclear Power
101
weapons increase oil both sides. The country which wishes to have only a minimum deterrent need have one that amounts to only a fraction of the opponent's striking force when the former's number of dispersed weapons is large relative to the number of weapons which would cause unacceptable damage to the opponent's cities.28 By similar reasoning, the country which contemplates a first strike can provide itself with the necessary capability only by resource expenditures very much larger than the expenditures the opponent must make to counter and re-estabIish the deterrent balance. Of course, improvements in accuracy, reliability, and payload of missiles would work against the stabilizing effect of increased numbers. On the other hand, increased hardening and dispersal would be stabi lizing.29 In general, it does not seem that a system of numerous hardened and well-dispersed sites at fixed bases would ever become seriously un stable in a structural sense (barring a substantial breakthrough in active defense against missiles), since it is hard to believe any aggressor would be so sure of the accuracy and reliability of his missiles and target intelligence as to be confident of eliminating each of the defend ing sites with a single missile. Two missiles per site would seem about the minimum for any aggressor who is not a pathological risk-taker. Even with this ratio, over-all instability could of course occur if the aggressive-minded side were allowed to gain a very large preponder ance in weapons. Statements of comparable precision regarding mobile systems, both land- and sea-based, are more difficult to make. A land-based mobile system may be highly invulnerable, and hence may contribute to over all stability, if methods for continuous aerial surveillance of its move ments are not devised. However, if space photography develops to the point where surveillance becomes possible, missile trains may become more vulnerable than fixed-base missiles. Since they are "soft" installa tions, lethal radius against them will be high; a moving train might not 28 A further conclusion might be that, as missile stocks increase, the "competi tion" between numbers of missiles and passive defense measures as means of preserving a minimum punitive deterrent comes more and more to favor the former, provided the accuracy and reliability of the enemy's missiles do not increase proportionately. 29 Hardening and dispersal are complementary measures. The greater the hardening, the lesser the degree of dispersal which is necessary to make sure a single attacking missile does not destroy more than one defending missile site. Thus hardening can produce administrative and logistical economies by allowing the closer clustering of missile sites.
102
Deterrence and Defense
be able to get out of range of the blast between the time an enemy space satellite transmits a "fix" and the enemy fires a missile with a highyield warhead against that fix. If missile trains carry several missiles (as in current U.S. plans) one successful attacking missile could elim inate more than one defending missile, in which case the mobile system might be considerably less stable than the fixed-base system. Also the generalization made above, about the balance of terror becoming more stable as it proceeds to higher levels of armament, might not hold if both sides relied heavily on rail-mobile missiles. At some point in his accumulation of missiles, an aggressor would reach the point where he could simply blanket the defender's railroad system with enough blast "overpressure" to destroy all trains. Missile-firing submarines are in a somewhat different category, since they are not subject to attack by missiles, but rather by other sub marines and other ships, as well as by aircraft. Here the crucial firststrike requirement is not for numbers of attacking weapons, but rather for the development of technical means of identification, tracking, and coordinated attacks. Since the enemy may develop such means in secret, sole reliance on Polaris submarines might create an unstable situation. Also, Polaris submarines do not benefit from the stabilizing effect of numbers, since each successful attack on a submarine would mean the loss of 16 missiles. With aircraft against aircraft, stability depends on the degree to which defending aircraft are protected and dispersed, the efficiency of the defender's warning systems, and the efficiency of the attacker's air defenses. If the defender's warning system is not reliable, and if his aircraft are unsheltered and only moderately dispersed, the situation may be very unstable, since one attacking bomber can destroy several bombers which are caught on the ground. But if the defender does have a very good warning system which cannot be penetrated by sneak attacks or evaded by "end runs," a minimum strike-back capa bility can be assured by an ability to get enough aircraft into the air, after receipt of warning, to penetrate the attacker's air defenses in sufficient strength to inflict unacceptable damage. The number of at tacking bombers is virtually irrelevant. With reliable warning, the bal ance is likely to be stable unless the attacker is confident that his air defenses can achieve a very high kill rate against the retaliating bombers.30 30
Of course, warning systems are vulnerable to destruction. The situation
with Strategic Nuclear Power
103
When the prospective attacker has large numbers of missiles and the defender relies entirely on aircraft, similar reasoning applies. If the defender does not have a good warning system against missiles, if his aircraft are unprotected and undispersed, and if the attacker has enough missiles to hit all the defender's airbases, the defender must have enough bombers on continuous airborne alert to inflict unacceptable damage after attrition by the attacker's air defenses. If the defender does have a good warning system against missiles, he needs the same number of aircraft on ground alert, but the degree of alertness must be much higher than when the attacker uses aircraft, because of the very short flight-time of missiles. When both sides have mixed forces, as we must expect in the fore seeable future, stability depends, of course, on some combination of the factors just discussed. Both structural stability and over-all stability are ultimately economic functions. Structural stability depends on the amount of economic re sources which an incipient attacker must invest to gain a capacity to eliminate a given amount of the defender's forces, compared with the amount of resources which the defender must invest to counter the attacker's increase. For example, if the cost of building a "hardened" missile site, requiring three enemy firings for a sure kill, were sub stantially less than the cost to the enemy of building three additional missiles, the balance would be fairly stable structurally. It is con ceivable that the cost of building the three attacking missiles would be less than the cost of the hardened missile, since missiles designed for attack do not have to be protected. Then the balance would be structurally unstable, since the defender would have to spend more than the attacker to offset the attacker's increase. Or if a 10 per cent increase in the attacker's air defense kill rate would cost him consider ably more than a 10 per cent increase in the defender's aircraft on air alert or high ground alert, this would make for stability. But over-all stability depends not only on structural stability but also on the total amount of expenditure which the attacker must make to gain a sufficient first-strike capability. This, of course, is a function of the margin of insurance which the defender has provided himself in building his strike-back force. could be unstable if the attacker were willing to gamble that destruction of the defender's warning system would not cause immediate retaliation.
104
Deterrence and Defense
H. Pre-emptive Attack We have left the question of pre-emptive attack for separate consid eration because it involves certain factors not pertaining to other forms of instability—principally its incentive, which does not stem from the prospect of gain, but from the desire to forestall losses from a first strike by the opponent which is believed to be imminent. In the usual analy sis of pre-emptive war, a spiral of mutual reinforcing fears and counterfears is postulated: A begins to fear that B intends to strike first, from which A derives some incentive to deliver a forestalling strike; B, realizing that A is subject to such fears, begins to fear A's forestalling attack, from which B develops some incentive to strike first even if, originally, it had entertained no such intention; A, recognizing that B fears a pre-emptive strike by A, has its fears of B's first strike rein forced, and feels even greater pressure to strike first; B, guessing that A feels this pressure, develops still greater fears and incentives to strike first,... and so on for as many regressive steps as either side might wish to contemplate. If such a spiral of expectations and incentives did occur, either side might be willing to accept much greater retaliatory damage after its own first strike than it would accept if the pre-emptive motiva tion did not exist. The following analysis will attempt to show that although pre emptive attack is at least conceivable when manned bombers are the mainstay of each side's strategic nuclear power, it is unlikely even then, and it becomes increasingly unlikely as both sides come to rely more and more on long-range ballistic missiles. If there were to be a serious danger of pre-emptive attack, it would have to follow from the existence of two conditions: (1) that there is a substantial if not decisive first-strike advantage—i.e., that if war is expected with certainty, it is much better to strike first than second; and (2) that one side believes the other side is very likely to strike first. The first condition is often taken as axiomatic. The advantage of get ting in the first blow in all-out war is said to follow from the benefits of surprise;31 from the advantage of being able to choose the most propi tious time of attack; and, above all, from the effect of the first blow in establishing an asymmetry of surviving forces favoring the attacker. The 31
Although surprise is usually mentioned among the advantages of striking first, surprise would be minimal under conditions conducive to pre-emption—i.e., when both sides have strong fears of impending attack by the opponent.
with Strategic Nuclear Power
105
defender must strike back with a disorganized and badly decimated force, in the face of air defenses and civil defenses which are alerted. Although the defender can inflict some damage in retaliation, this damage is as nothing compared with the damage which the "defender" could have inflicted if he had been allowed to strike first. The "advantage of striking first" which we are presently discussing is not the advantage which follows from having a sufficient first-strike capability. When a country bent on aggression has such a capability, it has an incentive to strike, simply because the victim's retaliatory blow would be tolerable—tolerable when measured against the expectation of gains from the attack. The "advantage of striking first," as it relates to the question of pre-emptive attack, means that even though a retalia tory blow would be unacceptable when measured against objectives other than pre-emption, it is nevertheless preferred to the costs of taking the opponent's first strike. It implies simply that if war is believed to be imminent, one's own losses are minimized by getting in the first blow. Whether there is an advantage in getting in the first blow depends on a combination of strategic and intrinsic considerations. In strategic terms, there is a strike-first advantage when the side striking first can destroy more of the opponent's strategic nuclear capability in doing so than it loses of its own. In intrinsic terms, a strike-first advantage exists when the destruction to one's population and economy which would be suffered in the opponent's first strike is greater than the destruction which the opponent could inflict in a retaliatory blow. There may be a strategic advantage in striking first when the bulk of each side's striking forces is made up of manned bombers, if one assumes that a considerable number of the defender's aircraft will be caught on the ground. One attacking airplane can eliminate several on the ground, since the defender is likely to have several planes at each base and since each attacking aircraft may be able to hit several bases in succession, carrying, as it may, a multiple bomb load. So the side which strikes first may be able to achieve a considerable preponderance of forces even though it may have struck from a prewar position of inferiority. How ever, even if such a strike-first advantage did exist in this sense, it would hardly trigger off a spiral of fears and counterfears leading to pre-emptive attack unless each side had a first-strike capability which could be construed by the other as being sufficient or nearly sufficient to make a first-strike rational for non-pre-emptive reasons. And even if this condition did exist, either side, rather than striking first because of
106
Deterrence and Defense
uncertain fears that the other intended to strike, would be more likely to take measures—such as putting more aircraft on a high alert status— which would reduce the other's first-strike capability. There may not be a first-strike advantage in an all-missile or pre dominantly missile environment if one makes four quite plausible as sumptions: that the targets of a first strike would be the opponent's missile bases; that missile sites are well dispersed and hardened, or mobile; that a given number of missiles aimed at missile bases would cause less damage to the population and economy than a smaller num ber of retaliating missiles aimed at cities; and that the primary targets of a retaliatory blow would be cities. Consider first a situation in which both sides rely entirely on fixedbase ICBM's, and missiles on both sides are equally accurate, reliable, and vulnerable. Suppose each side has 100 ICBM's. If accuracies, relia bilities, and vulnerabilities are all very high, we might imagine an ex treme case in which each protagonist would feel absolutely confident that a single attacking missile could eliminate one of the opponent's launching sites before its missile was fired. Then of course there would be an advantage in striking first, since, although the attrition of forces on both sides would be equal and complete, all the war damage would be concentrated on the side which received the first (and only) blow. But this is an extreme case which is hardly likely to be realized. The attacker probably will always have to lose more missiles in attacking than he is able to destroy of the defender's. Let us suppose, to give technology and the attacker's confidence the benefit of every doubt, that the missile-to-target ratio drops as low as 2:1. Then, with equal num bers of missiles on each side, it may not be advantageous to strike first. The attacker would lose two missiles for every one of the defender's which is destroyed. Firing off his entire stock of 100 missiles, the at tacker would have to expect retaliation by at least 50 missiles against his cities. Rather than accept this result, the "attacker" might prefer to let the other side strike first against his missile sites. For either side, a first-strike advantage would not exist if the damage from 50 retaliatory missiles on its cities were expected to be greater than the by-product damage to population which would be expected from the enemy's first strike with 100 missiles directed at missile sites, not cities. Or, following another line of reasoning, each side might well believe that it would be best to let the other use up all its striking power, leaving the attacked side with a residual force which could be used to compel the attacker to
with Strategic Nuclear Power
107
capitulate. Thus, even with the low missile-to-target ratio of 2:1, there may be disadvantages in striking first. Obviously, with higher missile-totarget ratios, the supposed "strike-first advantage" becomes even more doubtful. It is sometimes overlooked, in discussions of pre-emptive attack, that the act of pre-emption tends to convert the other side's target system from primarily counterforce to primarily countercity. If pre-emption cannot virtually eliminate the opponent's striking forces, this conversion may result in the pre-emptor's suffering more damage than he would have absorbed as the result of the opponent's first strike. Although pre emption then would reduce the number of missiles landing on one's own territory, the ones which did arrive would hit where they would do maximum damage. The primary object of getting in the first blow is to reduce the damage to oneself, as compared with what it would be if the opponent were allowed to strike first. But if it takes two or more missiles to elimi nate one, missiles may be more useful for reducing the enemy's damageproducing capacity when they function as targets than when they are used as instruments of attack. With a missile-to-target ratio of 2:1 for both sides, two missiles destroy one in attacking, but "destroy" four as targets. The strategic advantage in not striking first varies as the square of the missile-to-target ratio, when the latter is greater than 1:1, and roughly equal for both sides. The strategic disadvantage may be enough in itself to discourage pre-emption if an important objective is to gain the best possible position in the post-strike balance of forces for bar gaining purposes. But there may also be an intrinsic disadvantage in striking first if the gain from reducing the number of missiles impacting on one's own territory is overwhelmed by the extra costs incurred in making cities rather than forces the enemy's primary targets. Let us suppose, however, that each side prefers to strike first rather than accept a first strike by the other, possibly because neither can be sure that the other would avoid cities on a first strike. Continuing to assume a 2:1 missile-to-target ratio and equal forces on each side, the strength of the incentive for either side to launch a pre-emptive first strike would then depend on its estimate of the likelihood that the other intended to strike, as well as on the amount by which the damage suf fered from the other side's first strike was expected to exceed the dam age from the other side's retaliatory blow after one's own first strike. In other words, neither side would contemplate pre-emption unless it
108
Deterrence and Defense
believed that the probability of a first strike by the other ( p ) , times the population-economic costs which this strike would cause (c), was greater than the retaliatory damage (d) which would follow its own pre-emptive strike. Thus the minimum condition for "incentive to pre empt" can be expressed by the formula: pc>d. It must be emphasized, however, that this is a minimum condition. Even if pc were very much larger than d, the estimated probability of the other side's first strike probably would have to approach unity to create a very strong incentive to pre-empt. It is very hard to believe that any country would deliberately accept the certainty of severe retalia tory damage in preference to the uncertain prospect of being the re cipient of a first strike. As long as there existed any significant chance of avoiding war altogether, inaction would be preferred to striking first. It is possible, however, that one side—let us say, the Soviet Union— might have other than pre-emptive motivations for striking first, so that the addition of even uncertain fears of a first strike by the United States to the motives already existing might be enough to trigger a Soviet first strike. The greater the Soviets' non-pre-emptive incentives to strike, the less certain they would have to be of a U.S. intent to strike first in order to be willing to strike first themselves. But a pre-emptive motivation is hardly likely to arise unless both sides have a very substantial first-strike counterforce capability. Neither side can develop an incentive to pre-empt unless the other side appears to have enough strength to be willing to strike first for non-pre-emptive reasons. And if one side does begin to fear attack by the other, it cannot consider pre-emption unless it also has a large first-strike counterforce capability. In an all-missile environment, it would be possible for both sides simultaneously to have a sufficient first-strike capability only when the total number of missiles on each side was relatively low—relative, that is, to the number of retaliating missiles which each would tolerate as a price for the non-pre-emptive gains which it could expect from striking first—and when the missile-to-target ratio was also low. This condition would exist in the hypothetical case just presented: when each side has 100 missiles, when the missile-to-target ratio is 2:1, and when each is willing to accept retaliatory damage by 50 missiles. Then either might begin to fear an attack and might decide to pre-empt. But again, for pre-emption to seem rational, the pre-emptor would have to prefer 50 missiles on his cities to 100 missiles aimed at his missile-
with Strategic Nuclear Power
109
launching sites. If each side had 500 missiles, neither would have any fear of surprise attack (because each would know the other would not willingly accept a retaliatory blow by 250 or more missiles), so there would be no incentive to pre-empt. We must conclude that when both sides' strategic arsenals are com posed primarily of missiles, the danger of pre-emptive war is very small—first, because there may be no advantage in striking first even if one is sure the other intends to strike; and, secondly, because of the unlikelihood that both sides simultaneously will have a capability approaching sufficient first-strike proportions. Pre-emptive war would be more likely if both sides relied chiefly on manned bombers, but even in this case the danger would be rather small because neither side would be likely to trade the certainty of retaliation for the uncertainty of a first strike by the other, even if both sides had a substantial first-strike capability. We have assumed equal striking forces on each side. When the forces are unequal, there might still be no advantage for either side in striking first in a predominantly missile environment, so long as a first strike by the other side would be less costly than its retaliatory strike. Of course, the greater the preponderance achieved by one side, the less retaliatory damage it would have to expect, and the more likeli hood that it would see an advantage in striking first. On the other hand, the greater its preponderance, the less need for the stronger side to fear a first strike by the other; hence if it does strike first, it is not likely to be for pre-emptive reasons. If the foregoing argument is correct, some important consequences follow. In particular, it would not be dangerous for the United States to attempt to have at least a semblance of a first-strike counterforce capability to deter Soviet aggression in Europe. It is sometimes argued that such a policy would incur a serious risk that the Soviets would be tempted to strike first in a pre-emptive attack. But, as we have seen, in a predominantly missile environment, the Soviets would be so tempted only if they foresaw a virtual certainty that the United States would strike them first, and perhaps not even then. A probability of far less than certainty that the United States would strike after provocation should be sufficient to deter Soviet attack in Europe, and it probably will be impossible, anyway, for the United States to obtain a first-strike capability that would induce a Soviet expectation of near-certainty that the United States would strike first. Moreover, even the moderate
110
Deterrence and Defense
risk of a U.S. first strike which might arise in case of Soviet provocation in Europe could be avoided by the Soviets simply by refraining from provocative behavior. Similarly, the argument is sometimes heard that unprotected missile sites in the United States, or even "hard" sites in Europe controlled by the United States or NATO, might provoke a pre-emptive war because, since they would be knocked out in a Soviet first strike, they imply an intention to strike first. But if we do not have enough weapons to induce a near-certainty in Soviet calculations that we do intend to strike first, the danger of Soviet pre-emptive attack would not arise. Yet it may be desirable to have some of these soft or short-range missiles (even quite a lot of them) in order to present the Soviets with at least some risk that we would strike first if sufficiently provoked. In extension of the argument just mentioned, it is sometimes said that it would be in the interest of the United States for the Soviet Union to have nothing but highly invulnerable weapons. Then the Soviets would reason that we would have little fear of a first strike by them because their weapons were obviously intended to strike back rather than strike first; hence they would have no reason to fear a U.S. pre emptive attack, and no incentive to pre-empt themselves. This argu ment is based on an exaggerated image of the danger of pre-emptive war, and it ignores the fact that it cannot be in the U.S. interest, either in fighting an all-out war which the Russians start, or in deterring a Soviet move into Western Europe, for the Soviets to make their weap ons invulnerable to our counterforce fire. Perhaps the most plausible rationale for a Soviet pre-emptive attack would not be to forestall an all-out U.S. first strike which was considered imminent, but to forestall the gradual escalation of a limited war. It is entirely possible that a limited nuclear war might threaten to expand, in geographic limits and intensity, to a level more destructive to the Soviet Union and its allies than a counterforce all-out first strike by the United States. In such a situation, if the Soviets were deeply committed to continuing the war, a first strike at the United States might seem the best alternative to them.
I. Accidental War There can be no doubt that there is some danger of accidental all-out war, but the danger should not be exaggerated. There are two aspects
with Strategic Nuclear Power
111
to the problem: the chance that accidents may occur, and the chance that if an accident occurs, it will set off all-out war. The second is by far the more important of the two. There would seem to be two general kinds of accidents worth considering: the inadvertent or irresponsible firing of a missile or dropping of a bomb (or a salvo of missiles and bombs), and the transmission of a false warning that an attack is im pending or under way. Suppose a single unexplained nuclear explosion took place in the United States. Would the United States immediately retaliate all-out? It seems hardly conceivable, since the Soviets would not be likely to send their missiles or bombers in seriatim. They would attack massively or not at all. Of course, the single explosion might have been a pre mature firing of one missile which was intended to be included in a massive attack. But if this were the case, it would soon be confirmed by blips on the early warning radars indicating that additional missiles were on their way. If the single explosion were in fact an accident, it is extremely unlikely that the warning system would provide such con firming evidence. Of course, a whole salvo—say, eight or ten missiles— might be fired by accident. In this case also, it is unlikely (although perhaps more likely than in the case of a single explosion) that the United States would retaliate unless the warning system provided con firmation that a full-scale attack was in progress. Presumably, the same reasoning would apply to the accidental firing of a U.S. missile or salvo against the Soviet Union. The danger that the accidental firing of one or a few missiles will touch off the ultimate holocaust depends crucially on the degree of stability in the balance of terror—i.e., the degree to which each side's forces approach a sufficient first-strike capability. For example, if the Soviet Union did not have a large first-strike capability, the United States would be predisposed to believe that an unexplained explosion on its territory was an accident, or that a questionable signal on its radar screens was a false alarm. And if the U.S. first-strike force did not approach sufficiency, we would have to have very strong confirming evidence that an attack was under way before we would launch our own strike. Short of such evidence, the alternative of waiting and hoping that the event or signal was accidental would very likely seem preferable to deliberately (and perhaps unnecessarily) accepting the enormous costs of all-out war. Thus the danger of accidental war in the mature missile age is lessened by the same considerations that seem
112
Deterrence and Defense
to make pre-emptive war unlikely—notably, the difficulties of acquiring substantial first-strike capabilities. What of the danger of the sudden derangement of someone who has the authority to "press the button"? If the person is relatively low in the hierarchy—say, the commander of a missile squadron—he probably could not create a sufficiently large "accident" to trigger all-out war, for reasons just mentioned. The "crack-up" of a leading decision-maker, such as the President or the commander of SAC, would be an entirely different matter. One can conceive of a situation—when, for example, the stress of an international political crisis has been suddenly aug mented by a report of mysterious "blips" on the early warning screen— that might cause a President to lose his self-control and give a prema ture order to fire. It is also possible that a commander of SAC might become so seriously committed to the urgency of striking first that, upon learning from NORAD that some objects which might be enemy missiles were in the air, he might give the order to retaliate for fear the Presi dent would delay too long. We can only hope that the men in these responsible positions would be extremely well balanced, and that at the moment of decision they would be in the presence of other men who could prevent them or discourage them from taking precipitous action.32 The danger of accidental war from false warning, in the missile con text, is closely related to the degree of vulnerability of the missile sites. The greater the vulnerability, the greater the pressure to fire the missiles upon the first receipt of warning. In view of the extremely brief period between initial warning and impact (from 15 to 30 minutes), there will hardly be time to wait for confirmation of an early ambiguous warning; there will be extreme pressure on decision-makers to fire away in order to maximize damage to the enemy. On the other hand, with highly in vulnerable missile forces, there will be greater willingness to ride out the attack in the knowledge or hope that a large portion of the retalia tory missiles would survive. The pressure to fire on ambiguous warning would also be lessened to the extent that we were sure we had a comfortable margin beyond a minimum strike-back force. Then any warning which might be an indi32 Various systems are conceivable which would require the concerted action of two or more men to fire a missile, or to flash a firing order to field missile crews. For example, firing might require that two or three buttons on different sides of a room be pushed simultaneously.
with Strategic Nuclear Power
113
cation of an incoming missile strike would be viewed askance because of the pre-existing belief that an enemy first strike was highly unlikely. If the strike-back force were just over the "minimum," on the other hand, the probability that the warning was not false would increase, and hence the pressure to fire on the warning alone would be greater. This is another reason why it would be dangerous to rely on a deterrent ca pability which just barely met the minimum requirement. Even if we were confident of having a large enough surviving force to cause unacceptable damage to the Russians in retaliation, this would not completely eliminate the psychological pressure to fire before impact of the enemy's missiles. If the war were thought to be under way, pre war calculations of "minimum unacceptable damage" would be irrele vant because such calculations bear only on deterrence. After the war started, we might be interested in doing as much damage as possible to the enemy in order to achieve the best postwar power position vis-a-vis the enemy. The greater his degree of destruction, the less likely that he would be able to take over the world before we recovered suffi ciently to prevent it. Thus some pressure to strike on warning alone will always exist so long as we are interested in maximizing the enemy's damage. It might be, of course, that we would reject this aim in favor of a bargaining strategy; that we would prefer to accept the enemy's first blow, temporarily hold back our surviving forces, and then threaten a retaliatory blow against the enemy's cities, to persuade him to terminate the war. Of course, the risk of accidental war because of false warning will be small if there is little chance that the Ballistic Missile Early Warning System (BMEWS), or the companion Missile Defense Alarm System (MIDAS), will generate a false warning; if, in other words, these sys tems can distinguish with high confidence between a salvo of missiles and such natural phenomena as meteorite showers. There is a difference of opinion among those who are familiar with these systems as to their degree of precision in identification, and governmental policy does not seem to be settled on the related question whether missiles should be fired on warning alone. General Thomas S. Power, Commander-in-Chief of the Strategic Air Command, stated in February 1960 that the BMEWS radar warning system might "see things that are not there" and that therefore retaliatory missiles would have to ride out an attack.33 33 Preparedness Investigating Subcommittee, Senate Armed Services Commit tee, "Missiles, Space and Other Defense Matters," op.cit., p. 51.
114
Deterrence and Defense
There are many other indications, however, from Air Force and other sources, that BMEWS is expected to be able to distinguish a missile attack from other signals, and that missiles may be fired solely on warn ing of such an attack.34 Apparently the MIDAS system—an orbiting satellite which picks up signals from the infra-red rays of the exhaust of a missile which has just been fired—is expected to provide very reliable identification when it becomes operational in the late 1960's.35 However, even an absolutely reliable warning system would not elim inate the need to "harden" or "mobilize" missiles, since the order to retaliate might not be received in time to get the missiles off before impact of the enemy's strike—either because of mechanical failure of the communication system or because of delay at one of the human links of the system. For example, it is entirely possible that the President—the only person who can authorize the firing of ballistic missiles under pres ent procedures—simply would not be available to receive the warning when it was flashed by the Air Defense Command. Then, too, there is the possibility that the enemy would find a way to "beat" the warning system, perhaps by jamming it, or even by destroying portions of it. Of course, attacks on the warning system would be highly provocative, but the enemy might feel, if our offensive capabilities fell far short of first-strike "sufficiency," that we would be loath to carry out retaliation for anything less momentous than a full-scale attack on the United States that was known to be under way. If there is any possibility of either BMEWS or MIDAS generating false warning, it would seem essential, in order to avoid the danger of accidental war, to adopt a fixed policy that missiles will not be fired until after the impact of the enemy's missiles. Such a policy would have its costs—chiefly in greater need for protection of missile sites and in deliberately accepting the risk of some reduction in retaliatory power should the warning turn out to be justified. The alternative policy—to fire immediately upon receiving warning—raises serious risks of me chanical or human error. 31 For example, see testimony by General Thomas D. White, Chief of Staff of the Air Force, in ibid., p. 127. After the above passage was written, however, the Kennedy administration an nounced, in the spring of 1961, that it had decided that missiles would not be fired until after impact of the enemy's first blow. 35 Testimony of Lt. Gen. Bernard A. Schriever, ibid., p. 69.
with Strategic Nuclear Power
115
This does not mean that warning is totally useless as a means of pro tecting a missile strike-back capability. As long as there is a possibility that the attacking missiles will not arrive simultaneously—a very plaus ible possibility—warning can be used to take certain last-minute readi ness measures, such as starting the count-down on liquid-fueled mis siles. This would maximize the number of missiles which could be launched immediately after impact of the first attacking missiles. Mobile missiles could be placed in motion, Polaris submarines in port could put to sea, and the decision-making apparatus could be placed in readi ness. In general, warning time should be used not to make a decision to retaliate, but to take all possible alert measures, from missile and aircraft crews all the way up to the office of the President, to minimize destruction from the first attacking salvo, and to facilitate the fastest possible reaction once the first salvo has arrived.
J. Summary and Conclusions The central issue in choosing the proper level of strategic nuclear capabilities is whether we can be satisfied with a minimum strike-back punitive force or whether more than this is desirable for limiting dam age and influencing the military outcome of all-out war and for deterring Soviet tactical moves. If we do need more than a minimum deterrent, a second question is: how much more? Let us assume, for the sake of discussion, that the strategic environ ment in years to come will be quite stable. Just to make it concrete, assume that a minimum punitive deterrent force can be maintained with striking forces roughly one-half the size of the attacker's. Suppose this minimum deterrent force is large enough to hedge against a wide range of uncertainties—including the possibility of a major technological breakthrough, of miscalculation of the relation of forces by either side, of miscalculation by us of the enemy's level of unacceptable damage, of moderate degrees of irrationality in the enemy's leadership, and so on. Even thus well-hedged, there would be a considerable gap between the size of this force and the required size of a first-strike counterforce capability which would be adequate for deterrence of enemy moves short of all-out attack. This latter force would probably have to be larger than the enemy's forces. The forces necessary for limiting dam age, bargaining, and generally fighting the war most efficiently should
116
Deterrence and Defense
the enemy start it, would be somewhere in between the minimum deter rent force and first-strike capability. Would a minimum deterrent force be enough? To argue that it would be involves assuming that the enemy will believe we will or might act irrationally. For us to retaliate with an indiscriminate all-out blow at the enemy's cities after we had absorbed a Russian first strike might be wholly irrational. We would dissipate our remaining bargaining cap ability in so doing, and we would contribute nothing to limiting our damage—in fact, we might leave ourselves wide open to additional damage if the Russians had retained some forces. Thus, if the Russians knew we were rational, a minimum punitive strike-back force would be a very ineffective deterrent. After striking first at our forces, the Russians could be sure that we would be deterred, or could be deterred by postattack blackmail, from striking back with full force at Soviet cities. However, the minimum deterrent thesis holds: the Russians cannot be sure of our rationality; therefore, they face a risk that we will retaliate all-out, and they cannot afford to take this risk. The argument is a strong one, especially when the minimum deterrent force is not a niggardly one, when it threatens enough damage to remove risks of miscalculation and to discourage any Russian propensity to gamble that we might behave rationally and not retaliate all-out. Some marginal risks would remain uncovered; notably, those of accidental war and the rise to power in Russia of a very irrational leader of the Hitler stripe. However, as we have argued earlier, the danger of acci dental war is likely to be small in a stable technological environment. We might reasonably hope that if a Russian dictator should develop the extreme sort of irrationality that would be necessary to override the high risk posed by our well-hedged minimum deterrent, he would be restrained by his colleagues. It is often overlooked that a minimum deterrent force probably would itself provide some capacity for limiting damage and bargaining should deterrence fail. Such a force does not have to be fired reflexively and only against enemy cities. Although we discussed earlier several quali tative differences between typically counterforce and typically countercity capabilities, these differences are likely to narrow with further technological development—for example, with improvements in missile accuracy. Thus a strong minimum deterrent may provide enough coun terforce capability to do almost all that is feasible in the way of mitigat-
with Strategic Nuclear Power
117
ing our own damage by striking at residual Soviet strategic forces. Such a capability also would provide a considerable post-attack bargaining potential; just how much would depend, among other things, on how much of his striking force the enemy expended on his first blow. If he fired everything he was able to fire in his first salvo, our bargaining strength very likely would be superior to his; if he held enough back to have superiority in post-attack bargaining, at the price of forgoing some attrition of SAC, he would be risking very severe damage to his cities, should we opt for a strictly punitive and all-out retaliatory blow. Thus a well-padded, high-confidence minimum deterrent force would go far toward meeting moderate "defense" objectives. If our "defense" objectives were more ambitious—e.g., if we wanted to be sure of having enough superiority in the post-attack bargaining to enforce a demand for restoration of the status quo ante in the Eastern Hemisphere, or to strike heavily enough at the Soviets' tactical forces to make it impossible for them to move, or (even more ambitiously) to "win the war" or to destroy completely the Soviet Union's economy and paralyze its government and society—we would wish to have forces in addition to even a "high minimum" deterrent capability. But whether we would want to embrace these larger defense objectives and spend the extra resources in peacetime which would be necessary to realize them, would depend on how likely a Soviet first strike was thought to be. The value of all preparedness measures intended to reduce our losses or to make positive gains in all-out war tends to decline as the probabil ity of war declines. If we had the high-confidence deterrent we have been discussing, substantially insuring against all possible Soviet incen tives for a first strike, the chances of war would be very low. In view of this, it seems doubtful that the defense benefits from having forces well beyond those necessary for high-confidence deterrence would justify their peacetime costs. However, there is a purely psychological argument for having more than a minimum deterrent force. We have been assuming that such a force, even one large enough to cover all reasonable uncertainties and risks, would still be smaller than the enemy's forces in a situation of moderate to high technological stability. But a quantitatively inferior force carries a serious psychological liability. No matter how many times it is pointed out, with impeccable logic, that deterrence does not
118
Deterrence and Defense
require matching the Russians in quantities of weapons, that it requires only the capability to inflict unacceptable costs, national self-respect and prestige are bound to suffer from a policy which deliberately ac cepts a permanent inferiority in nuclear striking power. This is partly the legacy of centuries of experience in which national security, power, and prestige did depend, more or less, on matching or exceeding the enemy in either mobilized power or war potential. The greater the pub lic recognition that war potential is no longer significant in all-out war, the more difficult will it be for the public or even the knowledgeable leaders of a country to feel secure with only a fraction of the enemy's mobilized power. Apart from historical experience, there is something inherently unnatural about accepting a gross inferiority in any category of power—particularly, for Americans, in the field of airpower and nuclear technology, in which they are accustomed to casting their own country in a role of world leadership. National pride and prestige are of course legitimate values; it is en tirely logical to be willing to sacrifice some resources and comforts in their service alone. More importantly perhaps, doubts and insecurities (no matter how irrational) arising from a position of nuclear inferiority may subtly affect the nation's diplomatic bargaining positions. Our negotiators may be less firm in standing up to Soviet demands when we have only a minimum deterrent than when we have striking forces at least equal in number to those of the Russians. And the Soviets, for their part, have already given evidence of the increased diplomatic confidence which they feel as a result of having grabbed the lead in missilry. In short, the symbolic or psychological effect of "being ahead" or "being behind" is likely to affect political relations with the enemy, and prob ably with third countries as well, even though the asymmetry in forces is not enough to give the enemy a sufficient first-strike capability or any thing approaching it. There remains the question whether we should go farther and at tempt to have a first-strike counterforce capability for deterrence of enemy attacks in Europe and Asia. For effective deterrence, this capa bility would not have to meet the exacting requirement of "sufficiency" —i.e., the requirement of being able to limit the enemy's retaliatory blow to acceptable dimensions should he retaliate indiscriminately against cities. However, it probably would require exceeding the enemy's strategic forces. An attempt to have such a capability might touch off a spiraling arms race, and might possibly create some danger
with Strategic Nuclear Power
119
of pre-emptive attack. Aside from these risks, the desirability of having a force of this size would depend on its cost and presumed deterrent utility, as compared with the cost and effectiveness of alternative means to deterrence and defense in Europe and Asia. The next two chapters will consider these alternatives.
3. Deterrence and Defense in Western Europe A SURPRISE thermonuclear attack against the United States itself probably is the least likely contingency we have to face, although constant vigilance will be necessary to keep it unlikely. Much more probable are lesser aggressions around the periphery of the Soviet empire, ranging from a large-scale Soviet ground attack in the NATO area to minor satellite "nibbles" and "brush-fires" in Asia. Deterrence and defense against these challenges pose a quite different set of prob lems than those raised by the possibility of direct surprise attack. The reasons are fairly obvious: surface forces, especially ground forces, have a significant role to play, and the credibility of strategic nuclear retaliation is much lower than it is as a response to direct attack on the United States. The issue faced in this chapter is how to determine what combination of strategic and tactical forces is needed to obtain an optimum combination of deterrence and defense against Communist provocations short of all-out surprise attack, particularly in Western Europe. We shall begin by taking a critical look at recent NATO doc trine on deterrence and defense.
A. The Sword and the Shield in NATO Strategy The military strategy and capabilities of NATO are habitually de scribed by official spokesmen in the imagery of the "sword" and the "shield." The "sword" is composed of the U.S. Strategic Air Command, the Bomber Command of the British Royal Air Force, and other naval and air forces which are able to contribute to a long-range bombing offensive on the territory of the Soviet Union. The "shield" is the for ward line of ground forces and tactical air forces guarding the "central front" in Germany.1 According to the evidence available at the present 1 NATO, of course, has other ground and tactical forces at its disposal besides those in the shield. There are active forces on the northern front in Norway and Denmark and on the southern front (Greece, Turkey, and Italy). In addition, NATO countries have some active forces not deployed on the forward line which could be brought up to support the shield, as well as reserves which are mobilizable in varying periods of time.
in Western Europe
121
writing, the shield is composed nominally of 21 1/3 divisions, of which five are American and three British. However, apparently because of the dispersal, undermanning, and underequipment of some of the divi sions, the total force is the equivalent of only 16 or 18 effective, combatready divisions.2 Facing the shield, in East Germany, are about 20 modern Soviet divisions as well as six or seven divisions of East German "security forces." Back of this Soviet spearhead are perhaps about 120 more combat-ready Soviet divisions in other Eastern European coun tries and the Soviet Union itself, and about 40 or 45 satellite divisions outside of East Germany. In NATO's strategic doctrine, the sword and the shield complement each other in both deterrence and defense. The primary instrument of deterrence, of course, is the retaliatory force. It is also said to be the chief means of defense against major aggression. The line between major and minor aggression has never been defined precisely, but pre sumably any attack with ground forces sufficient to defeat the NATO ground forces would be "major." An attack of such magnitude, accord ing to the official policy, would be met with massive nuclear retaliation against the Soviet Union, the assumption apparently being that such all-out retaliation, with some help from the shield and other ground and tactical elements, would attain "victory." But the most interesting aspects of NATO strategic doctrine are the functions claimed for the ground force "shield." As described by Gen eral Lauris Norstad, Supreme Allied Commander in Europe, these func tions are three: (a) to hold or delay the advance of the Red Army until the "total weight of the retaliatory power can be brought to bear"; (b) to deter minor incidents and probing actions and also to help deter a major ground attack by making it clear to the enemy that "he would have to use substantial force to breach the shield, an act he knows would bring down upon him the full weight of the deterrent, including our heavy strategic power"; and (c) to provide a degree of flexibility which removes the need of having to choose between total war and acquiescence in enemy aggression, and affords a means of dealing with "less than ultimate" incidents with "less than ultimate" means.3 The first-named function would be the defense function of the shield 2 Roger Hilsman, "On NATO Strategy," in Arnold Wolfers (ed.), Alliance Policy in the Cold War, Baltimore: Johns Hopkins Press, 1959, p. 152. 3 Senate Committee on Foreign Relations, "Mutual Security Act of 1958," Hearings, 85th Congress, 2nd Session, p. 187.
122
Deterrence and Defense
in total war. It is the most dubious of the three. The official assumption is that the strategic nuclear exchange would be brief and decisive.4 Just how it is to be decisive has never been clearly explained—whether by eliminating the will of the Soviets to continue fighting through ex tensive destruction of their cities, or by destroying the sources of supply of the Red Army, or by establishing a bargaining superiority through counterforce attacks on Soviet nuclear installations which would induce the Soviets to call it quits. Perhaps the reasoning is a combination of all three. But what would the shield contribute to this end? The phrase quoted—that time is necessary to allow "retaliatory power to be brought to bear"—suggests several things. One is that U.S. "massive retaliation" would not or could not take place in a single blow. This may be plausible when the retaliatory strike must be made by aircraft, not all of which will have been in high alert status. It is much less plausible if the re taliation is to be carried out with missiles. Perhaps the reference to "time" means the time necessary to make the decision to retaliate. Further statements by General Norstad and others indicate that time is required to allow the retaliation to "take effect." This seems to con note something rather different: that time is necessary to allow for sufficient attrition of the Red Army's rear-area sources of supply to bring its advance to a halt, or perhaps to allow the hopelessness of its situation to be fully appreciated by the Soviet leadership and make it willing to capitulate. But, granted that these effects will take time, what is the point of delaying the advance of the Red Army during that time if the strategic exchange is expected to be decisive in the end? What does it matter whether Soviet ground forces are in control of one-third, one-half, or all of Europe by the time the Soviet government capitulates or its war effort collapses? Would the Soviet Army try to maintain an autonomous occu pation after its central command had been destroyed or had capitu lated? It hardly seems likely. It might be argued that holding the Red 4The amount of time required is variously estimated. Roger Hilsman states: "It is assumed that the Strategic Air Command would require no more than 30 days in which to do its work, and for this period of time it is believed that a 30-division force, fully supplied with atomic weapons and battlefield missile sys tems, could hold very near the western border." See his "NATO: The Developing Strategic Context," in Klaus Knorr (ed.), NATO and American Security, op.cit., p. 29. F. O. Miksche, on the other hand, says that NATO's official policy is based on the assumption of a six- to ten-day war. F. O. Miksche, The Failure of Atomic Strategy, New York: Frederick A. Praeger, 1958, p. 86.
in Western Europe
123
Army as far east as possible at least would minimize war damage in the countries to the west. However, it seems unlikely that France could avoid severe damage even if the ground combat were limited to West Germany, since the enemy surely would attack NATO's supply and logistics network in France. Nor would Britain and Italy be immune. Such attacks would almost surely be nuclear, since it is very hard to con ceive of the Russians exercising restraint in Western Europe when they were engaged in a thermonuclear struggle with the United States. If the NATO ground force shield has any significant value for fight ing an all-out nuclear war, it must be on the assumption that the stra tegic nuclear battle may not be decisive in itself. Conceivably, the shield might contribute something to victory if our war strategy were one of bargaining—a counterforce first strike, followed by an ultimatum and a threat to attack cities unless the terms of the ultimatum were met. The less territory the ultimatum required the Soviets to give up (especially if it called only for a return to the status quo ante), the more willing they might be to heed it. In other words, the more the Soviets' ground advance is delayed, the more likely that such a bargain ing strategy would succeed. Or, if we were to contemplate an attrition strategy rather than a bargaining strategy—i.e., try to stop the Red Army physically or destroy the enemy's ability to continue the war—a combination of ground force action and strategic nuclear action might succeed when neither, taken by itself, could do so. If there is some chance that the strategic nuclear exchange might not be decisive, then we should not write off the possibility of a "brokenbacked war"—a fairly prolonged struggle of ground armies for posses sion of territory after an all-out nuclear exchange. To win such a war would require ground forces in Europe larger, of course, than those required for a mere holding action. If one takes a pessimistic view of the kind of world that would exist after an all-out nuclear exchange, then it does seem somewhat unrealistic to imagine that armies would still wish to struggle for the charred embers of Europe when manpower was so desperately needed at home. Taking a somewhat more optimistic view, however, Europe might still seem worth fighting for. It is at least conceivable that the two superpowers would exhaust their stra tegic striking power without achieving a decision or that each would hold back for bargaining purposes a certain quantity of highly invul nerable strategic forces. Then the strategic battle would stop (except possibly for limited strikes to underline bargaining demands) and the
124
Deterrence and Defense
ground war in Europe would continue. Even after a tremendously de structive thermonuclear battle, both superpowers might recognize some value in maximizing their postwar power position by the conquest, or denial of conquest, of intervening territories. If the Red Army could establish control of Western Europe, the Soviet Union would very likely dominate the world, or at least the Eastern Hemisphere, after the period of reconstruction. The Soviets cannot be expected to give up this opportunity if they can maintain control and essential order at home, and the United States might very well be interested in blocking them.5 While it is conceivable, therefore, that the NATO shield could con tribute something to the waging and winning of an all-out war, it seems much more realistic to regard its proper defense function as one of limited war—defining this term specifically to include a large-scale World War II-type of war fought entirely with non-nuclear weapons. General Norstad states the limited war function of the shield (the third function mentioned earlier) as follows: "If we concentrate only on weapons and forces for general war, we deny to ourselves the capacity to dispose of lesser situations, and could suggest an opportunity for limited aggressions, which ultimately could bring us disaster. In other words, we must be able to respond to less than ultimate incidents with decisive, but less than ultimate, means." Then, in the next sentence, he apparently contradicts himself, by denying the possibility of limited war in Europe: "By this I am not suggesting that limited wars are pos sible along this sensitive frontier of NATO. It would be very unlikely, I feel, that a serious incident could remain limited."6 General Norstad states that the Western European area is so vital for both sides that a war which began in a limited way would inevitably become total. It has also been stated dogmatically by various spokes men that any war involving Russian and American soldiers could not be kept limited.7 The possibility of "probing actions," "border incidents," 5
It is interesting to note that the Soviets apparently are counting on a brokenbacked war. Their doctrine is that the strategic exchange, although very im portant, would not be decisive. Hence they foresee a need for large ground forces to terminate the war decisively. This may be a case in which the Soviet commit ment to an ideological doctrine—the importance of "permanently operating fac tors"—somewhat accidentally leads them to a conclusion which may be correct on coldly pragmatic grounds. See Raymond L. Garthoff, Soviet Strategy in the Nuclear Age, New York: Frederick A. Praeger, 1958, chap. 4. 6 Senate Committee on Foreign Relations, "Mutual Security Act of 1958," Hearings, op.cit., p. 187. 7 See testimony of General Thomas D. White, Air Force Chief of Staff, in Pre-
in Western Europe
125
and the like, which would be dealt with on limited terms by the shield forces alone, is admitted. In statements by General Norstad, one gets the impression that such occurrences do not fall within his definition of "limited war." The latter is implicitly defined as involving substantial bodies of troops on each side, on the Korean War model, with deter mined fighting aimed toward significant territorial goals rather than merely the probing of intent. According to the public declarations of NATO strategy, this kind of fighting could not take place without pre cipitating all-out war in rather short order.8 But the phrase "less than ultimate incidents," in the Norstad state ment just quoted, would seem to cover a far greater and more serious range of contingencies than is suggested by the terms "border incident" or "probing action," and "less than ultimate means" would seem to in clude responses to aggression which would fall into the category of fairly substantial "limited war." One plausible explanation for the ap parent contradiction in the above statement is that General Norstad wished to avoid undercutting the massive retaliatory threat, by sug gesting to the Russians that a lesser response was contemplated. On other occasions, General Norstad has clearly indicated that the shield forces would fight a limited war against all Soviet incursions that could be dealt with successfully at this level, which, of course, is only reasonable.9 A somewhat more subtle interpretation, which gains additional sup port from other statements by Norstad, is to the effect that we really do not believe a war in Europe could be limited, but we are not sure the Russians share this opinion. Therefore, there is need for a limited deterrent to backstop the Big Deterrent—as distinct from a limited re sponse to substitute for a massive response in case of failure of the de terrent. In other words, if the Russians might mistakenly gamble that they could make limited gains without precipitating all-out war, the chances of such miscalculation could be reduced by posing a limited denial deterrent for limited aggressions—one based on a capacity to paredness Investigating Subcommittee, Senate Committee on Armed Services, "Major Defense Matters," Hearings, 86th Congress, 1st Session, Part 1, p. 100. 8 Robert E. Osgood presents a perceptive discussion of certain contradictions between NATO's declaratory policy and operational capabilities in "NATO's Strategic Troubles," mimeographed, 1959. 9 House Committee on Foreign Affairs, "Mutual Security Act of 1959," Hearings, 86th Congress, 1st Session, Part 2, p. 466.
126
Deterrence and Defense
deny the enemy territorial gains rather than on the threat of drastic punishment. The most interesting of the three functions of the shield, as outlined by General Norstad, is the second—its deterrent role. Of course, the shield should be able to deter all minor Soviet attacks, which obviously it could contain. What is not so obvious, however, is the role claimed for the shield in deterring "major" aggression; in this role the shield does not directly deter by a clear ability to defend successfully, but rather complements the deterrent effect of the sword. General Norstad describes in the following way the shield's function in complementing the strategic airpower deterrent: . . . if this forward line of the NATO countries were undefended, or were held by only a token force, which would be practically the same thing, it would be quite a simple matter for the Russians, or their satellites supported by the Russians, to move across that line. It could be either as a result of a border incident or more likely as a re sult of a deliberate probing operation to put us in a position of politi cal disadvantage. Against an undefended line, this would be easy to do, and no substantial force would be required on their part; they would get something for nothing, and I don't have to point out to you that this would put us in an extremely difficult situation. . .. We, together with the other NATO members, would have the responsi bility of doing something about it. Should we fail to do something it is perfectly obvious what would happen to the NATO alliance.... If we did not respond . . . we would destroy something of tre mendous value. However, if we decide to carry out our obligations under the treaty, we would be confronted with the necessity of using substantial force ourselves, and I am talking now about heavy re taliatory forces. I believe that this country and the alliance would take the proper action in a case of this kind, but I do not have to suggest that this would be an extremely difficult decision to make. On the other hand, if the forward line were defended, and de fended with reasonable strength, a border incident of the kind I have described could be held in check. If there were a decision to set off a probing operation, then there would be force to stop it—not to hold an all-out attack, but to hold it for a short period of time. During that time, the Russians would have to give consideration, not only to the forces of the shield, which are relatively light, but also to the fact
in Western Europe
127
that this action on their part would compel them to use some signifi cant force, involved in which would be consideration of the full consequences of the retaliatory forces as well. As a consequence, we feel the most important function of the shield forces is really not to fight, not even to defend, but to complete the deterrent.10 The essential idea in the statement just quoted is that a strong shield would force the enemy, if he wished to make any gains at all, to take the initiative in starting a substantial war, a provocative act which would run a serious risk of touching off an all-out response. Of course, the initiative always rests with the aggressor and this fact tends to strengthen deterrence, because the aggressor must bear not only the moral onus for starting the war, but also the psychological burden of weighing the immense uncertainty concerning the opponent s intentions and the "fog of war," and overcoming the natural inertia which is likely to inhibit any movement from a state of peace to a state of war. But it is also desirable for the deterrer to raise as high as possible the level of conflict at which the aggressor must exercise his initiative, for this tends to multiply the uncertainties and risks. This deterrent function of the shield in "passing the initiative" is the result partly of its size and partly of its deployment. If it is deployed on the line, it will automatical ly be activated when the aggressive force collides with it; the freedom and the responsibility for choosing peace or war rest almost entirely with the aggressor. If it is small, or even if it is large but not on the spot, the enemy may be able to pass the initiative back to the deterrer by presenting him with a fait accompli. To illustrate: if the Soviets were to occupy Schleswig-Holstein quickly (which they could easily do if they faced only a flimsy shield), they would then be committed to defend Schleswig-Holstein against a counterattack by NATO forces. The West would then have to bear the psychological burden of initiat ing substantial conflict, a heavy burden indeed if the only available response were strategic nuclear war, but heavy enough even in a local response. An attempt to eject the invaders might well eventuate in allout war, since it would require the application of substantial force. If the expected costs were greater than the value of Schleswig-Holstein to the West, the imperatives of honor might not be sufficient to generate a response. The Soviets could be expected to do everything possible to 10 House Committee on Appropriations, "Mutual Security Appropriations for 1959," Hearings, 85th Congress, 2nd Session, pp. 563-64.
128
Deterrence and Defense
inhibit a response; presumably they would occupy the conquered area in considerable force so as to raise as high as possible the costs of re pelling them. They would no doubt proclaim that they had no more territorial ambitions. They would offer to negotiate and perhaps insist that their occupation would be only temporary. They might threaten to increase their commitment of forces and take more territory if NATO attempted to push them back to the original border; such threats might well include a little judicious nuclear blackmail. In such circumstances, as General Norstad says, the decision to respond at all would certainly be "difficult." But if the West maintained a strong shield force on the border, it would be committed in advance to a determined and costly defense of the territory, and the responsibility for deciding to set off a substantial conflict which could grow into all-out war would rest with the enemy. A closely related point is that a sturdy shield would prevent the Soviets from making an aggregate large gain by a series of "nibbles," no one of which was thought serious enough by the West to warrant an all-out response. The Soviets would have to consider the following chain of logic: the stronger the defending shield, the greater the forces neces sary to defeat it, the larger the resulting conflict, the greater the provo cation and the apparent stakes, and the greater the probability of strategic nuclear retaliation. In addition, "accidental" incursions, or "border incidents" caused by the unauthorized action of some minor Soviet or satellite commander, could not flower into a larger deliberate aggression because the shield would be able to nip them in the bud rather quickly, or at least force the enemy to make a deliberate decision for either peace or war. As General Norstad has stated: ". . . we must force a conscious decision on the part of the Russians that they either will or will not go to war. They can't slide into it, they can't back into it, they can't wake up some morning and find they have gone too far."11 Norstad says the most important function of the shield is to "com plete the deterrent." By "completing" he apparently means that a strong shield serves to link up the ground forces' capability to deter minor moves (by their capacity to block them successfully) with the sword's function of deterring major aggression by the threat of nuclear punish ment. If the shield were weak, there could be a dangerous gap in the 11 House Committee on Foreign Affairs, "Mutual Security Act of 1959," Hear ings, op.cit., p. 445.
in Western Europe
129
spectrum of deterrence. The enemy would be deterred from making very minor attacks, and if he made them, they would be thrown back. But he might be confident of being able to make a larger attack and seize a considerable piece of territory without having to commit forces large enough to touch off general war. To close this gap, the shield must be large enough to assure the enemy that any attack powerful enough to break through would bring about nuclear retaliation. What is still unexplained about this formula is why a certain level of intensity of violence in the initial ground battle is necessary to trigger an all-out nuclear response, or why the Soviets would consider it neces sary. One explanation is that in deliberately starting a war of some considerable scope and intensity, the Russians would automatically signal a major objective, since they would be deliberately accepting costs and risks appropriate only to such an objective. Our willingness to respond with strategic nuclear weapons presumably depends, in part at least, on our perception that the enemy's objective is serious enough to warrant our suffering the costs of all-out war in order to defeat it. For example, if the enemy had to punch through only a 15-division shield to make gains, he might claim that his objective was merely to occupy a certain province temporarily to punish the West Germans for some offense against the East Germans, and the claim might be plausi ble enough to deter nuclear retaliation. But if he had to break through a 30-division shield, this claim would be much less plausible, the West would be more likely to interpret the Russian aim as the total conquest of Western Europe, and would be more willing to retaliate. Whether strategic nuclear war is a rational answer to any Soviet objective in Europe is not the point. Chances are it is not, and will become even less so in the future, but the Soviets do not know this with certainty, nor can they be sure of the rationality of our leaders. They will assign some probability to the massive response and the probability may vary with the size of the objective which they think the West sees in their moves. This "revelation of major intent" idea is not fully sufficient, however, as an explanation of the complementary deterrent function of a strong shield. It seems hardly likely that the Russians, if they had decided to take Western Europe, would try to do it by a sequence of military "nibbles." Any successful nibble would set off a great conventional mili tary build-up in the West which would preclude further nibbles. Thus, if the Russians try to take Western Europe by military means, they will
130
Deterrence and Defense
do it in force. Their "major intent" would be evident as soon as large forces began crossing the East-West border, no matter how many shield divisions were there to meet them. A more plausible interpretation of the Norstad thesis (although "revelation of major intent" no doubt is part of it) is simply that in at tacking a strong (30-division) shield the Russians would be committing a highly provocative act. We tend to overestimate the importance of the strategic value of territory in international affairs and underesti mate the importance of more subtle psychological elements. As Thomas C. Schelling has suggested, the symbolic connotations of violent acts may be more important in determining the appropriate response than their apparent objective.12 After all, if the Russians were to attempt a breakthrough into a certain province of Western Germany with a force as large as 60 divisions (and tangle with a 30-division shield in so do ing), there would be no logical reason to believe that their objective went beyond the conquest of that province—at least, such an interpre tation would not depend on the existence of the shield. The important point is that expectations on both sides have crystallized around the idea that any "substantial conflict" in Europe between Russian and NATO forces would mean all-out war. This is a sort of psychological convention which we ourselves have done much to create by our de claratory statements. Since the Russians know that we share this expec tation, they know that if they precipitate such "substantial conflict" they run a serious risk of massive retaliation. The capacity to force the enemy to initiate a substantial conflict has some further deterrent implications. The higher the intensity of the initial ground warfare, the greater the likelihood that strong emotions favoring a nuclear retaliatory response will be generated in Western public opinion. This is really an extension of the familiar "trip-wire" argument: that the placement of even small contingents of American and British forces on the East-West frontier puts the Soviets on notice that any attack on Western Europe will immediately involve the United States and Britain, and bring into action the strategic nuclear forces of these countries. Some trip-wire effect could be realized with only token forces from the extra-Continental countries; an attack on such forces would implicate the honor, prestige, and other emotional 12 Thomas
C. Schelling, "Toward a Theory of Strategy for International Con flict," The RAND Corporation, P-1648, May 8, 1959, pp. 37fF.
in Western Europe
131
values of these countries to a greater degree than a mere treaty com mitment.13 But a heroic delaying action by larger forces, with attendant larger casualties, would provide greater emotional fuel, as well as time for these emotions to have their eifect on the Western leadership. Such "emotional mobilization" might well be the marginal factor which would persuade an indecisive leadership to order strategic retaliation. Even if the leadership did shrink from "massive" retaliation, it would experience severe moral, as well as popular and military, pressure to do something to help the forces resisting the Soviet advance—such as permitting the use of nuclear weapons "tactically" in the battlefield area, or undertaking some limited or selective form of retaliation against the Soviet Union itself. Contemplating either of these two possibilities in advance—both as to their direct consequences and as to the risk that they would spiral into all-out war—should give the Soviets pause. A shield strong enough to put up a stiff battle, even though a losing one, against a major Soviet attack would tend to enhance the apparent defense effectiveness of strategic nuclear action, and therefore increase the probability of the latter being applied. In other words, the official 13
The trip-wire concept is rejected by General Norstad and other spokesmen for NATO, presumably in part because this function alone could be served by very small forces, and its acceptance would thus detract from Norstad's arguments for a considerably larger force. The concept has otherwise been criticized on two grounds—first, that token forces are not necessary to show that an attack has taken place; and, second, that the Russians would be stupid to trip the wire deliberately by ground action and trigger off a nuclear strike against them; that before doing this they would take the initiative and strike first themselves with their strategic forces. For a statement of this argument, see Malcolm Hoag, "The Place of Limited War in NATO Strategy," in Klaus Knorr (ed.), op.cit., pp. 105-6. Both of these criticisms are illogical. As for the first one, the point of having a trip-wire is not to show that an attack has taken place—which of course would be evident as soon as the Soviet forces crossed the border—but to implicate extraContinental nuclear powers by means of something stronger than a treaty con tract. As for the second, it is true, of course, that if the Russians did expect the trip-wire to work, and if they were committed to attack, they would be foolish (if there were an advantage in striking first at the strategic level) not to begin their attack with a massive nuclear blow against NATO's strategic nuclear forces, thereby reducing the damage which the latter forces could inflict on them in reply. But the argument falls with the second "if." The Russians are not com mitted to attack. In fact, this argument really supports the concept of the trip wire, because if the Soviets believe they must themselves start strategic nuclear warfare to avoid the much worse consequences if we start it, this belief, in itself, is a very potent deterring factor, since there is a good chance the Russians will con sider no war at all to be a better state of affairs than an all-out war in which they strike first.
132
Deterrence and Defense
NATO strategic dogma—to the effect that "victory" can be achieved by strategic retaliation plus a holding action by the shield—is more plausi ble, the greater the ability of the shield to hold. It is not necessary for the Soviets to believe in this dogma; they need only to realize that Western leaders and strategists believe it, to increase their assessment of the chances that strategic retaliation will take place. On the other hand, if the Soviets face only a thin shield, they may well doubt that the Western leaders themselves really believe that a combined shield-sword action can prevent the overrunning of Europe. They may gamble on achieving a quick conquest of the Continent, thereby facing the West with the problem of undoing a fait accompli, which would be incom parably more difficult than preventing a conquest. In these circum stances, massive retaliation would very likely seem futile; but if it could be undertaken in the early stage of what was likely to be a prolonged ground battle, Western leaders might believe that retaliation could stop the Red Army's advance, either by drying up its sources of supply or by withering the Soviets' will to continue the fighting. Also, any at tempt to force the Soviets' withdrawal from Western Europe after they had already overrun it would probably require extensive nuclear bomb ing in Western Europe itself, which the British and American leader ship would find extremely repugnant. A strong shield—say, the NATO goal of slightly less than 30 divi sions—would force the enemy to undertake a noticeable build-up, both near the point of attack and in the general theater, before he could be sure of mounting a successful assault. This build-up would provide warning which, first, could be used to mobilize and transport reinforce ments and reserves to increase the defense effectiveness of the shield. Even if the enemy could move troops into East Germany faster than the Western allies could match them, any strengthening of the shield would contribute to the complementary deterrent effects which we have been discussing. Perhaps more importantly, the warning time provided by the Russian troop movement could be used to take civil defense measures, such as evacuation of city populations, which, by reducing the cost of all-out war, would tend to increase the likelihood of a retaliatory response. In addition, active air defenses and counterforce striking capa bilities could be alerted and augmented, with a similar effect. The time provided by forcing the Russians to mobilize in advance of war, and also by the shield's delaying action after the outbreak of war, might also be used to take certain provocative actions—i.e., beyond civil de-
in Western Europe
133
fense and preparedness measures—which, by indicating a willingness to run serious risks, would suggest to the Russians that we were willing to fight an all-out war unless they desisted from their attack. For ex ample, we might begin attacking Soviet ships at sea, submarines, and reconnaissance satellites. Such measures would be risky in some cir cumstances, because they might stimulate the Russians to launch an allout pre-emptive strike; on the other hand, the Russians might well feel that their more rational course was to stop the preparations for attack, or terminate the tactical war if it had already started. As mentioned, time would be required to make the full deterrent impact of such meas ures felt. And of course time would also be useful for diplomatic con tacts to supplement the military demonstrations and to attempt to negotiate a quick settlement. The enemy is most apt to think we are serious about carrying out massive retaliation if we clearly reserve this extreme response for con tingencies which seem to threaten our very survival. If, in other words, we provide for a limited response to fairly substantial but not full-scale ground attacks—at substantial cost in preparedness—we make it easier for the Soviets to believe that we really mean to retaliate with strategic nuclear airpower in the event of a major attack. Such obvious qualifica tion and limitation of the all-out threat will show that we are not relying entirely on bluff, that we are creating, at considerable expense, a capa bility for rational response to a variety of challenges. Perceiving this, the Soviets are likely to think that we regard massive retaliation as being also a rational response in those contingencies for which we have cre ated no effective alternative means of action, especially if our first-strike counterforce capability is fairly impressive. Too much reliance on stra tegic nuclear power and "massive" threats, on the other hand, will generate a certain amount of "threat inflation"—a depreciation of the value of all such threats even when applied to the most serious transgressions. The complementary effects mentioned so far have been those which have to do with enhancing the Soviet expectation of deliberate nuclear retaliation. Other effects flow from a raising of the chances that an at tack will generate conditions favorable to an accidental spiraling of the war to all-out proportions. In general, the Soviets must consider that the longer the ground battle is prolonged, the more time and the more occasions there will be for such accidents to occur. One of the more likely accidents might be the use of nuclear weapons by some battlefield
134
Deterrence and Defense
commander acting under the extreme stress of battle, or deliberately disobeying orders in order to save his unit. As Bernard Brodie has pointed out, the distinguishing feature of the limitation of modern war is that it involves the deliberate withholding from action of weapons of great military effectiveness.14 The more effective the delaying action of the ground forces, the more the military commanders are likely to chafe under such "unnatural" restrictions which, if lifted or ignored, would (it might seem to them) make all the difference between winning or losing the war. Not only the increase in available time, but also the rise in the in tensity of battle which would result from any strengthening of the shield would be conducive to the inadvertent initiation of full-scale war. The more violent the initial conventional battle, the greater the chance that one side or the other would begin using tactical nuclear weapons. If these weapons were used under some tacit or explicit limitation on permissible targets, there would always be the chance of some accidental violation of the limitations being interpreted by the other side as inten tional, which might lead to a fairly rapid raising of the limitations and ultimately, by a gradual loss of control, to all-out war. Such a spiraling or escalation process is perhaps the most likely manner in which the extreme disaster of all-out war would occur. Some tolerance of infrac tions could be counted on, since both sides would wish to avoid all-out war, but there is no assurance that any particular accident or mistake would be within the limits of tolerance. We have been speaking of ways in which the NATO ground force shield, by supporting the credibility of the retaliatory threat, tends to narrow the gap between the independent deterrent effect of the shield's capacity to defend territory and the independent deterrent utility of the strategic nuclear forces. In sum, the shield serves (in a degree de pending upon its strength) to increase the enemy's apparent objective, to force the enemy to commit a symbolically provocative act, to provide time for taking measures which would decrease our own costs in all-out war, to increase the probability that retaliation plus the shield forces could effectively defend Europe, and to create conditions conducive to emotional commitment, irrationality, and accidents—all of which should tend to raise the chances (in the enemy's calculations) that major ground aggression would precipitate strategic retaliation. 14 Bernard
Brodie, Strategy in the Missile Age, op.cit., p. 311.
in Western Europe
135
On the other hand, a counterforce strategic capability tends to nar row the gap from the other direction, by reducing the costs to us of all-out war. In other words, the shield and a counterforce strategic capacity are mutually supporting in strengthening the credibility of a massive response—the counterforce because it lowers the stakes or the degree of provocation which are likely to trigger strategic retaliation, the shield because it forces up the size of the stakes and the degree of provocation in ground warfare. NATO has a viable military deterrent when, in the Soviet image of NATO's cost-gain calculus, these two en tities overlap, or when there is a substantial probability that they over lap. The probability that is necessary will depend, of course, on the Soviets' valuation of a major gain in Europe, coupled with their capacity to weather a massive nuclear attack. Against the "complementary" deterrent utility of surface forces must be set their "depreciatory" effects—that is, the considerations which would support a tendency for them to reduce the credibility of the threat of massive nuclear punishment. In the preceding discussion, a number of reasons were advanced why the "cushion of time" provided by denial capacity may support threats of nuclear punishment in Europe. It is possible to argue on other counts, however, that the longer the time between the initiation and the completion of conquest, the smaller the chances that strategic nuclear retaliation will take place. A strong de laying action on the ground would provide more time for intra- and inter-governmental consultation about a retaliatory response. But nuclear retaliation as the result of a carefully considered decision might seem less likely than retaliation as a more or less automatic response, especially when the retaliation being considered is of massive dimen sions. The terrible consequences of all-out nuclear war would no doubt be imagined very vividly by leaders when they had time for serious de liberation—more vividly, perhaps, than if the decision had to be a quick one, in which case it would be more likely that prior plans and policy declarations would be carried out more or less automatically. Since most of these plans and declarations have been oriented around the principle of deterrence by massive punishment, it may well be doubted whether they are grounded on full and adequate consideration of the costs involved in carrying them out if deterrence should fail— costs which no doubt would be visualized with greater clarity in the process of making a deliberate decision whether or not to accept them. If a quick decision is needed, the civilian leaders of the United States
136
Deterrence and Defense
and Britain are likely to rely heavily on the counsel of their military ad visers, whose thinking will be strongly conditioned by the existing plans and who can be expected to favor the quick application of all available power. If some time is available, however, the voices of caution are more likely to make themselves heard and with greater effect. In addi tion, time would give the Soviets opportunity to inhibit a retaliatory response by suggesting negotiations, threatening counterretaliation, or promising "peaceful coexistence" with Britain and the United States. A thesis more often heard is that surface and tactical forces tend to erode the credibility of the retaliatory threat by providing the means for an alternative, less costly response. Without much doubt, this is correct with respect to those enemy moves which the ground shield can throw back successfully; there is obviously little point in suffering costs which one does not have to suffer. But these moves would be either deterred or defeated anyway. If the argument has any important valid ity, it must be on either of two possible grounds. One is that by having ground forces sufficient to deal with some threats, we reduce the enemy's expectation of our strategic nuclear retaliation in response to larger aggressions which the shield cannot block. But there is no good reason why this should be so. It would be just as logical for the enemy to suppose that we were providing ground forces to handle minor enemy moves which would not justify the costs of all-out war, but that we still intended to retaliate in case of larger aggressions. The other possible basis for this depreciatory effect has more validity, although it is rarely mentioned. It is that if we should decide that the massive retaliation threat was no longer credible enough to deter, and conse quently decided to build up a ground force sufficient to defend, the enemy might realize this by observing the build-up, and attack before the build-up was completed. The complementary and depreciatory effects of the shield are op posed and mutually offsetting but not mutually exclusive. That is, we have not stated here two sets of arguments, only one of which is valid, and between which the enemy must choose in assessing the deterrent implications of the shield. Rather, we have listed a number of different considerations, all of which are likely to affect the enemy's estimate of the probability of retaliation, although some work to increase and some to reduce this probability. What the net effect will be will depend on the enemy's weighing and balancing of both sets of considerations. It
in Western Europe
137
seems likely that the complementary effects exceed the depreciatory effects; that the shield, on balance, tends to support the credibility of the massive retaliatory threats as a response to those aggressions which the shield itself cannot handle. Quite a bit would depend, however, on statements of goals for the shield forces and the extent to which those forces came up to the goals. For example, the repeated declara tions by General Norstad that 30 divisions are the minimum shield requirement tend to create the impression that the West will not re gard aggressions that can be contained by 30 divisions as a sufficient provocation for massive retaliation; the actual existence of only about half this force may leave a range of intermediate-sized attacks (capable of overrunning the present shield) which the Russians might feel could be made successfully without serious risk of strategic nuclear retaliation. This illustrates that a declaratory policy which is necessary for persuasive purposes within the alliance may have an adverse effect on deterrence.
B. The Role of Tactical Nuclear Weapons15 The NATO shield now has a "dual capability." It is capable of tacti cal nuclear warfare by virtue of the provision of nuclear carriers and warheads to the U.S. forces in the shield and the gradual equipping of the other forces with missiles capable of carrying a nuclear warhead, 15 The distinction between "tactical" and "strategic" nuclear weapons has always been a fuzzy one. In current usage, tactical nuclear weapons are short-range weapons of relatively low explosive power, deployed on or near the battlefield area, to be used for striking at military targets in the combat area or directly behind it. The term "strategic nuclear weapons" is usually applied to more powerful long-range weapons intended for use primarily against targets on the homeland of the enemy—cities and industrial installations as well as the enemy's strategic forces. Strategic weapons tend to be associated with all-out war; tactical weapons with the concept of limited nuclear war. In this section, we shall follow this widely accepted distinction. However, as will be discussed below, a more useful and realistic distinction would be one based on the targets against which the weapons are used, rather than on the range or yield of the weapons themselves or their proximity to the battlefield. Thus, a tactical use of nuclear weapons would be the employment of nuclear weapons of any size against military targets directly supporting the enemy's ground-war effort, wherever these targets are located, even on the enemy's home territory. Long-range weapons deployed far from the im mediate combat zone might be used against such targets—i.e., they might be employed tactically. The strategic use of nuclear weapons would mean their em ployment against the enemy's cities or industrial economy, or against his longrange air and missile forces. Some forces ordinarily considered tactical, such as tactical air forces deployed in Europe, might be used for strategic purposes.
138
Deterrence and Defense
while yet retaining substantial capacity to fight by conventional means. The nuclearizing of the shield is usually justified by the as sertion that such weapons increase the defense capability of the shield. It is argued that the use of nuclear weapons tactically tends to favor the defensive side, and thus either reduces the manpower requirements to achieve a given defensive capability, or enhances the defense effec tiveness of a given level of manpower. The argument is quite obviously valid against an enemy who possesses little or no tactical atomic capa bility, but it becomes dubious when the enemy is plentifully supplied with nuclear weapons suitable for tactical use—as the Soviet Union no doubt is supplied. There seems to be great plausibility in the Army's contention that in bilateral tactical nuclear warfare, manpower re quirements are greater, not smaller, than in conventional warfare, be cause of much higher casualty levels and consequent need for replace ments, and the need to maintain a substantial logistical back-up for resupply of equipment. It is asserted that the defense has an advantage in tactical nuclear warfare even in a condition of parity because the danger of presenting a target for an atomic weapon will prevent the enemy from concen trating enough forces at one point to make a breakthrough. There are several rather convincing rebuttals to this. Even if it is granted that the tactical use of nuclear weapons would increase the ability of the de fensive side to hold territory, to initiate such warfare would greatly increase the costs of war and considerably increase the chances that the war would eventually become all-out, with vastly greater costs. In our broad definition of the term, "defense utility" is a function of the costs and risks of fighting as well as of the capacity to hold. It would seem that the much greater cost of fighting tactical nuclear war would more than offset any possible increase in denial capacity. Our European allies obviously are likely to believe this even if we do not. Therefore, many Europeans might prefer that the United States not honor its commitments to NATO if doing so means tactical nuclear warfare on their territory. But it is also questionable whether the use of nuclear weapons would increase the capacity to deny the enemy territorial gains. Even if the enemy cannot risk concentrating his forces, neither can the de fensive side, and effective defense must always include an ability to counterattack since the aggressor will almost automatically make an initial gain by virtue of the advantage of surprise and momentum. A
in Western Europe
139
further point is that if neither side can risk concentrating for a break through, the war is likely to be one of small-unit action, infiltration, and the like, verging on guerrilla tactics. If the enemy has an advan tage in manpower, he is likely to be able to make gains in this type of warfare. Moreover, if this image of future tactics is correct, the forces on both sides will be so closely interlocked in the battle area that our own nuclear explosions may do as much damage to our own forces as to the enemy's. A final point is that tactical nuclear warfare would in crease the advantage in reinforcement capability which the enemy now possesses by virtue of the land contiguity of his home base to the theater of war: in response to our bombing of his transportation and communication lines, he would be very likely to bomb European (and U.S.?) Atlantic ports, thus drastically reducing the possibilities of re inforcement and supply by sea from the United States. Amphibious landings of the World War II type probably would be impossible. In sum, there does not seem to be any defense advantage in preparing to fight a tactical nuclear war against an enemy who is similarly equipped. The only conceivable defense benefit would be to have a counter to the initiation of tactical nuclear warfare by the enemy. But there would seem to be little incentive for him to start tactical nuclear war fare as long as he enjoys superiority in conventional power. The utility of tactical nuclear weapons lies almost entirely in their deterrent function. First, they would help to deter the enemy from initiating the use of such weapons. In addition they would exert some deterrent effect against conventional aggression. Even though it might be irrational for NATO to use these weapons first, the fact that the shield forces have them significantly raises the costs and risks for the Soviets, especially when the Soviets note the strong U.S. and NATO commitment to their use in the event of serious aggression, and the extent to which NATO's detailed military plans apparently are premised on their use. The Soviets must not only expect higher direct costs should the war remain limited to the Central and Western European theater (than if NATO's intention were to use conventional weapons only), but also a much higher probability that the war would quickly degenerate into an all-out slugging match. Soviet military leaders (as well as their U.S. counterparts) seem committed to the idea that any use of nuclear weapons would lead straight to all-out war. Tactical nuclear warfare is much more likely than conventional warfare to give rise to accidents leading to the inadvertent explosion
140
Deterrence and Defense
of full-scale war. Even if NATO planned to fight a conventional war, and the war started at the conventional level, the possession of atomic weapons by the troops on each side would create possibilities of their accidental firing. The chance of accidental firing becomes greater as smaller weapons are developed, because the smaller the weapon, the lower the level of command to which it is likely to be assigned and the larger the number of fingers that will be on atomic "triggers." When and if a large number of atomic mortars get into the hands of platoon sergeants, the chance that at least one of them will be fired accidentally or irresponsibly rises almost to certainty, and once one is fired the symbolic strength of the distinction between conventional and nuclear weapons as a criterion for war limitation will have been gravely eroded. Once the war crosses the nuclear threshold, distinctions will be hard to find that are clear enough to serve as focal points upon which mutual expectations might converge regarding "natural" or "legitimate" limitations. The start of limited nuclear war need not immediately set up fears of imminent all-out attack, leading to pre-emptive attack at the strategic level, if the strategic balance is highly stable—i.e., if both sides have a very strong strike-back strategic capability. How ever, limited nuclear war can easily shade into a "spiraling" situation born of deliberate decisions to step up the intensity of the war just a bit to convince the enemy of the high costs that will follow his con tinued rejection of terms of settlement. The enemy will have to recog nize that the line between tactical nuclear warfare and limited strate gic reprisals (and between limited reprisals and not-so-limited repris als) is a blurred one and that we are more likely to carry out our threat of massive retaliation if we can approach its fulfillment gradu ally in a sort of "absence of mind," than if it requires a decision to in crease the intensity of the war from the conventional to the strategic nuclear level in a single move. If the Soviets were to attack conventionally a nuclearized shield, this would show a willingness to take risks which in turn would im mediately suggest a major objective which might justify a massive response. Thus, tactical nuclear weapons help to link up the deterrence of minor aggression, which can be accomplished with relatively small conventional forces, with the deterrence of major aggression exercised by the strategic retaliatory forces. They do so by ensuring, or at least increasing the chances, that any determined aggression would be in terpreted, at its outset or soon after, as having a major objective—since
in Western Europe
141
only far-reaching aims would justify the enemy in risking the high costs of tactical nuclear warfare. This linking-up effect of tactical nuclear weapons, and their deter rent value in general, would be strengthened if it could be made to seem that a tactical nuclear response would automatically follow any aggression, or at least any attack sufficiently powerful to break through the shield at the conventional level. Absolute automation is not possi ble, but an approach to it can be achieved by an obvious orientation of military plans and intentions around tactical nuclear weapons. NATO's present posture seems to be reaching a considerable degree of such "ad ministrative automation" through the elaboration of doctrine and the hardening of attitudes reflected in repeated statements from high sources that "nuclear weapons are becoming conventional" or that any major aggression would be met at the nuclear level. The NATO shield can still fight conventionally, but with the increasing equipment of the forces with atomic arms and the miniaturization and sophistication of tactical atomic weapons, the conventional response is likely to seem less feasible in practice and less natural in doctrine. These trends will not escape the Russians and will make it more difficult for them to believe that they can deter a tactical nuclear response by threatening greater damage against the West in retaliation. In other words, as NATO becomes more and more committed to a tactical nuclear re sponse, the burden of the initiative in starting a nuclear war shifts more heavily onto the Soviets.16 As mentioned earlier, not having the initiative is an asset for deterrence, although a liability for defense. The degree of automation could be increased by giving the military commanders advance authority to use tactical nuclear weapons if necessary to hold a line or save their unit. There is no evidence that such authorization has been given, but Secretary of State Dulles hinted 16 Secretary of State Dulles showed that he recognized this "passing the buck" function of tactical nuclear weapons when he said in a magazine article in October 1957: ". . . it may be that by the 1960 decade the nations which are around the Sino-Soviet perimeter can possess an effective defense against full-scale conven tional attack and thus confront any aggressor with the choice between failing or himself initiating nuclear war against the defending country. Thus the tables may be turned, in the sense that instead of those who are non-aggressive having to rely upon all-out nuclear retaliatory power for their protection, would-be aggres sors will be unable to count on a successful conventional aggression, but must themselves weigh the consequences of invoking nuclear war." John Foster Dulles, "Challenge and Response in Foreign Policy," Foreign Affairs, Vol. xxxvi, No. 1 (October 1957), p. 31.
142
Deterrence and Defense
at the possibility when he said on November 20, 1957, that if an attack were made on NATO forces "like that on Pearl Harbor" the decision to shoot back would be made by the commanders on the spot.17 It has been argued that when both sides are plentifully supplied with tactical nuclear capabilities, the most rational first move for the Russians (supposing they did not have an adequate first-strike strategic capability) would be to launch a massive strike at NATO's tactical nuclear forces.18 If these forces were not very well protected (so the argument goes), the first strike would establish immediately a drastic asymmetry favoring the Russians in tactical nuclear capabilities, since they could strike at short range and with very little warning. Hence, it is concluded, relying heavily on tactical nuclear forces can not be justified even as a deterrent, since any attempt to make these forces sufficiently invulnerable to withstand a surprise attack would be prohibitively expensive. No doubt there is, and will continue to be, an advantage in striking first at the tactical nuclear level, unless both sides go in for intensive hardening, mobility, and dispersal of their short-range nuclear forces. The side which absorbed the first blow would have its battlefield nuclear capabilities sharply reduced in relation to those of the op ponent. However, it could redress the balance of forces in the combat zone by bringing its long-range strategic forces into play against the enemy's tactical nuclear weapons and ground forces. This would re quire, if the Soviets struck first tactically, that the United States have at least enough strategic forces, beyond those necessary to deter a strategic attack on itself, to re-establish symmetry on the local battle field. To accomplish this might drain away a substantial part of our strategic capability, especially if this capability consisted largely of missiles, since we would have to fire at long range with a consequent depreciation of accuracy. This may be one of the best arguments for having a substantial first-strike strategic counterforce capability—to deter the enemy from a tactical nuclear surprise attack, not by a threat of all-out response on the enemy's homeland, but by having a capability to redress the tactical battlefield balance should the enemy upset that balance by a surprise first strike at the tactical level. The argument just mentioned—that unprotected nuclear weapons 17
New York Times, November 21, 1957, p. 1. is the position taken by Malcolm Hoag in "What Interdependence for NATO?" World Politics, Vol. XII, No. 3 (April 1960), pp. 369-90. 18 This
in Western Europe
143
in the battlefield area invite a tactical surprise attack—must stand or fall on two assumptions: (1) that the Russians are virtually certain that if they attack conventionally, NATO will use tactical nuclear weapons in response; and (2) that the United States (and other NATO members) does not have a surplus of strategic weapons sufficient to re-establish battlefield symmetry after the Soviets' first tactical strike. If the Russians think there is a good chance NATO will not strike first with tactical nuclear weapons, presumably they would prefer to make their conquest conventionally at much lower cost. Even if they expect ed, with high probability, that NATO would strike first tactically, they might well prefer to attack conventionally and take the risk, knowing that they could re-establish tactical symmetry by use of their own strategic weapons. Only if the U.S. strategic nuclear forces were just barely sufficient for a "minimum deterrent" against strategic surprise attack, and the Russians knew it, would a first strike at the tactical level make much sense for them. If the United States had a comfortable surplus over the minimum deterrent requirement, the chances that it would use this surplus "tactically" would certainly be much higher than the chances that it would retaliate with an all-out strategic blow at the Soviet homeland (assuming that the U.S. strategic capability is not so large as to permit a massive strategic counterforce blow at acceptable costs). Once the United States had brought its strategic weapons into play in the battlefield, the chances of a further escalation would be increased, because a conspicuous distinction—that between "battle field" nuclear weapons and long-range strategic weapons—would have been eroded. Therefore, the Soviets would be taking very serious risks in making a tactical first strike. It is misleading to distinguish between tactical and strategic nuclear weapons entirely on the basis of range and proximity to the battlefield area. Long-range strategic weapons can be used tactically—i.e., against the enemy's ground forces and their supporting installations. More over, such tactical use need not be limited to the local battlefield, as we shall discuss below. Therefore, if one side wishes to establish tacti cal nuclear superiority by striking first, it must consider striking at all the opponent's nuclear forces which may be used tactically. A tactical first strike thus may be virtually indistinguishable from a strategic first strike and may incur similar costs and risks for the attacker. If long-range weapons were introduced into a limited nuclear war initial-
144
Deterrence and Defense
Iy limited to short-range weapons, the area limitations of the war would be eroded because each side would feel a strong compulsion to strike at the other's long-range forces. This possibility, and the further possibility that all restraints would be abandoned once such reciprocal long-range strikes took place, should be a strong deterrent against the initiation of short-range tactical nuclear warfare, whatever the imme diate advantage of getting in the first blow in such warfare. 1. EXTENDED LIMITED NUCLEAR WAR The concept of limited tactical nuclear war usually includes, as one of its limitations, the idea of "sanctuary." In particular, it is assumed that the homelands of the two superpowers will be inviolate. This as sumption may be obsolescent. When and if the strategic balance be comes highly stable—i.e., when both sides have highly invulnerable and substantial strategic forces—the danger that such limited viola tion of the superpower sanctuaries would set off all-out conflict may substantially diminish. It becomes plausible, then, to envisage a "limit ed war" in which long-range nuclear forces would strike at such targets as troop concentrations, ammunition and fuel dumps, railroad yards, etc.—the traditional "interdiction" targets which contribute to the strength of armies—on the territory of the Soviet Union itself. And this image would have to include similar attacks by the Soviet Union on NATO's supporting facilities far to the rear of the battleline, possi bly including certain targets in the United States such as training camps and troop embarkation points. Targets on both sides might in clude some of the bases of the long-range nuclear forces which were carrying out such interdiction or were in a position to do so. As in the case of geographically limited tactical nuclear war, this kind of strategy probably would not be a rational one for NATO to initiate as a means of defense against a Soviet aggression with con ventional forces. The Soviets probably would have the advantage in this kind of a battle, considering the more complicated logistics sys tems of the Western forces and the limited number of ports of em barkation and debarkation. The cost in by-product damage to civilian populations would be high. But the threat or possibility of such a strategy would carry considerable deterrent value prior to the war. The Russians would be put on notice that they could not count on fighting a successful limited war in Europe with their own territory remaining inviolate. And we would wish to inform them also that we
in Western Europe
145
did not believe that strikes against tactical military targets on Russian territory would trigger a Soviet all-out response. This threat might be more credible than the threat of an all-out first strike because of its direct contribution to the battle for the control of territory. It would be a logical military strategy, in line with traditional concepts of war fare, and therefore one which the Soviet military leaders might very well expect their Western opposite numbers to initiate. In this conception, the "tactical" and the "strategic" tend to merge. Military leaders, including President Eisenhower, often have spoken of the lack of a clear dividing line between these two categories of warfare,19 and the growing stability of strategic deterrence may show them to have been more realistic than the civilian analysts who have tended to divide the tactical and strategic, or the limited and all-out, types of warfare into rather sharply defined categories. We might even go so far as to define limited or "tactical" warfare, in the age of stable strategic deterrence, as all warfare of whatever intensity in which both sides strike deliberately only at military targets and con tinue to reserve enough strategic nuclear forces to deter the opponent from beginning indiscriminate attacks on cities. This definition would permit inclusion within the concept of "limited war" of attacks by long-range strategic forces—including, for example, ICBM's—against the strategic nuclear forces of the other side which might be used for tactical interdiction purposes. This type of war certainly could be extremely destructive, with by product damage to cities and populations in both the United States and the Soviet Union possibly exceeding that which might have been regarded, before the war, as unacceptable. In fact, there might come a stage in such a war when any distinction between it and all-out war which began by a surprise thermonuclear attack would be purely academic, if we consider that the latter might be fought under some restraints in target selection. The only difference would be that it would have become all-out "from the bottom up" rather than "from the top down," so to speak.20 Even if both sides reserved an ultimate "Sunday punch" of invulnerable strategic forces, the degree of pop19 The Air Force, for example, regards SAC as having a limited war capability in addition to its primary role of deterring and fighting all-out war. 20 Incidentally, this may be the only kind of all-out war which makes much sense out of NATO's strategic plans to use small tactical atomic weapons on the local battlefield in all-out war; if the war is indiscriminate at the strategic level, it is also likely to be so in the battlefield area.
146
Deterrence and Defense
ulation and economic damage suffered in "extended limited nuclear war" could even exceed that which one side might have escaped with, had it used its strategic weapons at the outset in a massive counterforce strike at the enemy's strategic forces. If one side or the other should begin to fear during a limited war in Europe that the conflict eventually would escalate to an extended tacti cal nuclear war involving very high levels of destruction, it might prefer the consequences of striking first all-out to those of "riding the escalator," even though, before the war had started, the costs of the opponent's all-out retaliation to a first strike had been considered un acceptable. Here it must be pointed out that if this form of warfare—it might be just as appropriate to call it "semi-all-out" war as "extended limited" war—is regarded as a serious possibility, there is a requirement for much larger quantities of strategic striking forces than might suffice for a minimum deterrent. If we possessed only the latter, the Soviets could bang away at our tactical supporting installations—even those in the United States—with confidence that we could not rationally strike back all-out, and we could not match them in the tactical counterforce war because we would then dissipate our ultimate deterrent against all-out attack. In general, the side with the greatest surplus of forces over a minimum deterrent would have a strong point of advan tage in this type of war and could fight it with least fear that it would precipitate an unrestrained all-out war. Perhaps it is more likely that if tactical nuclear war does expand to encompass the homelands of the superpowers, it would take the form of selective, limited strikes, primarily for bargaining purposes rather than for attrition. At least it seems worthwhile to consider this possi bility, which we shall do in a later section of this chapter.
C. Future Military Alternatives for NATO NATO now faces a period of crisis during which hard choices must be made concerning future strategy and weaponry. The crisis stems from one primary cause: the imminent development by the Soviet Union of a powerful and sophisticated nuclear missile striking force. This development has brought into serious question what has been the dominant premise of all NATO strategic planning—the presumed effectiveness of the threat of American and British strategic retaliation
in Western Europe
147
in deterring major aggression. There is increasing doubt that this threat will retain sufficient credibility to deter, now that the Russians are apparently on the brink of attaining a missile striking power which could rain extensive death and destruction on the United States and Britain even after absorbing an all-out counterforce first strike by the strategic forces of these two countries. Certainly—barring extremely optimistic assumptions about the capability of the West to locate and knock out Russian missile sites, and barring also radical innovations in air defense or a massive civil defense effort on our side—the credibility of the retaliatory threat will come increasingly to depend on a Russian belief in the irrational recklessness of the U.S. and British leaders and this belief may not be sustainable. If the United States should appear to be placing increasing reliance on intercontinental striking power and less on bases and real estate in Europe, the credibility of the threat of massive retaliation would be further undermined.21 In this situation, NATO must decide whether it is going to take steps to strengthen the nuclear deterrent, or whether it is going to sub stitute measures for more effective defense. This is the essence of the strategic dilemma which NATO now faces. The fundamental alterna tives may be listed as follows: (1) The building of a substantial first-strike counterforce strategic capability by the United States. (2) An increase in the strength of the shield forces, reinforced sufficiently to permit them to fight and win a sizable conventional limited war in Europe, up to and possibly including a war fought with maximum conventional forces. (3) Independent strategic forces for some Continental NATO countries. (4) A jointly controlled NATO strategic force, or a strategic force controlled jointly by the European powers. (5) A strategy of limited strategic or tactical retaliation. (6) Some combination of these alternatives. 21 The Russians, and the Europeans, would do well not to overestimate this possible effect of concentration by the United States of more of its strategic forces within its own borders. The United States' valuation of Western Europe, and willingness to fight in its defense, do not rest entirely on Europe's strategic value, but also on emotional and moral factors which would not be affected by a redeployment of strategic forces.
148
Deterrence and Defense
1. A U.S. FIRST-STRIKE COUNTERFORCE CAPABILITY The first alternative has been discussed in the preceding chapter. We need only repeat here that a strategic force which could not only cause sufficient attrition of the Soviet nuclear forces in a first strike to make their retaliation bearable, but also either defeat the Soviet Union or prevent Soviet territorial gains, without the assistance of substantial tactical forces, would have to be very large and powerful. The future course of military technology is subject to great uncertainty, but as of now it appears that when the Soviet Union has a large long-range missile force, a sufficient first-strike force for the United States may turn out to be technically or economically unfeasible. This present outlook probably is not adequate reason, by itself, for not continuing to develop some counterforce capability, and especially for not con tinuing research and development in areas—particularly in intelli gence techniques, and in accuracy and reliability of missiles—which may eventually make practicable an adequate first-strike force. Progress in air defense against missiles and, in particular, an all-out effort in the field of civil defense would reduce the requirements for offensive striking forces. And it is well to remember that a first-strike force which the United States would not consider effective enough to use, but which was nevertheless quite formidable, might create some degree of expectation in Soviet minds that it would be used in some circumstances, and therefore might exercise a significant deterrent effect. Whether it would be desirable to create such a force for deter rence purposes alone, in preference to the other alternatives listed, would depend on whether its deterrent value was thought large enough to offset the losses which would ensue if deterrence failed and forces for an effective defense at bearable cost were not available. 2. A CONVENTIONAL DEFENSE CAPABILITY If a sufficiently credible strategic first-strike deterrent proves to be unfeasible or too costly, and if tactical nuclear warfare is an inefficient method of defense, one alternative is to increase NATO's' capability to fight conventional war. But how far should this increase go? Is there, or will there be, a range of very serious provocations which we can rely upon the nuclear threat to deter and against which we do not have to be prepared to defend? Or must we go all the way with ground
in Western Europe
149
forces, providing sufficient numbers to defend successfully against an all-out Soviet ground attack? The massive threat will retain some credibility, for reasons we have mentioned several times, even after the Soviets have attained a sub stantial nuclear missile striking force. It has been argued that the effect of its declining credibility is to narrow the range of contingencies which can be deterred to those at the higher end of the spectrum of "seriousness," and that the ground force shield must grow just enough to be able to defend successfully against those provocations which drop out from under the Big Deterrent umbrella, and to force the Russians, if they are to attack with any hope of success, to precipitate that high er degree of ground conflict which is now necessary to implicate our Big Deterrent plausibly. Implicitly, in this argument, the shield must be larger but not large enough to defeat a full-scale conventional ground attack by the Soviets, for this latter move is one which we would presumably consider serious enough to justify massive retalia tion; therefore it would be deterred. In this view, the shield must in crease only gradually as the credibility of the massive threat declines gradually, and more or less in constant ratio with the latter. This view point has been summed up by Henry A. Kissinger: "The line of demarcation between limited war and all-out war in Europe need not be determined in the abstract. The stronger the local forces of NATO, the less likely it will be that the Soviet Union will be tempted to adventure. The more effective the military establishment on the Con tinent, the larger must be the Soviet attack designed to overcome it. The more the required effort approaches the scale of all-out war, the clearer the challenge to our security and the more plausible our over all deterrent. In short, as the horrors of all-out war multiply and cripple the will to resort to it, the minimum objective of the forces in Europe must be to raise the scale of the Soviet effort required to defeat them to a level that can leave no doubt about its ultimate objective."22 But if the crucial determinant of our will to retaliate is our percep tion of the enemy's "ultimate objective," it could be argued that this perception would be just as clear were the Soviets to attack a 20- or 30division shield as if they attacked a defending force of many more divisions. A certain minimum number of divisions—let us accept Gen eral Norstad's objective ot 30—would be necessary to force the enemy 22 Henry A. Kissinger, "The Search for Stability," Foreign Affairs, Vol. xxxvn, No. 4 (July 1959), pp. 547-48.
150
Deterrence and Defense
to signal an intent greater than simply "probing" or "piecemeal nibbling" and to produce the other complementary deterrent effects previously described. But beyond this, no further clarification of intent or provocation or "emotional mobilization" would be required; a major attack would be under way, obviously intended to overrun the whole of Western Europe. In this view, it would not matter much whether the shield were 30 or 60 divisions strong. The Soviet estimate of the chances of massive retaliation would be roughly the same in either case and if the Soviets decided these chances were low enough, they would simply attack with the forces necessary to overwhelm the shield, whatever its size. The logical conclusion of this argument is that if the shield forces are large enough to force the enemy to disclose a major intent and to create a certain minimum degree of provocation, there is not much point in increasing the shield further unless it is enlarged to the point where it can defend Western Europe against the full force of the Red Army. Whether it should be increased to the latter level depends on the presumed credibility of the retaliatory threat. If the nuclear threat's credibility declines below a certain threshold, our valuation of Europe in conjunction with the increased probability of Soviet attack presents a risk which is more onerous than the cost of providing the means for effective defense on the ground. As long as the credibility of the threat stays above this threshold, the costs of pro viding a ground defense equal to the task of defeating the whole Red Army more than offsets the risk involved in continuing to rely on the nuclear deterrent, and the only ground forces required are those necessary to complement the threat of massive retaliation. The higher the cost of a sufficient ground defense capability, the lower the thresh old of presumed credibility which will be thought sufficient to war rant continued reliance on the nuclear deterrent. But once the threshold is reached, the only rational course is to provide forces to defend against a full-scale Soviet ground attack—unless, of course, measures are taken to shore up the credibility of the massive threat by an in crease in counterforce capabilities or by the diffusion of nuclear forces to the countries directly subject to ground attack—measures which might maintain credibility above the threshold. Let us suppose that the credibility of the massive retaliation threat has dropped so low as to lose all its deterrent effectiveness, and that a full-scale ground defense capability is therefore needed. How large would such a defense force have to be?
in Western Europe
151
It is clear that the present NATO ground forces would have little hope of defending successfully against a Soviet maximum attack. The Russians are reported to have about 175 army divisions, of which per haps 140 are combat-ready, with £0 or 22 in East Germany, perhaps eight in Polandf Hungary, and Rumania, and 75 in Western Russia. They could mobilize a total of up to 300 divisions in thirty days.23 The East European satellites have perhaps 40 or 45 divisions of question able strength and loyalty. Compared with these figures, NATO ground strength looks puny in deed. Along the Elbe on the "central front" are deployed nominally about 21 divisions from several NATO countries, including five United States divisions.24 However, their effective combat strength is probably the equivalent of about 16 divisions because of undermanning and various inefficiencies associated with their multinational make-up and non-standardization of equipment. Aside from the shield forces, Turkey maintains an army of some 20 divisions, France has about 10 divisions in the Metropole and Algeria which presumably would be available for NATO use in case of a large-scale Soviet attack, Italy maintains six divi sions under NATO command plus three or four understrength divisions, Denmark and Norway have perhaps the equivalent of a division be tween them, Greece has about six divisions, and Portugal has one. Great Britain maintains the equivalent of about three divisions at home and at various posts in the Empire. Practically all the active ground forces of the other European NATO countries—Great Britain, the Netherlands, Belgium, Luxemburg, and West Germany—are included in the central-front shield. In addition, there are three divisions in the United States (the Strategic Army Command) which could be avail able very quickly for transportation to Europe, plus three more under strength divisions. Whether the two United States divisions in Korea 23 The
figures are from Hanson Baldwin, The Great Arms Race, New York: Frederick A. Praeger, 1958, p. 36. Baldwin puts the number of Soviet front-line divisions in East Germany at 22 (plus 8 or 9 artillery and anti-aircraft divisions), but the figure may have dropped closer to 20 after recently announced troop reductions in the Soviet forces in East Germany. According to Alastair Buchan, although NATO officially calculates the Soviet Army strength at 175 divisions, only 140 of these are operational. Alastair Buchan, NATO in the 1960's, London: Wiedenfeld and Nicolson, 1959, p. 10. 24The country-by-country breakdown of the shield's forces is as follows: 5 U.S. divisions (plus 3 armored regiments); 3 British divisions; 2 French divi sions; 2 Belgian divisions; 2 Dutch divisions; 7 German divisions; and one Canadian brigade.
152
Deterrence and Defense
and the one division in Hawaii would be available for quick action in Europe would depend on whether we were willing to give up these positions in the event of a European war, and whether sufficient air transport were available. The U.S. Marines, with about three divi sions, might possibly make a contribution. Leaving out the United States forces in the Far East, this adds up to a rough total of about 70 active divisions. Information on the inactive reserve forces of the European NATO countries is hard to come by, but one writer estimates that under present programs the Allies could mobilize only six or eight divisions for reinforcing the central front within two or three weeks.25 Of course, the important comparison is between the forces which both sides would have available for action on the central front in West Germany. The Turkish, Greek, Norwegian, and Danish forces probably would not be available since they would be needed for guarding their own borders, although of course in doing so they presumably would immobilize an equivalent number of Russian divisions. The availability of the Italian forces would depend considerably on whether Yugoslavia fought on the Soviet side, the NATO side, or took a neutral position. Subtracting the forces of these five countries would leave something like 30 or 35 divisions available for action on the central front. Since the Russians presumably would not begin their attack before moving sub stantial additional forces into East Germany—a movement which would provide warning of impending attack—there probably would be time to bring the French forces back from Algeria, to bring in the three combat-ready divisions now in the United States, if an adequate airlift were available, and perhaps to mobilize some reserve forces in Europe.26 This picture is certainly not reassuring, especially when one con siders that the Russian forces apparently have made much more progress in modernization of conventional types of equipment since 25 F.
O. Miksche, op.cit., p. 58. Hilsman estimates that the Soviets would require at least thirty days to move 30 or 40 divisions into East Germany, and that this would provide NATO with at least three weeks to mobilize additional forces. See his "NATO: The Developing Strategic Context," in Klaus Knorr (ed.), op.cit., p. 33. Hilsman's estimate coincides roughly with that of Captain B. H. Liddell Hart, who also states that under present NATO arrangements the shield could be increased to 40 divisions in thirty days. B. H. Liddell Hart, "Shield Forces for NATO," The Guardian, February 16 and 17, I960; reprinted in Survival, Vol. n, No. 3 (MayJune I960), p. 108. 26 Roger
in Western Europe
153
World War II than have the Western allies. However, there are some offsetting considerations. There is an inherent advantage in the de fense, so that a defensive force can probably hold against a consider ably larger force.27 One long-time British student of military affairs, Captain B. H. Liddell Hart, suggests that a mobile and well-trained defensive force two-thirds the size of the attacking force could hold the line on the central front if about half of the defensive force were employed as a mobile reserve for quick counteraction against enemy breakthroughs. He indicates that a central-front shield of 26 divisions (13 as mobile reserves behind the front line) could hold against a Soviet force in East Germany which might be increased to 40 divisions within ten days.28 The Soviets would not be able to use all their forces on the European central front, since they would have to reserve some to guard various exposed points on their periphery. A big imponderable would be the attitude of Communist China. It is conceivable that China would join Russia in fighting the European war. Much more likely, however, is that China would busy herself picking up real estate in the Far East and South and Southeast Asia. If China thus fought an independent war of her own, she might pose a threat to Siberia which the Soviets would have to reserve troops to guard against. The reliability of the forces of the East European satellites is ques tionable, judging by the behavior of the Hungarian Army during the revolt of 1956. However, the troops of the more loyal satellites, such as Czechoslovakia and Bulgaria, probably would supplement the Soviet offensive. All told, the Soviets probably would have at least 100 active divi sions available for action on the central front, and they could mobilize great masses of reserves in rather short order. Thus, even if, following Liddell Hart, NATO's active shield could hold against the 40 divisions which the Russians could quickly bring to the front line, there remains the problem of dealing with the reinforcements which would be steadily coming in. The obvious maximum requirement for NATO is some combination of additional active forces plus a reserve mobiliza27 Malcolm Hoag has suggested that the advantage of the defense would be increased if we were to exploit the full potentialities of modern technology in our conventional forces—for example, by developing guided anti-tank missiles that would strike specifically at the weapon which is strongest in an ofFensive role. Malcolm Hoag, "What Interdependence for NATO?" loc.cit., p. 388. 28 Captain B. H. Liddell Hart, "The Ratio of Troops to Space," Military Review, Vol. XL, No. 1 (April I960), p. 14.
154
Deterrence and Defense
tion capability which could match the Russian build-up and rate of transport to the front lines. Again accepting the Liddell Hart 2:3 ratio of defense to offense, and assuming the Russians could build up a front-line force of 100 divisions in sixty days, this would mean a capacity to raise and bring to the front some 35 or 40 additional NATO combat-ready divisions in that time, plus an ability to approach the Russian rate of mobilization of reserves. This might be considered the requirement for carrying out NATO's "forward strategy" of holding at the eastern frontier. A strategy of "trading space for time"—accepting early Russian gains to be recouped later by a counteroffensive—would pose lesser initial requirements. However, it probably would mean higher total requirements, and higher war costs, because when NATO did take the counteroffensive, the defense-offensive ratio would favor the Russians, the war would last longer, and more casualties and dam age would be inflicted on the European civilian populations and economic assets. The choice rests mainly on the question whether the peacetime economies of maintaining fewer active forces and a lesser degree of readiness in the reserves would justify these higher costs in war. The assumed probability of war under either alternative would also affect the choice; the lower the chances of war, the lesser the degree of readiness that would be deemed acceptable—assuming of course, that NATO would be able to mobilize sufficient forces to halt the Russian offensive and begin a successful counterattack before Europe was overrun. To produce a ground force of the necessary strength and readiness to carry out either of these strategies is certainly not beyond the re sources of the West, contrary to the assertion often heard that the West cannot hope to match the Russians "man for man." The Atlantic Pact countries have a total population of more than 440 million as against about 300 million in the Soviet Union and the Eastern Euro pean satellites, excluding Yugoslavia.29 NATO's industrial capacity is about three times that of the Soviet bloc. The European NATO countries alone are not far behind the Warsaw Pact countries in popu lation and their industrial potential probably is greater. An even more striking fact is that the total number of men presently under arms in the NATO countries is larger than the active military manpower of the Warsaw Pact countries. On January 1, 1958, the Soviet bloc countries (not counting China) had just over 5 million 29 Encyclopedia
Britanmca, Book of the Year, 1960, p. 63.
in Western Europe
155
men in their armed forces, while the NATO countries had slightly over 6 million.30 Of course, a larger percentage of the Western allies' man power is employed in navies and air forces. In army manpower alone, the Warsaw Pact countries do exceed the NATO countries, but not by a large margin—4,091,000 to 3,162,40031—and the first figure in cludes 2,800,000 for the Soviet Union, which probably is too large in view of recently announced reductions. If certain neutrals, such as Sweden, Finland, and Spain (whose neutrality in case of an all-out Soviet invasion of Western Europe might be difficult to maintain), are included on the NATO side, Western European active ground forces exceed those of the Warsaw Pact. This wide discrepancy between the comparison of the fighting man power on both sides and the comparison of numbers of combat divi sions is remarkable. No doubt a large part of it can be accounted for by the much greater proportion of the Western forces which is devoted to logistics and administrative functions. This in turn may be due chiefly to the much higher standard of living which the Western soldier requires, although some of it may also be attributed to the complexities of coordinating armies of several sovereign states—as compared with the more monolithic command arrangements which are possible on the Soviet side. Mijksche estimates that without in creasing either military expenditures or men under arms, and by re ducing the standard of comfort of the Western soldier (although not down to the Soviet level), NATO active ground strength in Western Europe could be increased to 60 or 70 divisions.32 If this estimate is correct, it would seem that the minimum active forces necessary for an effective defense of Western Europe might be produced by austerity and reorganization measures alone. Whether such measures would be feasible politically is another matter; probably they would not be to the extent suggested above. Nevertheless, there would seem to be significant opportunities for increasing the "division slice" obtainable from present defense budgets and military man power, opportunities which would at least ease the requirements for additional build-up of active forces. Reducing the tremendous rate of turnover in the Western armed forces, by increasing required terms of service for conscripts and en couraging voluntary long-term enlistments, would also increase the 30 F.
O. Miksche, op.cit., pp. 62-63. Britannica, loc.cit. 32 F. O. Miksche, op.cit., p. 65. 31 Encyclopedia
156
Deterrence and Defense
combat effectiveness realizable from given manpower levels. It would do so in two ways: by reducing the proportion of the troops that were engaged in training, and by increasing the fighting effectiveness of those in combat units.38 There would still be a requirement for approaching the Soviet standard in speed of mobilization of reserves, and in modernity of con ventional armament. Obviously, a considerable increase in numbers of reserves and in their degree of readiness would be necessary to compete with the Soviet capability in this department. It would be a costly program, but of course not as costly as maintaining an equivalent number of active forces, and it would be in line with the Western tradition of the citizen soldier. Reserves necessary in the first two or three months of the war probably would have to come from Europe, although the requirement would be reduced by the number of active U.S. divisions which could be brought in from the United States. After this period, reserves mobilized in the United States might become available. Naval forces and tactical airpower would also have significant roles to play, of course, in a European war fought with conventional weap ons. Upon the United States and British navies would fall the major burden of transporting reinforcements of both men and supplies after the first few weeks of the war. The very large Soviet submarine force would present a serious obstacle to the performance of this function. In view of this threat, perhaps serious consideration should be given to the stockpiling of some equipment in Europe, plus much greater reliance on air transport for troop reinforcement, to take the strain off the sea-lift, and to hedge against a possible underestimation of the submarine threat or overestimation of the Navy's anti-submarine capability. The United States Army already maintains considerable "mobilization reserves" of equipment. Is there any reason why some of these reserves could not be maintained in Europe rather than in the United States? To do so would not only ease the supply problem dur ing the war but would also provide an additional earnest to the Europeans of our intention to fulfill our alliance obligations. 33 Liddell Hart estimates that a shield force of 13 regular divisions composed of long-service troops would be better able to check a sudden attack by the Soviet forces now in East Germany than the present shield force of 21 divisions, which is handicapped by its large proportion of short-term conscripts. See his "Shield Forces for NATO," loc.cit., p. 108.
in Western Europe
157
Tactical airpower in Europe would provide an offset to the Soviet submarine capability by interdiction of Soviet supply concentrations and transport routes. Alastair Buchan has suggested that the greatly increased mobility of the Soviet Army since World War II may have its favorable side for the West: by restricting the Army's movements to established roads, it may have made it much more vulnerable to interdiction by air than the less mechanized Soviet ground forces of World War II.34 3. INDEPENDENT STRATEGIC NUCLEAR FORCES FOR NATO COUNTRIES35 Analysis of the consequences of the spread of nuclear weapons technology to additional countries is often implicitly prejudged in a negative direction by referring to it as a "problem." It used to be called the "fourth-country problem"—now with the addition of France to novitiate membership in the nuclear club, and the realization that many countries are at least technically capable of producing atomic weapons, it is referred to as the "nth-country problem." Designation of the question as a "problem" is usually premised on the idea that the disadvantages and dangers for world peace and Western security which would follow from the widespread distribution of nuclear weapons far outweigh any conceivable benefits which might accrue. The problem is seen as one of preventing the spread of such weapons. Here we shall approach the question a bit more positively. While recognizing that nuclear diffusion does have its risks, we shall ask what benefits to United States and NATO security might follow from a wider distribution of nuclear capabilities within the alliance. Such distribution would attempt to solve NATO's strategic dilemma by in creasing the deterrent potency of the alliance rather than by increasing its capacity for defense. The most frequently mentioned incentives for the Europeans to de velop their own strategic nuclear capabilities are the following: (a) to obtain a deterrent against ground invasion as a substitute for, or com plement to, the threat of U.S. retaliation as a deterrent to this contin34 Alastair
Buchan, op.cit., p. 11. In the preparation of this section and the one following, I have profited from reading a manuscript by Klaus Knorr, "On the Organization of Nuclear Deterrent Power in the West," mimeographed, 1960, published under the title, "The Future of Western Deterrent Power: II. A View from the United States," in Bulletin of the Atomic Scientists, Vol. xvi, No. 7 (September I960), pp. 271ff. 35
158
Deterrence and Defense
gency; (b) to obtain a counter to Soviet attempts at "nuclear black mail" by providing a capacity to retaliate in case the blackmail threat is carried out; (c) to strengthen their abilities to pursue national in terests which do not coincide with the agreed interests and aims of the whole alliance, and which the alliance, particularly the United States, is not willing to support; (d) to achieve greater bargaining power vis-a-vis the United States and other allies in determination of the alli ance's goals, methods of implementing goals, and sharing of burdens; (e) to obtain a means to national security which is believed to be cheaper, both economically and politically, than the alternative of main taining large conventional defense forces; (f) to reduce a humiliating dependence on the United States; and (g) to enhance the general prestige and political status of the nation.38 Of these incentives, the first two—providing a deterrent against conventional invasion and nuclear attack or blackmail—are the ones which coincide most closely with the national interests of the United States and the common interests of NATO as a whole. What contribu tion would a system of independent European deterrents make to these 36 This list is similar to the one given by Robert E. Osgood in his excellent essay, "NATO: Problems of Security and Collaboration," American Political Sci ence Review, Vol. LIV, NO. 1 (March 1960), p. 124. Of course, there are also some disincentives from the Europeans' point of view. To have a nuclear capability probably would make them targets in an all-out war which otherwise (they might hope) might be carried on, at least at the thermo nuclear level, as a duel between the United States and the Soviet Union. No doubt many thoughtful Europeans recognize the dangers inherent in the further spread of nuclear capabilities, especially the risk that their spread to Europe might stimulate a desire to have them by governments much less responsible than their own. The strength of pacifist opposition in some countries might create a serious internal political problem. And some anxieties may exist that if the Euro pean countries were to acquire a strategic nuclear capability of their own, the United States might then take less seriously its obligations to NATO and retire to a "Fortress America" concept of defense, especially when it has enough ICBM's to make unnecessary the maintenance of bases in Europe for its own security. It is rather significant that no European country besides Great Britain and France has declared a firm intention or desire to have strategic nuclear capabilities of its own, although some (notably Switzerland, Sweden, and West Germany) are considering the acquisition of tactical nuclear capabilities. Turkey and Italy, of course, have U.S. ICBM's on their soil, but these are not independently con trolled, since the United States controls the warheads. Some countries, such as Denmark, have indicated a positive desire not to have nuclear weapons. Thus, there may not develop, in the foreseeable future, any strong clamor for new memberships in the nuclear club. Great Britain seems to be in the process of demoting herself to associate membership, and France may decide to resign its membership if a post-de Gaulle government finds the dues too high.
in Western Europe
159
objectives, and what would be the requirements for a useful contribu tion? What are the countervailing risks and disadvantages? We shall assume, as a basis for discussing these questions, that the United States has a secure strike-back capability sufficient to deter direct attack on itself, but not a first-strike counterforce capability which is adequate by itself to deter, with high confidence, Soviet tacti cal aggression or nuclear blackmail in Europe. However, we shall also assume that the Soviets have not written off entirely the possibility of U.S. retaliation, so that it still poses some small risk which they must consider in addition to the new risks which European-controlled nuclear forces would create for them. First, it is essential to recognize that independent strategic nuclear forces for any European country would have only a deterrent function. They would be worse than useless as a means of defense should deter rence fail, since no European country could hope to develop a strategic force which could defeat the Soviet Union in all-out war; their actual use in case of ground invasion would be rational only if the country preferred obliteration plus occupation to occupation; their use after nuclear attack would precipitate more nuclear damage to the victim and could not prevent its occupation. Hence, the deterrent value of these weapons rests on some Soviet apprehension that they might be used irrationally, accidentally, or perhaps because some countries might in fact prefer to be "dead than Red." This limitation of nuclear weapons to a strictly deterrent function perhaps is more clearly recog nized by thoughtful Europeans than it is in the United States; as we pointed out in the previous chapter, the much larger American nuclear capability gives some plausibility to the idea of "defeating" the Soviet Union in all-out war, although of course at great cost. By contrast, the British White Papers on Defense for 1957 and 1958, it will be remem bered, explicitly recognized that Britain was indefensible in all-out war and yet posited the need for an independent British strategic force as a deterrent, even against Soviet conventional ground attack on the Continent. From the European point of view, the danger inherent in the declin ing credibility of the American threat to retaliate in case of Soviet ground attack on Western Europe is not that the Americans will not retaliate, but that the Russians will come to believe they will not. For many Europeans, this fear may in fact be compounded by the fear that the United States will nevertheless retaliate, after the Soviets have at-
160
Deterrence and Defense
tacked on the ground in the miscalculation that the United States would not respond. In other words, it would be completely reasonable for a European to prefer that the United States not retaliate after his country is invaded for fear his own country would inevitably be caught up in the nuclear holocaust. If we assume the Europeans are subject to this paradoxical dual anxiety, then they would have two apparently contradictory motives for acquiring their own separate deterrent forces, especially when the contingency to be deterred is that of ground invasion. One would be to strengthen deterrence by presenting the Russians with an extra risk of a nuclear riposte which would supplement and possibly even strength en the risk already present in SAC. The other would be to reduce the chances of all-out war should deterrence fail and ground invasion take place. They might expect to accomplish the latter aim through the oper ation of a familiar theme in the American approach to foreign policy— that we will help those who show a willingness to help themselves. Then, by not retaliating themselves, they might release America from her moral obligation to retaliate. But they would hope, of course, that the Russians were either so obtuse, or so uncertain about the American psychological reaction, that they would not dare to gamble that the U.S. threat to retaliate had been negated by the establishment of inde pendent European strategic forces. These two motives are not really contradictory but merely manifestations of the central dilemma of the nuclear missile age: how to pose credibly the greatest possible prospect of cost for the enemy, thus reducing the chances of war, while at the same time minimizing one's own prospective costs should war never theless occur. To deter ground invasion by its own efforts, a single European country would need at least enough strategic nuclear forces to inflict destruction on Soviet cities more than equivalent to the country's post war value to the Soviet Union. This might be called a "minimum firststrike punitive capability." Of course, it would be completely unrealistic for any European country to expect to have a first-strike counterforce capability of any significance against the massively preponderant forces of the Soviet Union. But the threat of a strictly punitive retaliation with only such a minimum force probably would not seem very credible to the Soviets, since it could not prevent the country's occupation and could not prevent severe counterretaliation by the Soviet Union. Con sidering the possibility of irrationality and accident, the Soviets prob-
in Western Europe
161
ably would not discount the threat entirely, however. And they would have to consider a possibility that the European country might gamble on the chance that the Soviets would not counterretaliate, reasoning that the Soviets would not wish to damage assets which they expected to use themselves, and would wish to avoid the risk of triggering U.S. retaliation. If, in the Soviet calculations, severity of possible damage were substitutable for credibility, the European country might have an effective deterrent against ground attack if it could threaten consid erably more than minimum unacceptable damage—enough more to offset the Soviets' discounting of the threat. This criterion might pose excessive requirements, however, if the credibility of United States massive retaliation to ground aggression in Europe had not dropped to zero. As long as the Soviets fear some chance of U.S. retaliation, each European nuclear power need have only enough punitive capability so that the risk posed by its threat, when added to the continuing small risk of U.S. retaliation, presents the Soviets with an over-all risk which they cannot afford to accept. The greater the Soviet expectation of U.S. retaliation—which might depend on the apparent size of the U.S. first-strike counterforce capability—the smaller the forces which a European ally would need to push the Soviets' risk-cost calculation over the threshold of unacceptability. It is entirely possible, of course, that the Soviets would consider the risk of retaliation by the European country invaded to be less than the risk of a U.S. first strike. It does stretch the imagination to believe that a country would deliberately invite its own nuclear obliteration by an act which could not forestall its own conquest and occupation. The United States, not having experienced the horrors of war as vividly and as intimately as the Europeans, and perhaps feeling a strong moral obligation to honor threats and commitments, might in fact be more likely to retaliate. But this is of no consequence. It does not matter in the Soviet reckoning which set of risks and possible costs is the largest; it is their sum which creates the over-all deterrent effect. It is rather crucial, however, that the possession of an independent nu clear deterrent by a prospective victim of ground attack not undermine the credibility of a U.S. response. It is possible that the creation of inde pendent European capabilities might instill an attitude of "Let George fight his own battles" in the United States. If this effect occurred—more precisely, if the Russians believed it had occurred—the establishment of independent deterrents in Europe might cause a net reduction in
162
Deterrence and Defense
Western security. The very fact that the Europeans had acquired independent strategic forces would show that they doubted the U.S. intent to retaliate and would reinforce the Soviets' own doubts. How ever, it seems rather unlikely that the Russians would dare to base their actions on such speculative considerations. To be confident of deterring a nuclear attack on itself, a European country would need a minimum strike-back punitive capability—i.e., strategic forces which could absorb a counterforce blow and still inflict enough damage in retaliation to more than offset the Soviets' postwar valuation of its conquest. The forces for such a capability would have to be highly invulnerable and/or numerous. Missile-firing submarines, mobile land-based missiles, and perhaps some missile-firing aircraft on continuous airborne alert, or able to go on such alert status in periods of political crisis, would seem the best types of weapons; fixed-base missiles (unless they were extremely "hard") would be too vulnerable and aircraft on the ground would be too likely to be caught at their bases by a surprise missile strike, considering that warning time could not be much more than five minutes. Non-missile-firing aircraft which did survive the surprise attack probably would not be able to penetrate the enemy's air defenses in sufficient numbers, unless gaps could first be opened up in those defenses by missile attacks on the defense installations. As in the case of ground attack, the deterrent effect of the victim's strike-back capability, or of Soviet uncertainties about whether this capability could be eliminated in the Soviets' initial strike, would be supplemented by the continuing possibility of U.S. retaliation. In contemplating a nuclear attack on a European country, the Rus sians would have to give serious consideration to the possibility that the attack would "trigger" U.S. retaliation. Two types of possible trig gering are worth noting. One type might be called psychological or "emotional." A nuclear attack on a European ally of the United States would be extremely provocative. Even though the United States did not have sufficient first-strike counterforce capability to be willing to retaliate after ground invasion of an ally's territory, the emotional shock of a cold-blooded nuclear attack might stimulate the United States to respond. This shock would presumably be the greater, the more power ful the attack the Soviets were forced to launch to eliminate the ally's strike-back capability. Thus, if this triggering effect is to be relied on for deterrence, the nuclear ally might be well advised to base its forces
in Western Europe
168
on its own territory rather than on submarines at sea; the provocative effect of massive area attacks to eliminate mobile land-based missiles, and of high-megaton bursts to destroy hard-based sites, would be much greater than attacks on submarines. Here is another case where the logic of deterrence may conflict with the objective of minimizing war costs. Another type of triggering would be "attritional"—that which might result from the expenditure of Soviet forces in attacking a European nuclear country. That is, if the ally's forces were numerous, dispersed, and well protected, the Soviets might conceivably use up enough of their forces on their first strike so that the United States would then have counterforce capabilities sufficient to destroy enough of the re maining Soviet forces to make tolerable a Soviet retaliatory strike on the United States. This triggering effect, too, would depend on the ally's forces being land-based; the elimination of a submarine force (assuming that this were possible) would not cause significant attrition of the Soviet forces. The attritional triggering effect might be quite powerful if the Soviets were forced to fire several missiles at any particular target in Europe to be sure of eliminating it. Any number of missiles thus ex pended by the Soviet Union would reduce the United States first-strike missile requirements by some multiple of the number expended. Thus, the cheapest way for the United States to obtain a sufficient or substan tial first-strike capability with respect to a Soviet nuclear missile attack on Europe might be to deploy in Europe a portion of its forces above its own minimum deterrent requirement. Any given number of missiles de ployed in Europe probably could cause the attrition of a much greater number of Soviet attacking missiles than could the same number of missiles deployed in the United States and fired on a U.S. first strike.37 These two triggering effects would be additive, of course. Both of them, taken together, might add up, in the Soviet mind, to a consider able increase in the probability of U.S. retaliation over what this proba bility would be in the case of ground invasion alone. 37 Missiles deployed in Europe also would decrease the U.S. first-strike require ment for deterrence of ground attack (because of the greater accuracy of missiles at short range), if the United States retained control over the firing of the missiles, or if the degree of integrated planning between the United States and European powers was sufficient to guarantee or create a high probability in Soviet calcula tions that all the alliance's forces would be fired simultaneously in response to a major ground invasion.
164
Deterrence and Defense
Both triggering effects might also occur after a ground invasion, if the invaded ally responded with a nuclear attack on the Soviet Union and the latter replied in kind. In this case, the attrition of the Soviet forces would be that which was caused by the counterforce strike of the ally. The emotional triggering effect would be greater in this case if the Soviets found few counterforce targets for their return blow and struck at cities in reprisal. Of course, it would not be rational for the European country to launch a nuclear retaliatory blow after being invaded, in the expecta tion of triggering a U.S. response, unless it preferred nuclear devasta tion, plus a problematical chance of preserving its independence, to occupation with little or no damage. But it cannot be too often repeated that the Soviets can never be sure of the degree of rationality in their victim. As long as there were any significant chance of a Soviet move causing a nuclear exchange between itself and its victim, the Soviets would be restrained by the possibility that the exchange would danger ously increase the chances of a U.S. response. If the Soviets took seriously the danger of triggering U.S. retaliation, they might prefer a massive attack on all the nuclear forces of the en tire NATO alliance, including the United States, to the risks of attempt ing piecemeal gains. But if the European increment to the allied striking capacity made all-out war a condition of successful aggression, this in itself would be a potent deterrent—if the alliance possessed a powerful strike-back capability. The deterrent potency of the triggering effects is directly related to the size of the first-strike counterforce capability of the United States. The greater this capability, the less cost the United States would have to accept in letting itself be triggered, and the more likely the emotion al and moral provocation resulting from a Soviet nuclear attack in Europe would override the inhibiting effect of the cost expectation. Also, the greater the U.S. first-strike counterforce capability, the lesser the amount of attrition of Soviet forces in the European nuclear ex change which would be necessary to make a U.S. first strike rational. What of the utility of independent European strategic forces for resisting "nuclear blackmail"? Of course, this is one of the primary reasons the European countries want and need their own nuclear forces; even if their ground forces were adequate to deal with any de termined Soviet tactical thrust, they still would need an anti-blackmail
in Western Europe
165
capability if the U.S. deterrent capability was not a credible instru ment for this purpose. A country's capacity to resist blackmail threats depends on the as sumed credibility of the threat, which in turn depends on the threatened country's ability to retaliate should the Soviet threat be carried out. Of course, nuclear blackmail threats may assume any degree of sever ity and the demands may range from small political concessions to com plete capitulation. But, presumably, if the Soviets were to make good on their threat, they would do so by attacking all the nuclear forces of the country being blackmailed, so as to minimize its capacity to re taliate. If so, a capability to resist blackmail might be defined as a strikeback capability which could inflict deprivations on the Soviet Union serious enough to offset the Soviets' valuation of the acceptance of their demands. If it did come to the point where the Soviets' threat had to be implemented, however, the Soviet aim might expand to the complete conquest of the victim. Thus, a high-confidence capability to resist blackmail is equivalent to the forces needed to deter an all-out Soviet counterforce attack with the object of conquest. However, the problem is not quite this simple. The blackmailed country might prefer giving in to the Soviet demands rather than accept even a small risk of nuclear obliteration in case the demands were rejected. The bargaining would favor the Soviets, since they would have the means to devastate their victim completely, while the latter would be able only to threaten "unacceptable" damage. Moreover, even though the victim did possess a force invulnerable enough to be able to strike back severely, the Soviets might be willing to accept this damage because the gains which they would make from demonstrating a willingness to carry out their blackmail threat would ramify far be yond the value of the concessions demanded of the immediate victim. The resolution of other countries to resist blackmail threats would be seriously undermined by the Soviet demonstration that its threats were not bluffs.38 There are further complexities. Suppose the Soviet threat were a minimal one—for example, a threat to drop a nuclear bomb on an airbase being used by U.S. strategic forces unless the host country ejected those forces from the base. The Soviets might reason that if the threat were carried out, the victim would not retaliate with all of its nuclear 38 This point is made by Morton A. Kaplan, "Problems of Coalition and De terrence," in Klaus Knorr (ed.), op.cit., p. 34.
166
Deterrence and Defense
forces for fear of leaving itself defenseless against future blackmail or ground invasion. The more likely response, the Soviets might reason, would be a "tit-for-tat" retaliation against a Soviet airbase. The Soviets might be willing to accept this as the price of demonstrating to other countries its determination to carry out such threats. Of course, the Soviets would also have in the back of their minds the possibility of some limited U.S. retaliation. This possibility would exist so long as the United States had somewhat more than enough striking power to deter an all-out attack on itself. But the Soviets might not worry very much about this possibility, since they would also have the capacity to carry out a tit-for-tat limited counterretaliation against the United States, the expectation of which might very well deter the United States from acting. The upshot seems to be that an anti-blackmail requirement cannot be stated with much precision. A small capability, which the Soviets would be able to wipe out if they chose, might considerably stiffen the Europeans' will to stand up to blackmail, if they believed the Russians would not take the risk that a counterforce attack on the European victim would trigger a United States riposte. A larger or better-protected force, part of which could survive a Soviet counterforce blow, would further increase the Europeans' resolve. On the other hand, even such a substantial capability might not be sufficient to resist relatively minor demands—such as the denial of the use of European bases by U.S. forces, or ceasing military activity against some nonEuropean country—and it might not deter the Soviets from carrying out a massive counterforce blow if they placed a very high value on the demonstration effect of carrying out their threat. There can be little doubt, however, that the possession of some amount of strategic nuclear capabilities would strengthen the Europeans' resistance to blackmail. In their triggering role, independent European nuclear forces would help to forge the "link" between the independent deterrent qualities of the ground force shield and the deterrent effect of the American and British massive retaliatory power. As we pointed out earlier, perhaps the most important deterrent function of the shield is to raise the stakes in a ground conflict to the point where the nuclear deterrent power of the extra-Continental countries becomes implicated with significant credibility. The "links" we are talking about are the ground force shield used conventionally, the tactical nuclear capabilities of the shield, strategic
in Western Europe
167
nuclear forces controlled by the Continental European countries, and the strategic forces of the United States and Great Britain. Independent strategic forces on the Continent fit most naturally between the tactical nuclear and the American-British strategic links. The outbreak of tacti cal nuclear warfare would be more likely, perhaps, to touch off stra tegic retaliation by countries in which the tactical nuclear war was being fought than by American and British forces. Countries in the theater of war would already be suffering severely as a result of the tactical nuclear warfare; they might view the additional costs from a strategic exchange with the Soviet Union as only marginal, and might see an all-out (American-British) strategic attack on the Soviet Union, which they would expect to trigger, as their last chance for avoiding political extinction. Britain and America, themselves untouched by nuclear warfare, might feel great inhibition against moving from the tactical nuclear level to the strategic level, even if the allies were suffer ing defeat, but would lose this restraint once the Soviets started drop ping hydrogen bombs on France and Germany. Strategic weapons in the hands of Continental European countries might also contribute importantly to deterring the Soviets from initiat ing atomic warfare on the battlefield. In making such a move, the Rus sians might feel compelled to strike simultaneously at the strategic forces of the battlefield countries, thus incurring a serious risk of retali ation by the United States and Britain. Of course it is the Soviets' advance contemplation of this logic which is important for deterrence, not our own. If they think one or more of the links are weak or do not interlock with the others, they may be will ing to take large risks. But if they do recognize the interlocking—if they believe with high confidence, in other words, that at whatever level they initiate the fighting, the Western powers are substantially com mitted by the logic of their own interests, or by their pre-war planning, to keep raising the level of intensity before accepting defeat until the ultimate level of all-out war is reached—there is only one logical alter native for the Soviets to keeping the peace. That is to initiate all-out war with a massive counterforce strike at the very beginning. Limited aggression with limited means, in an attempt to make repeated gains while avoiding a degree of provocation which would touch off all-out war, would then seem a thin possibility and a dangerous strategy. It would be the Soviets, then, who would be caught in the dilemma of "all or nothing." Unless, by invoking "all" they were able to eliminate
168
Deterrence and Defense
enough of the West's strategic forces to make their own probable costs in all-out war commensurate with their expected gains, they would presumably choose "nothing." Significantly, the bridging of any gaps in the spectrum of deterrence —gaps which lie between a weak shield and the massive retaliatory capability of the peripheral powers—does not crucially depend on the creation of a capability to fight and win a limited war, at any level. It depends only on the creation of forces under a system of deployment and control such that (1) a Soviet attack at any level below all-out war will provoke at least one of the allies to "up the ante" to a higher level of intensity; that (2) if the Soviets match this bid, this will provoke a "raise" by the same or other allies; and that (3) unless the Soviets "fold" at some level, such raises will be continued up to the level of allout war. To be more precise, it depends only on a Soviet belief that such a sequence is likely to occur. Note that there is an inherent contradiction between this logic and the logic of limited war. It is the essence of limited war logic that de feat is believed preferable to warfare beyond some ceiling of war in tensity. In limited war, we want to communicate to the enemy that we intend to stay below this ceiling, in the hope that he will do so as well. We do not want to tell him that at whatever level he initiates the war, we will be forced to move to a higher level, and eventually to the ultimate level of violence. For example, we do not want the enemy to believe that once we have crossed the threshold between limited con ventional and limited nuclear war, we are prepared to cross further thresholds—i.e., to erode the limitations progressively—until he calls it quits. Rather, we wish to instill a belief in him that we recognize ceil ings in limited war—and identify them for him—which we will not violate even if the only alternative is defeat. Whether we choose to follow the linking logic rather than the limited war logic depends on how strong we consider the deterrent effect of the former and on whether, if deterrence fails, we believe the enemy will accept a negoti ated settlement acceptable to us, well before the level of total violence is reached. Against the deterrent advantages of independent European strategic forces must be set certain risks and disadvantages associated with the diffusion of nuclear weapons to the Continental European members of NATO. One of the obvious risks is that of the irresponsible use of nucle ar weapons—that is, their use by a European ally to obtain objectives at
in Western Europe
169
variance with those of the alliance as a whole. It is easy to imagine cases in which the possession of an independent nuclear force would tempt an ally either to use or to threaten to use these weapons against a non-nuclear country in, say, the Middle East or Africa. Or such a force, if it were well protected, might allow a European country to undertake conventional action against a non-European country, with confidence that a Soviet nuclear threat intended to deter such action could be safely ignored. It is said, for example, that one of the motives for the development by the British of their own hydrogen bomb stemmed from their experience in the Suez crisis of 1956, when they discovered that a military action which lacked unequivocal American support could not be continued in the face of a Soviet nuclear threat.39 Dependence on the United States to counter Soviet nuclear threats and capabilities has reduced the interests of the allies which can safely be pursued by force to a sort of common level defined by the interests of the United States. The acquisition of independent deterrents would have a liberating ef fect on an ally, in the sense that some of its interests which were di vergent from those of the United States might be advanced by force or threats of force. However, the danger of irresponsibility can be exaggerated. The European governments would have a sober appreciation of the dangers and possible terrible consequences for themselves if they were to fire their strategic weapons in any contingency short of a direct threat to their own survival or a major Soviet attack on Western Europe. Any attempt to use or threaten to use nuclear weapons for aggressive pur poses not sanctioned by the alliance would be very costly politically. In general, the very possession of such terrible instruments of destruc tion might tend to encourage a sense of responsibility and caution. Paradoxically, the "irresponsible use" of strategic weapons, from the United States point of view, might very well include the triggering function we have discussed as a factor supporting the credibility of the alliance's general massive retaliation threat. There are probably many circumstances in which the United States would prefer not to be triggered—indeed, this is the whole point of the triggering argument: that the United States would initially prefer not to retaliate to some challenges, but could be forced or stimulated to do so by either the actual use or merely the existence of strategic nuclear weapons in the 39 This is not to argue that there were not other factors which contributed to the collapse of the French-British-Israeli campaign against Egypt in 1956.
170
Deterrence and Defense
hands of an ally. Even in the case of a major Soviet ground force attack on West Germany, the United States might prefer to accept defeat in limited war rather than accept the costs of all-out war, but might find itself sustaining them nevertheless as the result of German initiative in firing nuclear long-range missiles at the Soviet Union. Another danger is that the ally possessing an independent deterrent would lose its willingness to contribute to the alliance's ground forces— especially if it had to develop the nuclear deterrent capability itself at great expense. A member might see little profit in helping to defend an ally with ground forces if it felt its own security was assured by its nuclear deterrent strength, and it would certainly not use its nuclear forces strategically and thus risk suicide in the service of another coun try's independence. Indeed, the alliance itself might collapse—in fact, if not in form—as a result of the disintegrating tendencies set in motion by independent strategic forces. Yet this need not happen if the allies keep well in mind that their strategic nuclear forces serve a deterrent purpose only; that in case of a failure of deterrence, the members all have a common in terest in coming to each other's aid for purposes of tactical defense. And, as will be discussed below, the separate strategic forces would have much greater deterrent potency if they were coordinated at least to a degree that would create some risk for the Soviets that an attack on a single country might be followed by retaliation by other countries in addition to the one attacked. The spread of nuclear weaponry to several countries in Western Europe might increase the difficulty of achieving a workable disarma ment or arms control agreement, if only because the more parties to a negotiation, the more interests which must be satisfied. Against this, it might be argued that the United States and the USSR could easily im pose upon other countries any arms control agreement reached between themselves, through the use of various forms of pressure at their dis posal, and that once the USSR had agreed to a policeable disarmament plan, much of the incentive for the West Europeans to have their own nuclear deterrents would have evaporated. Public opinion would also be a potent force working against their possession of these weapons once the superpowers had agreed to renounce theirs. DifiFusion of strategic nuclear power throughout Western Europe might be matched by similar dispersal of the Communist nuclear arse nal to at least the most reliable satellites in the Soviet bloc. Particularly
in Western Europe
171
frightening is the prospect of the magnification of Chinese Communist power and aggressive tendencies through possession of such weapons. The possession of a strategic nuclear capability by some members of NATO might stimulate incentives to acquire them among unaffiliated countries, either for prestige reasons or as insurance against being sub jected to nuclear blackmail by a European country. The Soviets might see an interest in passing out nuclear weapons to a Nasser or even a Castro to increase their nuisance potential vis-a-vis the West. And if Egypt, say, should acquire them, the clamor that would arise from Israel and other countries in the Middle East to have them also can be only too easily imagined. The possibility that establishment of nationally controlled nuclear forces in Europe might set off a competi tion in "nuclear sharing" with other, much more irresponsible, govern ments is certainly a disquieting thought. However remote the possibil ity may be, it deserves sober consideration. An increase in the number of nuclear powers might increase the chances of all-out war by accident. Certainly there would be more chance of accidents occurring, simply because of the greater number of independent control systems and of "fingers on the triggers." On the other hand, there might be less chance that an accidental explosion would touch off an all-out conflict (we have already argued that this probability is small in any case) because the country on whose terri tory the explosion took place could not immediately be sure of its source. During the time period necessary to ascertain the source, it might also be made clear that the explosion was an accident. The possibility of accidental firing might even be a positive benefit from the deterrent point of view; even though the Soviets doubted that a country subjected to invasion would fire off its strategic nuclear forces in response after sober consideration of the probable conse quences, they would have to consider that tensions and emotions might run so high in a war situation that an irresponsible or nervous com mander of a missile crew (perhaps one whose installation was about to be overrun by the enemy's invading forces) would press his button and thereby trigger off all the country's nuclear forces. A. NUCLEAR SHARING
If, on balance, the alliance would obtain a net gain in security by an expansion of the "nuclear club," the question arises as to how this ex pansion is to take place. Obviously, there are two alternatives: inde-
172
Deterrence and Defense
pendent development, and United States sharing of its nuclear strength with its NATO partners. There seems to be no question but that almost all NATO countries have at least the technical capability to produce, in time, their own nuclear weapons. A recent report to the American Academy of Arts and Sciences concludes that twelve countries, including Belgium, France, West Germany, and Italy, could produce their own nuclear weapons in the near future and that fourteen more, including all the rest of the NATO members except Iceland and Portugal, have the basic economic and technical capacity to do so eventually.40 Nevertheless, independent development of nuclear capabilities and delivery vehicles would be extremely costly and time-consuming. If nuclear diffusion is desirable, or if it is believed likely to take place anyway, there is a strong case to be made for the United States to share its nuclear "knowhow" and weapons with its European allies. This would enormously reduce the cost and the time required for the Europeans to acquire significant nuclear forces. The United States could use its bargaining power to channel the development into weapons which would be most desirable from the point of view of the interests of the United States and of the alliance as a whole. It could influence the sharing of burdens of development and production among individual countries according to some principle of equity, perhaps stimulate coordination in targeting, and encourage cooperative planning as to the contingencies in which the weapons ultimately would be used. The United States could make its aid contingent on European fulfillment of their pledges to maintain ground forces under NATO command, or perhaps even demand addi tional ground force contributions.41 U.S. nuclear sharing, by reducing the costs to the Europeans of developing nuclear forces, would of course ease the pressure on them to reduce spending for ground forces. United States nuclear sharing could take several forms: sharing of information on how to make bombs, warheads, and missiles, sending of technical missions of scientists and engineers to guide European de velopment efforts, relaxation of restrictions on the use of by-product plutonium produced by nuclear reactors made available for peaceful purposes under U.S. aid programs, outright gifts or sales of bombs and 40 Howard Simons, "World-wide Capabilities for Production of Nuclear Weap ons," Daedalus, Vol. 88, No. 3 (1959), pp. 385-409. 41 This quid pro quo has been suggested by Arthur Lee Burns. See his "NATO and Nuclear Sharing," in Klaus Knorr (ed.), op.cit., pp. 151-75.
in Western Europe
173
warheads, and gifts or sales of missiles or some of their components. Relaxation of present restrictions on information sharing and the use of by-product plutonium would be costless for the United States. Gifts of bombs and warheads would not reduce our own deterrent capability, since we have a great plenitude of these. It is delivery systems, particu larly missiles, which constitute the greatest bottleneck in the building of a U.S. deterrent force; therefore, any consideration of the transfer of missiles to the Europeans would have to be weighed against the re duction of the U.S. deterrent forces. However, when the Minuteman and Polaris missiles go into quantity production in two or three years, they are expected to be relatively cheap and plentiful; at that time transfers to Europe would not significantly strain the U.S. capacity to provide its own deterrent against direct attack. In the interim period, perhaps the United States could afford to transfer some air-breathing missiles, such as Snark, Matador, and Hound-dog (the latter to be mounted on aircraft), or the Europeans could rely on their own manned bombers placed on a high alert status, perhaps on continuous airborne alert. Although the primary utility of independent strategic forces in Eu rope would be that of deterring Soviet attacks or blackmail in Europe, it is worth passing mention that such forces also would contribute to the deterrence of a Soviet nuclear attack on the United States. An attack on the United States might trigger European as well as United States re taliation. At least the Russians would be taking a serious risk in leaving the European forces undamaged; it seems likely that if they did con template striking first all-out, they would strike at the total nuclear forces of the alliance. The forces in Europe would multiply the targets which the Russians would need to destroy to be confident of coming out of the war with a tolerable degree of destruction. Their air defense problem would be complicated. The very need to strike at numerous European nuclear installations would reduce their gains to be made from attacking, because the unavoidable "by-product" destruction of European economic assets presumably would reduce the value of Europe to them as a prize of war and would burden them, as conquer ors, with an enormous reconstruction problem in addition to the re building of cities and industry in their own homeland.42 Yet it must be 42 If
the Soviets believed they would have to strike at nuclear forces in Europe as well as in the United States in initiating all-out war, they would face the problem of whether to try to hit all targets simultaneously or cross all warning
174
Deterrence and Defense
realized that the European forces would not make a net additional contribution to the deterrence of general all-out war if their creation involved a reduction in U.S. strike-back capability. U.S. sharing would achieve this bonus effect only if the things shared were surplus to a U.S. strike-back capability. Presumably missiles would not be in this category, although some bombs and warheads might be, and technical information would be. 4. AN INTEGRATED NATO OR EUROPEAN NUCLEAR DETERRENT Many of the dangers and difficulties associated with a congeries of independent national nuclear forces within NATO would be eliminated or alleviated if the weapons could be placed under some sort of inte grated, supranational command. Two general types of such integration will be discussed here, although many variants of these types are con ceivable. Under one concept, several European countries would inte grate their nuclear forces under a joint command. The U.S. Strategic Air Command (and perhaps the RAF Bomber Command) would re main separate, although all three could maintain a loose connection, and perhaps coordination of plans, via the North Atlantic Treaty Organization. The United States might share some of its nuclear stock piles with the European group and might also provide technical assistance and information for the development and production of nuclear weapons and delivery systems by the group itself. This plan would be in the tradition of the proposed European Defense Com munity (and its substitute, the Western Exnropean Union), Euratom, and the European Coal and Steel Community; and it would be con sistent with other aspects of the general movement toward European, rather than Atlantic, integration. Opposed to this plan would be some form of integration on the At lantic model. Variants of this range all the way from giving the Euronetworks simultaneously. If they chose the latter, the impact on targets in Europe would occur first and would confirm the warning to the United States, which, if the fact of impacts in Europe could be communicated very quickly and if U.S. missiles could then be fired instantly, might allow the U.S. missile force to be fired before the explosion of Soviet missiles on U.S. targets, in certain knowledge that an attack was under way. If the Soviets chose to strike all targets simul taneously, the crossing of the U.S. warning line ahead of the European warning system would provide the Europeans with warning time equivalent to that avail able to the United States.
in Western Europe
175
pean members of NATO some voice in the disposition and use of SAC (which at present is outside the NATO military commands and strictly under U.S. national control) to the creation of a tight Atlantic Defense Union which would control all of the alliance's nuclear weapons and delivery systems—SAC and the RAF Bomber Command as well as whatever nuclear forces might be developed on the Continent.43 Another form of "integration" would require the United States to share its nuclear weapons with its European allies on a bilateral basis, under some system of dual control. This would be an expansion and refinement of the present arrangement whereby some countries receive tactical and strategic missiles from the United States, with their war heads remaining under the control of U.S. teams. A. EUROPEAN GROUP DETERRENT
From many points of view, a nuclear force controlled jointly by the European members of NATO would be preferable to their development of independent forces. Most of the risks and disadvantages associated with independent forces would be eliminated or reduced. There would be less risk of the use or threatened use of nuclear weapons for narrow ly national purposes outside Europe. This might reduce the incentive for countries outside Europe to acquire these weapons. If nuclear dis armament should become a realistic possibility, presumably it could be negotiated more easily if the European forces were controlled jointly than if they were controlled by separate national states. The members' pledge to aid each other at the nuclear level, as well as the joint institu tions of control which might be created, would tend to mitigate the divisive tendencies within NATO which might result if several members possessed independent nuclear forces, although the establishment of a European Group deterrent force not under NATO command might tend to bipolarize NATO around its North American and European halves. However, coordination of U.S. policy with a European Group nuclear command presumably would be accomplished more easily than with a number of independent nuclear powers. Costs of development and production of weapons for a collective force would be lower than for separate national systems because resources would be pooled and 43
Both forms of integration are discussed by Ben T. Moore, in NATO and the Future of Europe, New York: Harper and Brothers, 1958, chap. 9; and by Alastair Buchan, op.cit., chap. 5.
176
Deterrence and Defense
duplicating activities would be avoided. Thus more European resources would be available for maintaining tactical defense forces. But the greatest appeal of this proposal, as compared with nuclear laissez-faire in Europe, lies in its probable greater effectiveness as a deterrent. The extent to which this effectiveness would be increased would depend to a high degree on the apparent cohesiveness of the union which was established to control the weapons. Command and control arrangements will be discussed more fully below; it is enough to say here that even with the loosest form of union—e.g., a pledge by each member to retaliate when any member is attacked, plus some joint planning to this end—the Soviets would be taking a much greater risk in attacking a single European nuclear power than would be the case if the victim were relying solely on its own deterrent capability with some vague possibility of "triggering" other countries' retaliatory forces. The enemy would see much less chance of being able to pick off the sep arate national strategic forces by successive counterforce strikes. Assuming that the Group had sufficient apparent cohesion to cause the Soviets to believe with a high probability that an attack on one member would mean retaliation from all members, the Group's ability to deter attack would depend on the familiar question whether it could muster sufficient first- and second-strike capabilities—plus a somewhat more problematical capability to trigger United States strategic forces. Leaving aside for the moment the triggering possibilities, the Group would have a minimum second-strike force when the Soviets, in strik ing first at the Group's forces, could not preclude damage to themselves that would be greater than the value they attached to conquest of the entire Group. With very well-protected or mobile forces, such a capa bility might in time be achieved, thus effectively deterring a Soviet nuclear attack. Deterrence of a Soviet ground attack with high confi dence, or of a Soviet nuclear attack on one member, would require a sufficient first-strike capability to reduce Soviet nuclear forces to the extent that the Soviet counterblow, when spread over all members of the Group, could be tolerably absorbed, and sufficient also to defeat the Soviet ground attack or to bargain the Soviets into discontinuing the ground attack under threat of heavy additional punishment. A capabil ity of this size and nature probably would be beyond the resources of the Group even with nuclear sharing and development assistance from the United States. However, such high-confidence deterrence—stem ming from a virtual certainty in the minds of the Soviet leaders that the
in Western Europe
177
whole Group would retaliate—may not be required. Considering the severity of the damage which the Group would be capable of inflict ing—especially if retaliation were threatened against Soviet cities rather than forces—even a small probability of Group retaliation might be sufficient to deter, no doubt a smaller probability than that of retalia tion by a single victim in a system of independent national nuclear forces. A European Group deterrent would also have a greater capacity to trigger American strategic retaliation than would independent national forces. The Soviets would still have to consider some possibility of U.S. retaliation, of course, even after creation of a European Group deter rent, although we have been assuming that the credibility of U.S. re taliation, considered by itself, may be too low for effective deterrence of attacks in Europe. Triggering in the "attrition" sense could occur if, after a nuclear exchange between the Soviet Union and the European Group, enough Soviet forces had been used up so as to give the United States a sufficient counterforce capability. Obviously, more attrition to Soviet forces would take place in an exchange with the European Group than in an exchange with a single European nuclear power. Thus the United States would need less counterforce capability to be willing to be triggered, or, to put it another way, with a given U.S. counterforce capability, triggering would be considered more likely by the aggressor. Another reason why triggering would be more likely with a European Group deterrent is that, in attacking the whole Group, the Soviet Union would have signaled an ultimate objective so large that the United States might be willing to accept very large costs to frustrate it. In the case of Soviet attack on a single country in an unintegrated system, the U.S. threshold of acceptable cost would be lower, since the Soviets might plausibly indicate that conquest of this single country represented the limit of their ambitions. The degree of credibility and general effectiveness of the European Group deterrent would depend rather crucially on its command and control arrangements. A great variety of such arrangements may be visualized, ranging all the way from a simple treaty pledge by the Group's members to retaliate in case any one were attacked, plus joint planning to coordinate the retaliation, to the complete submersion of national sovereignties in a full European union. The last is probably too unrealistic to be worth considering. The pledge plus joint planning would pose, no doubt, a more effective deterrent than a system of com-
178
Deterrence and Defense
pletely independent national deterrent forces. But even if the Group possessed an adequate first-strike force (or did so in combination with the United States), there would exist strong anxieties among the mem bers that the others would fail to respond, which would inhibit the willingness of each to retaliate for fear that it would be alone in doing so and therefore that its own retaliation would be futile and would concentrate completely unacceptable retaliatory damage on itself. To weaken these anxieties and therefore to increase the credibility of a Group response, somewhat tighter arrangements, more restrictive of national sovereignty, would be required. One such arrangement would be to place the Group's nuclear forces under the immediate operational control of a single commander, responsible to some higher political authority. The presently existing Western European Union, or the European Atomic Agency, might be converted into such an authority. If it were still thought doubtful that the political organ would give authority to the military commander to act after an attack on a single member had taken place, such authority might be voted and delegated to the commander in advance of war. Such advance authority might specify the circumstances—i.e., nuclear attack on any member, or ground invasion of any member beyond a certain terri torial limit, or with a certain size of enemy commitment of ground forces—when the commander could order retaliation on his own initia tive. This would indeed amount to a considerable degree of abdication of national sovereignty and probably would be resisted by some mem bers of the Group. Presumably, agreement would be more easily reached, the narrower and the more serious the range of contingencies covered, and the more clearly they were defined. For example, it probably would be easiest for the members to agree to delegate retalia tory authority to the commander for the contingency of nuclear attack on a member or on the Group as a whole. On the other hand, limitation of the authorization to only a few very clearly defined contingencies might allow the enemy to tailor his aggressive moves so that they would not fit the specified conditions. However, in doing this, the Soviets would be taking a risk that the commander might interpret his instructions flexibly and retaliate anyway. If the Group members balked at fully abdicating their freedom of choice to a military commander, even under a narrow set of circum stances, the commander might be given a directive in advance of war, stating that under certain specified circumstances he would be given
in Western Europe
179
the authority to retaliate, but reserving the ultimate decision to the political organ.44 This would morally commit the political organ to order retaliation when the specified circumstances arose, a commitment on which the political authority might find it difficult to renege. An even stronger directive might grant the military commander the au thority to retaliate unless he received positive orders to the contrary within a specified time after the start of the aggression—a sort of re versal of the "fail-safe" system. The political organ would then have made it somewhat problematical that it would be able to stop retalia tion from taking place; it would have, in effect, given up a portion of its freedom of decision to a military subordinate; the military comman der might set the start of the aggression at an earlier point than the political organ, or the political organ might simply find it impossible to agree on a countermanding order within the specified time period. This arrangement would of course be communicated to the Russians; their recognition that the political authorities might not be able to stop retaliation from taking place would probably exert a strong de terrent effect. Still other arrangements are conceivable under which a portion, but not all, of national freedom of decision might be given up. For ex ample, the political organ might write its constitution so that a majority vote, rather than unanimity, was sufficient to unleash the response. Or the commander might be given advance authority to take certain preparatory actions, such as starting the count-down on missiles and holding them just this side of firing, or putting the Group's manned bomber force in flight toward its targets, subject to a final go-ahead order from the political organ. Such actions would signal to the Soviet Union that preparations for retaliation were under way and thus might cause the Soviets to cease and desist. They would also set up a certain momentum in the direction of retaliation which the political organ would find morally and psychologically difficult to override. They would increase the danger that the retaliatory force might be set off by accident. Such actions would involve the acceptance of risk that technological factors or human error might set off retaliation irre spective of the ultimate political decision; knowledge that such risk 44 There is a precedent for this sort of semi-automation in the NATO Council's decision of December 1954 to authorize SHAPE to base its war plans on the assumption that atomic weapons would be used from the outset of serious hostilities in Europe.
180
Deterrence and Defense
had been deliberately accepted by a decision made in advance of war should impress the Soviet Union with the Group's probable intent to retaliate. Finally, there is the possibility that the ultimate decision to order retaliation would devolve in advance upon the country actually at tacked. In effect, this would mean that the victim would have at its disposal the entire retaliatory force of the Group. When nuclear attack against a certain country occurred, or when the enemy's ground forces had advanced beyond a certain point, command of the retaliatory forces would automatically pass to that country. To achieve a high degree of certainty that this country's orders would be obeyed by the subordinate military commanders of the entire Group, each of the nuclear force's installations would be controlled by a multinational team, with the provision that a member from the attacked country would take command of the installation as soon as the aggression became clear. The extent to which the ground forces of NATO have achieved integration under supranational command (not to speak of the higher degree of integration proposed in the rejected European Defense Com munity) gives evidence that command and control arrangements of this sort are not beyond the bounds of possibility. Yet it must be recognized that the parallel is not exact. The multinational character of the ground forces deployed on NATO's forward lines insures that several members of the alliance would automatically become involved in case of Soviet attack. It is hardly conceivable that one of the mem bers would pull its forces out of the line and decide not to fight, once the attack was under way. At the strategic nuclear level, however, this kind of automation by deployment, which is triggered directly by the enemy's attack, is not feasible. Automation must be obtained by advance political decision and delegation of authority, in a manner which at least makes questionable the ability of political authorities to rescind the order effectively. And it must be admitted that a degree of collective obligation and automation that might freely be accepted by the member governments at the level of ground force action would be accepted with much greater reluctance at the strategic level, where military action would involve a much greater degree of destruction and might be ineffective in denying territory to the enemy. The polit ical obstacles in the way of effecting such integration are great, but they do not seem insurmountable. And it must be emphasized that
in Western Europe
181
the degree of collective cohesion and mutual confidence in each other's intent which the allies themselves might think necessary to justify their pledging their forces to the Group, and using them in case of attack, is likely to be far higher than the degree of integration neces sary to pose an effective deterrent. For deterrence depends on what the enemy thinks the Group might do, not on what the Group actually would do, or the degree of confidence among the allies that the collec tive arrangements would work in a crisis. The greater the Group's first-strike counterforce capability (and the greater the likelihood of either simultaneous action by the United States or the triggering of a U.S. retaliation), the less difficult it should be to obtain any given degree of advance collective commitment and automation, and the lesser the degree of such commitment and automation that would be required for effective deterrence. Theoretically, the idea of a European Group deterrent is not neces sarily dependent for its feasibility on a U.S. willingness to share its nuclear stockpile. The Europeans could decide to set up a Group de terrent and provide the weapons by their own efforts. Yet U.S. willing ness to share would enormously increase the nuclear capability avail able to the Group, and a U.S. offer to share would provide political leverage for influencing the Europeans in the direction of a Group deterrent rather than of national deterrents. Moreover, the United States could stipulate, as a condition for its gift of atomic bombs (and missile warheads?) and bomb and missile production blueprints, that the Europeans integrate their research and development and produc tion facilities so as to realize maximum total deterrent power from the available European resources. The United States might also be able to influence the command and control arrangements. If we have to accept the probability that the "nth-country problem" is not going to be solved by methods of arms control (at present, we can only con jecture about this), then it would seem wise to try to guide the diffusion of nuclear weapons into forms of supranational control. Although a multipolar balance of terror might be dangerously un stable, it is not necessarily true that a tripolar or quadripolar balance would be less stable than a bipolar one, and it might be more stable. Nuclear weapons controlled by countries directly in the line of ground attack, with other European countires pledged to aid the front-line countries by strategic retaliation, would appreciably increase the Soviet risks in attempting a major ground invasion of Europe, and
182
Deterrence and Defense
would practically foreclose nuclear blackmail of the European coun tries. As against this, a bipolar balance (if by this is meant a U.S. capability to deter nuclear attack upon itself, but not to threaten a first strike credibly) is accompanied by serious instabilities in the tactical balance of power which the Soviets can exploit by their supe riority in this realm. There are, of course, many other problems connected with the establishment of a European Group deterrent force besides the pre eminent problem of agreement on the contingencies in which the force would be used, and at least partial "automation" of that commitment. Agreement would have to be reached, for example, on how the Group's nuclear installations were to be spread around among the various members so as to distribute fairly the damage which would occur should the Soviets attack strategically. There would also have to be agreement on how development and production burdens would be shared—not only for those nuclear bombs which the Europeans would themselves produce, but also for their carriers. Whether prestige considerations would make the Group's members want to maximize their individual contribution to the Group's arsenal and emplacement of Group weapons on their soil, or whether they would wish to mini mize their contribution for economic reasons and to reduce their own damage in war, is difficult to say. The members' preferences no doubt would vary. There is also the problem whether the Group would be able to prevent some independent national production, or whether some members might be unwilling to place all their nuclear stocks under the command of the Group. It might be desirable to let the members reserve some weapons to their national control. This would pose an extra risk to the Soviets: the risk that if the Group deterrent should fail to be activated, at least the country directly attacked might respond and, in so doing, succeed in triggering off the forces under Group control. B. ATLANTIC INTEGRATION
An alternative to a European Group strategic force would be some form of integration of all the NATO strategic forces. In its most com plete form, such a scheme would put the Strategic Air Command and the British Bomber Command, plus whatever nuclear capabilities were developed on the Continent, under a military command respon sible to the North Atlantic Council. Administrative and operational
in Western Europe
183
control might be vested either in SACEUR—the present military commander of the ground forces committed to NATO—or in a separate commander with jurisdiction only over the strategic nuclear forces. The apparent attraction of this plan, from the European point of view, is that it would commit the United States to retaliation after a Soviet attack in Europe with roughly the same degree of firmness with which the European governments would be committed under the European Group organization suggested above. Arrangements for command, control, and automation similar to those suggested for the European Group might also be applicable to the Atlantic integration plan. However, integration of NATO's deterrent forces in this rather extreme form is subject to very serious objections, which probably must rule out its serious consideration. In the first place, the American government and public would never agree to release control of 9AC to a NATO commander, even if he were an American. Certainly, even if SAC were placed under a NATO commander for operational pur poses, the United States would never agree to any form of automation or advance authorization to the NATO commander to commit SAC in certain contingencies; it would insist on having the last word as to when SAC was to be used, through its veto on the North Atlantic Council. Thus, merely giving the European members of NATO a voice in the control of SAC would not solve the problem of the presumed non-credibility of the U.S. threat to retaliate under conditions of nuclear parity with the USSR. With the United States retaining a veto over SAC's use, the credibility of retaliation by SAC would still depend on how closely SAC's forces approached a sufficient first-strike counterforce capability, reckoned according to the degree of U.S. interest in Western Europe, or the degree of Soviet provocation in Europe that would be serious enough to make the United States willing to initiate all-out war. Giving the European governments a voice in determining when SAC was to be used very likely would reduce the credibility of retaliation by SAC. The more voices that had to be heard, the more difficult to make the decision, and the Europeans might very well be less willing to give their approval to the use of SAC in a crisis than would the United States, given the Europeans' more vivid experience with the horror of aerial bombing in World War II. In fact, one suspects that one reason why the Europeans are demanding a voice in the con trol of SAC—even though this reason may be more subconscious than explicitly recognized—is to be able to prevent SAC from retaliating
184
Deterrence and Defense
when it does not have the capacity to prevent devastating counterretaliation on both Europe and the United States. The arguments for placing SAC under NATO control are advanced as a remedy for the waning plausibility of a U.S. unilateral decision to retaliate. But with the United States retaining a veto over SACs use, the only logical reason for the Europeans to demand NATO control of SAC would be to give themselves a veto as well. As was noted earlier, the Europeans are caught between the desire to maintain the credibility of deterrence and the desire to avoid their own incineration should deterrence fail. If there is danger that the United States would retaliate unilaterally when it did not have a sufficient first-strike capability, the control of SAC by NATO has "defense" value for the Europeans because such control might be able to keep SAC on a leash. But their arguments for this arrangement must be presented in terms of strengthening the de terrent in order to avoid undercutting the credibility of retaliation in the Soviet calculations. Alastair Buchan has proposed a modified form of strategic nuclear integration under NATO, with only part of SAC under NATO com mand, and that part located in Europe. This plan would roughly parallel the present NATO arrangement for integration of the alliance's tactical forces, with most European members committing practically all of their active forces to NATO command, but with the United States making only a limited contribution, although its contribution does represent the strongest element in the integrated force. In the Buchan proposal, sufficient strategic nuclear forces, including SAC forces, would be located in Europe to constitute a sufficient strike-back capability to deter a Soviet all-out nuclear attack on Europe. The force would be operationally under the control of a single commander —not SACEUR and preferably not an American—who would be re sponsible to the NATO Council. Thus, NATO would have two military commanders—SACEUR, who would continue to command the tactical forces, and a Supreme Commander Deterrent (SACDET), who would command all the strategic forces under NATO. Buchan's case for this arrangement rests on four main points: (1) it would remove any Soviet incentive to attack Europe in the belief that SAC would not retaliate, since there would be in Europe itself a strategic force with strong strike-back capability; (2) the arrangement would increase the general cohesion of the alliance; (3) it would solve the nth-country problem as regards Europe, at least, by removing the
in Western Europe
185
temptation to develop separate strategic forces; and (4) it would "conform to the changing balance of economic strength as between Europe and America, and shift some of the burden of deterrence on the shoulders of the former."45 These are cogent arguments and there is much to be said for this proposal, provided certain questions of political feasibility could be satisfied. The United States government might find it hard to agree to place even a part of SAC under NATO command, since SAC is regarded in the United States as primarily an instrument for deterring attack on America herself, and whatever remains of American chauvinistic isolationism tends peculiarly to focus on strategic airpower. I am not sure whether Buchan proposes that those portions of SAC which would be transferred to NATO command should continue to be responsible to SAC as well, or that their organizational connec tion with SAC should terminate. He appears to contemplate that they would continue to be part of SAC, an arrangement which would parallel the existing command set-up for the American contribution of ground forces to Europe, under which these forces have a dual responsibility—to SACEUR and to their Chiefs of Staff at home. If this is a correct interpretation, there are serious doubts that the United States would agree to the plan, for reasons just mentioned. United States acceptance might be more readily given, I should think, to a plan whereby a portion of SAC's forces would simply be turned over to the NATO command and their responsibility to SAC dissolved, except as the American individuals who would participate in the multinational manning of the bases and their command superstructure were personally members of SAC. Then the question of the autonomy and integrity of SAC as an organization need not present so serious a problem. Another possible difficulty might be United States worries that the transfer of some of its nuclear capability to Europe would weaken SAC's capability to deter an attack on the United States. Such fears would be objectively invalid, however, if the United States maintained a sufficient strike-back capability on its side of the Atlantic. A more serious obstacle might arise from the U.S. strategic commitment not simply to a strategy of "minimum deterrence" of attack on itself, but to a strategy of "destroying" the Soviet Union—in fact, "winning the 45 Alastair
Buchan, op.cit., p. 71.
186
Deterrence and Defense
war" if war should occur. To maximize their capacity for striking back powerfully enough to accomplish this purpose, U.S. military leaders might prefer to have all the U.S. strategic capabilities in the United States and under U.S. control, since they would not feel certain that the collective NATO forces in Europe would act in the event of attack on the United States. The latter forces, after all, would be ultimately controlled by the NATO Council, in which an adverse vote by a single member could block their use. Finally, United States authorities might feel that the transfer of part of our nuclear and missile stockpile to Europe would weaken the alliance's capacity to deter a massive ground attack by the threat of a first strike. Without such transfers, the United States might be able to muster a fairly credible first-strike capability by itself; the transfer might weaken the credibility of this threat without at the same time creating a credible first-strike capability for the NATO force. This particular objection could be overcome only by a high degree of coordinated planning between SAC and the NATO force as well as political commitments between the NATO members to the effect that all the alliance's forces would strike first simultaneously. There might be objections to an integrated NATO deterrent force by some other members of the alliance, especially those, like de Gaulle's France, who wish to break away from what they consider a humiliating dependence on the United States and re-establish them selves as independent powers in world affairs. France might well prefer an independent European Group deterrent to one organized under NATO's aegis, although her highest preference no doubt would be for a completely independent deterrent force of her own. France may find the cost of the latter more than she can bear, however, and may turn to the European Group alternative. Would United States participation in the European deterrent organ ization through NATO complicate the problem of agreement on con tingencies when the force was to be used, and the related problem of arranging for some degree of automaticity in the response? There are considerations on both sides of this question. It can be argued that there is greater willingness in Europe to abdicate portions of national sovereignty in a common cause than there is in America. The evidence for this lies in the existence of such organizations as the Coal and Steel Community, Euratom, the Common Market, and the Council of Europe, in the proposed European Defense Community, and in the
in Western Europe
187
increasing degree of collaboration between European countries in the production of military equipment. The strength of the "third force" idea in Europe, aimed at creating a third center of world power which could be independent of, and mediate between, the two present "superpowers," seems also to point in this direction. If the American contribution to the NATO force were to have a dual responsibility—to SAC as well as to SACDET—it is difficult to imagine the United States agreeing to confer on a NATO commander advance authorization to retaliate in certain circumstances. The degree to which America has allowed her tactical forces in Europe to be integrated into the NATO command structure probably would not have been possible had SACEUR not been an American officer. On the other hand, if the United States retained under her own control at least a minimum strike-back capability to deter an attack on herself, she might not balk at agreeing to the minimum degree of automation—an automatic response by the NATO force in case of an all-out Soviet strategic attack on the whole of Europe. Considering the apparent firmness of the American commitment to retaliate in case of nuclear attack on any single NATO member, the United States might even more willingly agree to collective retaliation in that contingency, and possibly to some degree of automation, than some of the other NATO members. After all, it would be the European countries which would have to absorb the Soviet counterretaliation—again, so long as SAC-in-America main tained a sufficient strike-back capability, the Soviets would hardly dare retaliate against the United States, even though the United States, by its participation, would share the responsibility for the reprisal by the European-based forces. The same would apply if the collective commitment specified all-out reprisal in the event of ground invasion.46 Indeed, the United States might even be able to persuade the Europeans to go farther in such concerting of planning and automation of responses than they would go without U.S. participation, as a quid pro quo for an agreement by the United States to turn over part of SAC's weapons to NATO. It does not seem, on balance, that the political obstacles to the Buchan proposal for a NATO deterrent force are much less tractable than those which would confront a European Group deterrent disso ciated from NATO, although, of course, these obstacles are rather 46 Buchan apparently does not propose that the European-based forces would be used in this contingency.
188
Deterrence and Defense
serious in both cases. Buchan's proposal would seem to share all the advantages of the latter scheme over a system of independent strategic forces, and it has further advantages over the European Group idea that more than offset its possibly lower degree of feasibility. It would utilize an already existing command and political structure, thus avoiding the difficulty of forming an entirely new organization, which might create great complexity and confusion by cutting across firmly established NATO lines of command. It would include British partici pation, which might be doubtful, though essential, in the case of the European Group scheme. Above all, it would tie America more closely to NATO and avoid the possibility, inherent in the European Group plan, that the alliance would tend to split in half.47 Participation by American missiles and missilemen in the NATO deterrent forces would increase the chances that a Soviet nuclear strike on Europe would touch off full-scale retaliation by SAC-in-America; they would perform a trip-wire function similar to the role being played by the American component of the ground force "shield." American participation in the command structure of the NATO deterrent should encourage close cooperation with SAC, especially if the American facilities in the NATO command were responsible to SAC as well.48 This would strengthen the credibility of a coordinated retaliation by the alliance's total nuclear forces in case of a major conventional or nuclear attack in Europe. With a European Group deterrent, such full-scale retalia tion might have to depend on the rather chancy "triggering" effect, and even if triggering worked, the alliance's retaliatory blow would not be well coordinated. Integration under NATO command would facilitate coordination with the tactical air forces under SACEUR— some of which could participate in strategic retaliation. The greater degree of integrated planning with SAC that would be possible under the NATO deterrent scheme would increase the willingness of the individual members to fire weapons on their territory—or decrease their incentive, in a crisis, to seize these weapons forcibly from the multinational units operating them in order to prevent their firing49— 47 Buchan makes these three points in stating his case against integration on a strictly European model. Buchan, op.cit., p. 70. 48 If they were responsible to SAC as well as to SACDET, the multinational manning of individual bases and launching sites which Buchan proposes probably would not be feasible for the SAC portion of the NATO force. 49 This possibility of national seizure is mentioned by Professor P. M. S. Blackett in a dissent to Buchan's proposal which appears as an appendix in Buchan's book (op.cit., p. 127).
in Western Europe
189
since all members could feel fairly confident that SAC would strike simultaneously with, or soon after, their own blow. This would maximize the expected first-strike damage to the Soviet forces, and minimize the severity of the Soviet retaliatory damage. Besides reduc ing the aggregate Soviet retaliatory capability, a coordinated first strike would also spread the Soviet retaliatory damage over all the members of the alliance; the European members would feel less fear that Soviet retaliation would be concentrated against their own coun tries. Atlantic integration on the Buchan model probably would force the Soviets to choose between striking first at the forces of the entire alliance, or not striking at all. If the alliance as a whole could mount a completely unacceptable second strike (which the European Group alone might not be able to do), the Soviets presumably would choose not to strike. Coordination of plans would reduce or eliminate the danger that the SAC and the NATO deterrent force would duplicate targets, thus dissipating some of the alliance's nuclear strength. Without such coordination—i.e., with reliance solely on triggering—each of the separate nuclear forces would have to choose its targets on a firstpriority basis (since it would not know with certainty which targets the other force planned to hit, or even whether it would participate at all) and these first-priority targets would very likely be similar for both forces. If both forces were able to agree in advance on such a sharing of targets, it would greatly increase their mutual confidence that the other would act in concert, for both would know that a failure to act on either side would severely reduce the counterforce effective ness of the party which did act.50 If it should prove infeasible to have a collective NATO deterrent force, but feasible to have a European Group force—i.e., if the United States refused to participate in a collective deterrent for Europe—it would still be possible, of course, to coordinate planning between SAC and the European strategic forces. But the divisive effect on the alliance which probably would have followed the establishment of a separate European force would undermine the confidence on both 50 In the event of an all-out Soviet nuclear attack on the NATO collective de terrent, the latter's retaliation would perhaps be countercity, if it were uncertain of SAC's participation. But with integrated planning between SAC and the NATO forces, the strike would probably be counterforce, in which coordination in target ing would be essential. In retaliation after either nuclear or conventional attack on a single NATO member, both SAC and the NATO force would be "striking first" against Soviet nuclear forces.
190
Deterrence and Defense
sides that the other would act according to the common plan, and hence would reduce the credibility of the threat of a common re sponse. This would be especially likely if neither of the two organiza tions could muster separately a significant first-strike counterforce capability. As an alternative or supplement to joint planning, SAC and the European Group might increase the credibility of joint retaliation by deploying some of each other's forces on the territory of the other, retaining such forces under their own control. Then the Russians would know, for example, if they contemplated a first strike against the forces of the European Group, that they would have to strike at those forces of the Group which were deployed in the United States, with a strong chance that this would provoke retaliation by SAC. If they wished to strike at the United States, they could not avoid attacking SAC installations in Europe, which would seriously increase the risk that the European-controlled forces would join SAC in retalia tion. Such reciprocal deployment could also be applied to the Buchan plan for a NATO collective deterrent, further increasing the degree of integration and the probability of simultaneous retaliation by the forces of the whole alliance. Integration within a NATO framework would be subject of course to most of the problems involving siting, financing, and manning, and command and control problems generally, which were discussed in connection with the European Group proposal. The problem of where the forces were to be located would be particularly difficult Military prudence would indicate that they be stationed considerably to the rear of the front lines (to the extent they are based on land) in order to avoid their being captured by invading forces before the decision was made to fire. Siting them in the rear seems to indicate France and Italy as the most logical host countries; yet France is in no mood to increase her degree of collaboration with NATO and seems to want no nuclear weapons on her soil except those under her sole control. On the other hand, forward placement near the front lines might be best from a deterrent viewpoint, especially if some kind of arrangements for automatic response at military discretion had been prepared in advance; the military commander, fearing that the bulk of his nuclear installations would be overrun quickly if he did not act, might act before the political authorities had time to agree on countermanding his orders.
in Western Europe
191
c. BILATEBAL NUCLEAR SHARING WITH DUAL CONTROL An alternative to the two forms of integration which we have just discussed would be for the United States to share some of its nuclear and delivery capabilities on a bilateral basis, with the United States and the host country having joint control over the weapons and joint responsibility for the decision to fire them. The United States has already moved in this direction by supplying tactical and strategic means of delivery to some of its allies, while retaining control over the warheads, which are stocked nearby under the control of U.S. teams. U.S. control of the warheads is required by United States law. It is widely recognized, however, that this system is inefficient, since it might take overlong in a crisis for the warheads to be screwed on and the weapons made ready to fire. Hence, if the law could be changed, there is a strong case for having the warheads installed on the missiles, with the United States keeping its finger on the "safety catch" by some other method, and with arrangements to release the safety catch in appropriate contingencies. One possibility would be to put the war heads on the missiles, with the United States exercising its veto by a system whereby the missile would be fired only if two buttons were pushed, a U.S. team being in charge of one button. A more risky arrangement would simply require U.S. formal approval for firing; such a system would have a deterrent advantage over the one previous ly mentioned, since a host country would not have to use force to override a U.S. veto and hence might be more willing to do so in a crisis. Another plan has been suggested by General Pierre Gallois, a French officer.51 In the Gallois proposal, while the decision to fire the weapons would be subject to a double veto in normal peace time conditions, control would pass to the host country when that country was attacked or under serious threat of attack. This plan would mitigate or eliminate the dangers of irresponsible use of the weapons by the host country, while giving it, in effect, an independent retalia tory force in circumstances when its survival was directly threatened: If the United States did not have a first-strike counterforce capability which made its own threat to retaliate credible, but did have a surplus over its needs for minimum deterrence of an attack on itself, this ar51 Pierre M. Gallois, "New Teeth for NATO," Foreign Affairs, Vol. xxxix, No. 1 (October 1960), pp. 67-81.
192
Deterrence and Defense
rangement would create for the Soviets a risk of at least some retalia tion should they decide to attack the host country. It would also stiffen the resolve of the host country to resist blackmail attempts. However, if the Soviets had, prior to the establishment of such a system, feared U.S. retaliation with some significant probability, the establishment of the system might reduce the credibility of a U.S. response, since the United States would appear to be abdicating responsibility for retalia tion to the European partner. The desirability of this system would seem to depend, therefore, on the credibility of U.S. retaliation already being so low that the Soviets no longer felt any degree of restraint from this prospect. This scheme also would present the problem of agreement on the specific circumstances in which firing control would pass to the European host country, although presumably this problem would be more tractable than under a multilateral arrangement. The Gallois proposal could be applied to the European Group deterrent plan, or as a condition for U.S. participation in a collective NATO deterrent, with the United States sharing weapons while maintaining control of the "safety catch," but agreeing in advance to relinquish such control in specified circumstances. It might be desirable not to specify the conditions of relinquishment too clearly; then, provided the Russians believed that retaliation by the European Group (or by a single host country) was more likely than U.S. retaliation, the United States would have a flexible instrument for putting pressure on the Soviets in the earlier stages of a crisis by giving up its power to veto a firing order. To avoid giving the impression that it was backing out entirely from its own threat to retaliate, it might be well for the United States to declare that if the ally or the European command did decide to fire, U.S. retaliation would automatically follow. A variation on the Gallois plan would be a procedure whereby control of only part of the shared weapons would pass to the host country in a critical situation, and control of the rest would pass to the United States. Then the Soviets would face a dual risk. U.S. control of some of the weapons would be a hedge against failure of the host country to fire its weapons; one set of missile teams might fire and trigger the other in so doing. This sort or arrangement would be much less likely to indicate to the Soviets that the United States had abdicated its own commitment to retaliate.
in Western Europe
193
5. LIMITED RETALIATION Discussions of military strategy often tend to pose two broad alternatives, or categories of action: all-out thermonuclear war, and limited war. We recognize several gradations of limited war, from the "border incident" or "brush-fire," through larger conventional wars on the Korean model, to limited nuclear war fought with "tactical" nuclear weapons. And, as an earlier chapter of this book has sug gested, as well as some other authors' writings, all-out war might be fought under restraints which gave it an outcome considerably short of utter mutual annihilation. Yet one intuitively suspects that there remains a gap between these two categories which may embrace intermediate types of warfare that have hardly begun to be systemati cally explored. One of these types is a strategy of limited retaliation.52 Limited retaliation is defined as a strategy of single, small, successive nuclear strikes directly against the Soviet Union after the Soviets have initiated tactical ground aggression of major proportions. The pur poses of this strategy are, first, to deter the aggression by threatening such reprisals; and, second, by actually beginning to carry them out if deterrence fails, to persuade the Soviets to cease and desist under pain of an eventual accumulation of human and material costs which would more than offset the fruits of their aggression. It is to be hoped that deterrence would result from the enemy's recognition of our will to inflict ultimately unacceptable costs; but in case deterrence should fail, each separate reprisal would be moderate in destructiveness, since its purpose would not be to punish the enemy immediately to a degree commensurate with his transgression, but rather to demonstrate, at minimum cost and risk to ourselves, our will to make such reprisals, and to suggest a willingness to continue making them beyond the point at which the aggressive venture became unprofitable for the enemy. Keeping the initial reprisals small would also provide maximum time and opportunity for the enemy to learn that he had miscalculated our intent and to change his mind and discontinue his aggression before the costs of war for both sides became incommensurate with the stakes 52 See Morton A. Kaplan, "The Strategy of Limited Retaliation," Policy Memorandum No. 19, Center of International Studies, Princeton University, April 9, 1959; Thomas C. Schelling, "Surprise Attack and Disarmament," and Klaus Knorr, "NATO Defense in an Uncertain Future," both in Knorr (ed.), op.cit., pp. 205-8 and 292-94; and Glenn H. Snyder, "Deterrence by Denial and Punish ment," op.cit., pp. 10-11, 32-34.
194
Deterrence and Defense
at issue. Of course, the Soviets would return the reprisals against the United States. A limited reprisal war would be pre-eminently a bar gaining war, with each side attempting to convince the other that it should give up the fight, or come to terms, in order to avoid eventually intolerable damage. Limited reprisals could be threatened and carried out against the enemy's cities, against bases and elements of his long-range strategic forces, or against his ground tactical forces and their supporting in stallations. Some targets, of course, such as railroad terminals or ports, might involve damage to cities as well as to strategic or tactical capabilities. In the discussion which immediately follows, we shall assume that reprisals are threatened or directed against cities; counterforce reprisals will be taken up in a subsequent section. One reason why the limited reprisal concept has been late in receiv ing serious analysis is that the continuing national security debate has tended to bipolarize around the competing objectives of deterrence by the threat of maximum punishment, and defense by denying territory to the enemy at minimum cost to ourselves. Limited retalia tion was not congenial to the proponents of massive retaliation because obviously, it seemed to them, deterrence was most likely to be successful if the enemy were presented with the prospect of maxi mum costs. Nor did it appeal to the limited war advocates, because it did not seem compatible with either cost minimization or denial of territory. Limited retaliation would bridge this dichotomy by using the threat to inflict punishment both as a means to deter the beginning of the war, and as a means of ending it on acceptable terms. Because of this bipolarization of the massive retaliation debate of 1954 and 1955, certain statements by administration spokesmen sug gesting a limited retaliation strategy tended to be overlooked. Such statements were made chiefly by the late Secretary of State John Foster Dulles in his attempts, in response to criticism, to modify the "massive retaliation" doctrine as he had stated it in January 1954. The burden of these qualifications was that, although a capacity for massive retaliation was necessary as a deterrent, this did not mean that massive retaliation would actually be applied in case of any significant Communist aggression. The essential requirement, Secretary Dulles repeatedly emphasized, was that the potential aggressor be made to realize in advance that any attack would cause him to suffer punishment outweighing any gains he might hope to make. On
in Western Europe
195
November 28, 1954, for example, Dulles stated: "We must have the capacity to respond at places and by means of our own choosing. This, however, does not mean that any local war would automatically be turned into a general war with atomic bombs being dropped all over the map. The essential thing is that we and our Allies should have the means and the will to assure that a potential aggressor would lose from his aggression more than he could win. This does not mean that the aggressor had to be totally destroyed. It does mean a capacity to inflict punishing damage."53 In other statements, Dulles also stressed the punishment aspect of the strategy—implying that the enemy would be deterred primarily by the prospect of unacceptable costs, rather than by the prospect of failure to achieve his territorial objective. But the punishment, he said or implied, would be tailored to the degree of provocation. He occasionally spoke of "selective retaliation."54 On January 9, 1955, James Reston reported in the New York Times that a policy called "2X" was being discussed by the administration. This meant, Reston explained, that nuclear punishment would be delivered against the aggressor equal to twice the aggressor's potential gain. In Restons words, "the punishment will fit the crime, only double."55 Hanson Baldwin reported in April 1955 that a strategy of "measured retaliation" was being discussed, with particular reference to aggres sion by Communist China. It implied, he said, that in case the Chinese launched an assault on the Nationalist islands of Quemoy or Matsu, one Communist mainland airfield would be destroyed with nuclear weapons. The first strike would serve as a warning; if the aggression continued, another field would be destroyed, and so on. Although this concept apparently was limited to attacks on military targets, the primary purpose of the nuclear strikes would not be to make a contri bution to physically defeating the Chinese assault, but rather to discourage the enemy from continuing his attack by raising the prospect of intolerable costs.56 The critics of the Dulles doctrine, with its corollary of reduced con53
New York Times, November 29, 1954, p. 4. For similar statements by Dulles, see ibid., March 20, 1954, p. 2; ibid., March 17, 1954, pp. 4 and 6. 54 See, for example, Dulles' statement to the Senate Committee on Foreign Re lations, ibid., March 20, 1954, p. 2. 55 James Reston, ibid., January 9, 1955, rv, p. 8. 56 Hanson Baldwin, ibid., April 21, 1955, p. 1.
196
Deterrence and Defense
ventional ground forces, often stated that if the Big Deterrent failed to deter, the United States would be faced with two very uncom fortable alternatives: to carry out the threat of massive retaliation and accept the disastrous costs of all-out war, or to acquiesce in the enemy's conquest. In retrospect, it would seem likely that both of these extreme alternatives would be rejected in the moment of crisis— in the event, for example, of a major Soviet incursion into Western Europe. The decision-makers would then be forced to realize, in cruel clarity, that to carry out the massive threat would involve unacceptable costs for the United States if an adequate counterforce capability were not available, while to do nothing would mean a craven reneging upon our most solemn promises as well as passive acceptance of the demolition of the entire U.S. security position. Considerations of both honor and strategy would put them under enormous pressure to do something which would preserve a modicum of national self-respect and which would hold out at least a hope of stemming the enemy's advance without catastrophic losses to life and property in the United States. In groping for such a response, their thoughts would turn naturally to the idea of striking a limited nuclear blow at the Soviet Union, accompanying it with an ultimatum to the effect that further blows would follow if the aggression were continued. Of course, they would be fully aware that the Soviets probably would respond in kind, and probably against the United States itself. Nevertheless, they might feel that they could credibly present the Soviets with evidence of determination to continue the process of reprisals through several moves and that, faced with such determination, the Soviets might decide to call the whole thing off. At least, this prospect might well seem better to the U.S. decision-makers than either the prospect of certain national catastrophe or certain loss of the entire Eurasian continent to the enemy. The reason why it is especially appropriate to reconsider the con cept of limited retaliation at this time is that the threat of massive retaliation is rapidly declining in credibility with the Soviet accumula tion of a long-range ballistic missile capability. In the years ahead, neither side may find it possible significantly to regulate the other's behavior by threatening a massive first strike, because both will know that a first strike cannot prevent utterly devastating retaliation. And the free world may not have sufficient ground forces to hold the Communist armies at every point where they might attempt a penetra-
in Western Europe
197
tion. Under such conditions of stalemate at the massive level and inadequacy at the tactical level, threats of limited retaliation may take on increasing importance in the power struggle, restoring to strategic nuclear power a more positive or active function than merely deterring the massive use of strategic power by the opponent. Such threats may be fairly credible and effective deterrents, because it will seem at least plausible that the threatener does not believe the carry ing out of the limited reprisal must inevitably result in all-out war. Although the notion of combining diplomatic bargaining with the slow-motion lobbing back and forth of nuclear missiles seems a rather bizarre application of the Clausewitzian dictum about war being the continuation of politics with an admixture of other means, it at least does inject an element of rationality and politics into the conduct of strategic nuclear war. It deserves serious analysis as an alternative to deterrence by massive threats, which may be ineffective, and to de fense by tactical forces, which may be unavailable in adequate quantity. A. LIMITED RETALIATION VS. MASSIVE RETALIATION AS DETERRENTS
Suppose we threaten to respond to major Soviet tactical aggression with nuclear strikes against Soviet cities, no single one of which would inflict sufficient costs to make the aggression unprofitable, but several of which would do so. The threat to begin such reprisals is more credi ble than the threat to undertake massive retaliation when one obvious ly does not have a sufficient first-strike counterforce capability. The cost to us from the enemy's counterreprisal is not likely to exceed greatly the cost that the enemy suffered in our initial reprisal, and hence will be smaller than our valuation of the stakes at issue. The enemy's realization of this, and his realization also that we may expect him to give in after one or two moves, provides grounds for his believing that we may expect to come out of the war with costs below what we would have lost had we simply acquiesced in the enemy's conquest or fought a losing tactical battle. Hence it is more plausible to him that we will begin limited reprisals than that we will carry out immediate massive retaliation, which would very likely produce costs from the enemy's counterretaliation that were far beyond our valuation of the stakes, territorial and otherwise. Massive retaliation is purely a deterrent threat; under most foresee able circumstances, it will not seem rational to fulfill the threat if it fails to deter, because we must expect the enemy to counterretaliate
198
Deterrence and Defense
massively. Whatever credibility this threat retains will rest almost en tirely on the enemy's uncertainties about how much damage we are willing to accept and about how rational we are. Limited retaliation, however, may have more than a deterrent function; it may also be con ceived as a reasonable means of fighting a war and terminating it, on acceptable terms and at acceptable cost, after deterrence has failed. At least the enemy must consider the strong possibility that we will so conceive it. This element—that limited reprisals are more likely to be considered rational in the doing as well as in the threatening—supports their credibility as compared with the credibility of the threat of mas sive retaliation. The reason we may consider limited reprisals to be rational to carry out, as well as to threaten, is not just that the enemy's counterreprisal is likely to be tolerable in its cost, but that in carrying out a limited reprisal we would suggest a willingness to inflict further reprisals, which might persuade the enemy to terminate the war. A limited reprisal is not just the fulfillment of a previous threat, as would be the case with massive retaliation, but is itself a threat of the strong est kind, because it involves actually beginning to fulfill a threat. The enemy can arrest the process of fulfillment at any time simply by com ing to terms, or by ending his tactical aggression. Putting it another way, a limited reprisal strategy really merges a threat and its fulfill ment: we can threaten the enemy with eventual unacceptable damage and then fulfill the threat in stages; with each stage in the fulfillment, the credibility of the original threat is strengthened. We can demon strate at relatively small cost on the first strike that we do intend ulti mately to inflict costs upon the enemy greater than his valuation of his possible gains. The enemy's expectation that we will wish to make such a demonstration increases the credibility of the first reprisal.57 Also, our wish to avoid the humiliation of reneging on our threat might strongly impel us at least to begin the reprisals, in the hope or expectation that, after one or two exchanges, the enemy would be convinced of the unwisdom of continuing his aggression. We conclude, therefore, that the threat of initiating limited reprisals is a much more credible threat than the threat of massive retaliation, when we have much less than a sufficient first-strike force. It could be argued that, nevertheless, a threat of massive retaliation is more effec tive as a deterrent since deterrent effectiveness is a function of the 57
For a discussion of the deterrent value of dividing the fulfillment of a threat into stages, see Thomas C. Schelling, The Strategy of Conflict, op.cit., pp. 41-42.
in Western Europe
199
severity of a threat as well as of its credibility; the much greater costs which the enemy would suffer if we did carry out a massive threat, it might be argued, would more than offset the lower plausibility of such a threat as compared with one of limited reprisals. The enemy would not dare to accept even a slight risk that we would retaliate massively, in view of the terrible consequences if he miscalculated. In rebuttal, it must be pointed out that the threat of limited reprisals carries the implicit threat of eventual massive costs. We do not threaten to carry out one or two reprisals and then quit; we threaten to continue reprisals until the enemy restores the status quo ante. The enemy would know that, once the first strike or first exchange had taken place, his chances of avoiding further exchanges and eventually costs higher than the value of his objective would have declined sharply. With the first reprisal, we would have established a psychological momentum in the direction of further reprisals; it would be relatively easy for us to con tinue making reprisals—easier, perhaps, than making the decision to start them. It might be possible for us to carry the intensity of the con flict almost to the level of full-scale war, as the result of a series of acts, each of which, by itself, we might consider rational. We might believe, in other words, that after the next strike, or next exchange, the enemy would decide to discontinue his tactical advance. We could rationally harbor such an expectation at any stage during the trading of reprisals. A new cost calculation would be made at each stage; logically, the cal culation would focus primarily on expected future costs rather than on costs already suffered. However, past costs would be relevant to antici pated future costs in that they would be an index of the enemy's ap parent willingness to continue counterreprisals. Also, there is a natural human tendency to want to recoup past losses, and to overestimate the chances and underestimate the costs of doing so; this might stimulate a desire to hit the enemy "just one more time" in the hope that he would decide to quit, thereby justifying all the losses which we had already suffered. Any limited reprisal threat carries within it the implicit threat of all-out war, as the result of either side's possible inability to maintain the controls and restraints required of a limited reprisal strategy. Neither we nor the enemy can be sure that such restraints could be preserved beyond one or two exchanges. The longer the lim ited exchanges are carried on, and the greater the degree of destruction already suffered, the more likely that emotions would be generated which would override the restraints.
200
Deterrence and Defense
All these considerations would strengthen an expectation in the enemy that, once we had started limited reprisals, either we would continue them as the result of rational, calculated decisions, or the war would get out of control and explode to all-out dimensions. This, com bined with the distinct likelihood that, after having threatened such reprisals, we would feel impelled at least to begin them, makes a strategy of limited retaliation a rather potent deterrent. B. LIMITED REPRISALS AND LIMITED WAR
A strategy of limited retaliation might be considered a form of limited war. However, it differs from the usual concept of limited war, first, in that it involves nuclear attacks on the territory of the superpower op ponent; and, second, in that it is entirely a bargaining strategy, aimed at dissuading the enemy from continuing the war by setting up a prospect of unacceptable punishment. Although the prevailing concept of limited war includes a bargaining element, it also includes an ele ment of physical denial or restriction of enemy gains that is absent from the strategy of limited reprisals. In the usual image of limited war, the function of strategic nuclear weapons is a passive one; they are to be held in reserve as a deterrent to violation of the target or weapons limitations in the local battle theater. They are threatened sanctions, in other words, whose aim is not to contribute directly to the winning of the local war but to prevent an increase in its destructiveness to a level incommensurate with the value of the objective. In most previous discussions, it had been thought that the policing of the limits of limited war would be accomplished by the implicit or explicit threat to invoke all-out war if the opponent blatantly trans gressed the limitations. But when both sides have substantial invulner able strategic forces, the credibility of this threat is almost as low as the threat to retaliate massively at the outset of war. Henceforth, it may be possible effectively to police war limitations only by a firmly declared threat of limited strategic reprisals in case of violations. Lim ited strategic reprisals would perform this policing function most logically in a limited nuclear war; the enemy would be put on notice that the use of weapons with a greater than permissible "yield," or attacks on out-of-bounds targets, would cause a limited reprisal to be made against his homeland. In a limited war initially limited to con ventional weapons, violations might be more effectively deterred by the
in Western Europe
201
threat to use nuclear weapons tactically in the battle area, although this threat could be backed up by the additional threat of limited strategic reprisals if the violations continued. But limited strategic reprisals might be used in limited war for a different purpose: to contribute directly to the winning of the war. Previous conceptions of "limited war" and "graduated deterrence" have tended to emphasize what we would not do, in the interest of holding the war to a bearable level of destructiveness, rather than what we would or might do to bring the war to a satisfactory close. Limited reprisals might be used positively to deter the enemy from continuing the war and to persuade him to accept a settlement. In this sense, the strategy would not operate merely as a threat to deter the enemy from using the weapons or hitting targets considered out-of-bounds, but would deliberately be used in order to increase the enemy's expectation of the costs of continued fighting—enough to cause him to desist and accept our settlement terms. Thus, in the limited reprisal strategy, deterrence—in the sense of deterring further enemy fighting—becomes more or less amalgamated with the objective of defense—to hold terri torial objectives at reasonable cost. There is no intent to argue here that the inauguration of limited nuclear reprisals would be preferable to the commitment of additional ground forces in order to win a limited war. But the ground forces may not be available, and it may be entirely reasonable that they are not. After all, the peacetime maintenance of armies is costly; a rational preparedness policy must balance the cost against the probability and seriousness of the contingencies in which the ground forces would be useful. It may not be desirable to maintain armies sufficient to deal successfully with all conceivable limited war situations up to and in cluding a full-scale Soviet attack on Europe with conventional weap ons. Aggressions may occur, therefore, which are too big to be blocked with available ground capabilities. In these cases, limited nuclear re prisals may appear a reasonable alternative to both all-out war and futile ground action. As Thomas C. Schelling has pointed out, the development on each side of a relatively invulnerable nuclear striking force for all-out war should reduce fears that limited strategic reprisals would quickly trig ger an all-out war. Neither side is likely to answer a limited reprisal by firing off all its strategic forces, because that would mean leaving itself wide open to massive retaliation by the other side. Nor is either side
202
Deterrence and Defense
likely to interpret a limited reprisal as the first blow in an imminent all-out strike, since such an interpretation would mean believing that the side making the reprisal was preparing to commit suicide. All forms of limited war may be less inhibited in the future as the threat of all-out war becomes unavailable as a sanction to deter loosening of in hibitions; for this reason the threat of limited retaliation becomes more credible as the credibility of massive retaliation declines.58 c. THE THEORY OF LIMITED RETALIATION: A HYPOTHETICAL CASE The theoretical mechanics of a strategy of limited retaliation can best be explored in terms of a hypothetical case. Suppose the contin gency to be deterred is a major Soviet conventional attack on Western Europe. Suppose the Soviets have enough tactical superiority to take Western Europe with relative ease. Assume that the United States or NATO has declared that such an attack would cause the immediate destruction of a Russian city of 200,000 population or an equivalent degree of damage to a city of larger population, and that an additional urban area of this size would be destroyed each day until the Red Army stopped its advance and retreated to the line of the prewar status quo. Suppose valuations of Western Europe are symmetrical; each side values it at the equivalent of ten destroyed cities of the size specified. It is reasonable to assume that both sides would evacuate all their cities above 200,000 population as soon as the war started. The Soviets would have two alternatives on their first move: to attack, or not to attack. After the Soviet Army attacked, the West would have two alternatives: to make a reprisal as threatened, or not to retaliate. Suppose a reprisal was made. After suffering this first and all subse quent strikes, suppose the Soviets have three salient alternatives: (1) to strike back at a Western urban area of similar or greater size and con tinue advancing on the ground, (2) to retaliate and retire to the prewar boundary, or (3) to retaliate and offer a settlement at some line be tween the original status quo and the conquest of all Europe. Other Soviet alternatives are conceivable: for example, the Soviets might choose any of the three suggested but without counterreprisal. How ever, it seems plausible that they would insist on evening the score in destroyed cities; moreover, if the Soviets wished to complete their con quest or demanded a settlement involving substantial territorial gains 58 Thomas C. Schelling, "Surprise Attack and Disarmament," in Klaus Knorr (ed.), op.cit., p. 207.
in Western Europe
203
for them, a counterreprisal might deter further Western reprisals and would serve to increase Western cost expectations in refusing to accede to Soviet demands. At least, it simplifies the problem to assume that the Soviets would make a similar counterreprisal for every reprisal carried out by the West. Finally, we assume that if the Soviets complete their conquest, all reprisals will cease. Under these conditions, obviously the Red Army will not attack if the Soviets expect to lose at least ten cities before completing their conquest. Similarly, the West will not initiate reprisals if it expects the nuclear exchanges to continue for at least ten rounds. The costs of ground fighting would also enter into the calculations of both sides, but, for simplicity, let us assume that the ground warfare is costless. Once the war has started and reprisals have begun, each side's will ingness to continue the war, and the settlement terms it offers, depend on its estimate of the other's resolve to continue the reprisals and the settlement terms it is willing to accept, which may in turn depend on the latter's estimate of the first side's resolve and willingness to accept terms, etc. This is the familiar regression of reciprocal expectations which is inherent in any bargaining situation. Each side would be seeking—by verbal threats as well as reprisals—to influence the other's resolve and intentions by putting on a face of determination to continue trading beyond the level of cost which the other side could accept. The West might feel that its first blow against a Soviet city would so con vince the Soviets of the West's determination to carry out the "city-aday" threat that the war would end at that point. The Soviets might believe that their first counterreprisal would be sufficient to discourage the West from continuing the exchange. Clearly, if each side persisted for some time in the expectation that the opponent would capitulate after the next exchange, or the next few exchanges, an amount of city destruction approaching the dimensions of all-out war might result. In this event, of course, both sides would have "miscalculated"—they would have failed to appraise correctly the other side's determination and expectations. But they would not have been "irrational." This is a bargaining problem of the type which has no determinate mathematical solution. For this reason (and because of the author's limited mathematical expertise) it seems best to state in verbal terms, though as precisely as possible, the considerations which would seem to bear on the bargaining. Let us look at the problem first from the Soviet point of view. As-
204
Deterrence and Defense
sume that, after the war starts, the West sticks to its demand that the Soviets restore the status quo ante or suffer ultimately unacceptable costs. If the Soviets continued to believe that the West would give up making reprisals before exacting costs greater than the Soviet valuation of Europe, the Soviets might decide to complete their conquest. If the West was able to convince them that they would suffer greater reprisal costs than their valuation of Europe, by a margin greater than their prestige losses in capitulating to the Western demand, they would re tire to the original boundary. At some point in the progress of the war, however, the Soviets might feel they could maximize their net gain or minimize their net loss by offering a negotiated settlement. Suppose the Soviets considered offer ing a settlement at the demarcation line between the opposing ground forces. They would make this offer (while continuing to advance) if they believed that the West would accept it, and if they believed the costs of continued fighting would be greater than the difference be tween their valuation of overrunning the whole of Europe and their valuation of settlement at the ground demarcation line. If they were unsure whether the West would accept this offer, they might make an offer more favorable to the West so long as its acceptance would be a more favorable outcome than either fighting to the bitter end or giving in to the West's demand for restoration of the original boundary. In general, the Russians would make the offer which promised the greatest "expected value." The expected value of any offer would be the sum of the expected values for all the possible outcomes of the offer. For example, one possible outcome of a Soviet offer to settle at the ground demarcation line would be immediate Western acceptance. Its expect ed value would be the value of this settlement to the Soviets times the probability that the West would accept. Another possible outcome would be Western acceptance after one further exchange of reprisals. The expected value of this outcome would be the Soviet value of settle ment minus the cost of the additional reprisal times the probability of Western acceptance at this point. Other possible outcomes would be Western acceptance after two reprisals, three reprisals, etc., and their expected values would be calculated in similar fashion. Of course, if the settlement offered were the military demarcation line, this line would be moving forward during the negotiations and offsetting par tially the costs of continued reprisals for the Russians. Similar calculations would be taking place on the Western side. If
in Western Europe
205
the West believed that the Russians expected to make a net gain by fighting on and completing their conquest, or a lower net loss than in capitulating to the initial Western demand, the West might relax its original demand for restoration of the status quo ante, and offer terri torial or other concessions. The hope would be, of course, that the Rus sians would see in the acceptance of this offer, and ending the war, a better outcome than continuing to fight, even though continuing to fight promised a net gain. Once both sides have relaxed their original demand—the Russians for all of Europe, and the West for the status quo ante—then each side must weigh what it stands to gain or lose by sticking to its "last offer" (and threatening further reprisals if it is not accepted) against its gains and losses in capitulating to the other side's last offer. Both may feel that holding out for a time may persuade the other to make a better offer, and feel that the gains from, and the chances of, inducing this better offer are worth the reprisal costs suffered in inducing it. But after one or two more reprisals, one side or the other may realize that it has underestimated the opponent's determination, that the costs of holding out are likely to be too high, and consequently make another offer which is closer to the other side's position. Several offers and counteroffers might be made. At some point, one side may feel that acceptance of the opponent's last offer promises the greatest net gain or lowest net loss that it can expect to make. Then, theoretically, a settle ment will take place. The preceding remarks have attempted to describe the bare mechan ics of the bargaining process without reference to any special charac teristics of the situation—such as the general political milieu, geographical and historical elements which may focus expectations concerning possible terms of settlement, degree of rationality in the opposing leaderships, abilities to communicate or to block communica tion, and asymmetries in capabilities or values—all of which may tend to influence the outcome of the bargaining. We now turn to a discussion of some of these factors. Thomas Schelling has pointed out that certain characteristics of the situation which are more or less external to the valuation of the stakes and the costs of the sanctions may have a strong capacity to focus the expectations of both sides on a particular settlement.59 For example, it 59
Schelling, The Strategy of Conflict, op.cit., esp. chap. 3 and Appendix A.
206
Deterrence and Defense
was suggested that a settlement might be offered on the line of contact between the opposing ground forces. Both sides would be likely to feel that this was a "natural" basis for a compromise settlement if each be came convinced that its own original demand was not likely to be satis fied at acceptable cost. The side which was losing the ground battle would be unlikely to feel that it could successfully demand better terms, and the other side, realizing this, would be unlikely to offer or consider other terms less favorable to itself. Once conquest of a part of a political unit has been effected, the status quo has been at least partly redefined. In accepting the new status quo, the defender would lose prestige as compared with a settlement at the status quo ante, but could at least claim success in limiting the aggressor's gains. The ag gressor would lose face in having to admit frustration of his initial ob jective, but would at least avoid the much larger loss associated with withdrawal to the original boundary. If (contrary to one of our initial assumptions) the ground forces were symmetrical, so that the fighting might seesaw back and forth over the original political boundary, this boundary would also have a strong attraction as a line of eventual settle ment. With asymmetry in ground forces, other boundaries, of countries which the aggressor had overrun—e.g., the boundary between West Germany and France—would very likely be settled upon if the aggres sor's forces were at this boundary or not far beyond it at the time the bargaining came to a head. Geographical lines or barriers, such as the Rhine, might serve a similar function. Political issues which existed prior to the war—for example, the Berlin issue—might provide material for fashioning a settlement which both sides would consider the optimum one. If one side, prior to or during the conflict, were able to establish a predilection to irrationality, or communicate that it lacked the degree of restraint demanded by the limited reprisal strategy, or suggest credibly a belief that the other side would soon come to terms, its bargaining position would be enhanced. A similar advantage would redound to the side which was able to establish that it believed its valuation of the stakes of the conflict was higher, and its value of the assets damaged in the reprisals was lower, than the similar valuations of the other. The defender's bargaining position is probably favored by a higher valuation of the strategic prize. That is, I have probably understated the worth of Western Europe to the United States (or overstated its worth to the Soviets) in postulating a symmetrical valuation. While it
in Western Europe
207
is not a hard and fast rule, each side is likely to value the territory which it already holds, or the allies which it already has, higher than the territories under the control or protection of the enemy. The Soviets could strike in Hungary in 1956 without much fear of a Western mili tary response because of a reciprocally shared expectation that Hun gary was part of the Communist community. The West could not respond militarily because of the grave risk that the Soviets would interpret this act as the signal for general war. Conversely, an attack on Western Europe would jeopardize more values for the West than it could gain for the Soviets. Realizing this, the Soviets might well expect the West to be willing to lose more cities than the Soviets for the prize of Western Europe, and hence they would be more easily convinced than the West that continuation of the war would result in intolerable costs. In addition, U.S. or NATO prestige and honor would have been "placed in hock," so to speak, by the advance threat to retaliate against a city a day. It might be very difficult for the United States, for reasons of its own self-respect, to fail to carry out this threat right up until the completion of the Soviet conquest of all Western Europe. These factors would exercise substantial deterrent effect against Soviet initiation of the attack. Once the attack was under way, how ever, other factors would come into play which would tend to favor the bargaining position of the aggressor. Once having sent forces into West Germany, the Soviets would be substantially committed to continue the attack, perhaps just as firmly, or more so, than the West would be com mitted to retaliation. To halt the attack and retire to the original boundary would mean taking a large cut in prestige and political influ ence. Having once given in, the Soviets' bargaining position in future limited retaliation situations would be grievously weakened. The more territory overrun by the Red Army, the more difficult it would be for the Soviets to pull back. Once they had consolidated con quest of the prize, it probably would become very difficult to dislodge them by further city-trading. It would become progressively more diffi cult, the closer conquest was approached. This is a very important factor which may outweigh all other considerations. It follows that ground forces are an important supplement to a strategy of limited retaliation. The closer the West can come to balancing the Red Army in ground power, the more likely that a limited retaliation threat and strategy would succeed. It seems probable that if the West's ground
208
Deterrence and Defense
defenses were only of token or "trip-wire" size, its bargaining position would be inherently weaker than that of the Soviets. The Soviets might then think they could finish the job in a few days, before excessive re taliatory costs were suffered, and the West, knowing that the Soviets could do this, would be reluctant to begin retaliation. The Western ground forces must at least be large enough to insure that the Soviets must spend enough time completing the conquest to incur offsetting retaliation. Soviet superiority in ground forces would have another important consequence. The Soviets could bear the thought of losing evacuated cities with a fair degree of equanimity because they would offset their losses with territorial gains in Western Europe. The West would have to be certain that the Soviets would retreat to the prewar boundary before the West had suffered unacceptable damage in order to feel a comparable degree of confidence about beginning the city-trading. The West would always have in mind that initiating the reprisals (or continuing them) might turn out to be futile; they would face the prospect that the net result of the operation might be the loss of several cities in Western Europe and the United States and the loss of Western Europe. The Soviets, on the other hand, would find some comfort in the thought that even if their losses went beyond the value of their objective, they would be offset in part by a territorial gain. Mutual knowledge that this asymmetry existed would tend to strengthen the Soviet bargaining confidence and hence its bargaining strength at each stage of the reprisals, and weaken the bargaining position of the West. We must also consider the possibility that the aggressor would threaten to respond to a limited strike with a strike considerably more severe than the one inflicted on him. This would not be a very credible threat, for the aggressor would know that the defending side would not consider such retaliation to be legitimate and would be moved to retaliate with like severity in order to "even the score." If the aggressor held to his stated intention, he would retaliate again at a greater level of damage. Such a strategy would very likely cause a dynamic spiraling of the conflict to all-out or nearly all-out dimensions. The aggressor, if he is rational, would wish to avoid this as much as the defender. Nevertheless, the aggressor would likely make a threat of severe counterretaliation, even though he intended to respond with a strike of roughly the same severity as the one made by the defender.
in Western Europe
209
The threat might exert some depressing effect on the defender's will ingness to begin the reprisals. D. LIMITED COUNTEBFOBCE RETALIATION
Limited reprisals might be threatened or carried out against the enemy's military forces rather than against his cities. Limited counterforce reprisals could take two possible forms: attacks against the enemy's strategic forces, or attacks against his "tactical" installations —i.e., supply dumps, troop staging points, bases for tactical aircraft, lines of communication, and other targets supporting the enemy's army. Counterforce reprisals might be more plausible as threats and more rational as acts than limited reprisals against cities, for two reasons. First, they would be less likely than city reprisals to create emotional stimuli which might lead to a quick and uncontrollable spiral into all-out war. They would allow the initiation of reprisals at a low level of cost and provocation; thus they would provide a maximum of time for bargaining and mutual testing of intentions, before the reprisals rose to unsupportable intensity. Second, they would perform a dual func tion: not only would they constitute a warning of more severe attacks to come, including attacks on cities, if the aggression did not cease; they also would cause physical attrition of the enemy's military forces. Since the targets would be military in nature, the reprisals could more easily be interpreted by the recipient as simply moderate extensions of a limited war already in progress. Although such attacks would be violations of the superpower enemy's "sanctuary," they would not be such blatant violations of the norms of limited warfare as would de liberate reprisals against cities. Limited reprisals against "strategic" targets, such as ICBM sites, long-range bomber bases, reconnaissance satellites, or missile-firing submarines, perhaps would demonstrate more forcefully than tactical strikes that we might be considering moving to the all-out level of warfare. And if we perceived a serious danger that the limited war in progress would explode into all-out war, we might have an interest in reducing the enemy's strategic capability as much as possible before the explosion occurred. Of course, such an incentive would not arise unless we expected the rate of attrition to favor us. It would not favor us, for example, if we attacked land-based missile sites, because we probably would have to fire more than one of our missiles to be con-
210
Deterrence and Defense
fident of eliminating one of the enemy's. More fruitful targets would be bomber bases, air defense and warning installations, and missilefiring submarines. However, counterforce strategic reprisals would be more provocative than tactical reprisals because they would involve making a sharp, discontinuous jump out of the context of limited war. There would be some danger that the enemy would interpret such reprisals as being the opening shots in all-out war—for it is these targets which would have highest priority in such a war—and decide to launch an all-out first strike in response. Moreover, strategic counterforce reprisals would contribute little or nothing to improving our position in the ground force battle, unless of course the bases attacked were being used actively in support of the enemy's tactical war effort. Limited reprisals against tactical military targets inside the boundaries of the Soviet Union would seem more "natural" than strategic strikes, in that they could be considered simply an extension of a limited nuclear war which was going on in the forward ground combat zone. The targets would be traditional "interdiction" targets, attacks upon which would contribute directly to weakening the enemy's strength in ground combat. Such attacks would have a strong rationale in addition to that of simply inflicting punishment as a threat of more punishment to come; hence the threat to make such tactical reprisals would have greater credibility than the threat to make select ed attacks on cities. The actual carrying out of such reprisals would be less provocative, less likely to spiral out of control, than attacks on either cities or strategic counterforce targets, since they would not involve such a discontinuous jump from the previous level of limited war. The targets hit could be considered legitimate military targets. While not incurring a serious risk of an immediate all-out response, and while causing some physical attrition of the enemy's power to move on the ground, tactical reprisals would still serve the bargaining function by demonstrating a willingness to "up the ante" and to con tinue doing so until the other side agreed to settle the war. Beginning at the tactical level would give the West more scope for gradually increasing the pressure—by attacking larger and larger airfields, or by hitting targets involving increasingly greater casualties to civilian and military personnel, etc. Of course, the enemy would reply to such reprisals by counterreprisals to the rear of the combat zone in Europe and probably in the
in Western Europe
211
United States itself. Hopefully, the enemy would exercise the same restraint in his counterreprisals that we had exercised in our reprisals —i.e., he would attack only tactical military targets. We can only conjecture whether the Soviets would consider ports to be legitimate targets for counterreprisal. Certainly, U.S. and European ports used for embarkation and debarkation of troops and equipment would be attractive targets for the Soviet Union from the point of view of sheer attrition of the West's ground fighting capability. If the Soviets did attack ports, and if the Western ground effort had been heavily dependent on a capacity to bring in reinforcements from the United States and Britain, limited tactical reprisals very likely would favor the Soviets in rate of attrition. On the other hand, attacking ports would mean attacking cities; the Soviets might refrain from such attacks for fear of counterreprisals against Soviet cities. And it must be remembered that there would be no point in beginning limited nuclear reprisals of any kind against the Soviet Union unless the Soviets were already clearly superior in tactical capability. Adoption of a strategy of limited retaliation would mean prior recognition that we were unable to win at the local tactical level even with external reinforcements, or a prior decision not to have a winning capability at this level. A limited reprisal strategy is an alternative to such a capability. Although we require enough ground forces to delay the Red Army's advance until a settlement could be reached by limited reprisals and bargaining, such forces probably could be furnished, in the case of Europe, for example, by the present and prospective NATO shield and the extra forces which could be mobilized by the European countries themselves. Any strategy of limited counterforce reprisals, whether strategic or tactical, must stand or fall on its demonstrative or bargaining potency, not on its attritional effects. Either the rate of attrition will favor the Soviets, or the Soviets will have such a tactical superiority that any rate of attrition in our favor which we are able to establish will not be significant. The primary reason for starting limited reprisals against the enemy's forces rather than against his cities is not to wear down his forces (although this is a by-product) but to signal an even tual willingness to inflict more severe damage, with minimum immediate cost and risk to ourselves. In other words, the act of starting reprisals against forces creates a risk for the enemy that we will ultimately attack cities, or that we will step up the attacks against military in-
212
Deterrence and Defense
stallations to the point where the by-product economic and population damage is more than the Soviets can tolerate. Reprisals against forces, especially tactical forces, allow us to demonstrate this possible intent at minimum provocation and at minimum initial damage to our own economy and population. Limited counterforce reprisals might best be made against targets outside the enemy's territory—i.e., against his ships at sea, or perhaps his space satellites. Such attacks would be less provocative than attacks on land installations, because the enemy's territory would not have been violated. They would still serve the purpose of warning, of threatening to step up the intensity of the war. For example, to torpedo an enemy nuclear submarine would demonstrate a willingness to take risky actions which might eventuate in all-out war; yet it would not so engage the enemy's prestige that he would be forced to retaliate against one of our cities. It is possible that a limited local war on land might grow to include quite substantial nuclear fighting at sea or in space as each side sought to bring pressure on the other by demon strating its will to up the intensity of the war. Thus the concept of "limited retaliation" tends to merge into the concept of "force demonstration." E. LOCAL TACTICAL BEPRISALS
If the most likely targets for limited reprisals are tactical military installations, this strategy also merges into limited tactical nuclear war. The latter is most often discussed as an attrition strategy, intended primarily to defeat the enemy's military forces or deny them territorial gains. Yet it may be more reasonable to think of limited nuclear war as primarily a bargaining strategy rather than an attrition strategy, differing from limited strategic reprisals mainly in being restricted to the general battlefield area and its immediate rear. If it is reasonable to consider a bargaining strategy of limited reprisals directly against the Soviet homeland, then surely the rationale behind this strategy would apply, perhaps even more strongly, to geographically limited warfare. It would be illogical to apply this strategy to tactical targets inside the Soviet Union and not to tactical installations in closer proximity to the front lines. If we entertain this idea of limited reprisals in the battlefield area, we need to readjust our thinking about limited tactical nuclear war.
in Western Europe
213
Such questions as whether nuclear weapons favor the offense or the defense, whether they reduce the need for manpower, whether there is an advantage in using them first, etc., tend to become irrelevant when tactical nuclear warfare is cast in a bargaining rather than an attrition context. These questions implicitly assume that nuclear weapons would be used in quite large quantities, although in small sizes and mostly against military installations and forces. But as instruments of bargaining they might be used in small quantities, or one at a time, perhaps with each strike accompanied by a demand for, or offer of, terms of settlement. The targets might be considerably different than in an attrition-type war. For example, to strike at troop concentrations might be considered too provocative, and not in line with the spirit of a bargaining strategy, which is to demonstrate rather than to destroy and kill. Preferred targets might be such things as tactical airfields, naval vessels, and other targets which could be hit with minimum loss of life. If both sides were to accept the idea of using tactical nuclear weapons as bargaining instruments rather than as attrition forces, presumably the pressures to use such weapons first in a "massive" first strike would be reduced; this would reduce the danger that the tactical use of nuclear weapons would quickly become an all-out affair. It follows logically that such an interpretation, if credibly com municated to the enemy, would enhance the credibility of any threat to use tactical nuclear weapons in response to a serious conventional attack. It is not easy to see where a tactical nuclear bargaining exchange would eventually lead, but at least there would seem to be less chance of tactical nuclear warfare exploding into full-scale nuclear conflict, when it is started at a low level of initial destruction and clearly for demonstrative or bargaining purposes, than when it is highly destructive from the start. Perhaps it may be taken as a rule of thumb that the higher the destructiveness of tactical nuclear war at any particular time, the more likely it is to escalate to even greater de structiveness. Thus if nuclear weapons must be used, there is a great deal to be said for a strategy which allows them to be used sparingly and cautiously at first, thus causing minimum provocation, and providing decision-makers with maximum time in which to negotiate a termina tion of the war before the fighting gets out of control.
214
Deterrence and Defense
F. NUCLEAB BLACKMAIL
The threat of limited retaliation is the logical answer to nuclear blackmail when the United States can no longer credibly threaten an all-out first strike and when the country being blackmailed does not itself have a substantial nuclear deterrent force. Both the Soviet Union and its victim would know (or strongly suspect) that the United States would not care to initiate all-out war should the blackmail threat be carried out. A reasonable response would be a reprisal similar in weight of destruction to that carried out against the blackmail victim, and the credibility of such a response would be fairly high. Hence the Soviet Union might well be deterred from attempting the blackmail threat in the first place; if it were made, the recipient would likely consider it a bluff and feel safe in resisting it—at least safer than when the only counterthreat the United States could offer was that of massive retaliation. G. MILITABY BEQUIBEMENTS
Forces for limited retaliation would have to be additional to those required for "minimum deterrence" of all-out attack, since these latter forces would have to be held in reserve. In one version of the strategy —reprisals against cities from the start—the extra requirements might be small, since it is difficult to imagine more than from five to ten exchanges taking place without either the capitulation of one side or the explosion of the conflict to all-out dimensions. In the counterforce versions, the requirements would be higher, since many such exchanges might take place before a decision was reached. However, the number of exchanges which are likely to take place may not be an appropriate yardstick in either case. The outcome of the bargaining may depend on which side is able to threaten the most damage.60 If our surplus over our minimum deterrent forces were sufficient only to cause eventually unacceptable damage to the enemy—i.e., greater cost than his objective was worth to him—while he was able to threaten absolute destruction of the United States in return, the enemy very likely would have the advantage in the bargaining, and his realization of this would reduce the credibility of our limited reprisal 60 It would also depend, of course, on each side's willingness to accept damage, and on each side's image of the other's willingness to suffer costs, as well as on the comparative valuations of the strategic and political stakes.
in Western Europe
215
threat. Thus, while "matching" the enemy's forces is an inappropriate criterion for strategic force requirements when the only purpose of such forces is conceived to be the deterrence of all-out attack, or the deterrence of lesser attacks by the threat of an all-out response, this criterion seems much more plausible in the context of a limited reprisal strategy. Moreover, adoption of a strategy of limited retaliation, like a strategy of "all-out" first-strike counterforce retaliation, would tend to increase U.S. requirements for a minimum strike-back capability to deter an all-out Russian first strike. If the Russians feared the United States might start limited reprisals after being provoked, they would have a higher incentive to strike first than if they had no fear of any form of U.S. nuclear retaliation. The required size of our minimum strike-back capability, it can stand repeating, depends on the strength of the Soviet incentives to strike first at us, and one of these incen tives, perhaps the major one, stems from the Soviets' fear that we might strike first at them. It may be objected that our first-strike or limited retaliation threat is designed to deter the Soviets from tactical moves. They can remove their fear of a strike by us by simply abstain ing from these moves. But situations may arise over which the Soviets have little control that would make it imperative for them to act in some way to save some vital interest or their prestige. If there is a serious risk of U.S. limited retaliation when they do act, their potential cost from acting tactically is higher than if they faced no such risk. The alternative of an all-out first strike at the United States thus improves in attractiveness, compared with tactical attack. Thus, to keep the Soviets' cost of striking first higher than their costs of moving tactically, the United States needs a greater strike-back deterrent against all-out attack when it has credibly threatened limited strategic retaliation, than it does when its only plausible response is a tactical one. The actual initiation of limited reprisals would place a severe strain on the U.S. capability to deter all-out attack. The Soviets' psychological capacity to carry on a restrained, limited-retaliation war might be small; to convince them, and keep them convinced, after we had started limited reprisals, that continuing or terminating this type of war would be less costly for them than making an all-out strike, our deterrent against such a strike would have to be very powerful. There is still another reason why the minimum deterrent require-
216
Deterrence and Defense
ment against an all-out first strike might be higher in the context of a limited reprisal strategy. There would be less reason to expect that a heavy counterforce strike, carried out after several exchanges of limited reprisals had taken place, would touch off all-out indis criminate retaliation, than there would be when such a first strike was made "out of the blue" or even after tactical provocation. Both sides would have become accustomed to the idea of using limited strategic strikes for bargaining purposes. The enemy is more likely to believe or hope that his counterforce first strike would be followed simply by a continuation of the limited bargaining strikes. A heavy counterforce first strike might be conceived as a kind of interlude in a limited reprisal war, carried out to improve one's bargaining position, rather than to reduce the other side's damage-producing capacity to an acceptable level. Thus a greater strike-back capability would be re quired to increase the enemy's prospective retaliatory damage—in case we did retaliate indiscriminately and all-out—enough to offset his will ingness to gamble that we would not do so. The best type of weapon for carrying out limited strategic reprisals would be the long-range or intermediate-range ballistic missile. Air craft would be less suitable, because a large flight of bombers would have to be sent to make sure of getting at least one through the enemy's air defense. The enemy might misinterpret the approach of this flight as the initial blow in an all-out attack, however irrational such an attack might be. Moreover, if more than one bomber managed to get through the defenses, there would be a problem of communication with or between the planes to make sure that the strike did not exceed the planned level of destruction. There would be a possibility, of course, that the enemy would succeed in shooting down a single missile, but if this occurred others could be sent in quick succession. Foreseeable developments in defense against missiles do not indicate that losses from attrition of missile sorties would be great. A civil defense capability would be a very desirable adjunct to a strategy of limited reprisals. The side which was best able to put its population in places of safety—blast shelters in cities, or fall-out shelters outside large cities to which urban populations could be evacuated—would derive a bargaining advantage from this capability. Evacuation of cities, which may be thought questionable as a civil defense measure in all-out war, appears much more feasible in
in Western Europe
217
the context of limited reprisals, because plenty of time would be available to carry out the evacuations. By evacuation and other civil defense measures, carried out in advance of the reprisals or in their early stages, one could demonstrate that one really meant business and add considerably to the plausibility of one's limited reprisal threat. In short, such measures could be used as positive instruments of pressure as well as means of protection. If both sides had good civil defense programs, limited countercity reprisals could be initiated at a minimum level of casualties and provocation, thus increasing the chances that the exchanges could be kept limited, and increasing the credibility of a threat to initiate such reprisals. H. SOME CONCLUDING COMMENTS
I have tried to discuss the strategy of limited retaliation objectively, without taking a position either for or against its adoption. It is not at all clear whether the "pros" or the "cons" are the more weighty. What is clear, however, is that this strategy is entitled to serious considera tion, as one stage in the "spectrum of violence"—a stage which has been neglected in public discussion. In the past, there may have been some justification for denying the possibility of limiting strategic nuclear warfare, because of the widely held expectation that any nuclear attack on a superpower would in evitably touch off all-out conflict, an expectation which was derived, in part, from a technological milieu in which one side or the other might see an advantage in an all-out strategic first strike. If, however, the strategic forces on both sides become substantially invulnerable, so that neither side can rationally consider an all-out first strike, a limited strike is less likely to be taken as sufficient provocation for all-out war, particularly if the intent and rationale of the limited reprisal strategy are clearly understood. The primary purpose of the strategy is to deter ground aggression and nuclear blackmail, by providing a deterrent threat which is more credible than that of massive retaliation. Its secondary purpose is, by its actual application, to bring the war quickly to an end by convincing the enemy that his costs in continuing it are likely to be greater than his potential gains. Because of its cost and risk, the strategy is prob ably suitable for actual application only in response to attacks on areas —such as Western Europe—which the United States considers of highest importance to its own security, although limited reprisals
218
Deterrence and Defense
might conceivably be threatened as a deterrent to lesser contingencies. If the limited retaliation strategy is to provide an effective deterrent, the strategy must be declared and fully explained to the Russians. In case the Russians might otherwise think that their advance to a position of nuclear parity with the United States allowed them to undertake ground aggression with impunity, they must be clearly informed that we do have, and contemplate using, a strategy of punitive retaliation which would produce unacceptable costs for the Soviet Union. They must also be made to understand that we intend to exercise restraint, so that they will not interpret our initial reprisal as the first salvo in an indiscriminate nuclear war. Even if we do not wish to use the limited reprisal strategy as a deterrent or response to Soviet tactical moves, we may be driven to adopt it as a counter to its adoption by the enemy. The Soviets already have made several threats of limited strategic action against European countries, notably in the aftermath of the U-2 incident in May 1960, and even against the United States in the case of Cuba. The Soviets may attempt to use such threats to deter U.S. intervention at the con ventional level, or the use of nuclear weapons tactically, in future limited conflicts. We cannot be sure that the Soviets would not actually carry out such a threat at a relatively low level of provocation if the stakes were high enough; for example, one of the Soviet test shots in the Pacific might be allowed to drop on Johnston Island or a sparsely inhabited area in Hawaii. If we were intellectually unprepared to make any counterthreat or counterresponse other than the implausible and irrational one of massive retaliation, we might be at a serious bargaining disadvantage in such a situation. If we were then to dis cover suddenly that a single strike at some minor target in the Soviet Union was the only response which would preserve both honor and minimum rationality, our failure in advance to think through the rationale of a limited reprisal war would create serious danger that we would not be able to exercise the restraints necessary to keep the war from getting quickly out of control. If both sides understand that an exchange of limited strategic repris als is a possible outgrowth of limited tactical war, the possibility that an accidental strategic blow might immediately set off an all-out exchange would be much reduced. The recipient of the blow might then believe that the opponent was initiating limited strategic reprisals. It would respond in kind, but the side which had fired
in Western Europe
219
accidentally would be anxious to explain that the initial strike was an accident—therefore it might not counterretaliate but attempt to communicate instead. It is to be hoped that its explanation would be accepted, and that the strategic strikes would cease after one exchange.61 In sum, the strategy of limited retaliation would provide a means of credibly threatening unacceptable costs to the Russians, despite their achievement of a large strike-back capability for all-out war, which they might otherwise think made their own homeland immune to attack. It would extend the range of possible response by the West and preserve a retaliatory role for SAC beyond that of replying to an all-out attack on the United States itself. It would increase the risks for the Soviets in undertaking limited aggression and it would consti tute a valuable counter against nuclear blackmail attempts. Whether, after a conflict began, the threatened strategy would be rational to carry out, may be open to question, but at least it is more likely to be so than massive retaliation. A clear definition and explanation of it in advance would reduce the risks that it would spiral out of control. 6. COMBINATION OF ALTERNATIVES It is quite obvious, of course, that the various alternative forces and strategies we have just discussed need not be mutually exclusive. The best over-all strategy and force combination for NATO may be a judicious combination of several. It is important, therefore, to under stand how the different types of forces and forms of organization may support or perhaps weaken each other in the function of either deterrence or defense. As we emphasized earlier, NATO's basic choice in the face of the declining credibility of the first-strike threat of SAC is whether to strengthen the credibility of deterrence by the threat of strategic retaliation, or to strengthen its capacity to defend its territory by tactical means. The choice is not a clear "either-or": some minimum capacity to threaten strategic retaliation is necessary to resist nuclear blackmail whatever the size of the tactical forces, and some ground forces are necessary, however potent the nuclear deterrent, if only to handle small border incidents and probing actions. The problem is to determine what mixture of forces beyond these minimums is likely 61 Robert E. Osgood makes a similar point in "Deterrence by Strategic Nuclear Strikes," mimeographed, 1960.
220
Deterrence and Defense
to best satisfy NATO's strategic needs in the coming years, and how these forces should be organized and controlled. The two problems of deterrence of and defense against ground invasion, and resistance of nuclear blackmail, are closely interlocking. For example, it might be argued that the surest and cheapest way of dealing with ground invasion would be to have conventional forces which could surely hold the Red Army at the East-West frontier, at least long enough for reinforcements to be mobilized and brought to the front. But if, at the same time, the United States had just barely enough strategic nuclear power to deter a strategic attack on itself, if the Soviets had considerably more than this minimum, and if the European allies had no strategic forces of their own, NATO's conven tional defense capability could easily be disrupted by nuclear black mail, if not by nuclear attack in Europe. Fearing that the United States could not afford to use up any of its strategic nuclear forces in retalia tion against Soviet nuclear attack in Europe, individual European countries could be blackmailed into withholding their conventional forces from action in support of an invaded country. Or if a successful defense on the ground depended on the use of tactical nuclear weapons, Russia could use its strategic preponderance to deter the use of these weapons. It follows that the military security of Europe is questionable, even with a sufficient capability for tactical defense, unless the United States has a surplus of strategic power beyond a minimum strike-back capability, or the Europeans themselves, either individually or collectively, have minimum strike-back capabilities. We can be confident, however, that the United States will have somewhat more strategic power than the bare minimum which it must hold in reserve to protect itself against strategic attack—at least we can be confident that the Russians will think we have such a sur plus. It is at least conceivable that we might be able to build a strategic force large enough to make the Russians at least suspect that we had a sufficient first-strike capability. If the Europeans were con fident that the Russians did entertain some substantial fears of a U.S. first strike after the Soviet carrying out of a blackmail threat against a European country, the problem of blackmail would be solved. The U.S. first-strike capability might have to be significantly greater to deter Soviet major ground invasion in Europe than it would have to be to stiffen the Europeans' resistance adequately against blackmail. Up to a point, ground forces in Europe and a U.S. first-
in Western Europe
221
strike capability are complementary in the deterrence of Soviet ground attack; ground forces raise the necessary provocativeness of a Soviet conventional aggression, while increases in the U.S. first-strike capability lower the degree of provocation which (the Russians would fear) would trigger off a U.S. first strike. But when the credibility of a U.S. first strike drops below a certain threshold, no amount of ground force expansion can restore its necessary degree of credibility; then the security of Europe against ground attack may have to rest on ground forces able to defend European territory against the full strength of the invader. When and if this should occur, U.S. strategic forces beyond its own minimum deterrent requirement must be justi fied on grounds other than deterrence by the threat of an all-out first strike. There are several grounds, however, on which such addi tional strategic forces can be justified, aside from their value in "defense by offense" after a Soviet strike against the United States. One of these possible justifications would be provided by adoption of a strategy of limited retaliation. A U.S. capacity for limited retaliation might provide the Europeans with enough resolve to resist nuclear blackmail if the United States' surplus of forces beyond its minimum strike-back capability were large enough to destroy Russian assets (including lives) beyond the Soviet total valuation of Western Europe. It would have to be at least this large (preferably larger), because otherwise the Soviets might be willing to carry out their blackmail threat against one victim with high confidence that this demonstration would collapse the will of other European countries to resist blackmail threats. It is possible that a capacity for limited retaliation might deter Soviet ground invasion, provided the Soviets could be convinced of the United States' determination to carry out the strategy, and pro vided they thoroughly understood its rationale. The credibility of the threat of limited retaliation would be sensitive, however, to the degree of quantitative symmetry between the U.S. and Soviet strategic forces. If the United States had only enough beyond its minimum reserve (necessary to deter attack on itself) to inflict damage off setting the Soviets' valuation of their conquest of Western Europe, but the Soviets had the ability to threaten and inflict a great deal more damage on the United States without compromising their mini mum deterrent reserve, the Soviets might believe the United States would not hold out for more than two or three exchanges. Thus, if a
222
Deterrence and Defense
strategy of limited retaliation is adopted, there is something to be said for quantitatively matching, or at least approaching, the enemy's strategic forces as a proper criterion for equilibrium in the balance of terror. It would be extremely risky to rely on a strategy of limited reprisals as a full substitute for an adequate ground force defense capability, even if U.S.-Soviet strategic capabilities were symmetrical. Substantial ground forces are required as a complement to a limited reprisal strategy, to delay the enemy's tactical advance long enough to allow negotiations to take place and to minimize the enemy's sacrifice of gains and prestige in agreeing to terms. If the United States does not have a sufficient first-strike counterforce capability, but yet possesses a substantial one, its existence would strengthen our bargaining position in a strategy of limited reprisals because the Russians might fear that at some point in the limited exchanges we might become sufficiently provoked to make an all-out first strike. In general, the greater our first-strike counterforce capability—provided it is large enough to establish at least the possi bility of an all-out first strike in some circumstances—the greater our capacity to bargain at lower levels of violence by threatening to "raise the ante" or by actually doing so in small gradations. If future NATO strategy is to include the diffusion of strategic nuclear weapons among Continental members of the alliance, on either an individual or a collective basis, there are significant ways in which the deterrent effects of such distribution could be supple mented by ground forces in Europe and by the strategic power of the United States. There is no need to repeat here the ways in which European strategic forces might trigger a U.S. all-out first strike, although it is worth reiterating that this triggering effect does demand a fairly substantial U.S. first-strike counterforce capability. A U.S. threat of limited retaliation might strengthen the will to resist black mail of those countries in Europe which opted for non-membership in the nuclear club, or those which were not certain that any of their strategic forces could survive a Russian first blow. Fairly strong ground forces in Europe would perform the same complementary deterrent functions vis-a-vis European strategic forces as they presently do in relation to the United States strategic deterrent. For example, a multinational shield of ground forces would involve several European countries in the tactical war, making it more
in Western Europe
223
likely that they would use their strategic forces in aid of the invaded country. In the case of a collectively organized European strategic force, this involvement of their own armies would make it more diffi cult for members of the group to renege on their prior agreement to fire the strategic force simultaneously under specified conditions. A large collective ground force would narrow the range of contingencies which would have to be covered by the European Group strategic deterrent, thus easing the problem of prior agreement on the number and type of contingencies in which the Group's strategic forces would be launched. There is a countervailing consideration, however: by delaying the enemy's advance, ground forces would allow more time for the Group's political organ to develop cold feet and scrap the prewar agreement to fire under stated conditions, or to countermand the prewar delegation of authority to a military commander—arrange ments which would have been made, after all, for purposes of deter rence, not as rational war plans. Rationality, after the war had begun, might dictate not firing the strategic forces, and the delay enforced by the ground forces would provide time for rational considerations to prevail and for the enemy to point out convincingly why retaliation would not be rational. In such circumstances, the Group deterrent might disintegrate into a system of national deterrents; then, after taking over their initial victim, the Soviets could feel less risk in continuing on to further conquests, for each European nuclear power would have demonstrated its unwillingness to incur the costs of nuclear war. This consideration would reinforce the reluctance of individual European countries in the rear of the probable initial battlefront to join a Group deterrent scheme; by maintaining nuclear independence they would avoid having to reveal, soon after the enemy's initial move, their own lack of resolve. They might then hope to deter the enemy from crossing their borders after conquest of the initial victim, especially if the latter had been so foolish as to fire its strategic weapons, thus indirectly demonstrating to the enemy that he had underestimated the will of the European countries gener ally to carry out retaliation. On the other hand, in peacetime, the existence of a strong integrated ground force might help to hold the NATO alliance together against the disintegrating effects of the possession of independent strategic nuclear forces. As Robert E. Osgood has pointed out, "Only insofar as NATO has a strategy that provides its members with other options
224
Deterrence and Defense
besides total war and ineffective resistance will they have sufficient incentive for relying on NATO rather than on their own nuclear power."62 Integrated planning at the tactical level might foster inte grated planning at the strategic level and generally help develop that sense of community which would be essential to the credibility of an alliance pledge that an attack on one member—whether conven tional or nuclear—would be regarded as an attack upon all. A strategy of limited retaliation to ground invasion might be fol lowed by a European Group deterrent force. The threat to do so might be more credible than a similar U.S. threat because (1) the European countries would be anxious to find some rational use for their strategic forces before they were overrun, and (2) being faced with a threat to national survival itself, and already suffering considerable war damage, they might feel less qualms than would the United States about beginning a round of limited reprisals. A limited reprisal strategy for a collective European force would not be subject, in the same degree as for the United States, to a bargaining disadvantage if the Soviets had a considerable quantitative preponderance in weapons. Even though the Soviets could threaten more damage to the European countries than they could threaten to inflict on Russia, the Europeans presumably would value their own independence much more highly than the Soviets would value its extinguishment. Thus a large asymmetry in values at stake might offset an opposing asymmetry in ability to inflict damage. In cases of nuclear blackmail against a single member of a collective European deterrent organization, the threat of limited reprisal would be a more credible counter than the threat of an all-out response.63 62 Robert E. Osgood, "NATO: Problems of Security and Collaboration," loc.cit., pp. 127-28. 63 This point is made by George W. Rathjens, Jr., "NATO Strategy: Total War," in Klaus Knorr (ed.), op.cit., p. 88.
4 Deterrence and Defense in the 66Grey Areas" WHEN we turn away from the area of most vital importance to the United States—Western Europe—and look to areas such as the Middle East, Southeast Asia, and the Far East, two points of difference stand out sharply. One is the distinctly lower credibility of the U.S. threat of "massive retaliation." The other is the different character of the enemy threat to these areas: the greater danger of "indirect aggres sion" or political subversion as compared with the risk of overt military aggression. Since the primary danger is not military, U.S. interests in the "grey areas" must be protected chiefly by non-military methods. Nevertheless, military forces do have a role to play, not only in deterring and defending against overt military aggression, but also in dealing with more ambiguous challenges which may combine political, economic, and military elements.
A. Overt Attack It is hardly believable—even with the maximum conceivable counterforce capability or the maximum degree of irrationality in the U.S. leadership—that the United States would deliberately initiate all-out war in answer to any attack in Asia. A possible—but only barely possible—exception would be that of a major attack on Japan by Soviet forces. Japan is a special case, not just because of its strategic value, but chiefly because of the United States' moral and emotional involve ment with Japan as a consequence of the occupation stewardship. The Soviets could not afford to write off the possibility of immediate all-out retaliation if they should invade or attack Japan with the obvious intent of conquest; the probability would be large or small, depending primarily on the counterforce capability of the United States. We might not regard Japan as being of the same order of vital interest as Western Europe, and yet still feel that the degree of bold ness exhibited by the Soviet Union in attacking Japan presaged an imminent attack on Western Europe and perhaps the United States itself—and thus be motivated to strike first in an all-out war which
226
Deterrence and Defense
seemed inevitable or very probable. Something similar could perhaps be said about an unambiguous Soviet attack on India, and there are conceivable Soviet moves in the Middle East which would carry some risk of triggering a massive response. There is also a purely logical argument which tends to support the idea that the massive deterrent threat is still useful even with respect to these areas of lesser importance. The credibility of the threat is very low, of course, because the enemy will know our valuation of the areas is not anywhere near commensurate with the costs we would suffer in carrying out the threat. But the enemy may also place little value on these areas. Consequently, only a little credibility may be enough to deter him. He may not want to take even small risks with respect to them, especially since it is generally in these areas that his non-military means of expansion are likely to be most fruitful. If the enemy's valuations are symmetrical with ours, it follows that although the massive retaliation threat is much less credible in Asia than in Europe, it may not be less effective. This type of logic should not be pushed too far, however, and it may be entirely misleading. The chances are that below some rather low level of "likelihood," the probability of our retaliation becomes in significant in the Communist calculus—for practical purposes equal to zero—just as above some high level of subjective likelihood it may be considered a practical certainty. The obvious willingness of the Communists to engage the West on various secondary and tertiary fronts since World War II—despite the U.S. monopoly of atomic weapons—seems to bear this out. Thus, the Soviets probably feel, considering the massive retaliation threat alone, that there is a range of minor ventures which they can undertake with impunity, despite the objective existence of some probability of retaliation. Of course, the most likely enemy in Asia is Communist China, or one of the lesser satellites—North Vietnam or North Korea. The threat of limited nuclear retaliation or limited tactical nuclear war fare may be a fairly effective deterrent against major, overt, aggression by Communist China, so long as China does not have a nuclear capability of her own. But would it be in the interest of the United States actually to use nuclear weapons in Asia? Some circumstances are conceivable in which they might have considerable military utility —for example, a Communist Chinese attack in Southeast Asia, where port and airfield facilities are likely to be poor, and where conventional
in the "Grey Areas"
227
forces in the area probably would be inadequate. However, the political costs of their use would be serious. Moreover, if we initiated nuclear warfare, there is a strong chance that the Chinese would acquire nuclear weapons from the Soviet Union. They might even get longrange strategic weapons and use them against targets outside the battlefield area—for example, against our bases in Japan, Okinawa, and Hawaii, or even against the continental United States itself. If we exploded atomic bombs on Chinese territory, it would be difficult for the Russians to deny the Chinese the capacity to retaliate in kind against U.S. territory. The risks of escalation in this type of situation are only too evident. If at some future time we became engaged in an all-out war with a China plentifully supplied with ICBM's, we would run a serious risk of triggering an all-out first strike by the Soviet Union as a result of our having depleted our strategic capability to the point where the Soviets possessed a sufficient first-strike capability. Confronted with such a situation, or the possibility that it might arise, we probably would be deterred from initiating nuclear war even at the tactical level. A limited strategic reprisal strategy, applied against a nucleararmed China, runs into the objection that the Chinese would surely have the advantage in the bargaining because of their casual attitude concerning the value of human life. The threat to initiate such reprisals might have some deterrent effect if the Chinese had sufficient con fidence in our irrationality, but to carry out the threat if China had a substantial intercontinental nuclear capability would be very dan gerous. If she did not have such weapons, there might be some chance that the Soviets would not supply them, fearing that they might even tually be turned against Russia, but we could hardly afford to act on such a speculative consideration. Against such doubts concerning the feasibility of using tactical nu clear weapons or a limited reprisal strategy in Asia must be placed a countervailing factor: if the United States demonstrated a willingness to use nuclear weapons in Asia, considerable support would be given to its threats to use them in Europe. This gain might be enough to off set the risks and political costs of using them in Asia. The following thought would cross the minds of the Soviet leaders: "If the United States is so committed to a nuclear strategy that it will use nuclear weapons in conflicts where it has so little at stake, it will surely use
228
Deterrence and Defense
them if we start a conventional war in the much more critical area of Western Europe." Nevertheless, for defensive action in Asia we shall probably have to rely chiefly on conventional types of ground, sea, and tactical air forces. What criteria should determine the requirements for such forces? A simple answer would be that we should have a capability to deny the enemy any territorial gain even if he activated all his limited war capabilities at once. But even if there were no possibilities of nuclear deterrence, such an "absolute" defense capability would be excessive. If we possessed capabilities of this size, they could be deployed at every trouble spot in significant quantities; the enemy would then face a shield at each point which not only insured that we would respond militarily, but could deny him success with that portion of his forces available for action at that point. If he attempted to redeploy to strengthen his offensive power at one point, he would leave himself open to counterattack at the points from which the reinforcing troops were removed. Alternatively, assuming that our facilities for movement were as good as his, we would redeploy to match his redeployments. Facing such strength, the enemy would very probably be deterred, and if he were not, he would be defeated. This force would be excessive because it would involve practically no risk at all, and, as we have pre viously argued, there is always a range of contingencies in which risk is rationally substitutable for preparedness. Short of this "absolute" capability, we might concede the enemy an initial success if he attacked some areas (presumably the least vital ones) while posing the threat of counterattack following additional mobilization. In other words, as compared with the "absolute" posture, we would accept perhaps a somewhat greater risk of attack plus higher costs in defense (since counterattacking is probably more costly than defending a line) in exchange for lower costs of peacetime pre paredness. Alternatively, we might decide in advance not to defend certain areas (even though, if we had the forces, we would be willing to defend them), hoping that the possibility of our post-attack mobilization and counterattack would deter the enemy from attacking them. Or we might rely on a possibility that the enemy would be deterred by various uncertainties: concerning whether he really had sufficient superiority in the balance of forces to count on victory, whether we would fight
in the "Grey Areas"
229
and inflict substantial costs even though we could not win, or whether the prewar political alignments would stay put during the war. We might rely on the enemy's having enough hope for success by non-mili tary means so that he would wish not to absorb the costs of war, even though he were sure of victory. Or we might count on an indirect de terrent: the enemy's fear that a successful aggression in Asia would spur the United States to undertake a really massive program of economic and possibly military aid—thus vastly reducing the Communists' chances of future successes by political and economic penetration. If limited war forces are well below the "absolute" level necessary to defeat any conceivable limited aggression, a choice is raised as to whether the available force should be deployed at the trouble spots or in a central strategic reserve. Forward deployment exercises a substan tial deterrent effect by insuring that the United States will fight for the area. Like the NATO shield, they pass the initiative with the accom panying burden of responsibility for assessing the uncertainties and risks of war. The certainty of involvement of U. S. forces raises at least the specter of nuclear war. In the absence of forward deployments, the enemy might try to deter our response by initially committing forces of such size that they raise the probable cost of our response higher than our peacetime valuation of the objective. The enemy must recognize that as a result of the commitment of our prestige in the form of Ameri can combat forces, we would be willing to suffer considerably higher costs in defending an area than would be warranted by the value we would otherwise attach to it. The two U.S. divisions in Korea and the patroling of the Taiwan Strait by the Seventh Fleet are the prize ex amples of forces in Asia which exercise this "trip-wire" or "plate-glass window" deterrent effect. If U.S. forces are deployed on the territory of a country threatened with rebellion, they may help to deter such rebellion by providing a potential back-up for the local police forces and army. If a rebellion does break out, they can help to squelch it in its early stages. Another advantage of forward deployment over strategic reserves is that it per mits American troops to get accustomed to the climate and terrain where they may have to fight, and it allows combined training and maneuvers with the local forces. Finally, forward deployment reduces dependence on sea and air transport, which may prove to be inadequate for effective early intervention, and may be subject to serious inter diction.
230
Deterrence and Defense
But forward deployment may detract from the over-all defense utility of the available forces by reducing their flexibility. Forces lo cated at the armistice line in Korea, for example, may not be as readily movable to Southeast Asia, say, as troops on Okinawa or Hawaii which are perhaps closer to central transportation facilities and not committed to the defense of a specific sensitive front. The appropriate balance be tween forward deployment and central reserve depends on a variety of factors, including the available transportation, the degree of threat facing the area from which forces would be moved, and the ability of local indigenous forces to hold a line temporarily. High mobility would seem to enhance about equally the deterrent effects of forward trip wires and the defense capabilities of central reserve troops. In the former case, the enemy's contact with the trip-wire is more likely, with fast transportation, to bring about the projection of sufficient U.S. forces into the area to defeat the aggression. And, of course, there would be little point in having a central reserve of active forces if these forces could not move to a trouble spot until after the enemy had already consolidated his conquest. The degree of threat facing an area obviously affects the ability of forces to move out of that area to some other conflict point. For ex ample, if the U.S. shield in Korea were depleted to deal with a renewed war in Indo-China, this would dangerously increase the risk of another outbreak of conflict in Korea. Local indigenous forces will exercise some deterrent effect if they are strong enough to defend successfully or to hold until outside as sistance can arrive. The strength of such local forces, which may de pend on how much aid in equipment we have given them, naturally affects the size of the U.S. forces required for successful defense. Of course, the deterrent value of having U.S. forces deployed "under the gun" must be weighed against the political cost and feasibility of such deployment. Past experiences in Europe, Japan, and Okinawa in dicate that local populations do not always take kindly to the presence of American troops. This would be an even more serious obstacle, per haps, in countries such as Thailand, Pakistan, and Iran, where nation alistic feelings are strong. And it would be completely out of the question in neutralist countries such as Burma and India which—al though the United States interest in defending them is probably as strong or stronger than in defending some of the other countries with which we have signed formal alliances—have refused to associate
in the "Grey Areas"
231
themselves with these alliances or to accept military aid in materiel. Finally, the presence of U.S. forces on the territory of an ally would tend to give some point to Communist propaganda charges that the professed U.S. interest in the ally's security is merely a front for oldfashioned "imperialism." If the U.S. contribution to the ground defense of some countries in the "grey areas" must come from a central strategic reserve, located nearby or in the United States, a strong case can be made for the build ing of local stockpiles of equipment which U.S. reinforcements would use when they did arrive on the scene. Such stockpiles could consider ably reduce the sea and airlift requirement for moving the U.S. forces, and might also furnish replacement equipment for the local forces. Their value appears especially great when one considers the risk that Soviet submarines, operating clandestinely under the flag of an Asian proxy, might play havoc with our sea transport. The choice between local deployment and deployment in a strategic reserve also depends on the availability of airports in the threatened country, and their location relative to the zone of battle. Airports which are likely to be overrun cannot be counted upon. And, of course, the availability of a U.S. airlift also affects the choice. We shall not go into the airlift debate here except to note that such capability now appears to be well below what would be required for quick action in a situation where substantial numbers of U.S. troops would be needed.1 If there is some degree of warning before the fighting begins, or if the local troops can fight a strong delaying action, the strain on an airlift is reduced.
B. Indirect Aggression The problems of deterrence and defensive action in the Middle East and Asia are vastly complicated by the enemy's opportunities to practice indirect or ambiguous aggression. The existence of large disaffected elements within a country may make it possible for the Soviets or the Chinese to advance their objectives without committing a clear-cut, overt aggressive act such as is traditionally regarded in the West as a casus belli. Nationalist parties, or parties with grievances against the established government, can be encouraged by propaganda, they can 1 According to General Lyman L. Lemnitzer, Chief of Staff of the Army, the United States has enough airlift to move one division from this country to the Middle East within from ten to fourteen days. Honse Committee on Appropriations, "Department of Defense Appropriations for 1961," Hearings, op.cit., Part 2, p. 418.
232
Deterrence and Defense
be provided with leadership by infiltrated agents or domestic Com munists who have been trained in the Soviet Union or China, and they may even be clandestinely supplied with arms. Once a rebellion or insurrection is under way, the established government may be intimi dated, and inhibited from effective action, by the threat of outside as sistance to the rebels, by "volunteers" or otherwise. The United States and other Western powers may be inhibited from providing assistance to the government by the fact that the disturbance appears on the sur face to be an internal affair. Communist adeptness in this sort of activity was well demonstrated in the Greek Civil War, the Czechoslovak coup d'etat of 1948, the war in Indo-China, and, more recently, the subver sion of the armed forces in Syria. Perhaps the best way to combat such indirect aggression is to nip it in the bud by well-conceived programs of counterpropaganda, eco nomic aid, and other non-military measures. There are certain ways, however, in which a strong capability for military action may also con tribute. For example, a U.S. capability for effective intervention to put down a rebellion may deter the rebel leaders from acting, as apparently was the case when the United States demonstrated an effective capacity and will for action during the Lebanon crisis of 1958 and the Jordanian crisis in the preceding year. Such a capability may also counter the Soviet or Chinese threat to aid the rebels, thus giving the government confidence in standing up to such a threat, depriving the rebellious forces of the capability to use the threat credibly, and reducing their confidence in being able to call on external support. Thus reinforced, the local government may feel able to act with greater dispatch and firmness against the rebels themselves. Presumably, such deterrent and defensive benefits deriving from U.S. capabilities to intervene would be greater if our forces were actually stationed on the soil of the coun try threatened with insurrection. Military materiel aid to the established government may also strengthen the government's ability to deal with internal threats. Spe cial types of aid, such as tear-gas guns and small arms, may be needed for this particular purpose, although larger and more dramatic types of weapons, such as tanks and jet planes, may exert a sobering and deter ring effect on the rebels. Special training may be given to the country's armed forces and police to help them deal with internal revolution. Military aid may also induce a spirit of loyalty to the West among the country's military leaders, and prevent their subversion by Commnriist
in the "Grey Areas"
233
elements. When a government is able to deal effectively with internal disorder, this disorder is unlikely to reach the dimensions necessary to encourage its exploitation by the Soviets or the Chinese. Finally, if the government contains pro-Communist officials who may attempt a coup d'etat or attempt to press their policy demands on the government leaders by threatening to invoke "the street," over which they have control, such threats would be futile in the face of a government's con spicuous ability to deal with "the street." In some cases, failing deterrence of the rebellion, or failing success ful early action against it by the local government, U.S. forces might be used in support of the local forces. If participation by external forces on the insurgents' side were not clearly evident, such action would have to be weighed against possible political costs arising from charges of "imperialistic intervention." If the action were taken in an alliance context, in concert with forces from other local governments in the area (e.g., action under SEATO auspices), these political costs would be lower. Such intervention should probably be undertaken as early as possible in the fighting, while the local government still is the dominant political authority in the country. If action is delayed until the govern ment has collapsed or is on the point of collapsing, or until the rebel forces have already established themselves as a "government" with con trol over the majority of the population, the intervention will have lost much of its moral basis, and there may be greater danger of provoking a larger war. After the establishment of a new "people's democracy" claiming legitimate authority over the country, the war will have lost most of its ambiguous character, and the Soviets or Communist Chi nese may find it more difficult to withdraw their support or permit the new government to be overthrown. The war will then have taken on a more normal international character, with the Chinese or Soviets com mitted to support an allied government; intervention then will be con sidered much more provocative. When an aggression takes these ambiguous forms, the debilities of the threat of nuclear reprisals as a deterrent become even greater than those previously mentioned in discussing the case of overt aggression. When, for example, the aggression takes the form of covert assistance to an indigenous rebellion, it would be psychologically very difficult for the United States or the Western powers to carry out nuclear re prisals against the country which was aiding the rebels or thought to be the real instigator of and command center for the rebellion, because the
234
Deterrence and Defense
evidence that it was helping or was the instigator would not be clearcut. The triggering of nuclear reprisals requires a very clear-cut and provocative act of aggression, so that the locus of responsibility for the act and hence the moral basis for the reprisal are clear to all, including those who have threatened to inflict it. Nuclear reprisals take on to a peculiar extent the character of punishment for a crime, and, as in domestic criminal law, when there is "reasonable doubt" that the ac cused is in fact guilty, the community—in this case, the international community—will not give its moral sanction to the punishment. Am biguity about the locus of responsibility reduces the credibility of a threat of reprisal in still another way. An important ingredient in the incentive to carry out nuclear reprisals is the moral and prestige costs that would be suffered in not doing so, owing to a previous declaratory threat specifying the act which would incur the punishment. When it is not clear that the enemy has in fact performed an act which meets these specifications, the moral and prestige costs in reneging on the threat are lower, since the threatener can justify his failure to retaliate by pointing out that the evidence that the specified "crime" has been committed is inadequate. In many cases, the threatener may be glad to have such an excuse, since the implementation of the threat may result in costs from counterretaliation which exceed the value of the objective at stake. Since the potential instigator of "indirect aggression" will realize this, and since it will be easy for him to perform the act in a way which does not meet the specifications of the threat, the threat will not seem very credible. Even if the threatener himself has clear evi dence that the specified act has been performed, execution of the threat will be inhibited if this evidence is not clear to the rest of the world— for the moral and prestige costs of reneging depend on what the rest of the world thinks, not upon the certainty of the evidence in the threat ened own mind. Another closely related point is that the government which is threat ened with reprisals must clearly have the capacity, in advance of the aggression, to prevent it from taking place. If it can appear not to have such control—e.g., by suggesting that the action would be taken by "volunteers"—the effectiveness of the threat to inflict reprisals will be undermined, since the responsibility of the government will be uncer tain. Again, the important point is not whether the threatener believes the government has control or not, but whether the rest of the world so believes.
in the "Grey Areas"
235
Even though a particular government is universally recognized to have had control of the forces making the initial aggressive moves, if it can appear to have lost control after the initial moves, the will to exact reprisals will be weakened, since part of the rationale for beginning the reprisals (besides the value of avoiding the moral and prestige costs of reneging) will be to persuade those in control to desist from their aggres sive activity. If it can be made to appear that control has shifted to an autonomous band of "volunteers," reprisals against the home country may have little effect, since the government will derive a bargaining advantage from its claim that it no longer exercises control over the forces in question. Reprisals against the "volunteers" themselves may be infeasible, since they may not present good targets, the reprisals would have to be made on the territory of friends or allies, it may be impossible to establish communications with the "volunteer" leaders for bargaining purposes, and because, after all, it is the home govern ment which is responsible, so effective bargaining can only be carried on with that government. A conventional ground force capability is not subject to most of these infirmities as a deterrent to ambiguous aggression. The deterrent effect of such forces is related to the capability to prevent the aggressive or rebellious forces from achieving their ends. The ultimate source of di rection and control of the aggression or disturbance does not have to be ascertained; at least such determination is not as crucial to the will to carry out the response, and consequently to its credibility, as in the case of the threat of nuclear reprisals. Such forces are not used on or against the territory of a country outside the fighting theater which is thought to be the directing center of the aggression; therefore any un certainty about this country's degree of control of, or responsibility for, the aggression is not crucial. The forces are used directly against the opposing ground forces, on the territory of the country where the fighting is taking place. The initial act of aggression does not have to be clear-cut. The mission of conventional ground forces is simply to prevent the loss of a position which we consider important to our se curity; the circumstances surrounding the initiation of the conflict are therefore not nearly so relevant to the credibility of this response as to the credibility of a threat of limited reprisals. For the same reason, the problem of clearly locating the central command center of the opposing forces for purposes of bargaining is not critical. In addition, of course, the costs of fighting and the risks of spiraling are much less than in a
236
Deterrence and Defense
limited reprisal strategy, thus increasing the relative credibility of a conventional response. Finally, apart from the distinction between the two strategies for the purpose of deterrence, conventional ground forces provide a capability which is much more likely than nuclear reprisals to be able to prevent losses and enemy gains should deterrence fail.2 If the Communist side derives a military advantage in Asia by its capacity to fight "ambiguously," through the use of volunteers, proxies, covert aid, disguised forces, and the like, are there any possibilities for the United States to counter this advantage by developing similar methods? One advantage of ambiguous methods for the other side is that it allows backing down and retreating with minimum loss of face. The United States, in using only organized forces, finds it much more difficult to accept limited defeats or to bargain for a compromise settle ment during the course of the war. To some extent also, U.S. use of organized forces, heavily symbolic of the national prestige and honor, tends to decrease the willingness of the other side to "back down," be cause the involvement of our own national presence and prestige tends to involve theirs as well. The world might be a safer place to live in, if, at these minor levels of conflict, both sides could find a way of fighting which lowered the engagement of national honor and prestige. This would decrease the extent to which stimulation of nationalistic emo tions would hinder rational bargaining in limited war and would de crease possibilities for limited wars expanding in violence until their costs to both sides became incommensurate with the original objects of the conflict. Of course, this line of reasoning follows the logic of defense—mini mizing losses and maximizing gains should war occur—not the logic of deterrence. For deterrent purposes, we would want our national pres tige involved to the greatest extent at every trouble spot, to increase the provocativeness of the other side's moves and to raise the chances that we would be forced to respond in a way which would produce unacceptable costs for the enemy. We would want "trip-wires" of com bat troops wherever trouble might occur, we would want our support ing tactical air forces to have U.S. markings clearly painted on their wings, we might want to put naval vessels in the ports of threatened 2 Robert E. Osgood makes similar points in "A Theory of Deterrence," op.cit., p. 11. See also my "Deterrence by Denial and Punishment," op.cit., pp. 32-33.
in the "Grey Areas"
237
countries with the flag flying high, and we would want to make clear declarations of our commitments to defend these countries. But suppose we explore the obverse logic of having a capacity to apply force when needed with minimum symbolism. Are there any possible American counterparts to the Communist "volunteers," for in stance? Such might have been useful in Hungary in 1956 or in Indo nesia in 1957. Our military aid missions abroad might serve some such function. These missions normally do not exert any significant trip-wire deterrent effect. In the symbolism of international military life, they are technicians; they do not invoke the national presence and prestige as do fighting troops, although of course they do exercise some deterrent effect by indicating U.S. interest in the defense of the country where they are stationed. A question worth asking is whether these missions might clandestinely operate as combat forces in case of attack on their host countries. Their assignment to the defense of a sector of the host country's frontier in case of war need not interfere unduly with their peacetime training activities. Or perhaps they should be mixed through out the host country's troops in case of battle, to avoid their being identified as organized American units. Individual members of these missions might secretly be assigned positions of subordinate command in the host country's armed forces, provided, of course, that this was politically acceptable to the host country.3 If such an idea were to be accepted, there would be a case for considerably expanding the military missions. Similarly, military aid in equipment "involves" us less than overt participation by American fighting personnel. It would be useful, in many cases, to have a capacity quickly to expand shipments of military aid to a government threatened by indirect aggression. However, the manpower and skills of the recipient may limit the amount and types of aid which could usefully be given. Particularly in the case of heavy and complex equipment, we might wish to expand our contribution by sending along American personnel, ostensibly as "technicians" or "ad visers" but actually as operators of the equipment. For example, the tactical air force of the friendly government might unobtrusively be augmented by fighter aircraft flown by American pilots. Small naval vessels might also be furnished with American personnel in positions 3 In
the early days of the Korean War, some members of the U.S. military mission assumed de facto command of the South Korean units to which they were assigned.
238
Deterrence and Defense
of command, and perhaps serving as the nucleus of a crew. Experts in sabotage and intelligence techniques might clandestinely join the "mili tary mission," and even some American combat troops might be brought in under the organizational umbrella of the mission during a crisis. Such suggestions may seem fanciful, but if there are likely to be cases where we would wish to match the Communist great powers in the ambiguity of our involvement, they would seem to be worth con sidering. Such a capability perhaps would be most desirable when the group we wished to aid was a rebel army operating against a Commu nist government—e.g., in North Vietnam or North Korea. Overt inter vention by regular American forces would be too provocative, but we might be able to give substantial aid to the rebels by covert methods, at minimum risk of becoming involved in a formal war with a Communist great power, from which it might be difficult for either side to disen gage itself.
5 Beclaratory Policy and Force Demonstrations NOW that the costs of all-out war have become incommensurate with almost any conceivable political objective and even limited uses of force carry considerable risk that the protagonists may not be able to restrain themselves from using the most powerful weapons at their disposal, it is indeed probable, as Roger Hilsman has said, that nations in conflict will approach each other warily, like "porcupines in love."1 Yet conflicts of interest will remain, and if violence is becoming more and more costly and risky as a means of protecting interests and re solving conflicts, we can expect nations to be inventive in developing substitutes. One substitute might be an increased reliance on threats to use force and the demonstrative use of force, short of war, to influence an opponent by working on his fears of the consequences of war. In other words, calculations of reciprocal intent, and attempts to influence such calculations, are likely to become more important as compared with calculations of relative capabilities, and the actual clash of capa bilities in war. Now that the possible costs of war have virtually sur passed the limits of human imagination, it has become relatively less important to base decisions on what the enemy can do, and the apprais al of what he will do or might do has become crucial. The "balance of power" may become primarily a "balance of intentions" centering on mutual calculations as to the amount of cost which the opponent is willing to risk or accept, and his image of one's own willingness to risk or accept costs, with respect to a particular issue. In such a situation, highly intangible impressions of "nerve," "will power," and "irrational ity" may become vastly more important than they have been in the past. "Declaratory policy," in its broadest sense, involves decisions as to what we shall communicate to the enemy about our intentions and capabilities and how we shall communicate it. We shall discuss in this chapter only those aspects of declaratory policy which relate to the communication of intent, although some reference will be made to communications of capability, to the extent that they bear on intentions. 1Roger Hilsman, "On NATO Strategy," in Arnold Wolfers (ed.), Alliance Policy in the Cold War, op.cit., p. 182.
240
Declaratory Policy
As is often the case with an attempt to set up categories of behavior, the line between declaratory policy and the military aspects of national security policy is not clear-cut. Violent military action is often under taken partly, or even primarily, to signal future intentions. The United States intervention in the Korean War is a case in point; that interven tion was undertaken partly to inform the Communist enemies that we could be expected to respond in cases of future aggression in Asia, and to prevent them from developing expectations that future attacks in that area could be carried out with impunity or at low risk. The simple existence or building up of military forces cannot help but communi cate something about our intentions; if one devotes costly resources to preparation for certain types of action, the chances are increased that the will or intent exists to take the action. Deployments and maneuvers of force can function either to show an intention or to provide a neces sary capability to meet an enemy move. Yet the distinction seems to be clear enough to provide a useful framework for analysis. Declaratory policy may be either verbal or non-verbal, the latter category including not only the use of force short of war to demonstrate or communicate intentions, but also other organizational and personal maneuvers with intent-laden significance. For simplicity, we shall henceforth apply the term "declaratory policy" only to the verbal forms. The manipulation or use of forces to suggest intent will be discussed under the heading of "force demonstrations."
A. Declaratory Policy Declaratory policy may take an almost infinite variety of forms. It may include, for instance, speeches by high government officials, press conference statements, budget and State of the Union messages from the President, Congressional resolutions, diplomatic notes, and releases and "leaks" to the press. Treaties are among the most solemn and pow erful forms of declaratory policy. Of course, the enemy will build his image of one's intentions out of the whole range of information available to him. This will include not only what he observes about our military preparedness and previous military actions, but also what he knows about the American "charac ter" or psychology, the temperament of leading officials, the degree of sanctity we seem to place on alliance obligations, etc. Verbal communi cation is only one source from which the enemy derives his image. And,
and Force Demonstrations
241
of course, only a part of the verbal communication emanating from the United States is subject to governmental control or manipulation. Governmental statements must be made to a variety of audiences and for a variety of purposes, some of which may conflict with the objective of projecting to the enemy the most desirable image of our intentions regarding the use of force. Declaratory policy, as defined here, seeks to shape, restrict, or supplement the "normal" flow of information with the objective of strengthening deterrence, to the extent that this is com patible with other objectives and values, and with the democratic process. Since declaratory policy is concerned chiefly with deterrence, it may be well to recall certain points made in the first chapter about the logic of deterrence. Deterrence is essentially a function of the enemy's image of our own estimate of our costs, gains, and risks which would follow from particular responses to the enemy's aggressive acts, and his estimate of his own costs, gains, and risks resulting from the re sponses. If there are any responses available to us which the enemy thinks would provide us with a net balance of gain over cost and risk, and if such responses would leave the enemy with a net balance of cost and risk over gain, the enemy should be deterred. The enemy is confronted with a whole array of possible responses to his aggressive act, for each of which he calculates a balance of cost, risk, and gain for himself, and attempts to guess our estimate of our costs, risks, and gains. His image of our cost-risk-gain estimate deter mines the credibility (probability) of each of our available responses. His estimate of his own costs, risks, and gains from a particular re sponse determines the degree of credibility of that response which is necessary to deter the aggressive act. Declaratory statements may affect any of these elements in the potential attacker's calculations. Probably the most important genus of declarations is what we refer to as "threats." A threat contains two components, either by implica tion or by explicit statement: (1) a contingency or enemy move which will bring about a certain sanction or response; and (2) a description of the sanction or response, or—more generally—an indication of the consequences or costs which the enemy must expect as a result of his move. Threats should be distinguished from "commitments." A commit ment is a real intent to resist an enemy attack, codified or possibly cre ated by some agreement or declaration.
242
Declaratory Policy
Threats may simply clarify commitments or intentions which already exist. That is, the deterrer may already calculate, before threatening a particular response, that the response promises him a surplus of prob able gain over probable cost; therefore it is rational for him to make this response whether or not he threatens it. The purpose of threaten ing in this case is to make sure, or to increase the chances, that the enemy does not miscalculate. More interesting cases, however, are those in which the response threatened would not have been rational to carry out in the absence of the threat, but becomes rational as a result of the threat. The threat creates a commitment which did not previously exist. It does so by implicating additional values—honor, self-respect, prestige, and the credibility of our future threats and outstanding statements of intent— values which would be lost in failing to carry out the threatened sanc tion, and which would not be lost, at least not in the same degree, if the threat were not made. We become "committed" as the result of a threat when this involvement of additional values is sufficient to raise the cost of reneging higher than the net cost of fighting.2 For example, the United States might have preferred not to defend the islands of Quemoy and Matsu prior to the 1958 crisis, but as a result of certain state ments made during that crisis we probably became committed to their defense; because of these statements (and various force demonstra tions) the moral and political costs of acquiescing in their conquest probably became more serious than the costs and risks which would have been incurred in defending them. The Chinese realization of this probably was the major reason why an invasion never took place. A third class of cases covers those in which a commitment or intent to react did not exist before the making of a threat, and still does not exist after the threat is made. Such threats obviously are bluffs. They implicate values and create an additional cost in non-fulfillment, but not enough to create a balance of gain over cost in actually carrying out the threats. Yet such threats may have deterrent value because the aggressor3 does not have full information about the deterrer's cost-gain calculations. 2 For a discussion of this point, and other aspects of threat-making, see Daniel Ellsberg, "The Theory and Practice of Blackmail," loc.cit. 3 Throughout the rest of this section, the word "aggressor" means "potential aggressor," or "the party whom the deterrer is trying to deter."
and Force Demonstrations
243
The problem faced by a potential aggressor after a deterrent threat has been made is to determine whether the deterrer has become com mitted thereby. If he (the potential aggressor) believes this to be the case, he must then decide whether the costs of fighting are likely to exceed any gains he might expect to make, and, if so, whether the net cost will be greater than the loss he must accept in backing down.4 If both conditions hold, he is deterred, provided he is rational. However, deterrence does not depend on the potential aggressor be ing certain that the deterrer has committed himself. In fact, the aggres sor may have his doubts; all he knows for sure after a threat is made is that it has now become more costly than it was before for the deterrer to fail to act; he does not know whether the costs of reneging now seem higher to the deterrer than the net costs of fighting.5 But after the threat, it will seem more likely (to the aggressor) that the deterrer is com mitted to respond as threatened. What he must decide is whether the increased risk of the response, considered in relation to his possible costs and possible gains, is more than he is willing to accept. Threats chiefly affect the aggressor's image of the deterrer's apparent gains in making the threatened response (if the cost of not responding can be considered part of the gains from responding). Other types of declarations may affect in other ways the aggressor's assessment of the deterrer's estimates of cost and gain. Thus, the deterrer might suggest that he is rather indifferent to the costs of war, or that he thinks the aggression can be repelled very quickly, or that he believes the aggres sor will sue for peace soon after the battle begins. He might make very optimistic statements about his civil defense capabilities. Or he might indicate that he is confident of eventual victory, in case the other side 4
In theory, this is a fairly complicated calculation. The aggressor will be un certain about the costs; he must consider not only the cost of fighting a war at the level of response specified in the threat, but also the risk that the war may expand to higher cost levels. He may also be uncertain about his chances of gain ing his object. What he really faces is a series of possible "outcomes," each con taining a gain and loss component, and each discounted by an appropriate prob ability which gives an "expected value" for each possible outcome. The sum of expected values over the whole range of possible outcomes is his over-all expected value (or expected cost) for the aggressive venture. If this is negative, and the "expected cost" is greater than the intangible losses suffered in backing down in the face of the threat, he is deterred. 5 The deterrer's "net costs" of fighting are calculated, like the aggressor's, as outlined in the preceding footnote. They represent a combination of expected war costs, expectations of being able to deny the aggressor his object, and perhaps expectations of positive gains.
244
Declaratory Policy
might otherwise think he doubted his capacity to win. To illustrate, re peated statements by the United States military leaders to the effect that the West would "win" an all-out war precipitated by a Soviet at tack on Western Europe may enhance the deterrence of attack in West ern Europe, whatever the validity of this contention or whatever the Russians think about its validity. If there is danger that the aggressor might think that the immediate territorial object of an attack was too trivial to justify a response—i.e., if he might think he could obtain his objectives "piecemeal"—there are declaratory ways by which the deterrer can increase his apparent valu ation of the object, or the likelihood of a response to an attack upon it. The probability of action against relatively minor enemy encroachments may be strengthened by creating the impression that one believes in the "domino" theory, or believes that a small capitulation would encourage the enemy to additional bolder aggressions. Thus Secretary of State Dulles said during the Chinese offshore islands crisis of 1958: If we, the strongest of the Free World powers, were to show inde cision and weakness in the face of this challenge, we would merely confirm the rulers of the Sino-Soviet bloc, the rulers of international communism, in their hope that by threatening anywhere and every where around their circumference they could compel submission or surrender. If we must meet such a challenge, it is better to meet it directly and at the beginning rather than after our friends have become dis heartened and our enemies over-confident and miscalculating.6 Such statements provide a plausible rationale for an intent to defend objectives which are insignificant in themselves, by stressing one's con ception of the deterrent value of the response, rather than the value of the territory attacked. Similar in effect are statements that an enemy attack on a minor ob jective may be regarded as indicative of major intent. The enemy is put on notice that any "nibble" may be regarded as the opening stage or part and parcel of a large bite. Thus the Formosa Resolution of 1955 authorized the President to resist an attack on areas "related" to For mosa (the Quemoy and Matsu islands) if in his judgment such re sistance were necessary to the defense of Formosa and the Pescadore Islands. e Netu
York Times, September 26, 1959, p. 3.
and Force Demonstrations
245
If the deterrer can create the impression that his intentions are based on "all-or-nothing" moral principles, which cannot be compromised whatever the cost, deterrence is strengthened. Several statements by President Eisenhower and the late Secretary of State Dulles have indi cated that they would not stop to count the costs of nuclear war if such war appeared necessary to preserve U.S. rights and obligations. For example, the President, discussing the Berlin crisis at a news conference in March 1959, said (1) that "we are certainly not going to fight a ground war in Europe," (2) that a nuclear war would be a "self-defeat ing thing for all of us," but (3) that "we are never going to back up on our rights and responsibilities."7 The deterrer may also be able to influence the aggressor's estimate of the costs and risks to the latter which would result from a particular response. For example, by our repeated declarations that limited war in Europe is impossible, we may increase the Russians' subjective risk in attempting a conventional invasion of Europe, by creating anxiety that we might not exercise the restraint necessary to keep the war limited. (Similarly, by their repeated declarations that limited, or "tac tical," nuclear war is impossible, the Soviets may hope to deter us from using tactical nuclear weapons.) The deterrer may also exert some declaratory influence on the ag gressor's expected gain from attacking. The gain to the latter may in clude avoidance of the humiliation which would be suffered in backing down, to the extent that his prestige has become involved. Such poten tial humiliation might be reduced by providing some verbal loophole which would allow the aggressor to back away with minimum prestige losses. It is worth noting that threats may have anti-deterrent effects. They may engage the honor and prestige of the party threatened as well as those of the threatener. People (and nations) do not like to "back down" before a threat made by an opponent who is presumed to be their juridical, social, or organizational "equal." The Anglo-Saxon peoples seem particularly likely to experience a stiffening of will when faced with an explicit threat. Chamberlain was willing to negotiate away a piece of Czechoslovakia in a context of "rational discussion of the issues" but he was unwilling to capitulate to the German threat of force. Similarly, the United States in 1959 re fused to negotiate the Russian demand to make West Berlin a "free 7
Ibid., March 12, 1959, p. 12.
246
Declaratory Policy
city" as long as the demand appeared to be a "threat" or "ultimatum." President Eisenhower and others stated that the United States would not negotiate on this issue while a threat of force or "time limit" ex isted; the Soviets therefore were forced to announce publicly that no threat or ultimatum was intended. 1. CLARITY VS. AMBIGUITY Threats may be made with varying degrees of ambiguity. In general, the credibility of threats is enhanced by clarity and reduced by am biguity. If a threat is stated clearly, whatever values would be lost in non-fulfillment of the threat are involved with maximum firmness. Am biguity allows the deterrer to renege on his commitment with minimum losses to his prestige and honor and to the credibility of his future threats; he can plausibly say, after the aggressor has moved, that the move does not fit the conditions of the threat or that a military response had not really been threatened. And noting in advance of his aggressive act that the deterrer's threat has been ambiguously worded, the ag gressor will suspect that the deterrer does not intend to execute the threat and has deliberately made it vague in order to minimize the cost to him of non-fulfillment. Thus, ambiguity tends to reduce deterrence, even though it may be useful as a hedge against the failure of deterrence. The essential purpose of ambiguity is to create at least some anxiety in the oppenent, some worry that a certain move by him will have undesirable consequences, but to do so without actually committing oneself. The reason for avoiding commitment may be that one is bluff ing and wishes to minimize the cost of being "called," or it may be that one is uncertain how one will respond when the occasion actually arises and therefore wishes to retain maximum flexibility. Of course, if we are already sure how we will respond to given enemy acts, and if we think the enemy will be deterred if he is aware of our intent, we will want to threaten this response with maximum clarity. Even if we are bluffing, we may want a threat to be unambiguous, if the greater deterrent value of clarity is believed to offset the somewhat greater loss if the bluff is called. Governments have been extremely inventive in thinking of ways to make ambiguous threats. It may be stated that a particular area consti tutes a "vital interest," or that an attack on it would have "grave con sequences." Or a government may say it cannot "remain indifferent" to
and Force Demonstrations
247
certain events, as when the Soviet Union stated during the landing of U.S. forces in Lebanon in 1958 that it "could not remain indifferent to acts of unprovoked aggression in a region adjacent to its borders."8 In the nuclear age, a common way of threatening ambiguously is to refer to the possibility (or probability) that armed conflict beginning on a limited scale could not be kept limited. Thus war in Europe is said to be impossible or highly unlikely. Such statements invoke "nature" as a substitute for one's own supposed intent—i.e., they assert the exist ence of some mechanism or process, not subject to human control, or only partly controllable, which will very likely produce the undesirable outcome if it is triggered by enemy acts. The Soviet Union made sev eral statements of this sort during the U.S. landings in Lebanon in 1958—e.g., that because of the U.S. action the world had been placed "on the brink of catastrophe," and that the action could "set off a chain reaction which it would be impossible to arrest."9 Besides raising the prospect of some undesirable consequence with minimum engagement of one's own prestige, such invocations of "nature" also minimize the engagement of the prestige of the recipient of the threat, since the statement is designed to appeal to his reason rather than to place him under the threat of force. Besides alluding to forces in nature which may supersede the will of both participants, one can seek to represent oneself as a "force of nature," not subject to influence. Thus President Eisenhower has re ferred to the determination of the United States to maintain its rights in Berlin as being like an "immovable stone." Premier Khrushchev has said that if the United States were to use force against the East Ger mans, Soviet forces would "automatically" come to the latter's rescue.10 If the provocation and the sanction are described with high precision, the threat probably will carry greater weight with respect to that par ticular contingency than if it is merely included within a broad class of contingencies and sanctions stated in a general threat. However, too much precision may sometimes work against deterrence. If the act which will provoke the response is stated too specifically, the enemy may be able, fairly easily, to make his attack in such a way that it does not quite correspond to the description in the threat. Then if the deterrer has a motive to renege on this threat, he can do so with minimum 8
Ibid., July 19, 1958, p. 1. a Ibid ., July 21, 1958, p. 1. 10 Both remarks appeared in ibid., July 8, 1958, p. 1.
248
Declaratory Policy
loss of prestige and minimum erosion of his apparent will to act in the future. If the sanction is stated precisely, the enemy may be able to take measures in advance to reduce his losses. Precision in specifying the sanction may also be unwise if there is some doubt whether the sanction threatened would deter. For example, a U.S. threat to respond to a Communist attack on Quemoy with conventional means might not be sufficient to deter the attack; it might be wise at least to keep open the possibility of a nuclear response even though the threat of such a re sponse were not made explicitly. A specific statement of the sanction implies a promise that severer sanctions will not be applied. This is a weakness in proposals for agree ment upon, or unilateral statement of, the "ground rules" for limited war, especially limited nuclear war. It has been urged, for example, that the West unilaterally declare that it would not use hydrogen weap ons, or attack cities with atomic weapons (except possibly those in the front line of battle and those with important military targets directly alongside), unless the other side did so first.11 Whatever the desirability of thus limiting nuclear war, which of course is great, such an advance declaration might encourage the enemy to attack if he were confident of winning at acceptable cost within the limitations declared. Thus the greater deterrent effect of a general threat—e.g., to respond with suffi cient force to win, or to inflict enough costs on the enemy to offset his potential gain—must be weighed against the possibility that, should the tear occur, its costs might be held down by specific limitations de clared in advance. Highly specific statements of the contingency which will activate the sanction may create doubts as to whether other contingencies not specified, or covered only by a more general obligation, will evoke a response. Thus the Locarno Agreements of 1925, in which Britain guar anteed France's frontier with Germany, may have reduced the credi bility of Britain's obligation to invoke sanctions against aggression elsewhere under the League of Nations Covenant. In other words, they implied that Britain did not consider herself fully committed to act under the Covenant or that she believed the potential aggressor did not consider Britain committed, and that therefore a more specific state ment of intent was required. The Locarno Agreements probably tended to create the expectation that Britain would not act to resist aggression 11 Cf. Rear Admiral Sir Anthony W. Buzzard, "Massive Retaliation and Grad uated Deterrence," World Politics, Vol. vra, No. 2 (January 1956), pp. 228-37.
and Force Demonstrations
249
against any country not covered by similar specific guarantees. In the same fashion, Secretary of State Acheson's specific delineation of a U.S. "defense perimeter" in the Far East, leaving other Asian areas under the much more general protection of the U.N. Charter, tended to re duce the credibility of U.S. obligations under the latter. 2. REITERATION OF THREATS Somewhat similar results might follow from the reiteration of threats. To repeat a threat constantly may create a kind of "threat inflation" in which threats not repeated tend to lose their force. There may even be a weakening of the threat which is repeated, as a result of the psychologi cal mechanism expressed in Shakespeare's line, "The lady doth protest too much, methinks." In other words, constant repetition may create the impression that one believes the threat requires reiteration, that one doubts the plausibility of the threat in the enemy's mind. There is probably a balance to be struck between no repetition and overrepetition. Threats not repeated at all may weaken in deterrent value with the passage of time. The "happy medium" may be to call attention oc casionally to previously made threats, but only on appropriate occa sions, and in a manner which does not suggest doubt about the inherent credibility of the threat. Along the same line, the repeated official statements that deterrence of Soviet attack in Europe requires a "will" as well as a capability to retaliate with strategic nuclear power may have an anti-deterrent ef fect. Of course, the statement is correct, but use of the word "will" may imply to the enemy that a real effort of will may be required to offset our rational perception in the crisis that the costs of all-out war are unacceptable. If it were really in our interest to retaliate, no special "will power" would be necessary. Or the statement may imply that one must develop a firm intent to retaliate in advance of the occasion for doing so, an intent firm enough to override whatever obstacles of "rea son" get in the way when the casus retaliati occurs. Rather than stress ing will as a component of deterrence coequal in importance with capa bility, it might be better to speak as if the intent to retaliate were to be taken for granted. 3. WHO SHOULD MAKE THE THREAT? A threat carries greater weight, the more authoritative the source. Declarations of intent made by the President will be most readily be-
250
Declaratory Policy
lieved, those by the Secretary of State will carry almost as much credi bility, statements by the Secretary of Defense, members of the Joint Chiefs of Staff, and lesser officials will have somewhat lesser impact, and so on down the line. This rule is subject to qualification, depending on personalities and circumstances. For example, Secretary Dulles took such a leading role in the formulation of foreign policy that his state ments were almost as authoritative as those of the President, and in some cases perhaps more so. Since subordinate officials embody the na tional prestige to a lesser degree than the President and his top foreign policy adviser, and since statements by such officials can be disavowed with minimum embarrassment, they may be used to make pointed, explicit threats which would be too provocative or too prestigeful when made at the topmost level. In the Quemoy crisis of 1958, the severest U.S. threat was made by Secretary of the Air Force James H. Douglas. He said, with State Department approval, that U.S. aircraft in the Formosa area "are capable of using high-explosive bombs or more powerful weapons, if necessary," a statement which was widely inter preted as a threat to use nuclear weapons, if necessary, to protect Formosa.12 Threats supported by express Congressional approval tend to carry more weight than those made by the Executive branch without formal Congressional sanction. Thus the Formosa Resolution of 1955 and the Eisenhower Doctrine of 1958 (expressing our determination to defend countries in the Middle East against Communist-controlled aggression) gained deterrent potency by their expression as Congressional resolu tions. The Congressional action put the world on notice that the ex pressed intentions had the strong support of the American public; furthermore, it removed in advance any possible question regarding the President's authority to order the use of military forces in these areas without waiting for a specific Congressional authorization. 4. PUBLICITY VS. SECRECY Threats can be made either publicly or secretly. The prestige and honor of the nation are committed to a much greater extent if they are made publicly. Therefore public threats generally will have more de terrent value than private ones. If the deterrer does intend to carry out the action threatened, the threat should be made publicly, in order to make sure the aggressor understands that a commitment has been made. 12
New York Times, September 27, 1958, p. 1.
and Force Demonstrations
251
If the threat is a bluff, or if the deterrer is not sure he will be willing to carry out the threat when the contingency arises, it is perhaps best to state the threat privately, through official or unofficial diplomatic chan nels. The deterrer thus minimizes his loss of face should the enemy's challenge occur and the threat not be carried out, even though such a threat carries less deterrent value than one publicly proclaimed. 5. THE CONTEXT The circumstances in which a declaratory statement is made will affect its forcefulness and credibility. A sentence in a solemn, deliber ately prepared statement such as the President's State of the Union message will be given more credence than an off-the-cuff remark at a news conference. Thus if a government wants to commit itself unmis takably, the more formal the forum the better. But if it desires to threaten drastic sanctions pointedly, while avoiding a hard and fast commitment, the informal context will be more appropriate. The Presi dential news conference has yet to be fully exploited in this connection. The need to make judgments and express them quickly, the possibility of the President misunderstanding the reporters' questions, or the re porters misunderstanding the President's answers, the fact that the initiative in choosing the subject matter lies with the reporters rather than with the President, and the human tendency to be less cautious in face-to-face conversation than in formal written statements all tend to make press conference statements less authoritative, less prestigeful, and more ambiguous than, say, statements in a written television speech. Therefore, threats made or implied at a press conference will commit the government less firmly than the same threats made on more formal occasions. There is less potential cost, therefore, in declaring an intent which one does not or may not intend to carry out. A remarkable example of the declaratory function of the news con ference occurred in December 1950, during the Korean War. This was the occasion when President Truman declared, in answer to a reporter's question whether the United States was "considering" the use of atomic bombs in Korea, that, of course, the government always had "under consideration" the use of all its available weapons. Perhaps the President did not mean this as a deliberate threat, but the exchange could not have been better handled for this purpose if it had been planned deliberately. The statement raised the possibility that atomic
252
Declaratory Policy
weapons would be used, but with a minimum degree of commitment and minimum provocation to the Communist enemy. Use of the word "considering" injected an element of ambiguity: did the President mean that the government was considering the use of atomic weapons as a distinctly possible new departure in the Korean War strategy, or did he simply mean that all forms of capability naturally entered into the na tional security deliberations of the government? The fact that the sub ject was brought up by a reporter added to the ambiguity: the Presi dent could hardly deny that use of atomic weapons was being consid ered, without reducing the degree of threat which the Communists were already under. Was the President merely trying to avoid this re duction of threat, or was he stating a new threat? It may also be recalled that the first nuclear threats which were made by the Soviet Union against Britain and France during the Suez crisis in 1956 were made through the Soviet news organs, and were distinctly more belligerent and sharper in tone than the later diplo matic notes which those governments received. A deliberate "leak" to the press is another way of threatening with a minimum of commitment. An unattributed statement naturally is less authoritative than one which is attributed; nevertheless, the news story will have a certain threat value. But the threat is ambiguous because there is always the possibility that the reporter has misunderstood his informant, or has exaggerated or sharpened the meaning of the state ment to increase the news value of the story. And, of course, the gov ernment can easily disavow the story later if it decides not to carry out the implied threat.
B. Force Demonstrations There are many ways other than verbal declarations by which states may communicate their intentions. Such moves as visits of heads of state, recall of ambassadors, and "force demonstrations" have a long history as devices to demonstrate either a friendly or a hostile intent. We shall discuss here only the last-mentioned device—the non-violent (or perhaps violent) use of military force for demonstrative purposes. It may be useful to start by recalling various historical forms of force demonstration.13 Mobilizations, the calling up of reservists, the holding 13 For a useful historical survey of force demonstrations, see Alfred Vagts, Defense and Diplomacy, New York: King's Crown Press, 1956, chap. 7.
and Force Demonstrations
253
of maneuvers near the frontiers of the country threatened, are actions involving ground forces which have often been used and understood as symbolic threats. Naval forces, because of their greater flexibility than land forces, have been more often used for demonstration purposes. Such usages have included the movement of squadrons or fleets in the direction of the opponent's shores or naval forces. The stationing of the bulk of the U.S. Pacific Fleet at Pearl Harbor in 1941, rather than at U.S. West Coast bases, was motivated in part by a desire to impress the Japanese, although it did not go as far as was intended by President Roosevelt, who wished to send detachments of the fleet to Singapore and the Philippines, and generally to have ships "popping up here and there, and keep the Japs guessing."14 Another move is to have a single warship drop anchor in an enemy's port or off his coast, or at the scene of diplomatic conflict, as exemplified by Germany's use of the gunboat Panther in the Agadir crisis of 1911. Naval visits to ports of friendly powers have also been used to demonstrate the solidarity of an alli ance—here an outstanding example is the exchange of visits by the French and Russian fleets in 1891 and 1893. A prominent instrument of demonstration by the United States since World War II has been the Sixth Fleet in the Mediterranean. This fleet, and its augmentation and maneuver in certain times of crisis, have played a role in the deterrence of Soviet attack and even "indirect aggression" in the Middle East. The threat to King Hussein's govern ment in Jordan in 1957 quickly subsided after the Sixth Fleet carried out a demonstration maneuver in the Eastern Mediterranean. The landing of troops in Lebanon in 1958 was a spectacular instance of demonstration by the Sixth Fleet. The Seventh Fleet in the Pacific, particularly in its patrolling of the Taiwan Straits, has performed an analogous function. It was claimed by the Eisenhower administration that a force dem onstration was instrumental in bringing the Korean War to a close in 1953. This demonstration had several facets. The first move, taken on February 2, 1953, was to rescind the standing order to the Seventh Fleet to prevent Formosa and the Pescadores from being used as a base for operations against the Chinese mainland. Then, in his State of the Union message, President Eisenhower announced that he was "giving immediate increased attention to the development of additional 14 Hearings Before the Joint Committee on the Investigation of the Pearl Harbor Attack, Washington, D.C., Government Printing Office, 1946, Part 16, pp. 2163-64.
254
Declaratory Policy
Republic of Korea forces." In July, an additional Marine division was sent to Korea and additional modern air units were sent to Korea and nearby bases. Atomic missiles were sent to Okinawa and a portion of the Strategic Air Command was alerted for action. According to Secre tary of Defense Charles E. Wilson: "We were not too careful about whether that leaked out or not and we got the armistice."15 1. THE EFFECTS OF FORCE DEMONSTRATIONS The psychological effects of force demonstrations, when used to indi cate intent, differ from the effects of verbal threats in various ways. Demonstrations cannot be as explicit as threats; they are inherently somewhat ambiguous and diffuse. They do not specify the contingency which will activate the force, or how much and what kind of force will be used, although these things may be understood, and, of course, demonstrations may be accompanied by verbal threats which do speci fy these things. The ambiguity of demonstrations may allow bluffing, and retrenchment when the bluff is called, at the cost of less humilia tion than when one has to renege on an explicit verbal threat. Also, demonstrations can always be disavowed later as the unauthorized ac tion of military commanders or described as "routine maneuvers" or "planned in advance." In fact, it is quite customary to disavow demon strations in this way even while they are taking place; everyone knows that the disavowals are hedges against being too provocative, or against the possibility of having to back down; that what is really intended is to "throw a scare" into the opponent. Despite their ambiguity, force demonstrations have more deterrent potency in some situations than declaratory threats. Partly, this is due to the accretion of custom and precedent. It is a convention of inter national life that when a country starts maneuvering its forces clearly for demonstrative purposes, its patience is wearing thin, or it really "means business." In part, this is because military forces are visible; by bringing its military forces into view and moving them about, a coun try shows that it may be considering moving from the realm of diplo matic negotiations to the realm of military action. Another difference between force demonstrations and verbal threats is the variety of effects which can flow from demonstrations. Threats strengthen deterrence by implicating one's honor, prestige, and future 15 House Committee on Appropriations, "Department of Defense Appropriations for 1957," Hearings, 84th Congress, 2nd Session, p. 267.
and Force Demonstrations
255
bargaining power. Demonstrations can do this and much more besides. They can put in motion maneuvers which lead in the direction of vio lent action, thereby establishing a certain momentum which may be difficult to stop and making it easier to overcome the inertia of moving from a condition of peace to a condition of war. They may increase the chances that one's forces may become involved in some conflict between other countries, as a result of being fired upon or as a result of unau thorized or accidental action by one's own military commanders. They may increase one's readiness to act and reduce one's costs of acting. Since they are bound to be somewhat provocative, they may indicate a willingness to run risks, which may strengthen one's general bargain ing position. Some types of demonstration tend to merge into violent action, and some kinds of violent action may be carried out purely to demonstrate future intentions to use violence. Demonstrations may al low the prolonging of threats over time, since they may involve con tinued movements, or a continued presence over a period of time. Thus, the Soviet missile-firing range in the Pacific, and repeated test shots, constitute a continuing reminder to the United States of the vulnera bility of the West Coast complex of missile bases and missile and air craft manufacturing plants. Of course, different types of demonstrations produce these effects in varying combinations and degrees. Some demonstrations, which might be called "passive," are not much more than threats made in a different language; they produce little risk of accident, they do not increase readiness, etc. In this category one might put the appearance of a single naval vessel in a foreign port purely for symbolic purposes, the drop ping of a missile or bomb close to the enemy's shoreline, and aircraft "flyovers" and "flypasts." Such passive demonstrations may merge into more active forms which border on violent action or may actually involve a minor act of violence. In the days of American gunboat diplomacy in the Caribbean, U.S. ships might not only appear in foreign ports but also fire their guns, or even bombard harbor installations or seize the installations. In the modern day, instead of dropping a missile near the enemy's shoreline, one might be dropped on an uninhabited island belonging to the enemy. Or an enemy reconnaissance satellite might be shot down. Or one might have a nuclear accident "on purpose"—i.e., explode a bomb over enemy territory and then quickly claim it was an accident, allowing the other side to suspect, however, that it was not. It is con-
256
Declaratory Policy
ceivable that in some future severe war of nerves, the Soviets might let one of their Pacific test shots stray onto Johnston Island or Hawaii. Or in some future limited conventional war, one side might drop an atomic bomb on a small enemy naval vessel, to signal that it would "raise the ante" to the nuclear level unless its demands were met. The purpose of such actions would be to inspire a belief in the enemy that one would move to a more-than-demonstrative level of violence unless the enemy backed down. The act would be provocative, of course, and probably the enemy would feel it necessary to reply in kind. But in taking the action, in the knowledge that it would be pro vocative, one would have demonstrated a willingness to take risks and accept costs which might strengthen one's bargaining position. Before initiating such violent demonstrations, one would have to be confident, of course, that one could establish dominance in the war of nerves be fore the costs got too high or the exchanges got out of control. An important characteristic of some force demonstrations is that they allow gradual fulfillment of a threat. They may allow a sequence of preparatory moves obviously leading in the direction of violence but not violent in themselves. With each stage, the commitment to ultimate violence becomes stronger. A momentum is established, the movement becomes more widely known, and one's prestige becomes more and more committed. This allows a sort of calibration of the response to parallel the gradually more threatening or provocative posture of the enemy. The enemy becomes more and more worried, but he can stop our movement toward violence at any time simply by desisting from his provocative behavior. He has time during which to consider wheth er he has misjudged our intentions, time between his initiation of ag gressive acts and an irrevocable involvement with the United States, time to decide to pull back. He may be able to do this with relatively small loss of face if he has not threatened explicitly or been threatened by us. In this way, force demonstrations can form a sort of bridge between pristine peace and actual war. The bridge is formed by an ever-increasing probability that violence will take place. If the demonstrations merge into minor acts of violence, the bridge continues in the form of small actual war costs and a steadily increasing probability of largescale war. As was noted earlier, the decision deliberately to move from a state of peace to a state of war is a very difficult one for any nation to make. If it can make this decision gradually—i.e., "slide into war," so
and Force Demonstrations
257
to speak—the credibility of its threats to make war is higher. No particular move in the process is a discrete, discontinuous jump requir ing great effort of will. Some types of demonstrations may strengthen deterrence by creating some possibility that an enemy attack would involve our own forces. The clearest kind of such "demonstration by deployment" is the familiar ground force trip-wire. Wherever the United States maintains such trip-wires (Europe and Korea), the enemy knows that an attack will automatically bring the United States into the war. Naval forces can perform this trip-wire function in a modified way by putting themselves in a position where they may get involved in fighting. A recent example was the convoying of Chiang Kai-shek's supply vessels to Big Quemoy island by units of the Seventh Fleet during the Quemoy crisis of 1958. This operation created a certain possibility that our vessels might be inadvertently fired upon, that they might fire back at the shore batteries, and that as a result of a sequence of actions by local commanders we might find ourselves fighting a minor war with the Chinese Communists which could expand into something much bigger. Our landings in Lebanon in 1958 also created a risk of involvement, which probably was important in deterring the Soviets from encouraging the rebellion or in deterring the rebellious forces themselves. In both of these cases, the actions also demonstrated our interest in the areas, partly because they showed our willingness to accept the risks of becoming involved.16 Demonstration value accrues not only from the actual movement of forces toward the scene of a possible attack, but also from readiness measures taken at home. When we call up Army reserves, activate National Guard divisions, take ships out of "mothballs," start crash mobilization programs, start stockpiling supplies in danger areas over seas, etc., we demonstrate a willingness to fight by increasing our readiness to fight. At the strategic level, we can "threaten" a first strike by evacuating cities, by augmenting our air defenses, by putting SAC on full alert, by moving medium-range bombers to forward bases in Europe, by "counting down" missiles to a point just short of firing, or by starting to jam the enemy's radar. Such measures, besides indicating a will to fight by increasing readiness to fight, might also increase the 16 T. C. Schelling describes such actions as "threats which leave something to chance." That is, we deliberately give up some of our capacity to choose peace or war by creating the possibility of inadvertent involvement. See The Strategy of Conflict, op.cit., chap. 8.
258
Declaratory Policy
chances of accidental war. For example, if missiles are counted down to a point where simply a flick of a switch can send them on their way, the chances of accident are appreciably increased. We may deliberately accept an increased chance of accident as a way of putting pressure on the enemy by working on his fear of accident. He can relieve the pres sure by desisting from whatever behavior has provoked us. Of course, the enemy can also demonstrate in the same way. To offset his demon strations we would want to do things to augment our strike-back ca pacity, to the extent that the measures just mentioned had not already done so. For example, we could increase the number of Polaris sub marines on station, decreasing the number in port, disperse portions of SACs aircraft to civilian airfields, put some planes on airborne alert, and start land-based mobile missiles moving. If a conventional limited war is going on, we could threaten to "raise the ante" not only by the measures just mentioned, but also by demon strations suggesting a possible intent to initiate tactical nuclear war fare. For example, we could ostentatiously move short-range nuclear missiles to points in the rear of the battle area, or we could move in an aircraft carrier with planes known to be capable of carrying nuclear bombs. Or nuclear weapons could be put into the hands of lower-level military commanders, thus increasing the chances that they would be fired irresponsibly or accidentally. If we wanted to take full advantage of the potentialities of deterrence by demonstration, we might not want to have our forces in the highest state of readiness, quite apart from the high costs of such a posture. That is, we would want to have a maximum number of things we could do, preparatory to throwing the forces into actual battle, so as to have a capacity for gradually increasing the pressure and to provide the enemy with maximum opportunity for changing his mind. But, of course, if these measures were essential to effective defensive action, we would want to be able to do them quickly, if necessary.
6 The Reconciliation of Deterrence and Defense HOW do we choose among the alternative means of preparedness, strategy, and declaration which we have been discussing? Is there any systematic method available for making these choices? Is it possible to devise any criteria for evaluating the over-all military worth of differ ent types of military force, according to some common denominator of their respective deterrent-defense values, and for weighing military worth against the non-military values which must be sacrificed to pro vide military forces? We will argue here that although it certainly is not possible, in practice, to assign precise numbers to these values, there is a logical method for describing, estimating, and comparing them which can be illustrated numerically and graphically. The ex plicit recognition of this method may provide a useful check on judgment. The worth or utility of military forces is, of course, ultimately politi cal, if "political" is defined as advancing or preserving the values or objectives of the state in international politics, the prime value being, of course, the survival of the state itself. It is often said that both the use of military force in war and the allocation of resources to military forces in peacetime must be governed by the objectives of national policy; yet the logical nexus between "policy objectives" and military force "requirements" and strategy is not at all clear. Those who have attempted to explain the connection between these two entities have often had recourse to economic analogies. Thus, in 1943, Walter Lippmann wrote that "a foreign policy consists in bringing into balance, with a comfortable surplus of power in reserve, the na tion's commitments and the nation's power," just as the assets of a business enterprise should always be more than sufficient to cover its liabilities. On this criterion, Lippmann asserted that American foreign policy since the Spanish-American War had been "insolvent"; we have consistently failed to maintain enough mobilized power to make good on our foreign policy commitments.1 Similar economic imagery was 1 Walter Lippmann, U.S. Foreign Policy: Shield of the Republic, Boston: Little, Brown and Co., 1943, p. 9.
260
The Reconciliation of
invoked by General Matthew B. Ridgway, during his stormy tenure as Chief of Staff of the U.S. Army between 1953 and 1955. Ridgway as serted that foreign policy commitments were like "sight drafts" which might be presented for payment at any time; just as a bank has to main tain sufficient cash reserves to pay such drafts, so must a nation have sufficient ready military forces to meet all its commitments, in view of the enemy's capability to challenge those commitments.
A. The Requirements Approach vs. the Limited Economy Approach The Lippmann-Ridgway formula is sometimes referred to as the "re quirements approach" to military policy. Essentially, there are three stages in this approach: (1) to look at "national policy" or "commit ments" (presumably as defined in treaties, National Security Council documents, Presidential directives, and the like); (2) to survey the enemy's capabilities for challenging policy goals and commitments; and (3) to determine what military forces are needed to meet each possible challenge successfully. This approach produces an absolute and clear-cut definition of "requirements." This much is needed—no more and no less. The requirements approach implicitly denies that security is variable; that we can have more of it or less of it, depending on how much we are willing to sacrifice. It asserts that either we are secure or we are not and that we must have those forces which are necessary to make us secure. The requirements approach usually is associated with the military services. It can be found in statements of doctrine and apparently it is used, in modified form, in preliminary budget planning. However, in its pure form, it is more frequently advocated by civilians, especially Congressmen, than by military leaders. Sophisticated military officers are likely to think of "requirements" in terms of a range of relative de sirabilities or priorities. They are accustomed to taking a "calculated risk," although they would be the first to admit that this phrase hardly ever reflects a "calculation" in any systematic or theoretical sense, but simply means that a deliberate decision has been made to accept some degree of risk.2 They are also accustomed to modifying preliminary 2 The political usefulness of the term is in its connotation of deliberateness; if the risk-taker can say that he has taken account of possible adverse consequences and deliberately decided, nevertheless, to accept the risk of them, he is more or
Deterrence and Defense
261
requirements estimates to get "inside the ballpark"—i.e., within the range of what they can reasonably expect the fiscal and budgetary au thorities to be willing to allocate for military use. Directly opposed to the absolute requirements approach is the "limited economy" approach. This approach holds that there is an ob jectively determinable limit to "what the economy can stand" in the way of governmental expenditure and taxation, and that this places an absolute ceiling on the amount of resources which can safely be allo cated for military use. This notion perhaps is characteristic of the fiscal authorities, but it is also believed by a surprising number of military men. Military force levels, force composition, and budgets usually reflect a compromise between these two conflicting approaches. Yet there is no solid evidence that the compromise is reached in any systematic way. And even if it were, it would not assure good decisions, because it would be a compromise between two absolute criteria which are them selves both seriously defective. The limited economy approach does have the virtue of recognizing that costs are important. Yet it does not consider costs as simply sacri fices of civilian or non-security values to be weighed against the secu rity gains realizable by making these sacrifices. Instead, it emphasizes a presumed adverse effect of military expenditure on the stability of the economy. It holds that U.S. economic stability is part of national se curity—in fact, the bulwark of the security of the whole free world. It tends greatly to exaggerate the dangers to economic stability that re sult from governmental spending. In an extreme form, it asserts that we would go "bankrupt" if a certain level of expenditure were exceed ed. Most economists agree that there is no way of objectively determin ing "what the economy can afford." Of course, the greater the govern mental expenditure which is not offset by taxes, the greater the possibility of inflation, and, at some level of taxation, business invest ment incentives may begin to decline. But most reputable economists agree that United States military budgets since World War II have been well below the point where they might begin to create either of these dangers.3 This notion of the absolutely limited economy begs the less protected against later charges of carelessness or incompetence, should the adverse consequences materialize. 3 For an excellent discussion of the economic impact of defense spending, and also of the requirements approach vs. the limited economy approach, and other
262
The Reconciliation of
really important economic question in defense matters, which is: How much in the way of material comforts are we willing to sacrifice in order to have how much security? Because of its dominant focus on costs and economic stability, the limited economy approach inevitably tends to be biased in favor of deterrence, as opposed to preparations for effective defense. If war can be deterred at relatively low cost, why undertake the unnecessary extra expense and economic risks of being ready to fight it? Strategic nuclear deterrence seems to offer an easy road to economy. It is rapidly becom ing less plausible that a strategic nuclear force which can effectively deter enemy moves in Europe and Asia is less costly than the conven tional ground forces which would be necessary for effective defense. Yet it is beyond question that the heavy emphasis on deterrence by the threat of massive retaliation and other forms of nuclear action which characterized the Eisenhower administration was closely related to that administration's belief in the limited economy theory. The great virtue of the requirements approach is that it correctly stresses the connection between the nation's objectives in international politics and the values obtainable from military forces. Nevertheless the approach has serious deficiencies. An obvious one is that it tends to ignore the importance of economic costs; it emphasizes values to be gained by preparedness—in fact, takes them as imperatives—and leaves out the values which must be sacrificed. Another source of distortion is that policy objectives tend to be formulated by different persons and organizations than those who determine military requirements, and without much attention to the military support implications. Then these objectives and commitments become "givens" (however vague) to which the military services attempt to tailor their estimates of require ments. A more rational procedure would be to establish policy goals and military force levels simultaneously. Among the variety of values which may be lost as a result of enemy attack, the questions of which ones will be defended, in what sorts of contingencies, and in what way, cannot rationally be decided apart from the anticipated cost of fightstrategic issues, see Henry Rowen, "National Security and the American Economy in the 1960's," Study Taper No. 18, Joint Economic Committee, U.S. Congress, 86th Congress, 2nd Session, January 30, 1960. For a case study of policy-making when these two concepts clash, see my forthcoming study, "The 'New Look' of 1953," to be published with other studies by Paul Hammond and Warner Schil ling, under the aegis of the Institute of War and Peace Studies, Columbia Uni versity.
Deterrence and Defense
263
ing and the resource cost of preparing to do so. In other words, policy goals or "action" intentions do not determine preparedness costs and the costs of war. Intentions are subject to modification in the light of the costs of buying and using capabilities; there is a certain degree of "trade-off" between the intrinsic and strategic values which are directly subject to enemy attack, and the cost of military action and prepared ness to defeat the attack. There may be some places in the non-Communist world which we would prefer not to defend because they do not bear sufficient value to warrant the war cost and preparedness cost of doing so. There may be others for which we are willing to fight only up to a certain limit of war cost and preparedness cost; there may be some for which we are willing to fight a losing battle in order to salvage a certain amount of political, deterrent, and intrinsic value—the cost of preparing to win, and winning, being thought too high to warrant the additional value saved. A second deficiency in the requirements formula is that it over whelmingly emphasizes "defense" and slights "deterrence" by taking enemy capabilities as its yardstick and largely ignoring the enemy's in tentions and the effect of our own military posture upon them. It tends to overlook possibilities for deterring some attacks by threatening re sponses which pose unacceptable costs and risks for the enemy, as a substitute for having a force sufficient to deny the enemy territorial gains or to destroy his capability to fight. It is asking too much of pre paredness policy that it be ready to block all conceivable enemy moves and win all possible wars. Such a policy would provide too much se curity because it would provide a high level of deterrence as well as capability for denying all enemy gains. There would be perfectly legiti mate demands to reduce the forces, accepting a somewhat higher probability of some attacks, or higher losses if war should occur, or the possibility of not being able to frustrate certain enemy moves, all in the interest of reducing the peacetime economic burden. For example, if a nuclear deterrent threat is thought to be highly effective against some enemy moves, even though too costly to carry out because of the risk of the enemy's retaliation, it may not be neces sary to be ready with an alternative force which could defend at toler able cost. As noted earlier, the reason why nuclear deterrence can be effective short of an effective and rational defense is that the enemy can never be sure how we estimate the costs and consequences of nu clear war, or how much we value the object of conflict. Nor can he be
264
The Reconciliation of
sure how rational we are. Therefore, we may rationally threaten and prepare for action which we would not wish to carry out after the ag gression took place, and after it took place we might wish we had capa bilities which (quite rationally) we had failed to provide. Actually, the substantive military policy of the United States has recognized this truth, although it has never been fully accepted in official doctrine. We continue to threaten "massive retaliation" in response to a major Soviet ground attack in Europe, for example, when it would very probably be irrational to carry out this threat in view of the Soviet counterblow, and we have failed to provide sufficient conventional ground forces to de fend effectively at acceptable cost in several other areas—notably the Middle East. This may be perfectly rational policy if deterrence is thought to be effective enough. Another deficiency (which has already been touched upon) is that the requirements approach has an unjustified aura of absoluteness about it. It creates the impression that if we do not fulfill the "requirement" for any particular type of capability, we have nothing of any value, and that any forces beyond the stated requirements would be redundant. There are certain elements of absoluteness regarding particular forces, in the sense that marginal increments to some forces when they are below a certain level may not be worth very much, and increments above a certain level may have sharply diminishing marginal utility; examples of these will be given below. But the requirements approach vastly exaggerates the degree of absoluteness, and tends to hinder a rational balancing of marginal costs against marginal utilities. The requirements approach may also err in assuming fixed enemy capabilities. Existing enemy capabilities tend to be used as the yard stick, with little or no consideration given to how the enemy may change his capabilities in response to our preparedness decisions. Clearly, the deterrent and defense value of increments to our own capa bilities must be discounted by the probability that the enemy will counter by increasing his. National security policy-making would be improved if a system of analysis could be devised which avoided the absoluteness and onesidedness of both the "requirements approach" and the "limited economy" approach. In such a system, policy objectives, military "require ments," and economic limitations all would be determined in a single, unified intellectual process. In such a process, the urgency of support ing, or even having, political objectives would be directly balanced
Deterrence and Defense
265
against the economic and other costs of supporting them; ideally the end-result would be a set of military forces which reduced the over-all degree of military threat to ourselves and our allies to a level which was just warranted by the sacrifices entailed in buying and maintaining these forces. In other words, such an analytical process would show us just how much security we were justified in buying, in view of our comparative valuation of security and the things which would have to be given up in the interest of security. A clear distinction should be made among action policy, prepared ness policy, and declaratory policy. Action policy includes decisions concerning what enemy challenges the United States will resist and how—i.e., our intentions concerning the use of military force. Pre paredness policy concerns the determination of appropriate capabilities for deterrence and defense. Declaratory policy comprises decisions about what we will communicate to the enemy and the rest of the world about our action intentions and how we will communicate it. While it is useful to make this distinction, these three categories are really inter locked in a seamless web; decisions in any one category cannot be made rationally without reference to the other two.
B. The Balancing of Deterrence, Defense, and Resource Cost At first glance, it might seem that action policy takes precedence: we first decide what we have to do or want to do militarily if the occasion arises, then we decide what capabilities we need to do the job, with perhaps some discounts or "calculated risk" to give effect to cost con siderations. This is the essence of the modified "requirements approach." Yet action policy cannot be made without reference to technological possibilities, the deterrent and defense value of various kinds of capa bilities, the costs of using them in war, and the costs of providing them in peacetime. Nor can declaratory policy be formulated apart from a determination of what' our intentions really are and how credible our declarations are likely to seem in view of the capabilities we decide to have. Thus decisions in all three categories should be made simultaneously. Before establishing the most desirable level and form of preparedness, we must attempt to predict the effectiveness of alter native action responses in various contingencies—for example,
266
The Reconciliation of
whether our losses in all-out war would be minimized by a counterforce strategy, a bargaining strategy, a tactical retarding strategy (or some combination of these), or simply a strategy of maximizing Soviet damage. Whether we determine to have forces for a limited retaliation strategy, an all-out first strike, a tactical nuclear war, or a conventional defense to deal with lesser Soviet challenges would de pend in part on how we estimate the costs and effectiveness of each of these strategies. But preparedness policy must consider deterrent possibilities as well as defense or action effectiveness, and in striking the balance between these two sets of considerations it must also consider the peacetime preparedness cost of alternative combinations. Some forces which are useful for deterring enemy challenges may not be the most desirable forces for action in response to attacks. Thus preparedness policy must be based on a tentative analysis of action responses, modified by considerations of deterrence and the probability of enemy attacks, and the peacetime resource cost. Logically, action policy is subsumed within preparedness determinations; i.e., action policy is our most rational (least costly or most gainful) set of re sponses, given the military capabilities we have decided upon, after consideration of deterrent values, defense values, and peacetime costs. Declaratory policy is really subsidiary to the other two categories: given the capabilities and intentions we have decided upon, it deter mines what declarations and demonstrations would maximize deter rence, consistent with the risking of our honor and prestige in com mitments which we do not intend, or might not be able, to fulfill.4 Stated most concisely, preparedness policy must attempt to strike a balance among three objectives: mitigating war costs and other losses in case of enemy military challenge (defense), reducing the probability of challenges (deterrence), and minimizing the sacrifice of resources in peacetime. The costs and losses in case of enemy attack, and the probability of attack, can be combined in the concept of "expected cost." The expected cost for any given military establish4What I have just said here assumes, more or less, "a clean slate." Of course, at any particular time we will have some forces in inventory, the utility of which will have to be considered in deciding what additional forces to have; the only cost of the forces already on hand will be their maintenance and operating costs. Similarly, at any particular time, we will have certain commitments outstanding, and we may have little flexibility for changing these commitments. This inheritance from the past creates, to some extent, an "action policy" to which present and future preparedness and declaratory decisions must conform.
Deterrence and Defense
267
ment is calculated as follows: for each possible enemy move5 multiply the assumed probability of that move by the net value losses which we would expect to suffer as a result of that move, assuming we made the response which we thought would minimize our over-all losses of all kinds.6 Total expected cost would be the sum of these products for all conceivable enemy challenges. The expected cost calculation can be represented by the following formula, where E is expected cost, the numbers 1, 2, 3 . . . η are the aggressive moves open to the enemy, ρ is the estimated probability of each move, and c is the cost expected to result from each move, when followed by our most appropriate response: E = Pc1 plus pc2 plus pcs pcn7 We speak of net losses and costs, and "expected cost" rather than "expected value," on the assumption that the United States, in its foreign policy, is interested only in preserving values—not in acquiring 5 The word "move" refers here to a decision regarding the violent use of mili tary force by the enemy, not to "preparedness" moves, or decisions concerning the size, composition, and peacetime deployment of the enemy's forces. 6 In cases where we did not intend to respond, because our prospective war costs appeared greater than the values that would be saved by resisting, the losses would be in the form of prestige, future deterrent capacity, future capacity to attract allies, etc., as well as the power and intrinsic values associated with the territory concerned. 7 Mathematically knowledgeable readers will recognize that the formula simply applies the principle of "mathematical expectation" to the problem of national security. Obviously, the formula is a great simplification of reality. A very large number of possible action moves are available to the enemy. A literal and detailed application of the formula would mean visualizing all the possible moves and assigning a probability to each. Then the consequences of the entire war which would follow each move, and our best response, would have to be estimated. For each initial move-and-response, there would be a wide range of eventual con sequences, each of which would have to be assigned a probability. Thus, a fullscale Soviet conventional attack on Western Europe which was met by a tactical nuclear response might end quickly, or might last for varying periods of time at varying intensities of tactical nuclear warfare (depending in part on our own war plans), or might spiral to all-out dimensions. The "consequences" of any Soviet move would have to be visualized as a "cluster" of possible outcomes, each dis counted by an appropriate probability. And each of the possible consequences would be a mixture of territorial changes, changes in future Soviet intentions, war costs, prestige effects, and in general all the effects on all the values implicated by the enemy act. Obviously, any attempt to estimate all these factors with pre cision in advance of the war (or even during the war) would be foolish. Yet preparedness policy cannot be formulated effectively without some image of the "likelihood" (in lieu of a precise probability) of at least the most salient possible enemy moves, or without some estimate of the consequences which would follow from various possible responses to these moves.
268
The Reconciliation of
more. Thus, the consequence of aggression, and resisting aggression, will usually be a net cost, because even if the territorial status quo is preserved, we suffer the costs of preserving it. We fight because we think the net losses from resistance will be less than the costs of nonresistance. Admittedly, resistance may sometimes result in a net gain —e.g., if it substantially enhances national prestige, or increases future deterrent capability, or rolls back the Communist frontiers. However, it is plausible, and simplifying, to assume that, over-all, E is negative for the United States. Ordinarily, any increment to military forces will reduce expected cost. In terms of the formula, the deterrent effect of an increment is its effect in reducing the p's; its defense effect is its contribution toward reducing the potential c s for all the possible enemy moves. Both effects tend to reduce E or expected cost. The extent to which a given increment reduces expected cost is a rough measure of its mili tary worth or utility. Its total utility is the sum of its "deterrent utility" and its "defense utility" for all enemy moves in which the added forces would have utility of one kind or the other. The grand objective of military preparedness is not, however, simply to reduce or to minimize expected cost. The costs of preparedness must also be considered.8 Therefore, the object of preparedness policy can best be described as obtaining the minimum possible aggregate of expected cost and preparedness cost—i.e., obtaining the best possible combination of deterrence and defense capability, over the entire range of possible contingencies, within the limit of the resource ex penditure thought to be justified by the resulting reduction of expected cost. In speaking of "sums," "products," and "aggregates" I am, of course, speaking metaphorically, since "expected cost" and "preparedness cost" (and their constituent elements) cannot be made commensurate by any mathematical criterion. But the judgment used in making na8 Of course, preparedness costs are not limited simply to economic sacrifices. There are other "intrinsic" costs in terms of interrupted careers and separation from family and friends for individuals drafted into the armed forces, perhaps increased governmental controls on the economy, etc. On the other hand, prepared ness may yield intrinsic gains in the shape of stimulation of economic activity, by-product industrial uses of military research, and increases in national prestige. These factors should be considered in estimating the net costs of preparedness. And naturally the economic costs must include the cost of maintaining and op erating the forces, as well as costs of development and production.
Deterrence and Defense
269
tional security decisions, if it is good judgment, must be informed by a logic similar to that just outlined, even though the entities involved can only be intuitively estimated and compared. 1. A MATHEMATICAL EXAMPLE The following example, using simple mathematics, is intended only to illustrate the essential logic and involves no commitment either to the numbers or to the results. Suppose the United States government must choose among buying more strategic nuclear forces, or buying more conventional ground forces which would cost a similar amount, or doing neither, or doing both. Suppose we think we have the strategic nuclear forces necessary to deter surprise attack, but would like to have somewhat more cer tainty of deterrence. Assume there are only two possible Russian moves: surprise attack on the United States and a full-scale ground attack in Europe, and that the NATO ground forces presently avail able are not sufficient to hold against a full-scale Soviet invasion. To put it even more concretely, suppose the choice is between buy ing χ more Atlas and Titan ICBM's for emplacement during the period of the "missile gap," and buying y more divisions of ground forces, when either alternative would cost roughly the same. The ICBM's would increase our strike-back capacity following a surprise attack, and therefore increase our confidence in deterring a surprise attack. They might also increase somewhat our capacity for defense in all-out war—i.e., for striking at the enemy's nuclear forces which might be used in follow-up attacks. They might also increase our capacity to deter a major ground attack in Europe by increasing the likelihood (in the Russians' calculations) that we would then have the capacity and will to execute a first strike in response to Russian ground invasion in Europe. Let us assume, however, that after adding this increment we would still not be willing to initiate all-out war in response to Soviet ground attack in Europe—these additional forces, therefore, have no "defense value" with respect to the latter contingency. Assume that the alternative of increasing ground forces would give us the capacity to hold against a major Soviet conventional surface attack, at least until sufficient reinforcements could be mobilized in Europe and the United States. Assume that these additional ground forces would contribute nothing to the deterrence of surprise attack on the United States. They would, however, contribute substantially
270
The Reconciliation of
to both the deterrence of and defense against a ground attack in West ern Europe. Finally, assume (a rather drastic assumption!) that the enemy does not increase his own forces to offset our own increase. The situation before the addition of either type of force might be illustrated by using matrices of the type presented earlier in the dis cussion of deterrence. The numbers are merely suggestive, of course. They represent the anticipated cost to the United States for each combination of Soviet move and U.S. response. The numbers in the boxes shown in Figure 7 represent losses to the United States with each move-and-response. The probabilities of S UNITED
O
V
I
E
T
U
N
I
O
N
No attack
Nuclear attack on U.S.
Ground attack on Western Europe
No response
0
-500
-100
Massive response
0
-400
-400
Ground force response
0
0
-150
STATES
Probability of Soviet moves Expected cost
.60
.10
0
—40
Preparedness cost
.30 —30 = —70 —50
Expected cost plus preparedness cost
—120
FIG. 7 attack are calculated, theoretically, by estimating the Soviet "risk calculus" as discussed above in Chapter 1. That is, an attempt is made to guess the Soviets' estimate of their "payoff" for each attack-andresponse and the probabilities they associate with each possible re sponse by the United States—which gives a resultant expected value (gain or loss) for the Soviet Union for each of its possible moves. The U.S. estimates of the Soviet expected values, and our degree of con fidence in these estimates, are the source of the probabilities associated with each Soviet move.
Deterrence and Defense
271
The United States will respond to each Soviet move in the manner which promises the least cost. The least-cost response to nuclear attack on the United States is massive retaliation. The U.S. strategic forces have some defense value in this contingency, since their use against the Soviet forces not used in the first wave can reduce total damage by 100. These forces have no defense value for meeting a Soviet con ventional attack in Western Europe, and neither do the ground forces —the use of the strategic forces would precipitate more costs for us than the area is worth, and the ground forces cannot hold. There exists no rational defense, therefore, for the contingency of ground attack in Western Europe—therefore no defense will be attempted. Expected cost for each contingency, of course, is the product of its probability, times the costs which would follow on the most rational —i.e., least costly—response. The total of expected cost is the sum of the expected costs for each contingency, or 70, and preparedness cost is 50, an aggregate of 120. Now suppose the addition of the ICBM's is considered. With this alternative, the matrix shown in Figure 8 might be visualized. S
O
V
I
E
T
U
N
I
O
N
No attack
Nuclear attack on U.S.
Ground attack on Western Europe
No response
0
-500
-100
Massive retaliation
0
-300
-300
Ground force response
0
0
-150
UNITED STATES
Probabilities
.67
Expected cost
0
.08 —24
Preparedness cost
.25 —25 = —49 —60
Expected cost plus preparedness cost
—109
FIG. 8
Both the probability and the cost of all-out war are reduced—i.e., the additional ICBM's have both deterrent and defense values for this contingency. The probability of Soviet attack in Western Europe is
272
The Reconciliation of
reduced slightly by our belief that the added strategic forces increase the Soviet expectation that we will retaliate massively in this con tingency. But there is no change in the consequence of a Soviet ground attack—i.e., these forces have no defense value and hence will not be used in this contingency. The most rational response to ground attack in Western Europe is still "no response." Over-all expected cost is re duced by 21, at the price of 10 in preparedness. Since the sum of expected cost and preparedness cost will be lower when these forces are added, the addition of the ICBM's is desirable. If we consider the alternative of adding the ground forces, also at a preparedness cost of 10, we might arrive at the matrix shown in Figure 9. S
O
V
I
E
T
U
N
I
O
N
0
-500
-100
Massive retaliation
0
-400
-400
Ground force defense
0
0
-30
Expected cost
.80
.10
.10
0
-40
I
Probabilities
I
No response
OO
Ground attack on Western Europe
STATES
II
Nuclear attack on U.S.
CO
No attack
UNITED
Preparedness cost
-60
Expected cost plus preparedness cost
-103
FIG. 9
In this case, there is no change in the probability and cost of all-out war, but both the probability and the consequences of ground attack in Europe are sharply reduced because of the probable recognition by the Soviets that Western Europe can be denied to them, and be cause the ground forces now available can defend successfully at a cost less than our value of Western Europe. This alternative is the preferred one, because the total of expected cost and preparedness cost is lower than it would be if the ICBM's were added instead of the ground forces.
Deterrence and Defense
273
We could go on to consider whether the addition of both sets of forces would be desirable. It would be if an additional preparedness cost of 20 produced a reduction of more than 20 in "expected cost."9 Our model could have been made much more complex—for exam ple, by the introduction of uncertainty in the U.S. calculation as to whether the larger ground forces could hold Europe successfully, or by considering the possibility that the larger strategic forces might stimulate the Soviets' fears of a U.S. first strike, which in turn might increase their temptaition to strike first themselves—lthus perhaps reducing rather than increasing the deterrence of all-out war. Or we might have introduced a "tactical" nuclear strategy as a fourth defense alternative, or the consideration that a war begun on a conventional or tactical nuclear level might eventually spiral out of control. The latter consideration presumably would enhance the deterrence of a ground-force attack in Europe, but would also increase the expected cost of a defense. A similar logic might be applied to other types of choices. For ex ample, suppose the government is considering whether to spend $10 billion on a civil defense program, or to put the money into additional conventional ground forces for use in Europe. The civil defenses would 9 Of course, any probability number assigned to a given enemy move is a "subjective" probability which is bound to be subject to a wide range of possible error. The estimate of "expected cost" for each contingency probably should in clude some adjustment to compensate for error on the side of overoptimism. The amount added presumably should depend on our own degree of willingness to take risks as well as on our degree of confidence in the probability estimate. Logically separate from the range of possible error is the possibility of fluctua tion in our perception of probable Soviet intentions, through the acquisition of new intelligence information. Given the secretiveness of the Soviet regime, any small bit of new information may be sufficient to cause a substantial change in the perceived likelihood of aggressive moves. And even a small change in "like lihood" may cause a large change in our own "expected cost." Such possibilities should be considered in relation to the "lead time" for producing weapons. Given a lag of several years between the perception of increased threat and the availability of additional or difFerent forces to meet it, it may be desirable to maintain extra forces as a hedge against it. Also, the more diversified our own military establishment, the less risk is involved in the possibility of a sudden drastic change in perceived Soviet intentions. When prediction is difficult, when the stakes are high, and when errors in predictions and preparedness are cor rectable only over an extended period of time, it makes sense to take out insurance for a fairly wide range of risks. For a discussion of the relation of un certainty to preparedness choices, see Klaus Knorr, "Is the American Defense Effort Enough?" Memorandum No. 14, Center of International Studies, Princeton University, December 23, 1957.
274
The Reconciliation of
have most utility for defense in case of nuclear attack on the United States, and possibly some minor utility for deterrence of such an attack. They might have some value for deterring a major ground attack in Europe, by indicating to the enemy that we might have reduced our all-out war costs to a level at which we would be willing to carry out massive retaliation. If, in fact, the civil defenses would make retalia tion rational, they would also carry defense value for the contingency of a ground attack on Western Europe. The sum of these various types of utility would then have to be compared with the combined deterrent-defense utility of the additional ground forces. Theoretically, a complete calculation would set up a matrix for each possible Soviet move, but since the number of possible enemy moves is close to infinite—including, of course, limited attacks in Asia as well as in Europe, with varying commitments of forces—such a com plete calculus is impossible. However, at least the major contingencies can be identified and the deterrent and defense utility of particular increments of force for each contingency can be estimated, if only intuitively. The relative reliance on nuclear deterrence and ground force deterrence and defense in various contingencies might be sharply different. For example, strategic nuclear airpower might be relied upon to deter Soviet major attacks (either conventional or nuclear) in Europe, with conventional ground forces being provided to deter and defend against minor encroachments in Europe and limited aggres sions in Asia. Alternatively, the optimum allocation might turn out to be that of providing enough conventional forces to hold the Red Army in Europe without the necessity of nuclear action, with reliance on the threat of limited nuclear retaliation to deter enemy moves in Asia. The rationale for this latter variation might be that Europe is so im portant that means must be provided to hold it at minimum cost should deterrence fail, even though nuclear deterrence may be more effective in Europe than in Asia. The corollary reasoning for Asia would be that if the threat of nuclear punishment should fail to deter, the losses would be limited to areas not absolutely essential to the security of the United States (assuming that the threat would not be carried out), and that if the threat of punishment were carried out on a limited scale, the chances of keeping the war limited would be fairly good.
Deterrence and Defense
275
C. The Marginal Utility Criterion The rule of minimizing "expected cost plus preparedness cost" could be expressed in terms of the familiar economic principle of marginal utility.10 This principle simply says that in any allocation of scarce resources among competing uses, when each use is subject to diminish ing marginal returns as additional resources are applied to it, no increment should be given to any one use when it would yield a greater return in some alternative use. In the optimal allocation, the "last" or marginal increment to each use would yield a return exactly equivalent to its marginal cost, with marginal cost defined as the utility which could have been gained by applying the increment to other alternatives. When this condition holds, the maximum total utility is being obtained with the available resources. This criterion can be applied either to the broad allocation of re sources between non-military activities and the military establishment, or to the allocation of given resources within the military establish ment. The marginal utility of any increment to the military forces is virtually equivalent to the reduction in over-all "expected cost" which results. Additional resources should be allocated for military purposes only so long as each increment reduces expected cost more than it reduces non-security values; when the marginal reduction in expected cost is equivalent to the marginal sacrifice in competing values, the criterion of minimizing "expected cost plus preparedness cost" is met. Among competing uses within the military establishment, available resources should be allocated so as to minimize expected cost. Of course, this suballocation between military alternatives must take place tentatively before the broad allocation between the military and non-military sectors is determined. Actually, the allocative problem is much more complex than this criterion suggests. The first complexity is that marginal increments must be valued in two different currencies—deterrence and defense. If we visualize deterrent utility and defense utility as curves generally 10 So far as I know, the first published suggestion for applying the marginal utility principle to military problems was Bernard Brodie's "Strategy as a Science," World Politics, Vol. i, No. 4 (July 1949), pp. 467-88. A more recent treatment was Malcolm Hoag's "Some Complexities in Military Planning," World Politics, Vol. xi, No. 4 (July 1959), pp. 553-76. My discussion here draws heavily on these two incisive articles.
276
The Reconciliation of
rising as forces increase, the two curves may have different shapes for any particular type of force, and the relationship of the two curves will be different for different types of forces. Another complicating factor is that the utilities of different kinds of forces interact; the mar ginal utility of an increment to any one type cannot be evaluated with out reference to all the other forces which are available or planned. Thus we must differentiate between independent and interdependent utilities. Still another complexity is that capabilities cannot easily be segregated according to contingency or function—"all-out war" forces vs. "limited war" forces, for example. Forces primarily useful to resist or deter one enemy move may have a different degree and sort of value, or even a "disutility" with respect to other moves. Also, some forces may be subject to the phenomenon of "increasing marginal re turns" while others are subject to "diminishing returns." Finally, all estimated marginal returns must be discounted by the probability and extent of enemy countermoves. Let us examine, first, the shapes of the deterrent utility and defense utility curves for various kinds of forces, considering only independent utilities, assuming constant or diminishing marginal utility for all forces, and assuming that enemy capabilities remain constant. 1. THE INDEPENDENT UTILITY OF STRATEGIC NUCLEAR FORCES In discussing strategic nuclear forces, it is convenient to refer to "primary" and "secondary" deterrence and defense.11 The primary deterrent utility of such forces is their value for deterring a Soviet all-out first strike on the United States; their primary defense utility 11 After carefully weighing the disadvantage of introducing new terminology when other terms have attained wide usage, I have decided to use the terms "primary" and "secondary" in place of "passive" and "active" deterrence and defense, or Herman Kahn's "Type I" and "Type II" deterrence. Although "pas sive" and "active" have richer connotations than Kahn's terminology, they are rather misleading. The act of all-out countercity retaliation after taking a first strike is hardly a "passive" act; logically it requires a definite willful decision which may be more difficult to Take than a decision to retaliate after an enemy tactical move, if the latter "active" retaliation can take the form of something less than an all-out "first strike." The terms "primary" and "secondary" avoid these mis leading connotations and yet are not completely neutral; they correspond to the undoubted fact that we are primarily interested in deterring a nuclear attack on ourselves, and only secondarily interested in deterring and defending against enemy attacks elsewhere.
Deterrence and Defense
277
is their value for mitigating damage and minimizing other losses after such an attack. The "secondary" deterrent utility of strategic and other nuclear forces is their value for deterring enemy attacks, nuclear and otherwise, on other countries, and their secondary defense utility is their value for frustrating such attacks or limiting our losses. The size and relationship of these utilities depend rather critically on the degree of structural stability. For the purpose of the following discussion, we shall simply assume that stability will be fairly high— that it will be possible for each side to have a "minimum primary deterrent" at some fraction of the opponent's striking forces. Strategic nuclear power begins to yield primary deterrent value when it is considered probably just sufficient to pose a prospect of unacceptable damage for the Soviets. Forces below this level may have little or no deterrent value. Since we cannot be sure of the proper size of the minimum deterrent force, and may wish to hedge against Soviet miscalculation, irrationality, technological breakthrough, and other uncertainties (see Chapter £), some additional increments be yond the point of minimum deterrence will add to the deterrent utility of the force. At some higher point, however—say, a sure capacity to destroy 50 or 75 Soviet cities in retaliation—the curve of deterrent value begins to level off. In other words, the marginal deterrent utility of additional increments begins, at this point, to decline toward zero, and further increments eventually become redundant for primary de terrence. In the economist's language, the curve of marginal primary deterrent utility for strategic nuclear forces is a "kinky" one. The curve for primary defense utility has a much more constant slope, however, and continues rising far beyond the point where the deterrent utility curve levels off. Even forces below the requirement for mini mum deterrence will have some value for defense, since they may be applied against counterforce targets to limit our own damage or might possibly be used as a bargaining lever to induce the Soviets to limit their gains. Defense value continues to accrue for additions to forces beyond the point where they no longer have significant deterrent utility. It is very difficult, even abstractly, to identify the point where the primary defense utility curve begins to level off, but it is obviously much higher than the level-off point for primary deterrence. Its loca tion will depend on what assumptions are made concerning whether the Soviets will be amenable to bargaining in a total war situation, what counterforce targets will be available to hit after the Soviets'
278
The Reconciliation of
first salvo, and what degree of economic and population destruction will be necessary to destroy the Soviet society or capacity to continue fighting if the Soviets refuse to terminate the war by negotiation. If our defense objective is only to limit damage to ourselves by making counterforce strikes, additional capabilities beyond the deterrent re quirement may reach sharply diminishing returns fairly quickly. If we expect to bargain after the Soviet strike, we will see some value in making sure that our residual forces are as large as the Soviets' residual forces, and forces even beyond this point would have some value, although at a gradually diminishing rate of marginal utility. If we see no possibility of bargaining, and expect the Soviets simply to do as much damage to us as they can, then we will want to inflict maximum damage on them, and increments to strategic forces would have at least some value up to a very high level. It must be emphasized, however, that the primary defense utility of forces beyond the level-off point for primary deterrence must be discounted by the assumed probability of a Soviet attack. That is, even though it would be desirable to have large forces beyond those neces sary to deter in case deterrence should fail, our willingness to sacrifice resources in peacetime in order to have such forces would rationally be small if we were very sure deterrence would not fail. Under our assumption of fairly high stability, it would not be hard for us to muster an adequate deterrent, and our having such a limited force would pose no threat to the Soviets, so there would be little danger of a Soviet pre-emptive attack. The urgency of having additional forces for limiting damage and winning or stalemating the war would be slight, simply because there would be little chance of war. The argu ment for such forces would then have to depend on the chances of accidental war (which would be low in a stable technological environ ment) and on the chances of massive irrationality in the Soviet leadership. By way of contrast, should the technological environment turn out to be unstable, we would find it difficult to maintain an adequate de terrent force, hence even more difficult to have a damage-limiting or win-the-war force for the contingency in which the Soviets strike first; at the same time it would be more urgent that we have the latter type of force (than in a situation of high stability) because of the greater danger that the Soviets would strike.
Deterrence and Defense
279
In a stable environment, there would probably be no attainable level of strategic forces which would carry defense utility against Soviet aggressions in Europe and Asia—that is, the United States, under these conditions, probably would find a sufficient first-strike capability beyond its reach. But forces short of this standard, yet well in excess of the required deterrent against a Soviet first strike, might suggest a possible U.S. intent to strike first all-out in case of a major Soviet attack in Europe, even when we did not have this intent. The point in the accumulation of our forces at which this type of "second ary" deterrent utility began to accrue would depend largely on the degree of structural stability, the Soviets' impression of our degree of rationality, and their appreciation of the possibilities of escalation, which might depend on whether we had credibly communicated an intention to fight a limited nuclear war. It would also depend on whether the Soviets increased their forces to match our increases. It seems likely that the enemy would not foresee any significant prob ability of a U.S. first strike unless our forces were considerably beyond the necessary level for primary deterrence, and probably larger than the Soviet forces. Theoretically, from this point on, additional forces would carry secondary deterrent utility at a fairly constant rate up to the point where we thought the Soviets attached sufficient probability to our first strike to be deterred from provoking us. Beyond this point, the marginal secondary deterrent utility of further accumulations would diminish, but at a gradual rate. If the Russians expected our strategic forces to be used massively or not at all, the only value of the forces in the "gap" between the level-off point for primary deterrence, and the point where secondary deter rence began, would be their value for limiting damage and bargaining after a Soviet first strike. However, secondary deterrence does not re quire the assumption that a U.S. first strike would be massive or that it would be counterforce in nature. Threats of limited retaliation against Soviet cities or tactical forces, and possibly against some types of strategic forces, may have deterrent value, and these may be the only types of first-strike threats which either the United States or the Soviet Union will be able to make credibly if the strategic balance becomes very stable at the all-out level. In contrast to a strategy which envisaged only the massive use of strategic weapons, a limited retaliation strategy might invest such weapons with both secondary deterrent and secondary defense utility
280
The Reconciliation of
at force levels only marginally higher than the requirement for primary deterrence. These utilities, in other words, would tend to fill the "gap" between the point of sharply diminishing returns for primary deter rence and the point at which the threat of an all-out first strike began to carry some credibility. For purposes of limited retaliation, incre ments to strategic forces would yield fairly constant or perhaps slightly diminishing returns up to the point at which our own strategic forces were roughly equal to the enemy's—assuming that in a war of limited reprisals the side which was quantitatively inferior might be at a dis advantage in the bargaining. Beyond the level of the enemy's forces, additional forces would run into more sharply diminishing returns for limited retaliation purposes. 2. THE INDEPENDENT UTILITY OF TACTICAL CONVENTIONAL FORCES In contrast to strategic nuclear forces, the independent deterrent utility of tactical conventional forces tends to vary more or less directly with their defense utility; their deterrent effect is mostly a function of their capacity to deny territorial gains to the enemy, rather than of their punishment capacity.12 Also in contrast to strategic capabilities, such forces begin to yield some utility at a rather low level. We have something of value if we have limited war forces capable only of deterring or defending against a single minor satellite aggression, such as a new attack by the North Koreans, unsupported by Communist China. Additional forces would accrue additional value up to a very high level; the extreme, or leveloff, point would be active conventional forces which, along with those of our allies, could defend successfully, without post-attack mobiliza tion, against the maximum forces which all of our Communist enemies could commit simultaneously at any or all points. Just what level of forces, between these two extremes, is the desirable one is not some thing that can be determined with anything like the degree of precision 12 In some circumstances, conventional ground forces may yield some "punish ment deterrent value, simply because of the costs they could inflict on the mili tary forces of an attacker, even though they were clearly not capable of defeating an attack. This would be the case when the objective is not valued very highly by the prospective attacker, and when the attacker believes the defender may fight against a superior force, either because the latter is irrational, or because he hopes to salvage some prestige or moral values, or because he expects to bargain the attacker into making lenient terms of settlement.
Deterrence and Defense
281
which can be applied in determining, say, the "minimum deterrent" against a Soviet first strike on the United States. The question is com plicated by the many and varied contingencies in which such forces might be useful, by uncertainties concerning the size of the forces which the enemy is likely to commit to any particular challenge and how many challenges he is likely to make at any one time, by possibili ties (for the enemy as well as ourselves) for shifting forces from one deployment to another and for post-attack mobilization, and by the possibility of substituting nuclear deterrence for conventional denial forces as security against some contingencies. For the present, let us "assume out" nuclear deterrence—i.e., assume that our nuclear weapons, including tactical ones, have no value except to deter and fight a nuclear war begun by the Soviets. Interdependent utilities between nuclear weapons and conventional forces will be discussed below. Despite the complexity of the problem, certain benchmarks are discernible. One would be that the allied and U.S. forces which are deployed in or near all threatened areas must be sufficient to fight a holding action against the maximum possible enemy force until addi tional active forces can be brought in from the United States and other countries that have not been threatened, and that this expanded force must then be able to hold until materiel and manpower reserves and new units can be mobilized by the United States, by the attacked ally, and by other allies, and transported to the battlefield in sufficient strength to turn the tide and win the war. This criterion falls short of the maximum one mentioned above in that it assumes that the enemy would attack only one area at any one time, and it relies on post-attack mobilization. In practice, the requirement would be for active and reserve forces sufficient to defend that area in which the enemy was able to commit maximum force—undoubtedly Western Europe. We would still face the danger of concurrent attacks at other points— particularly by Communist China in Asia. Additional forces would provide deterrent and defense value at a rather constant rate as these further contingencies were covered. Presumably the marginal deterrent utility of additional forces would begin declining once we were fairly sure of being able to hold off, with post-attack mobilization, a simul taneous attack by the entire Soviet and Chinese Communist armies at all points of danger. However, additional active forces beyond this level would have some marginal defense utility because a delaying action,
282
The Reconciliation of
followed by post-attack mobilization which did not become effective until after the enemy had made large gains, would be a more costly war than one fought entirely with ready forces which were either on the line at the start, or able to move in quickly and throw back the attack. It seems probable, however, that the marginal cost of such additional active forces would be considered too high to justify their marginal defense value, especially considering that the enemy would very likely be deterred by the knowledge that he would ultimately be defeated. As in the case of damage-limiting forces for all-out war, the utility of spending prewar resources in the interest of lesser cost in war must be discounted by the probability that war may not occur. Thus, with respect to possible enemy attacks which are considered very unlikely, it may be reasonable to rely entirely on post-attack mobilization, or accept the possibility of a defeat or the necessity of a costly reinvasion and liberation, rather than to bear the expense of ready forces for an adequate denial action. 3. INTERDEPENDENT UTILITIES The utilities of nuclear weapons and conventional forces cannot be estimated and compared without taking account of various ways in which they interact and complement each other in both deterrence and defense. To take an obvious example, the NATO shield may sup port the credibility of a threat of massive retaliation by insuring that any conventional Soviet assault on Western Europe would have to be launched in great force, with unmistakably major intent. This effect would not operate, of course, unless U.S. or NATO first-strike capabili ties were substantial. Nor, even if the latter were substantial, would the shield provide such complementary deterrent utility beyond a certain range of ground forces. It takes only so many ground troops to force the enemy's attack to be maximally provocative and indicative of major intent. Large additional forces in the shield might depreciate the deterrent utility of the strategic forces by suggesting that we in tended to fight a conventional ground action. On the other hand, large ground forces might strengthen the strategic forces' deterrent value by suggesting that a strategic first strike, combined with a groundforce holding action, might successfully defend Europe, or that we might believe this to be possible.
Deterrence and Defense
283
When strategic forces are obviously short of a first-strike capability, their deterrent value in a threat of limited retaliation would also be complemented by the existence of a strong ground delaying capability which would strengthen our bargaining position in an exchange of limited reprisals. In this context the strategic forces and the conven tional forces would also be complementary in defense; even though neither the ground forces nor limited reprisals by themselves could prevent Soviet conquest, in combination they might be able to do so, or at least to limit Soviet gains. Even when they are unused, nuclear weapons may complement the defense utility of conventional forces. Held in reserve as a threat, they may deter overt Soviet assistance to a satellite aggressor, thus limiting the opponent's commitment of conventional forces to a level which our own conventional forces can deal with successfully. Conventional ground forces may contribute something to deter a full-scale nuclear attack on the United States. Since a primary aim of such an attack undoubtedly would be to make conquests in Europe and Asia, if the United States and its allies had ground forces in these areas which might be able to hold off the Red Army even during a thermonuclear exchange, the Russians might see only faint prospects of gain to offset their costs from such an exchange. Although it is widely believed in the United States (outside the Army) that ground forces would have little or no value in an all-out war, the Russians think differently, and for deterrence it is what the Russians think that counts. If the Russians happen to be right, NATO ground forces would also carry defense value in case of all-out war initiated by the Russians; they could strengthen our bargaining position after the Soviet first strike and make it more likely that we could negotiate an end to the war on terms short of the Communization of Europe. Interdependence is particularly complex among the various ele ments of "strategic" capability—offensive nuclear striking forces, active air defenses and warning, and civil defenses. For example, the marginal deterrent utility of active (shooting) air defenses against the contingency of direct attack depends on the size and degree of vulnerability of the retaliatory force. If the latter is considered large and secure enough to deter any deliberate attack without the protec tion of air defense, air defenses would have no primary deterrent value. However, if there is some question whether the retaliatory force is sufficient, air defenses would help to relieve such doubts either by
284
The Reconciliation of
forcing the enemy to expend some missiles against air defense installa tions on his first wave, or by causing attrition to the attacking missiles or bombers, or by injecting an additional element of uncertainty into the enemy's planning. The air defenses would be complementary, in primary deterrence, to other means for insuring an adequate strikeback force, such as sheltering of aircraft, hardening and "mobilizing" missile launchers, or increasing the numbers of dispersed retaliatory vehicles. However, active air defenses also can provide direct defense value, in ways that retaliatory forces and passive protective measures cannot. Active defenses, even if they are deployed only to protect retaliatory forces and not cities, may reduce the costs of all-out war by destroying enemy attacking vehicles before their bombs and warheads impact, and by protecting retaliatory vehicles to be used for counterforce action after an enemy first strike. If the active defenses are of the dual-purpose type, so that they can mitigate damage to cities as well as to strategic installations, their defense value is greater. When de ployed for the defense of cities, not strategic installations, the only value of active defenses for the contingency of an enemy first strike is that of defense. When active defenses have both deterrent and de fense utility, their defense value continues to accrue for additions far beyond the point where their primary deterrent utility diminishes or vanishes. Warning systems derive deterrent utility from their function of warning and alerting the retaliatory force. In other words, they are complementary to, or interdependent with, retaliatory forces for the function of deterrence of direct nuclear attack. This is practically the same as saying that warning is substitutable, to some extent, for retalia tory forces. We would not need warning for deterrence if the retaliatory missiles were numerous enough, or well enough protected, to insure unacceptable retaliation without warning; we might not need warning if we had a continuous airborne alert of bombers. Thus warning is not an absolutely essential requirement, but only some thing which must be weighed against other means and combinations of means for assuring an adequate strike-back capability. But warning may also have defense value, not only because of the protection it affords to retaliatory counterforce vehicles, but also because of the time it provides the civilian population to get into shelters. A warning system and a civil defense system are interdependent in their defense
Deterrence and Defense
285
utility. The utility of warning in this latter capacity is entirely de pendent on the prior existence of a civil defense program; the defense value of a civil defense program is largely (though not entirely) dependent on warning. The value of civil defense programs for deterring a Soviet first strike is questionable. Civil defenses may yield some primary deterrent utility by giving credence to our expressed determination to retaliate all-out against Soviet cities. On the other hand, they might suggest that our primary interest in all-out war would be to limit our own damage; hence the Russians might bank on our retaliation being relatively re strained and directed against counterforce targets; believing that damage to their cities and economy would be relatively low, their incentive to strike might be strengthened. On balance, each considera tion seems about equally plausible; hence, civil defenses probably should not be credited with any primary deterrent value. The central value of civil defenses is simply that of saving lives should war occur. In this defense function, increments to civil defense programs will continue to yield returns almost indefinitely, although marginal re turns will diminish gradually after the most urgent measures are taken. In this primary defense role, civil defenses are interdependent not only with warning, but also with active air defenses. Strategic counterforce capabilities, active air defenses, and civil de fenses also complement each other in secondary deterrence, but in a slightly different way. They are all three part and parcel of a firststrike capability, since they are all means of limiting our own damage in all-out war. Of course, increments to none of them will produce any secondary deterrent utility until their combined effect is such as to produce some probability (in the enemy's mind) that we believe our costs in all-out war would be low enough to warrant our initiating such war when sufficiently provoked. Beyond this point, increments to each would carry deterrent value by reducing our own prospective damage, although the value of active air defenses may be well below the other two measures if technology does not produce an effective anti-missile defense. For this objective of secondary deterrence alone, the aim of policy would be to achieve, with given resources, that com bination of all three measures which would produce the highest estimate by the enemy that we might strike first when provoked. There are, of course, differences in the way each alternative would affect the enemy's calculations. Certain types of civil defense, such as
286
The Reconciliation of
evacuation, can be used demonstratively in a crisis to underline a de termination to strike; it is rather difficult to demonstrate with missiles. Offensive counterforce capabilities not only can mitigate our own costs in war, but can also threaten to inflict costs on the enemy in a poststrike bargaining situation and strike directly at the enemy's tactical facilities; civil defense measures, of course, can only mitigate damage, although in their damage-limiting role they would complement strate gic forces in post-strike bargaining. The pervasiveness of interdependence, throughout the military establishment, gravely complicates the problem of determining the value of a given increment to any particular type of force. The mar ginal utility of an increment to any one system depends on all the other systems existing, planned, or potentially available. In short, rational allocations of resources—both within the military establishment and between military and non-military uses—must be founded on com parisons of whole combinations of various types of forces. Before making the comparison, the total deterrent-defense utility of each combination should be maximized (for given resources) by thorough consideration of all the interdependencies within it. Malcolm Hoag has made the important point that many kinds of military forces are subject to "increasing returns"—below a certain level of total capability, the value of marginal increments is small, but becomes large once this level is reached.14 Thus it is necessary to com pare large rather than small increments of different kinds of forces. For example, to compare the marginal utility of a billion-dollar invest ment in preliminary research for an anti-ballistic missile defense system with the marginal utility of the same investment in limited war forces would be very unfair to the former. One has to compare the utility of the completed anti-missile system with that of the limited war capabilities which could be had with the same resources. It should be recognized, however, that whether the phenomenon of increasing returns applies to a particular capability may depend crucially on whether its defense utility or its deterrent utility is being considered, and the consideration of both types of utility simultan eously may substantially mitigate the sharpness of the increasing returns. Take, for example, active air defenses. Air defense may con tribute nothing to the deterrence of all-out attack until it can provide sufficient protection to allow the survival of a retaliatory force suffi14
Malcolm Hoag, "Some Complexities in Military Planning," loc.ctt., pp. 561fE.
Deterrence and Defense
287
cient to cause unacceptable damage to the enemy. In this deterrent role it is subject to increasing returns—the value of marginal incre ments for deterrence increases sharply as a minimum strike-back capability is approached. But as a means of defending cities its marginal returns are likely to be relatively constant through a wide range of forces, or they may diminish gradually as defenses are pro vided for smaller and smaller cities. Then there is always the enemy to be considered. The deterrent and defense value of any preparedness measure depends on the countermeasures available to the enemy and the probability that he will take them. If we increase our missile striking forces or their degree of protection, and the enemy increases either the numbers or the accuracy and reliability of his forces to a degree sufficient to offset our increases, we obviously have gained nothing in our capacity to deter a direct nuclear attack and might better have abstained from the expenditure. But to realize that the enemy can take countermeasures, even if he can do so at less cost than our own measures, is not sufficient to justify a prediction that he will do so. Whether he does or not depends on a variety of factors, including his total resources, the urgency of nonmilitary demands on those resources, the degree of threat which he sees in our measure, the cost of countering, his prediction as to whether his countermeasures will provoke further increases by us, etc. Thus it is too simple to say that dispersal of SAC bases, or hardening of fixed missile sites, is futile because the enemy can produce enough additional missiles to take out the additional bases, or the hardened sites, at less cost than we would incur by dispersal or hardening. The question of whether and how the enemy will counter our preparedness moves is of course a great imponderable, but it is clearly crucial for our own decisions. There would seem to be certain appropriate rules of thumb. For instance, we should probably serve any given purpose —such as protection of our retaliatory forces—by measures which will be the most difficult for the enemy to counter, consistent with the possibly greater cost for us of these measures as compared with others. Thus, it may be so expensive or difficult for him to counter a measure such as undersea deployment of missiles that he will not try to do so. On the other hand, it may be relatively easy for the enemy to offset an expensive air defense improvement by a simple electronic device in his missiles. Another rule of thumb would be to take mea sures for our own security which are least threatening to the enemy's
288
The Reconciliation of
security. Another would be to have a diversified force which would require a many-faceted countering effort. Of course, there are many other "political" aspects to national security policy than the determination of the types of military chal lenges to which we will respond by force. A detailed consideration of these is beyond the scope of this book, but the present discussion would be misleading if we did not note that what might seem the most desirable action and preparedness policies under a "pure" deterrentdefense analysis, based only on the expected cost or marginal utility formula, might have to be qualified after consideration of the political side-effects. Thus, even though a policy of limited tactical or strategic nuclear retaliation might seem the best policy in some areas on grounds of deterrence and defense, it might have to be rejected if it would arouse fears, resentments, or tendencies toward neutralism which in the long run would seriously undermine the U.S. security position. Or a policy of nuclear sharing might have to be accepted in the interest of allied solidarity, even though, from a military point of view, its risks seemed to outweigh its potential benefits. Again, even though the deployment of U.S. troops or bases abroad might provide an optimum combination of deterrence, defense, and preparedness cost when its effects vis-a-vis the prospective enemy alone are considered, this alternative might have to be rejected if it would risk serious domestic political conflicts in the host country, threatening its Western orientation. Such considerations might be made semi-quantitative by asking ourselves a question such as the following: If this deployment or strategy, which is desirable on deterrent-defense grounds, would create political costs and risks, how much additional preparedness cost would be required to obtain the same degree of security with the next best alternative which would not incur such political liabilities? Would this increased cost be more or less onerous than risking the polit ical costs of the optimum strategy? The answers would at least be helpful in making our decision. The concepts we have been discussing—expected cost, marginal utility, interdependence, and especially the distinction between deter rence and defense—are, of course, no more than analytical tools of limited usefulness. They are not substitutes for hard thought, thorough empirical research, and seasoned judgment. There is no claim here that all national security problems can be easily solved simply by "applying" these conceptual tools. The most difficult tasks of policy
Deterrence and Defense
289
analysis are, and will remain, those of determining the facts, probabili ties, and values to which the analytical tools are to be applied. All we have attempted and hoped to do is to create an analytical framework which some may find useful. This chapter will have served its purpose if it has clarified the following points: that deterrence and defense are separate values which are promoted in varying proportion by different types of military forces; that the end-goal of preparedness policy is to achieve the best possible balance among deterring enemy attacks, mitigating war costs and other losses if deterrence fails, and minimiz ing peacetime sacrifices; that these disparate values must be made commensurate in some way before optimum decisions can be taken; and that optimum preparedness decisions are those which meet the criterion of minimizing expected cost plus preparedness cost, after suitable modification to allow for political side-effects.
INDEX Academy of Arts and Sciences, report on world-wide capabilities for produc ing nuclear weapons, 172 accidental war, IlOfi., 218-19 Acheson, Dean, 35 action policy, 265ff. active air defenses, deterrent and de fense utility, 283-86 advantage of striking first, 104fi. Agadir crisis of 1911, 253 alliances, in balance of power and bal ance of terror, 144-45 all-out war, deterrence of, 53ff.; alterna tive strategies for, 68ff. ambiguous defense, as counter to am biguous aggression, 236-38 Atlas missile, see missiles attacker-to-target ratio, 98 automation, 6, 24; in NATO strategy, 141-42 balance of intentions, 50, 239 balance of power, 41£E. balance of terror, characteristics as com pared with traditional balance of power, 41ff.; industrial potential in, 44; alliances in, 44-45; multipolar, 45; stability of, 97ff.; effects of various weapons characteristics on stability, 101-2
Baldwin, Hanson, 151n. ballistic missile early warning system (BMEWS), 113-14 bomber, utility for deterrence, 90-91; utility for defense, 91-92; utility for first strike, 93 Brodie, Bernard, 275n. broken-backed war, 123 Buchan, Alastair, 151n., 184ff., 188n. Burns, Arthur Lee, 172n. "calculated risk," political usefulness of concept, 260n. China, attitude in European war, 153 Chinese offshore islands, debate, 35-36 circular error probable, 87n. civil defense, 95-97; deterrent and de fense utility, 285-86 clean bombs, 8
commitments, 241 conventional ground forces, as comple ment to strategy of limited retaliation, 207, 283; as complement to European Group deterrent, 222-23; value for co hesion of nuclear-armed NATO, 22324; U.S. requirements in Asia, 228ff.; deployment, 229-31; as deterrent to ambiguous aggression, 235-36; as complement to first-strike threat, 282; contribution to deterrence of nuclear attack on U.S., 283 counterforce, action, 6, 21, 69, 73-74; capability, 53-54, 115ff., 148; sec ondary deterrent utility, 285-86 declaratory policy, 240ff., 265 defense, definition, 3; in all-out war, 5-6, 63ff.; theory of, 30ff.; strategies in allout war, 68ff.; primary, 276-77; sec ondary, 276-77 defense perimeter, in Far East, 249 defense value, 3-4; components of, 31ff.; vs. deterrent value, mathematical model of choice, 269ff.; primary util ity, shape of curve, 277-78 deployment of ground forces, nuclear vs. conventional strategy, 7 deterrence, definition, 3; minimum, 5, 54, 55ff., 62, 64, 115ff.; comparison with concept of "defense," 3ff.; theory of, 9ff.; credibility of, 13, 19; by pun ishment and denial, 8, 14ff.; mathe matical model, 17ff.; required cred ibility, 19; effectiveness of, 19; rele vance of Soviet ideology, 27; primary, 276-77; secondary, 17, 276-77 deterrent value, 3-4; of military action, 32, 33ff.; vs. defense value, mathe matical model of choice, 269ff.; pri mary utility, shape of curve, 277; sec ondary utility, shape of curve, 279 Dulles, John Foster, 35, 141n., 194-95 Eisenhower, Dwight D., 35, 35n., 86n., 247 Eisenhower Doctrine, 250 Ellsberg, Daniel, 17n., 19n., 242n. enemy countermeasures, 287
292 European Group deterrent, 175ff.; deter rent effectiveness, 176; capacity to trigger U.S. first strike, 177; command and control, 177ff.; U.S. nuclear shar ing with, 181 expectational value, 40 expected cost, 266-68 fail-safe system, 90 falling domino theory, 35 first strike, capability, 53-54; Soviet in centives for, 57ff.; asymmetries be tween U.S. and USSR, 81-82; credi bility of ,U.S. threat, 82-84; advan tages, 104ff.; as provocation for pre emptive attack, 107-10 force demonstration, 50, 252ff.; during Korean War, 253-54; effects, 254ff. Formosa Resolution of 1955, 250 Gallois, General Pierre, 191-92 Garthoff, Raymond L., 124n. "grey areas," deterrence and defense in, 225ff. Hilsman, Roger, 122n., 152n., 239 Hoag, Malcolm, 131n., 142n., 153n., 275n. Hounddog missile, see missiles ICBM's, see missiles independent strategic nuclear forces for NATO countries, 157ff.; European motives for acquiring, 157-58; Euro pean disincentives for acquiring, 158n.; value for resisting nuclear blackmail, 164-66; function in "link ing up" deterrent capabilities of SAC and NATO ground forces, 166-68; risks and disadvantages, 168ff.; risk of irresponsible use, 168-69; effect on European's will to contribute ground forces, 170; effect on cohesion of NATO, 170; effect on disarmament negotiations, 170; effect on stimulat ing diffusion of nuclear weapons with in Communist bloc, 170-71; effect on chances of accidental war, 171; con tribution to deterrence of attack on U.S., 173-74 independent utilities (vs. interdepend ent utilities), 276
Index independent utility of strategic nuclear forces, 276ff. independent utility of tactical conven tional forces, 280-82 industrial potential, in balance of power and balance of terror, 44 increasing returns, 286 indirect aggression, 231ff. initiative, 127 intentions and capabilities, 11, 49, 239 interdependent utilities of military forces, 282ff. interpersonal comparability of utilities, 23n. intrinsic values, 31-32 IRBM's, see missiles Japan, Soviet attack upon, as provoca tion for U.S. first strike, 225-26 Kahn, Herman, 63n., 67n., 96n. Kaplan, Morton, 165n. Knorr, Klaus, 157n., 273n. Korean War, U.S. entry, 34 Khrushchev, Nikita, 247 Lemnitzer, General Lyman L., 231n. lethal radius, 99n. Liddell Hart, Captain Β. H., 152n., 15354, 156n. "limited economy" approach to military policy (vs. the "requirements" ap proach), 260ff. limited nuclear war, 137ff.; use of strate gic nuclear weapons in, 143-44; ex tended limited nuclear war, 144ff. limited retaliation, 15, 38, 69-73, 193ff.; definition, 193-94; Dulles, John Fos ter, statements on, 194-95; Baldwin, Hanson, report on, 195; Reston, James, report on, 195; comparison with mas sive retaliation, 197-200; comparison with limited war, 200ff.; hypothetical case, 202ff.; counterforce, 209ff.; as type of local tactical nuclear war, 212ff.; as counter to nuclear black mail, 214; military requirements, 21416; relevance of civil defense, 216-17; Soviets threats of, 218; and danger of accidental war, 218-19; in combina tion with other strategies, 221-22; as strategy for a European or NATO
Index nuclear force, 224; against China, 227; in Asia, 233-35 limited war, 124-26; comparison with limited retaliation, 200ff. Locarno Agreements, 248 manpower under arms, NATO vs. War saw Pact, 154-55 marginal utility, 275ff. massive retaliation, 4, 6 mathematical expectation, 16, 267n. Miksche, F. O., 122n., 155 minimum deterrent, 5, 54, 55ff., 62, 64, 115ff. minimum first-strike punitive capability, for European country, 160 minimum strike-back punitive capability, 54; for European country, 162 Minuteman, see missiles missile defense alarm system (MIDAS), 113-14 missiles, Polaris, 87-89; Minuteman, 8788, 93-94; land-based IRBM's, 88, 92; Hounddog, 91; Skybolt, 91; Atlas, 92-93; Titan, 92-93 Moore, Ben T., 175n. multiplier effect, on stability of balance of terror, IOOn. NATO, tactical nuclear weapons, 6, 137ff.; automation in strategy, 14142; sword and shield concept, 120ff.; forces on central front, 121, 151-52; Soviet forces facing, 121, 151; shield, functions of, 121ff.; shield, deterrent role, 126ff.; shield, effect in depreciat ing threat of massive retaliation, 13536; military alternatives, 146ff.; con ventional defense capability, 148ff.; forces necessary to hold central front, 154; capacity to mobilize conventional ground forces, 154; manpower under arms, compared with Warsaw Pact, 154-55; independent strategic nuclear forces for, 157fi.; integrated strategic nuclear forces for, 174ff. NATO group deterrent, 182ff.; includ ing SAC, 183-84; including only part of SAC, 184ff.; command and control, 186-87; advantages over independent nuclear forces and European Group deterrent, 188-89
293 "nature," invocation of, as substitute for threat of intent, 247 "New Look," in U.S. military policy, 262n. Norstad, General Lauris, 121, 122, 12426, 128,131n. nuclear sharing, 17111.; with European Group deterrent, 181; bilateral, with dual control, 191ff. nuclear weapons, use in Asia, 226ff.; as complement to defense utility of con ventional ground forces, 283. See also tactical nuclear weapons Osgood, Robert E., 125n., 158n., 219n., 223-24, 236n. Pacific fleet, U.S., deployment prior to Pearl Harbor attack, 253 "plate-glass window," 7, 229 Pokrovsky, Major General G. I., 88n. Polaris, see missiles political aspects of national security policy, 288 political value, 32-33 positive control, 90 Power, General Thomas S., 91n. power values, 32-33 preparedness policy, 265ff. probabilities, 28ff. pre-emptive attack, 6, 27, 60,104ff. punishment vs. denial, as aspects of deterrence, 8, 14ff. punitive capability, 53-54 Rathjens, George W., Jr., 224n. rationality, 14,19, 25-27 Read, Thornton, 99n. "requirements" approach to military policy (vs. "limited economy" ap proach), 260ff. risk calculus, aggressor, 12; deterrer, 13 Rowen, Henry, 261n., 262n. Schelling, Thomas C., 23n., 198n., 201, 205, 257n. Skybolt, see missiles Soviet ground forces, 121, 151 spectrum of violence, 47ff. stability, in balance of terror, 97ff. Strategic Air Command, dispersal to
294 civilian airfields, 8; relationship to NATO group deterrent, 183ff. strategic capabilities, as distinguished from tactical capabilities, 52n., 137n. strategic force requirements, qualitative, 85ff. strategic reserve vs. forward deploy ment, 229-31 strategic value, 32, 33ff., 65 sufficient first-strike counterforce capa bility, 54 sword and shield concept, see NATO tactical nuclear weapons, in NATO strategy, 6, 137ff.; defense value of, 138-39; deterrent value of, 139-40; as deterrent link between conventional "shield" and nuclear "sword" of NATO, 140-41; Soviet first strike against, 142fi. Taylor, General Maxwell, 88n. threats, 23ff., 37; components of, 241;
Index clarity vs. ambiguity, 246ff.; examples of ambiguity, 246-47; general vs. spe cific, 247-48; reiteration of, 249; source, 249-50; publicity vs. secrecy, 250; context, 251 threshold of acceptable damage, Soviet Union, 57 Titan, see missiles "triggering," of U.S. first strike, 162-64 trip-wire concept, 130-31, 229, 257 Truman, Harry S., 34n. U-2 incident, 50 uncertainty, 27ff. Vagts, Alfred, 252n. value inventory, 10 warning systems, deterrent and defense utility, 284-85 White, General Thomas D., 77n., 80n., 114n.