An Epistemology of Noise 9781350011786, 9781350011816, 9781350011793

Citation preview

An Epistemology of Noise

Also Available from Bloomsbury Noise Matters, Greg Hainge Epistemology, Archaeology, Ethics, edited by Sebastian Luft and Pol Vandevelde Speculative Realism, Peter Gratton Genealogies of Speculation, edited by Armen Avanessian and Suhail Malik Metanoia, Armen Avanessian and Anke Hennig Introduction to New Realism, Maurizio Ferraris Exceptional Technologies, Dominic Smith

An Epistemology of Noise Cecile Malaspina Foreword by Ray Brassier

BLOOMSBURY ACADEMIC Bloomsbury Publishing Plc 50 Bedford Square, London, WC1B 3DP, UK BLOOMSBURY, BLOOMSBURY ACADEMIC and the Diana logo are trademarks of Bloomsbury Publishing Plc First published in Great Britain 2018 This paperback edition published 2020 Copyright © Cecile Malaspina, 2018 Cecile Malaspina has asserted her right under the Copyright, Designs and Patents Act, 1988, to be identified as Author of this work. Cover design by Catherine Wood Cover image: ‘inter esse’, Berlin 1985–87 © Maria Sewcz All rights reserved. No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage or retrieval system, without prior permission in writing from the publishers. Bloomsbury Publishing Plc does not have any control over, or responsibility for, any third-party websites referred to or in this book. All internet addresses given in this book were correct at the time of going to press. The author and publisher regret any inconvenience caused if addresses have changed or sites have ceased to exist, but can accept no responsibility for any such changes. A catalogue record for this book is available from the British Library. A catalog record for this book is available from the Library of Congress. ISBN: HB: 978-1-3500-1178-6 PB: 978-1-3501-4176-6 ePDF: 978-1-3500-1179-3 ePub: 978-1-3500-1180-9 Typeset by Integra Software Services Pvt. Ltd. To find out more about our authors and books visit www.bloomsbury.com and sign up for our newsletters.

To Andrea

Goya Series: They Do Not Agree, 1997 courtesy of John Baldessari

Contents Foreword Acknowledgements Note on Text List of Abbreviations Introduction Part 1

x xiv xvi xvii 1

Concepts: Information Entropy, Negentropy, Noise

13

I

How to Draw the Line between Information and Noise

15

II

Entropy as ‘Freedom of Choice’

23

III

Information Entropy and Physical Entropy

27

IV

The Idea of ‘Potential Information’

29

V

Physical Concepts of Information and Informational Concepts of Physics

35

Information as Process Rather Than Content

39

VI

VII To Think about Information as a Process of Individuation

43

VIII Redundancy and Necessity

51

IX

Logic and Freedom of Choice

57

X

Noise as Spurious Uncertainty

61

XI

Negentropy

65

XII

Complexity on the Basis of Noise

71

XIII The Astigmatism of Intuition

79

XIV The Path of Despair

85

viii

Contents

Part 2

Empirical Noise

91

I

On the Transduction of the Concept of Noise

93

II

Accidental Information, Predictable Noise

97

III

Ready-Made Information

103

IV

Cosmic Background Radiation

109

V

Noise in the Gap between Narratives

115

VI

Noise in Finance

119

VII Statistics: The Discipline of the Prince

133

VIII The Man without Qualities

139

IX

Noise Abatement: The Dawn of Noise

143

X

Noise Pollution

149

XI

Toxic, Viral, Parasitic

155

Part 3 I

The ‘Mental State of Noise’

165

The Crossroads: Mathematical, Technical, Empirical and Subjective Noise

167

II

Internal Chaos, Terror and Confusion

169

III

The Vicious Whir of Sensations

179

IV

Keat’s Negative Capability

181

V

Closure to Noise and the Paradox of the Declining Life

187

VI

The Catastophic Reaction to Noise

191

VII Anxiety

195

VIII Order

199

IX

203

Control

Contents

ix

X

The Helmsman Metaphor: Kybernetes

207

XI

The Helmsman in Plato’s Alcibiades Dialogue

213

Bibliography Index

219 228

Foreword To turn noise from an object of thought into ‘a variable within the process of thought’: this is the goal of Cecile Malaspina’s philosophical investigation of noise – philosophical because it entails transforming noise from an empirical phenomenon into a condition for the possibility of empirical conceptualization. Taking Claude Shannon’s notion of ‘information entropy’ as her starting point, Malaspina shows how the phenomenon of noise harbours a profound philosophical paradox. Information entropy is a measure of the degree of uncertainty or ‘freedom of choice’ about the state of a message. By aligning information with unpredictability, Shannon aligns it with uncertainty. But uncertainty implies ignorance. Thus the concept of information entropy entails this vexing consequence: if uncertainty indexes information then certainty indexes noise. But how could certainty, the apex of cognitive aspiration, be a symptom of lack of information? In contrast to Shannon’s twinning of information with entropy, Norbert Wiener’s characterization of information as the negation of entropy or negentropy seems intuitively plausible. Wiener sidesteps the troubling affinity between information and disorder by confirming our spontaneous identification of noise with disorder. Yet the curious reversibility between information and noise remains unaddressed. Focusing on this reversibility, Malaspina shows how from its inception the concept of noise as the obverse of information rests upon an equivocation between order and disorder. This is not merely an equivocation but rather an essential ambiguity, one that is symptomatic of a latent contradiction in the concept of noise. Rather than seeking to expose it as a flaw, Malaspina sees in this contradiction the clue to a deeper truth about noise. Her approach is dialectical, and the contradictoriness of noise as a concept is the key to its reality as a phenomenon. Working through this contradiction, Malaspina patiently unravels the superficial oppositions of order and disorder, certainty and uncertainty, knowledge and ignorance in all the theoretical contexts where the distinction between noise and information has been deployed. Her demonstration traverses information theory, cybernetics, thermodynamics, biology, psychiatry and sociology, drawing upon such diverse thinkers as Claude Shannon, Norbert Wiener, Gilbert Simondon, Michel Foucault

Foreword

xi

and George Canguilhem. By engaging these sources, Malaspina achieves a philosophical optic that is genuinely transdisciplinary. (Her achievement in this regard is perfectly complemented by that of Inigo Wilkins, whose work resonates beautifully with her own.1) This does not mean fashioning conceptual hybrids from disparate theoretical discourses in an eclectic and ultimately opportunistic fashion. Malaspina constructs a philosophical concept of noise by pinpointing the decisive fault lines in the workings of the various theoretical concepts whose functioning she carefully delineates. But the dialectical cast of her thinking renders her stance constructive rather than deconstructive. It allows her to integrate concepts from the mathematical and natural sciences alongside those from social and cultural theory. This transdisciplinary remit makes it possible in turn to articulate the epistemology of noise with its ontology, which is to say that it encompasses both what noise is and how it relates to knowing. What is important, from Malaspina’s perspective, is that this does not so much subvert as politicize the conventional distinction between epistemology and ontology: The conceptualization of noise is thus no longer limited to the classical philosophical problem of determining what we can understand of the reality of noise ‘in itself ’ or even ‘for us’. It is irreversibly contaminated by a political problem, which is the possibility of deliberate or accidental distortion also of our critical faculties through noise. (p.162) From this follows what is perhaps Malaspina’s most striking insight: Noise, beyond the reference to unwanted sound, thus reveals itself to be conceptually polymorphous because it has never been about types, classes or measures of phenomena that qualify noise as a particular type of disturbance, but about the relation between contingency and control. (p.203)

For Malaspina, ‘noise’ is not just the name for the force scrambling the recognizable outlines of phenomena; it designates the anomaly from whence the distinction between sense and senselessness originates. It is not merely a natural phenomenon or kind because its co-articulation with information is the consequence of an act of judgement, rather than the registration of a fact. Thus noise is a normative rather than a natural category, which is to say that it is made not given. The empirical discrimination of noise presupposes the normative establishment of its difference from information within a given disciplinary framework. But this difference – between control and contingency, determination and indetermination – follows from what Malaspina calls ‘a suspension in indecision’ or ‘unthinkable freedom of choice’ that is not of the order of structure or destruction. Precisely because it

xii

Foreword

cannot be naturalized, objectified, catalogued or classified, this anomalousness, from which judgement proceeds, can only be acknowledged by shifting from a perspective that treats noise as object to one that treats it as subject, or in Malaspina’s words, ‘the noise of cognition constituting itself, against the always looming crisis of its dissolution’ (p.173). Malaspina distinguishes two ‘subjects’ of noise here: the subject of cognition constituting itself and the experiential subject undergoing the legislated difference between information and noise. The subject of cognition enforces conditions of disciplinary regulation through which information is first identified. The experiential subject is the locus for the psychological, cognitive and affective ramifications of the experience of noise, which Malaspina explores through a particularly inventive reading of Steven Sands and John Ratey’s seminal 1986 article, ‘The Concept of Noise’. Ultimately, Malaspina’s book is an epistemic intervention. Knowledge is ‘of ’ uncertainty in both the subjective and objective senses of the genitive. It is uncertainty that knows. The subject of cognition exercising the power of judgement requires that the normative preconditions of cognitive judgement determine the constitution of empirical facts. This is the grounding power of judgement. But this grounding power is itself based on ungrounding as condition of normative grounding. This ungrounding is a consequence of the ineliminable role played by contingency in the exercise of determining judgement. This is why, for Malaspina, ‘noise is, like disorder, an inconceivable freedom of choice’ (p.187). Knowledge is not the reduction of uncertainty because it is constituted by it. Thus knowledge is neither solely predictive nor exhaustively fallible, no more than it is either verifiable or falsifiable. It is ‘of ’ uncertainty because it is rooted in this ‘inconceivable freedom of choice’. Malaspina uncovers the productive, form-generating powers of epistemic noise as the occluded source of cognition’s predictive capacities. Cognitive invention is not rooted in the ‘negentropic’ negation of contingency; it proceeds from the ‘negation of the negation of contingency’ (p.183). Regulation is not the precondition for innovation; innovation gives rise to regulation through the lawless collapse of regularity. Invention, whether cognitive, aesthetic or political, is not the negation of disorder but the negation of its negation, which is to say, the negation of order. Malaspina achieves a properly dialectical resolution of the tension between the negative physical characterization of noise as formdestroying entropy and the positive aesthetic valorization of noise as formgenerating novelty: radical transformation (whether cognitive, aesthetic or political) can only arise through the unconditioned judgement that affirms the irreconcilable tension between the destruction and generation of form. This

Foreword

xiii

judgement, constituting cognition against the backdrop of its dissolution, is a function of the noise that enables the process of thought.

Note 1 Inigo Wilkins, Irreversible Noise, Falmouth: Urbanomic, forthcoming.

Acknowledgements My first expression of gratitude goes to Prof Alain Leplège and Dr Iain Hamilton Grant (U.W.E.) for the freedom they granted me and for their unwavering trust in supervising the doctoral thesis on which this book is based. Prof Ray Brassier, Prof Emmanuel Picavet and Dr Matthieu Saladin are warmly thanked for their questions and insightful comments. Frankie Mace at Bloomsbury, as well as Deepakraj Murugaiyan have my heartfelt thanks for their support and inexhaustible attention. I thank also the photographer Maria Sewcz for generously putting her work ‘inter esse’ at our disposition for the cover of this book and John Baldessari, for letting us use his ‘They do not agree’ as the frontispiece. Catherine Wood paid particular attention to the artists wishes in designing the cover. The Reverberations conference organized by Dr Benjamin Halligan, Dr Paul Hegarty and Dr Michael Goddard at Salford University in 2010 has been determining for the transdisciplinary perspective of this book. I am grateful for their editorial support going into the publication of my contribution to Reverberations, the Philosophy, Aesthetics and Politics of Noise with Continuum in 2012. The intense exchange about noise and art with Dr Michael Schwab and all collaborators in the Data-Rush symposium organized in Vienna in 2016, especially Prof Mauricio Suarez, and also Dr Paulo de Assis and Tiziano Manca at the Orpheus Institute in Ghent have been immensely enriching. I am also especially grateful for a host of new references and ideas I owe to the generous suggestions of Prof Christian Walter and Prof Emmanuel Picavet at the Chaire Ethique et Finance, College d’Études Mondiales, Fondation Maison des Sciences de l’Homme, Institut des Sciences Juridiques et Philosophiques de la Sorbonne (UMR 8103). I am deeply grateful also to Dr Anne Lefebvre, for years of collaboration and dialogue, for her invitation to speak at the Van Eyck Academy in Maastricht in 2012 and for her initiative, to which I owe the opportunity of testing early ideas at the transdisciplinary seminar at the Ecole Normale Superieure in Paris in 2010, whose organizers Prof Claude Debru, Prof Jean-Charles Darmon and Prof Frédéric Worms are also warmly thanked. Also the European Meeting for Research in Systems and Cybernetics (EMCSR) in Vienna in 2012, 2014 and

Acknowledgements

xv

2016, as well as the Schelling Grundlagen Seminars at the Institute for Design Science, Munich, and have been important milestones. Prof Rainer Zimmermann is warmly thanked, alongside Stefan Blachfellner, director of the Bertalanffy Center for the Study of Systems Science, as well as Dr Jose Maria Diaz Nafria and the BITrum consortium. Dr David Rousseau, editor at Systema: Connecting Matter, Life, Culture and Technology, has my gratitude for his editorial advice on the publication of my article on epistemological noise, which has enabled me to articulate one of the core ideas going into this book. Not least do I thank all those not mentioned here and not directly cited, whose thought has illuminated the questions I could tackle, but also those questions that motivated me and that remain in the undergrowth. Amelie Mourgue-D’Algue, Sissi Taseva, my parents and brothers have indefatigably supported, if not freed me to work on this book. To my sons Federico, Paolo and Olivier I owe the greatest debt of gratitude, for enduring the period of research and writing and for encouraging me at critical times. To Andrea I owe the insouciance of beginning this work and the courage of finishing it.

Note on Text All quotations referencing texts with German or French titles are translated by myself.

List of Abbreviations ILFI

Simondon, Gilbert. 2005. L’individuation à La Lumière Des Notions de Forme et d’information. Grenoble: Editions J.r.me Millon.

METO

Mode of Existence of Technical Objects, Gilbert Simondon, trans. Cecile Malaspina and John Rogove, Univocal Publishing, 2017

MTC

Mathematical Theory of Communication, Claude Shannon and Warren Weaver, University of Illinois Press, 1964

NP

The Normal and the Pathological, Georges Canguilhem, trans. Carolyn R. Fawcett and Robert S. Cohen, Zone Books, 1991

WHO

World Health Organization, http://www.who.int/about/en

Introduction

It has become commonplace to use the word noise, almost with inverted comas, in a host of contexts unrelated to sound, often in opposition to information. It is thus not the din of the trading floor that interests us when we talk about noise in finance, but the uncertainty related to random variations in the stock exchange. Noise has become a concept intrinsic to the statistical analysis of the variability of data in almost every domain of empirical enquiry. Even acoustics can be argued to have fully emerged only during the 1950s, when noise could be represented as graphs of the frequencies and amplitudes of transitory signal changes over time (Castellengo 1994). That these two dimensions of the conceptualizations of noise, as sound and as random variation, speak to each other without being reducible to one another is what this book is about. This new, statistical meaning of noise is first and foremost the expression of one of the most profound methodological transformations of the modern sciences. Predating cybernetics and information theory, the source of today’s understanding of noise as inextricably linked to variation or even error goes back to the origin of the calculus of probability in games of chance, notably in the work of Pascal and Bernoulli’s law of large numbers – also called the ‘law of possibilities’. Interpreted by Laplace as an adequate representation of ‘errors in measurement in astronomy’, the ‘law of possibilities’ has subsequently been called the ‘law of errors’ (Desrosieres 2006). However, the definition of statistical noise we inherit from the ‘law of errors’ must not obscure the fact that in modern statistics precision itself has become a question intimately linked to noise: Precision is a measure of random noise. (S. Smith 2002, 34)

The special sense of the word noise thus implies both a methodological transformation and a new scientific status of the notions of uncertainty, probability, and error in relation to statistical averages. Thus enriched, the subsequent definition of noise in cybernetics and information theory has come

2

An Epistemology of Noise

to retrospectively encompass also the concept of physical entropy, and more generally of uncertainty, statistical variation and error. No longer considered only as a factor of disturbance, detrimental to information like ‘static noise’ in the channel of communication, the evolving concept of noise also becomes constitutive of new forms of knowledge and of new ways of understanding organization. This new connotation of noise has entered ordinary language as a side effect of the ‘information paradigm’ we have inherited from information theory and cybernetics (Malaspina 2012a; Morange 2006). Although Raymond Ruyer noted already in 1954 that cybernetics had failed to impose itself as a transdisciplinary scientific paradigm, the notion of noise has nevertheless steadily gained prominence over the past decades as a notion bridging disciplines: not only in relation to computer science and even complexity theory, but across the natural and human sciences and the arts, ideas of ‘order from noise’ or ‘complexity on the basis of noise’ are steadily rising to greater prominence (Atlan 1979; Mersch 2013; Nunes 2012; Ruyer 1954). The term inforgs has since been coined to emphasize the idea that we no longer inhabit only an ecosphere, but also an infosphere (Floridi 2002). Yet recent developments appear to suggest that, far from the dawn of an Information Enlightenment, we seem to sleepwalk into an era of noise: levels of stress and depression are rising as we pine under both noise pollution and information overload (Bawden and Robinson 2009; Berglund and Lindvall 1995). Even intelligence services suffer from too much information. Big data means that noise can be harvested through data mining, but vast amounts of data once more become noise as soon as we lack pertinent criteria to transform them into information (Watkins 2011, 31). A curious reversibility of information and noise thus becomes apparent: too much information, and also the repetition of the same information ad nauseam, becomes noise, whereas information that is radically new falls on deaf ears when context and criteria of pertinence are lacking to adequately distinguish information from noise. Despite the ever more apparent complexity of the relation between information and noise, the latter is often taken for granted as the mere opposite of information, based on the intuitive analogy with acoustic noise disrupting communication. What risks being overlooked in this simplistic opposition between information and noise is a palimpsest, a rich layering of intuitive notions of the still and the perturbed, the clear and the turbid (from Latin turba: crowd), opaque or confused. This opposition is also rich in ideas that

Introduction

3

have a proud history at their heels – such as order and disorder, work and futility (the latter indicating a leaking, untrustworthy vessel in medieval alchemy) (Watkins 2011, 31) – and rich also in mathematically formalized concepts, like Ludwig Boltzmann’s formalization of statistical entropy. In this palimpsest of concepts, notions and ideas, noise always appears to occupy the negative place of a dichotomy, be it in that of order and disorder, of physical work and the dispersion of energy in the state of entropy, or of the norm and the abnormal. In other words, noise is at best associated with the absence of order, of work or of the norm – be it the statistical, moral or aesthetic norm – and at worst, noise is identified as a threat to the norm and subversive of work and order: a perturbation, a loss of energy available for work, a parasite. Noise is thus a word that implicitly plays on the whole register of notion, idea and concept and does so by mobilizing linguistic, historical, sociopolitical and not least of all epistemological registers. If we are to understand the new fortunes of the previously reviled and now revisited idea of noise – from physics to information theory and cybernetics and beyond – then we must not only disentangle these notions, ideas and concepts, but also analyse the subtle ways in which new concepts of information have rewired our conceptions of noise: starting with the concepts of ‘information entropy’ and negentropy, which is what the first part of this book sets out to do, before looking at some cases of empirical noise in Part II (from the discovery of cosmic background radiation to noise pollution and the historical origin of Statistik as the nomenclature of knowledge necessary for the sovereign) and finally at the role of noise in the process of cognition itself, by focusing (in Part III) on the idea of ‘the mental state of noise’, developed in 1986 by S. Sands and J. Ratey to describe an internally experienced state of crowding and confusion created by a variety of stimuli, the quantity, intensity and unpredictability of which make it difficult for individuals so afflicted to tolerate and organize their experience. Attempts to do so may only add to confusion and psychotic phenomena. (Sands and Ratey 1986)

Yet before fanning out the whole spectrum of resonance of today’s notion of noise, it is important to seize the precise moment noise erupts as a key concept in science and technology. Claude Shannon famously devised a mathematical theory of signal transmission that paved the way for the effective elimination of noise from the channel of communication. According to Claude Shannon’s Mathematical Theory of Communication (MTC), information can be defined, in terms of

4

An Epistemology of Noise

its probability, as a measure of ‘information entropy’. Let it suffice to say here that ‘entropy’ indicates what Warren Weaver, in his introduction to MTC, calls ‘freedom of choice’, relative to the unpredictability of a message. What this means is, critically, that a piece of information informs us only, if it is not redundant, in other words, if it contains a margin of unpredictability and hence uncertainty. The status of this physical concept, ‘entropy’ – coined by Rudolf Clausius to describe the loss of available energy in thermodynamic terms – requires careful examination, especially when we consider the inflection that information theory has given to the way we use the word noise in the natural and the human sciences. Today the dictionary defines entropy as a state of molecular ‘disorder’ (Larousse 2017). Ludwig Boltzmann’s great innovation during the nineteenth century was to devise the statistical formulation of molecular entropy, which became the basis for Shannon’s formalized concepts of both information and noise. Yet despite the high degree of formalization, the intuitive idea of disorder continues to colour our idea of entropy and, consequently, both Shannon’s concepts of information and noise. The persistence of a persuasive and intuitive idea of disorder when we refer not only to noise but also to ‘information entropy’ may help explain why, despite the striking clarity of Shannon’s idea – namely that information must tell us something new, something we could not predict – his definition of information as ‘information entropy’ has failed to impose itself outside the mathematical theory of communication (MTC). What appears to have dampened the reception of Shannon’s concept of ‘information entropy’ is that the correlation between novelty of information and disorder also threatens the clear-cut conceptual opposition between information and noise. Norbert Wiener’s concept of information as, on the contrary, the negation of entropy has been adopted across the natural and human sciences through the neologism negentropy, coined by Leon Brillouin. A new understanding of information imposed itself as the negation of entropy, and more generally as the negation of disorder, meaning negation of everything contingent or unpredictable. The value of entropy thereby becomes a measure of unwanted variability, imprecision or error – in any case, a value to be eliminated for the sake of efficiency and certainty: entropy henceforth becomes synonymous with noise. Through the broad success of Wiener’s cybernetic theory of selfregulating systems with feedback, the concept of negentropy has found its way quite naturally into our thinking of organization and information in general. Indeed, the very notion of a system, any system, can be put in cybernetic terms as a set of organized constraints on contingency, in other words, as

Introduction

5

the organized negation of noise. As negation nestles at the very core of the cybernetic concept of information, information comes to reflect the level of organization of any system, insofar as it is apt to negate its spontaneous progression towards entropy. Noise, in turn, becomes a metaphor for entropy as the chaotic dispersion of energy, as disorder, if not as the entropic ‘death’ of a system. Such an impoverished view of information, impoverished because lacking in the complexity that entropy contributes, of course fails to adequately represent the theoretical wealth of Wiener’s own approach, and of the subsequent development of cybernetics into second-order cybernetics (i.e. the cybernetics of self-observing, self-regulating systems with feedback) as well as of more recent developments in complexity theory. The point is, and will be throughout, that concepts circulate through general discourse and that general discourse in turn leaves its mark on the circulation of concepts: as the metaphor of noise as ‘parasite’ in the channel of communication started thriving, the idea of negentropy disseminated itself across the natural and human sciences and general discourse, often without its mathematical formulation, and frequently without being directly acknowledged. This early formulation of one of the key concepts of cybernetics has thereby contributed to polarize our epistemic field in its relation with the unpredictable and the improbable. Now widely diffused, what subsists in general discourse of Wiener’s idea of information as negentropy subtly inflects our thinking about organization: from the organism to the ecosphere, from sociopolitical to economic relations, from networks to the idea of globalization. As a result, by emphasizing the negation of contingency our idea of information has become tethered to predictability and consequently antithetical to noise as the unpredictable. And yet our narratives of prediction and self-regulation have failed both spectacularly and catastrophically before the crises that have inaugurated the twenty-first century (Walter and de Pracontal 2009) – heralding a ‘post-truth’ era of politics, catastrophic crises in finance, war and migration. The time has come, it seems, to re-evaluate the epistemological import of Shannon’s entropic idea of information, and to do so in light of this new protagonist concept: noise. A closer look at both information theory and cybernetics reveals that the opposition between Shannon and Wiener’s mathematical approach to information and noise is far subtler than it seems. But even then, the consequences of this difference in thinking about the relation between information

6

An Epistemology of Noise

and noise remain significant. They inform the widely diffused integration of the concepts, both negentropy and noise, into theoretical contexts that are neither fully mathematized nor reducible to the technical idea of ‘noise in the channel of communication’ (Morange 2006). To understand the conceptual ramifications of noise thus requires a careful evaluation of the relation between physical entropy and ‘information entropy’. Only when the moral and perhaps even ideological connotations of the notions of ‘organization’, ‘work’ and ‘order’ are elucidated in their relation to predictive certainty can we begin to understand how noise, alongside concepts like ‘metastability’ or ‘non-linearity’, could become common parlance in business management and political lobbying alike. Take, for instance, David Cummings reflections on the ‘Vote Leave’ campaign, which he directed leading up to the Brexit referendum in June 2016. Cummings explains that the success of the campaign was driven by new communication technologies and the targeted use of social networks. His account of the success of the ‘Vote Leave’ campaign is laced with the words ‘non-linear’, ‘interdependent’, ‘unpredictable’, ‘irrational’, ‘complex’ and even ‘noise’: A news broadcast now contains much less information content and much higher noise than reading. The only way to improve this is experimenting with formats in a scientific way. (‘Dominic Cummings: How the Brexit Referendum Was Won’ 2017)

The politics of information and noise are thereby elevated to a pseudo-scientific status. However, what this pseudo-scientific smokescreen dissimulates is the oldest trick of the trade: xenophobia, the fear of the other. To acknowledge and analyse the interwoven nature of scientific, technological, moral and political components of the conceptualization of noise is therefore indispensable. It is all the more important, therefore, to stop and pause before the rapid proliferation of the idea of noise across the natural, the human sciences and public discourse. It is important to pause here for two reasons. As we have seen, the noise metaphor endows discourse with a scientific aura, lending it authority beyond the limits of its rational means. What is perhaps less obvious are the moral and ideological inclinations that can, in turn, also affect scientific discourse – for instance, when noise is associated with irrelevance, abnormality or disturbance. The second reason is more properly philosophical because it alerts us to an aspect of the theory of knowledge that can no longer be sidelined as marginal. The ambiguity that accompanies the flurry of conceptualizations of noise does

Introduction

7

not belong in the margin of error of scientific discourse, if we recognize instead that ambiguity is the space that we must negotiate as part of the moving frame of debate shared by contemporary science and general discourse. It is no longer just the question of how we define empirical noise, but of how the theory of knowledge handles ambiguity. In other words, we are looking at an epistemological problem of noise. The first thing that becomes apparent upon closer inspection is that a shared formal definition of noise is lacking. This lack opens a space for metaphorical reverberation within scientific discourse, and even more so in the straits between the natural and the human sciences, technology and the arts. Inigo Wilkins notably mounts a thorough critique of the metaphorical distortions of the concept of noise in the humanities in his Irreversible Noise (Wilkins forthcoming). Wilkins thereby aims to redress the fetishization of noise as randomness and chance, and its reduction to the unintelligible. By situating the concept of noise within the context of contemporary science, and by re-centring it around mathematical definitions of randomness in the context of complex dynamical systems, Wilkins builds a case for a concept of noise that can act, instead, as an index of intelligible constraint. Thus re-enforcing the mathematical and scientific parameters of the concept of noise, Wilkins emphasizes both formal (logical) and empirical (scientific) criteria by which to assess the critical potential of conceptualizations of noise in the humanities. The conceptual building site of Wilkin’s approach could be said to coincide with the present approach insofar as both problematize the multiple conceptualizations of noise. However, the problem addressed here differs in perspective. Rather than making metaphorical ambiguity the target of elimination, it is here taken to be a relevant philosophical problem in its own right. The objective here is thus not to eliminate ambiguity in the conceptualization of noise, but rather to make it explicit so that we can get to grips with it as a form of epistemological noise that accompanies the transformation of the epistemological field, especially in the context of today’s increasingly translational and interdisciplinary platforms for research. Metaphor, of course, remains a dirty word in the context of scientific and even of much philosophical discourse. Yet it is necessary to acknowledge that metaphors are used abundantly, if artlessly, in scientific and philosophical discourse. Being in denial about the critical role of metaphors, especially in the communication of specialist knowledge to non-specialists, and in the context of growing interdisciplinary consortia, is to submit uncritically to their rhetorical power.

8

An Epistemology of Noise

Metaphors of noise are not sufficiently subjected to critique, notably because the humanities are kept at bay. The much talked about rapprochement of the natural and the human sciences is currently still confined to making the humanities, especially psychology and sociology, more empirical, weighting analysis towards quantitative, rather than critical discursive, analysis. Yet what we need is precisely a return to the critical lessons coming from the literary end of the humanities: what we need is to learn how to use metaphors critically, purposefully and artfully. Rather than allowing the metaphor of noise to burrow its way into the cracks of scientific discourse, like a repressed and unacknowledged fear or desire, we must learn to cultivate a critical use of metaphors in public discourse within and on science and technology. We thereby honour the epistemic humility already practised in the sciences but also in the arts, where it has long been established that reality speaks to us through the art of framing, as much as through what is framed. At stake in this book is the shifting boundary between information and noise and the sprawling of the idea of noise, in the guise of its many technical and nontechnical definitions. The epistemological, moral and political implications of its impermanence are the red thread that connects the three parts of this book. It is precisely the difficulty in its conceptual delimitation, especially in light of the idea’s projection across the academic and professional boundaries of diverse theoretical and experimental fields, that makes noise a privileged philosophical problem. This restlessness of the conceptualizations of noise, the shifting boundaries between what we consider to be information and what we discard as noise, requires that we think about this restlessness as a form of epistemological noise. Meaning quite simply that the communication between diverse theoretical and experimental fields is not only subject to conceptual ‘noise in the channel of communication’, but also generates epistemological noise. The very movement of the idea of noise across disciplinary boundaries, the conceptual distortions provoked by this movement, is a form of epistemological noise accompanying its dissemination and transformations. In other words, the unstable concept of noise is itself an example of epistemological noise in the communication of concepts across theoretical boundaries. The conceptual distortions this sprawling provokes, and the metaphorical and ideological inclinations it reveals, could be said to act as what the French epistemologist Gaston Bachelard called ‘epistemological obstacles’. Yet rather than chastising interdisciplinary or transdisciplinary discourse for its imprecisions in the use of the term noise, or even its relapses into proto-mythological thinking (for instance, in the rapprochement between noise, disorder and chaos), the

Introduction

9

emphasis here is on the epistemological necessity of cross-fertilization between the diverse theoretical and experimental fields, in other words: on the coconstitutive role of noise in the formation of knowledge. This is what will be called epistemological noise. In order to understand the proliferation of ideas and concepts around noise, the focus thus cannot be exclusively technical. This would be to overestimate the fidelity of conceptualizations of noise to mathematical formalization, or even to the technical applications the term noise has found in computational logic and in new technologies of communication. However, also literary and artistic evocations of noise are insufficient on their own, if they are limited to a Romantic indulgence in noise as indicative of the incommensurability of Being. If noise becomes the placeholder concept of a philosophical Other, as that which does not submit to reason, then also hopes invested in its revolutionary potential risk petering out in enthusiasm. What is more, the temptation to indulge in noise as the mere negation of limit (embracing only the Greek apeiron, the unlimited) or of established norms (as the abnormal) squarely inhabits the conservative logic of negation. In short, it fails to subvert the very logic that the idea of negentropy occupies in the cultural and scientific imaginary. The present conceptualization of noise owes much to the attention paid in French epistemology and contemporary French philosophy to the ‘shifting sands of emergent truths’, as Alberto Toscano aptly expresses it in his translator’s introduction to Alain Badiou’s Being and Event: For millennia, philosophy has attempted to ground itself on One Eternal Necessity such as the prime mover, or the dialectic of history. Here it consciously chooses to ground itself on the shifting sands of emergent truths. (Badiou 2005, xxiii)

Readers may recognize echoes of Badiou’s insistence on the question ‘what counts as One?’ when we enquire into the question: what counts as information, and what can be discounted as noise? Also a post-Cartesian perspective on the subject, such as it animated a certain generation of French philosophers and theorists, will come into play when we consider the ‘mental state of noise’ in light of John Keats’s definition of the poet’s negative capability. However, the way noise will be problematized here, as a polyvalent and polymorphous concept, will not base itself on equating mathematics with ontology. Necessarily fanned across a wide range of topics, the problem considered here, i.e. the moving boundary between information and noise,

10

An Epistemology of Noise

will be treated more modestly like a minimal gesture in philosophical terms, a way of reiterating the act of drawing a line from various angles, rather than an epic tableau of noise in the grand genre of systematizing philosophy (Badiou 2005, 15). This minimal gesture of repeatedly drawing the line between information and noise in various contexts, has the advantage of revealing a margin of conceptual indeterminacy between diverse fields of knowledge. To take stock of the epistemological noise that arises from the inevitable sharing of concepts and from the unavoidable recourse to common language, helps us understand the growing complexity of the field of knowledge as a whole, in analogy to the way in which the philosopher and bio-physicist Henri Atlan speaks of ‘complexity on the basis of noise’ (Atlan 1979). Such an approach to the theory of knowledge, as engaging all fields of knowledge in their plurality, takes in its stride the shift from an ideal of knowledge without noise, indebted to the Cartesian Method, going towards an idea of knowledge that gains in complexity by being exposed to epistemological noise. To re-evaluate noise as a problem of epistemic complexity is to acknowledge the functional role of uncertainty and ambiguity in the process of concept formation. Michel Foucault’s introductory words to George Canguilhem’s re-edition of The Normal and the Pathological here help to express the motivation for this book: Error is at the root of what makes human thought and its history. (Foucault 1989, 22)

This is how noise can be understood, ultimately, as a radical concept, in the sense that Foucault and Canguilhem understand error, as touching the root of human thought and of its historical irreversibility. At stake is the relation between thought and contingency, which has become emblematic for a certain way of thinking about the philosophy and history of the sciences. Yet, while the swarming interest in noise makes it an imperative to engage with it conceptually, the synthetic view that is called for is, by definition, condemned to fail in making even a dent in any of the individual fields of knowledge and practice that gravitate around the notion of noise: this book will not improve stochastic models of noise, it will not resolve new problems of noise in big data, nor will it improve propositions to tackle noise pollution – least of all will it attempt to tell artists and musicians, or cultural and critical theorists, how to conceptualize noise. In fact, it cannot even begin to do justice to the extent of diffusion of the notion of noise to other disciplines, which is in a process of active fomentation, expansion and dispersion.

Introduction

11

While this book owes everything to the growing wealth of literature on noise, the conceptual movement I am after is not primarily concerned with the knowledge of diverse phenomena understood as noise, but with the idea of noise in the relation between the known, the unknown and the differently known. This is why, unlike Greg Hainge’s journey through the many dimensions of noise in his Noise Matters (Hainge 2012), this book cannot aim at what he calls an ‘ontological taxonomy’ of noise, because this is predominantly an epistemological enquiry, rather than an ontological one. Rather than aiming at a phenomenology of noise (Cage et al. 2012; Voegelin 2010), and despite benefiting hugely from the vast spectrum of literature on acoustic noise, on the cultural (Bijsterveld 2001; Boutin 2015; Gibson and Biddle 2016; Schafer 1994; M. M. Smith 2004) and even military history of noise (Volcler and Volk 2013) and the impact of cybernetics and information theory (Bunz 2012; Mersch 2013) as well as the psychology of perception (Bawden and Robinson 2009; Manson 2014; Shenk 1997), the problem of epistemological noise, as posed here, is ultimately co-extensive neither with a phenomenology of noise, (because the question posed here starts from Shannon’s counter-intuitive relation between information and noise), nor with its cultural history, (insofar as it focuses on the conceptual implications of thinking about noise). And although noise music and noise art are what opened my mental shutters to the prospect of thinking about noise, this book is about neither, leaving this avenue open for future projects (Attali 1985; Brassier 2007; Hegarty 2007; LaBelle 2006). This book looks also, albeit obliquely, at the emerging field of the philosophy of information, and more specifically at its effort of reconciling information concepts underlying science and technology with the humanities. Information philosophy has recently established itself as a specialist discipline in philosophy. It comprises new fields that have arisen from thinking about electronically mediated information, dealing with the transmission, circulation and conversion of one form of information into another (Dodig Crnkovic and Hofkirchner 2011). Its teeming developments are so diverse that they can barely be overseen and do not yet appear to constitute a coherent whole in the eyes of its contributors – encompassing fields as varied as logic and computation, cognitive and neurocognitive sciences, dynamical systems and actor network theories, cybersemiotics and biosemiotics, information systems and epistemology, and information culture and information ethics (Capurro and Hjorland 2003; Diaz Nafria 2010; Floridi 2010). Leaving many references, precious ideas and references out, indeed cutting large chunks of the work that prepared this journey into noise was the necessary

12

An Epistemology of Noise

sacrifice so as to allow the form of the argument to emerge. This cut is but the performative aspect of the problem this book ultimately faces: how do we draw that line that makes the form of an argument emerge, even an argument about noise? What can we afford to exclude? How much variety, and hence how much uncertainty can we retain, without dissolving the very movement of thought, whose emergence we only begin to comprehend? In this sense, we will ultimately come to think of maximum noise as an unthinkable freedom of choice. While this book can claim none of the academic fields it visits as its own, it seeks to understand the problem of the conceptualization of noise as a problem that relates them, without reducing them to any single dominating view. The oblique relation between these multiple domains requires that we understand the resonance of the idea of noise as something that, like the reverse of a carpet, reveals the messy connections that sustain the neatly separated forms of the academic organization of knowledge. To look under the carpet no doubt implies also a certain impertinence towards the well-established and well-deserved boundaries of specialist knowledge, at the risk of necessarily exposing one’s ignorance in comparison to those who have laboured hard to establish a more secure basis of expertise in any one of these fields. Perhaps the risk implied in being – not unlike noise – excessive of boundaries of discourse, resonates with George Canguilhem’s lightly humorous concession that ‘the philosopher is indiscrete everywhere’ (Canguilhem 1993, 19).

Part One

Concepts: Information Entropy, Negentropy, Noise

I

How to Draw the Line between Information and Noise

Draw a straight line and follow it. La Monte Young, ‘Composition 1960 #10 to Bob Morris’ The conceptualization of noise takes a new turn in relation to Claude Shannon’s definition of information as ‘information entropy’ – this much is certain. Just what it means to rethink noise in relation to ‘information entropy’ is the question posed in this first part of the book. The aim here is not to elucidate the concepts of information and noise to the engineer who is in no need of such speculations for practical purposes, but to understand how this new way of thinking about information feeds into the broader scientific and cultural understanding of noise. The latter turns out to be no passive receptacle of technoscientific concepts, but feeds back into the new technologically inspired understanding of noise, creating a new culture for theoretical and experimental practices. Shannon’s audacity consists quite simply in correlating both information and noise with uncertainty. Both concepts are henceforth derived from the statistical unpredictability he associates formally (mathematically) with physical entropy. While information entropy clearly implies a degree of desirable uncertainty, i.e. the novelty of the message, Weaver will say that noise can be discarded as ‘spurious uncertainty’. Yet it is, in both cases, unpredictability that is expressed via the calculus of probability and statistical analysis, constituting what is called ‘entropy of the message’. As Weaver explains in his introduction to the second edition of Shannon’s Mathematical Theory of Communication (MTC) of 1964: The quantity which uniquely meets the natural requirements that one sets up for ‘information’ turns out to be exactly that which is known in thermodynamics as

16

An Epistemology of Noise entropy. It is expressed in terms of the various probabilities involved – those of getting to certain stages in the process of forming messages, and the probabilities that, when in those stages, certain symbols be chosen next. (Shannon and Weaver 1964, 19)

Before getting a better grasp of the status of the concept of ‘entropy’, both as a concept in physics and as a metaphor in statistical analysis, we will compare two statements further down, one by Warren Weaver in his introduction to Shannon’s MTC and one by Norbert Wiener in his book Cybernetics (Wiener 1961). These two statements show that there was no disagreement between those generally acknowledged as the founders of information theory and cybernetics respectively, regarding the method of calculating information probability; however, they also reveal the fact that the same mathematical method nevertheless justifies two diametrically opposed definitions of information: one of information as ‘information entropy’ and the other, on the contrary, of information as the ‘negation of entropy’. Noteworthy is that these radically opposed definitions of information did not appear to constitute a problem even worthy of mention by either Shannon or Wiener. The introduction to Shannon’s MTC in fact begins by acknowledging Shannon’s conceptual debt, not only to Wiener’s mathematical work, but to his philosophy (Shannon and Weaver 1964, 3, n. 1). And yet, it proceeds to define information positively as a measure of entropy, and entropy as a ‘measure of one’s freedom of choice’. Contrary to Wiener’s definition of information as the negation of entropy, for Shannon, greater information goes hand in hand with greater uncertainty. A completely predictable message, by contrast, has only one possible outcome and is therefore redundant; it tells us nothing new. In Warren Weaver’s words, [I]nformation is a measure of one’s freedom of choice (p.9) […] in these statistical terms the two words information and uncertainty find themselves to be partners (p.27) […] entropy (or the information, or the freedom of choice […]) (p.13). (Shannon and Weaver 1964, 9–27)

In other words, the redundant message presents no ‘freedom of choice’, because it contains no ‘information entropy’. Information is null, if there is no uncertainty about the state of the message. Norbert Wiener, too, acknowledges the shared origin of the statistical conception of information in his own and Shannon’s work, amongst others: This idea [of developing a statistical theory of the amount of information, in which the unit amount of information was that transmitted as a single decision between equally probable alternatives] occurred at about the same time to

Concepts: Information Entropy, Negentropy, Noise

17

several writers, among them the statistician R. A. Fisher, Dr. Shannon of the Bell Telephone Laboratories, and the author. (Wiener 1961, 10–11)

However, Wiener and Shannon arrive at diametrically opposed ideas of what information is, because Wiener defines information precisely as the opposite of ‘information entropy’, namely as the negation of entropy (which the physicist Leon Brillouin later entrenches as the dominant technoscientific definition of the concept of information, by inventing the neologism negentropy): The notion of the amount of information attaches itself very naturally to a classical notion in statistical mechanics: that of entropy. Just as the amount of information in a system is a measure of its degree of organization, so the entropy of a system is a measure of its degree of disorganization; and the one is simply the negative of the other. (Wiener 1961, 10–11, emphasis added)

For Wiener information is precisely the reduction of freedom of choice, and thus the reduction of uncertainty. Information, in Wiener’s cybernetic theory, is a measure of increased constraint, associated with ideas of organization and order, bound to decrease entropy. Here entropy is not a measure of information, as with Shannon, but, on the contrary, a measure of its presumed opposite, i.e. disorder or noise. There is a startling matter-of-factness in the way both mathematicians provide diametrically opposed definitions of information, without mentioning their fundamental divergence. This may be indicative of the low priority that discursive definitions have for the two mathematicians. The real emphasis is instead on the mathematical innovation, which both share without disagreement, even complimenting each other. The concepts of information and noise are treated as theoretical tools that must be not only fit for purpose, meaning the communication of mathematical theories to a broader public, but also tailored to different needs, be it the transmission of a trans-Atlantic telephone conversation, or a successful targeting of a self-directing missile, to name just two of the most frequently cited examples of information theory and cybernetics. What this tacitly accepted dichotomy between ‘information entropy’ and the ‘negation of entropy’ reveals is, first of all, that there is freedom of choice in the discursive interpretation of mathematical formalization. This freedom of choice is nothing other than the ambiguity of non-mathematical concepts. It is thus not at the level of mathematical stringency itself that ambiguity arises about the conceptualization of information and noise, but at the level of freedom of choice in its discursive interpretation. This source of discursive ambiguity is

18

An Epistemology of Noise

important to underline, since it is at the level of discourse rather than at the level of mathematics, that the concepts of information and noise are translated into other scientific domains – notably biology (Morange 2006), and from biology to sociology and economy etc. – often via a tacit adoption also of the cybernetic paradigm of self-regulating machines with feedback (Mersch 2013). It is here, at this crossroad of conceptual circulation, that we must be most attentive, because concepts reveal themselves to be more than just theoretical tools: they are prisms through which we see and discover the world at the same time as being the tools with which we transform the world. Their consequences go well beyond mere functionality in a theoretical apparatus for this or that technological or scientific purpose: concepts contribute to shape cultures and precondition value judgements, while being in turn also imbued with cultural preconditions and slanted by pre-existing value judgements. Claude Shannon’s definition of information as ‘information entropy’ has the singular merit of having prepared the ground for a philosophy of noise that evades the Manichean opposition between information and noise, echoing that between order and disorder, life and death. It also evades the mere relativism according to which what we define as information or noise is a question of individual perspective. To demonstrate the cultural relevance of this conceptual feat, we will tackle the difficulty that arises when the concept of noise is no longer applied only to the channel of communication, but also to other domains, where the distinction between information and noise is not a given. In vivo, rather than in the well specified and controlled situation of the channel of communication, the distinction between information and noise is never readymade, but always presents itself as a vital decision or as an epistemological problem. At stake, hence, is the difference between information and noise in the making, i.e. the moment when information must be selected prior to the transmission of a message, necessarily implying a decision, an act of selection whereby information can stand out from noise. Is not the challenge of every form of research the problem as to how we can identify what counts as information and what, in turn, can be discounted as noise? The dividing line between information and noise is so fundamental to all forms of enquiry and experimentation that the consequences of Shannon’s ‘entropic ideas’ vastly exceed any technological framework, making the conceptualization of information and noise philosophical problems in their own right. Shannon’s ‘entropic ideas’ require us to rethink our most basic attitudes concerning information and noise. Rather than opposing noise to information, as the presence of entropy to its absence, he divides the presence of ‘information

Concepts: Information Entropy, Negentropy, Noise

19

entropy’ from the presence of ‘noise entropy’. The dividing line between information and noise now runs within entropy, rather than between entropy and its negation. This is a subtle but fundamental shift that effectively challenges the principle of the excluded middle, according to which a proposition is either true, or its negation is true, and which implicitly underscored the analogy of the information/noise opposition with that of sense/non-sense, and even organization/chaos. A new division between desirable and spurious uncertainty now competes with the classical opposition between truth and error or, as in the excluded middle, between the truth of a proposition and its negation. The philosophical consequences are profound, for the process of information can now also be understood as a cut across the fabric of uncertainty. Information becomes the process whereby this cut progressively gives rise to a form of measurable uncertainty. Shannon’s ‘entropic ideas’ thus have a profound philosophical and, more broadly, cultural importance, if only we are willing to consider their conceptual relevance beyond the technical realm. Common criticism instead holds that Shannon’s concept of information applies only to electronic signal transmission, and is utterly misleading in any other context. Complicit with this criticism is the equally common position that opposes culture and technology. Endowing only ‘cultural’ artefacts with signification, this view reduces all aspects of technology and even of science to their mere utility. In its extreme form it represents a technophobia that pits culture, and even nature, against science and technology in a relation of hostility. The widespread cultural condescension towards the mere utility of the sciences and technology corresponds in kind to the technocratic dismissal of culture as mere recreation (or product of consumption). In fact, both attitudes are but two sides of a coin. Both fail to recognize the cultural potential of Shannon’s deconstruction of the traditional opposition between information and noise and the revaluation of uncertainty that it entails. Just as technocracy implies an aberration of cultural values, so technophobia fails to rescue culture, because it is itself the symptom of a redundant, conservative idea of culture. It is redundant because it wilfully ignores the fundamental role that forays into mathematics and more broadly science and technology have always played in the visual arts, literature and music. The fascination with mathematics, science and technology has characterized the art of Greek Antiquity no less than the metrics of Arabic poetry, it has fuelled the European Renaissance five hundred years after driving Bagdad’s cultural prominence, finally it has been an indelible aspect of twentieth-century art, literature and music, and is even more so of the artistic practices of new millennium.

20

An Epistemology of Noise

The redundant opposition between technology and culture atrophies not only the quality of engagement between the arts, the sciences and technology, but in turn also atrophies the status of creativity attributed to science and technology, by denying it its cultural relevance beyond its utility. French philosopher Gilbert Simondon was right to speak of an enslavement of technology and to see in it a factor for mutual alienation in culture. To place Shannon’s ‘entropic ideas’ within this cultural frame of debate thus means overcoming the consensus that there is an opposition between technology and culture. The first task is to work against this alienation, so that we can recognize Shannon’s as a minimalist definition of information and noise of the highest cultural relevance. It is minimalist insofar as it deals with the conditions of possibility of information, precisely by bracketing out signification: it separates out signification from both the means and the process of transmission – thereby revealing the structural and procedural conditions of information processes, much like minimalist art did with artistic expression in an industrialized world. Taken outside the narrowly technical context of signal transmission, we can begin to see that Shannon’s ‘entropic ideas’ also offer an iconoclastic definition of information and noise, one that breaks the spell fusing signification with the means and process of its transmission as if they were one. Some of the most beautiful words regarding the reconciliation of culture with science and technology have been written by Gilbert Simondon in On the Mode of Existence of Technical Beings (Simondon, trans. Malaspina and Rogove 2017, 15–16): Culture has constituted itself as a defense system against technics; yet this defense presents itself as a defense of man, and presumes that technical objects do not contain a human reality within them. […] The most powerful cause of alienation in the contemporary world resides in this misunderstanding [caused] by its absence from the world of significations, and its omission from the table of values and concepts that make up culture.

Guiding us here is the ethos, rather than the method deployed by Simondon in METO (On the Mode of Existence of Technical Objects), where he gives a genetic account of the modalities of technicity across what he calls the evolution of technical individuals, technical elements and technical ensembles. It is not the objective here to construct a genetic analysis that would be in any way comparable to what Simondon did for the concept of technicity. No comparable historicizing claim will be made about the splitting of the idea of noise across technics and religion, and across theory and ethics. Nor will there be an attempt to determine the role of aesthetics in mediating such a split.

Concepts: Information Entropy, Negentropy, Noise

21

The objective here is more modest. It is to test two widely held presumptions about noise, and to do so in a number of different contexts, so as to reveal their intrinsic relatedness. The first is the implicit presumption that we can rely on an intuitive notion of noise, in order to bridge its definitions across different techno-scientific and cultural settings. The second presumption is that, rather than intuition, it is a formal, i.e. mathematical, definition that presides over the multiple uses of the concept of noise across the spectrum of scientific discourse. What emerges instead are far from uniform conceptions of noise, some of which profoundly counter intuitive. Although ubiquitous, both the idea of noise and information reveal themselves to be conflicted, both displaying a fundamental ambivalence towards novelty and change, as signaled by Shannon and Wiener’s mathematically identical, yet discursively opposed definitions of information.

II

Entropy as ‘Freedom of Choice’

To equate information with unpredictability is intuitive enough, if it is to tell us something new, something that does not follow automatically from what came before. It is equally easy to accept that a message we can fully predict is redundant, if it gives us no new information. What is much less intuitive are the consequences Shannon and Weaver draw from this unpredictability. By aligning the concept of information with uncertainty and by quantifying it as such and without concessions, we arrive at the apparently paradoxical conclusion that more information means more uncertainty. This appears paradoxical, in the sense that it contradicts the equally common assumption, even the doxa (opinion or dogma), that information is what reduces uncertainty, rather than increasing it. Quantifying information according to the degree of uncertainty it presents, according to the ‘entropy of the message’, has therefore caused alarm (Janich 2006). Shannon appeared to have fundamentally misunderstood what we mean by information. The fundamental role he gives to contingency in information contradicts what we commonly associate with the purpose of information, namely that information is reliable only if it reduces uncertainty and makes experience less contingent. Shannon’s definition of information thereby appears dangerously close to that of noise. In his article ‘What Is Information?’ José Maria Diaz Nafria’s makes the implications of this difference between the ordinary sense of information and Shannon’s definition of information entropy abundantly clear. If the only criteria for the quantity of information is its unpredictability, expressed in terms of entropy, then a critical signal consisting of few bits would be discounted as low in information, while the high entropy of irrelevant background noise would measure the greatest quantity of information: Just one binary digit may tell us if the universe is about to collapse, thus being very informative, and all millions of terabits on the web could just as well

24

An Epistemology of Noise be generated by the whim of electrons in a rheostat, being thus completely uninformative. (Diaz Nafria 2010)

It must be clear from this example that Shannon’s entropic ideas about information are not a mere extension or deepening of the ordinary notions of information and noise, but a challenge to the ordinary conception of information of the highest order, since, as Diaz Nafria’s example makes plain to see, nothing distinguishes outwardly ‘information entropy’ from what we would ordinarily call noise. While information is, as a matter of course, meant to tell us something new, the logical consequence that this novelty decreases predictability and thus increases uncertainty appears to be going one step too far. Shannon’s quantitative measure of information has since been interpreted almost as a form of sacrilege against the ‘true’ understanding of information, which ought to increase certainty. It is also discounted as incapable of telling us anything about what matters, which is not quantity, but quality of information, and which is a prerogative of its signification. We could say, on the other hand, that Shannon’s understanding of the relation between information and contingency is indeed paradoxical, but not because of the misplaced conceptual ambition. It is paradoxical in the sense that it is free of cultural pre-conceptions and therefore offends such pre-conceptions, transgressing their doxa: in this sense the conceptual innovation inherent in Shannon’s concept of ‘information entropy’ indeed acts as a form of conceptual noise, when it is exported from its technological application to other domains. Let us be clear, Shannon’s definition of information as an ‘uncertainty relation’ does not contradict itself, but the doxa according to which one ought to obtain from information simultaneously both novelty and a reduction in uncertainty. Shannon’s definition of ‘information entropy’ instead frustrates this paradoxical need (novelty and certainty) and thereby enables us to think about contingency as belonging to the conditions of possibility of all processes of information, including but not only of those processes we associate with signification in the semantic communication between sapient beings. What, then, is the relation between uncertainty and information, and hence also between information and noise? The answer to this question is not as obvious as it might at first seem and unfolding it may change the way we think about both noise and information. It is this question that is posed, in mathematical terms, by Shannon’s MTC. Shannon gives an engineer’s answer to this question, which Warren Weaver translates for a broader readership, expressing it in the following way in his 1964 introductory essay to the MTC:

Concepts: Information Entropy, Negentropy, Noise

25

[I]nformation is a measure of one’s freedom of choice […]. (Shannon and Weaver 1964, 9)

This definition is intentionally shortened here in order to indicate the importance we must attribute to it in the context of our enquiry into the conceptualizations of noise. As a technical definition it is, first of all, abstracted from the habitual association of information with signification. The common criticism of Shannon’s concept of information is therefore that it is a purely quantitative measure that is indifferent to the signification of a message, which consequently ignores the inherently qualitative aspect of information, which is its signification. The criticism levied against Shannon’s ‘entropic ideas’ is certainly important and valid at the level of interpretation and evaluation of a message, but it has also detracted attention away from two very subtle philosophical gestures that Shannon’s approach implies: one concerning information as a process rather than a given, the other concerning the role of contingency, and hence of uncertainty, in this process. It is necessary to render these explicit, as the concept has de facto been translated and applied to a great variety of disciplines in the natural and human sciences, often as a language that facilitated profound methodological upheavals, like the shift from classical to non-classical mechanics, from classical physics to quantum physics, from classical biology to molecular biology and biophysics. Shannon’s definition of information as ‘entropy of the message’ offers us a contribution to our understanding of contingency, which challenges both the doxa according to which information is a correlate of certainty and the certainty with which we can distinguish information from noise when the concept is taken outside the technical paradigm of the channel of communication, for instance in biology or economic theory, where the scenario of the engineer who transmits a ready-made message as information no longer prevails. Shannon’s contribution, which follows from the fundamental realignment of information and uncertainty, is fundamental insofar as it enables us to place information and noise on an equal footing, where both represent a measure of ‘entropy’ or unpredictability, prior to the assignment of signification, purpose or representation; prior, in other words, to the levels of decoding, interpretation and evaluation. If we follow through with Shannon’s ‘entropic ideas’, our fundamental assumptions about information must be rethought, taking contingency and hence noise into account, not only as that which impinges on the fidelity of the message, not only as that which obstructs the decoding and interpretation of information, but as an uncertainty fundamental to the process of information

26

An Epistemology of Noise

itself. This attention to contingency and uncertainty is what will enable us to rethink the definition of noise, to take it outside the channel of communication, in other words to think about noise in vivo, where the distinction between information and noise is always a process in the making. The emphasis is here on the conceptual consequences that must be drawn from Shannon’s alignment of information with contingency, and more specifically concerning the less Manichean, less oppositional relation between the concept of information and that of noise. We can now ask, in light of the indelible aspect of contingency in the information process, on what grounds can we draw the line between ‘information entropy’ and noise? Meaning quite literally that the distinction between information and noise is a problem of ground or foundation of knowledge. If both noise and ‘information entropy’ are measures of ‘entropy’, understood as in purely probabilistic terms as ‘freedom of choice’, then how can we be sure which measure of choice informs and which exceeds and deforms? If uncertainty increases with both information and noise then, in this greater uncertainty, with what certainty do we draw the line between information and noise?

III

Information Entropy and Physical Entropy

The conceptual operator upon which the idea of information uncertainty hinges, is entropy. But how are we to understand the idea of entropy, when it is no longer a concept bound by the theoretical and empirical constraints of the field in which it arose as a key concept: thermodynamics? Shannon’s theory of information is said to have emerged from his work on Boolean logic, applied to electrical switches. It owes more, in fact, to Norbert Wiener’s use of the calculus of probability in cybernetics, than to a direct engagement with Boltzmann’s statistical theory of physical entropy (Atlan 1979). The resulting formalism, however, is not just metaphorically, but formally analogous with the statistical expression of physical entropy. Let us see how Shannon’s designation of information as ‘information entropy’ builds entropy as an indispensable metaphor into this ontologically arbitrary concept of information. Shannon’s choice of expressing the unpredictability of the message as ‘information entropy’ (H) reflects the fact that its mathematical formulation in information theory, is indeed almost identical with Ludwig Boltzmann’s statistical formulation of molecular entropy in thermodynamics (S): H = – Σ pilog pi, S = – K Σ pilog pi Both information [H] and the physical system [S] measure the number of possible states, either as a message or as a physical entity. This probability, attached to the number of possible states, is the sum of probabilities of the ‘presence’ [p1, p2, …, pn] of ‘signs’ or particles [i], multiplied by a logarithm [log]. [H] is thus a measure of the uncertainty over the occurrence of one amongst all possible events H(p1, p2, …, pn). If all probabilities [pn] are equal, then the greatest possible ‘freedom

28

An Epistemology of Noise

of choice’ corresponds to the greatest possible uncertainty regarding the actual state of either system or message, with respect to all its possible states. More simply put, in the state of maximal entropy the movement of particles is determined by nothing but their random collision. The more random the movement of particles is, the greater is the number of their possible positions, speed and direction. Maximum entropy thus corresponds to the equal probability of all possible states of the system, which is often illustrated with the idea of the random collision of molecules in a canister of gas. All configurations of free particles can thus be said to occur with equal probability. Conversely, the probability of predicting the state or behaviour of the physical system, meaning the exact configuration of its micro-complexions, including the positions, speed and directions of its particles, is at its lowest. Entropy and noise indeed become identifiable, and to this extent more predictable, thanks to a better knowledge of the statistical framework which encompasses them, which we owe notably to the important work of Boltzmann and Shannon, no less than Wiener and many others. Nevertheless, it would be wrong to overstate the statistical determination of noise and call it predictable, because ultimately entropy and noise remain a measure of ‘freedom of choice’, characteristic of an undetermined state of the system or message. The quantity of information in Shannon’s sense is analogous to the probability with which the observer of a physical system can predict what Max Planck called the microcomplexions of a given system, and the probabilities of finding the system in any of these complexions: Maximal disorder corresponds to the greatest number of possible complexions with equal probability for all […]. (Atlan 1979, 31, n.1)

Disorder here means that no external order imposes any form of constraint that would compel particles to behave in one way rather than another. Both entropy and ‘information entropy’ must thus be defined by sophisticated statistical measures expressing the receiver’s uncertainty as to the determination of the system, message or event. Increased quantity of information, in this sense of ‘information entropy’, is thus not the equivalent with increased certainty about the system, even if certain forms of noise have identifiable and reproducible characteristics in statistical terms.

IV

The Idea of ‘Potential Information’

To predict the probability with which signals occur in a message, Shannon uses a mathematical expression that is almost identical to Boltzmann’s. What is significant, however, is that he leaves out the term ‘k’. ‘k’ is the physical constant that expresses the calorific value of flows of energy, understood as displacements of thermal charges, wherever a disparity exists between energy levels, for instance in electrical currents or in flows of matter. This algorithm, ‘k’, is what anchors Boltzmann’s formula in physical reality. Physical potential arises from unequal energy levels that compel a process of equalization. Potential therefore expresses a form of constraint on the system, which is forced by the disparity of energy levels to evolve in such a way as to equalize this difference. Potential thus increases the probability of a system to evolve in the direction of this equalization of energy levels, and its way of doing so will depend on the interaction between its constituent elements. Potential thus effectively reduces the number of possible states of a physical system: compared to the state of maximum entropy where all possible states occur with equal probability (for instance, the molecules randomly bouncing off each other in a canister of gas), potential obliges the system to actualize an equalization of energy levels. Now, if potential reduces ‘freedom of choice’ by compelling a process of equalization of energy levels, then the greater the disparity of energy levels – the greater the potential – the more powerfully the system is entrained to evolve in a particular way, as for instance in the flow of an electric charge. Even if a margin of indeterminacy persists as noise, potential is what reduces the number of possible states of the system, forcing it to evolve according to its constraints. To transpose the idea of potential information to Shannon’s ‘entropic ideas’ thus runs us into difficulties, if we want to preserve the idea of ‘freedom of choice’. To understand potential as a form of constraint no doubt offends common sense. It forces us to pause before the usual idea of potential as synonymous

30

An Epistemology of Noise

with opportunity in the sense of ‘freedom of choice’. When we say that an event has the potential to occur, we usually mean that it may or may not occur: potential thus represents an added option and hence greater ‘freedom of choice’, as when one says ‘he has the potential to become a great pianist’. When we say a child has potential, what is meant is that it has the opportunity to develop certain capabilities. On the contrary, a child seen as lacking in such potential is presumed to have fewer opportunities and thus a lesser ‘freedom of choice’. The child’s perceived aptitude, however, is better understood as an optimum fit with given requirements. The potential to become a great pianist, for instance, compels a child onto a certain path of development. Because potential is valued as increased ‘freedom of choice’, we ignore the constraints that turn perceived potential into a constraint: the child with perceived potential is compelled to succeed. But what is success, if not the growing constraint to proceed from one level of achievement to the next? It becomes an almost mechanical sequence of expectation and compliance – which ultimately feeds into the social reproduction of relations of power. If to succeed, more often than not, increasingly narrows the noose of the expectation not to fail, then is the ‘freedom of choice’ we attribute to potential not ultimately greater for the one who is not pressed into the mould of expectations, corresponding to a perceived potential, or who deliberately confounds these expectations? Art critic Martin Herbert, for instance, in his collection of essays Tell Them I Said No, looks at the work and lives of ten artists who, in different and sometimes extreme ways, refused to play the game of celebrity that enchains artists to an ‘overly educated’ and ultimately conservative audience (Herbert 2016; Judah 2017). In a market driven society it is undoubtedly heretical to question the link between potential and ‘freedom of choice’. It is only when we turn the idea of potential around and express a negative potential that the compelling nature of potential as constraint becomes more evident, and we can begin to think about potential as reduction of ‘freedom of choice’. If we say, for instance, that a certain group of underprivileged children will potentially fail to thrive in society, according to available sociometric parameters, then the idea of ‘potential’ reveals its negative characteristic of constraint more readily. In both cases, however, expectations are present, be they valued positively or negatively, and potential can be said to reduce ‘freedom of choice’ insofar as it represents criteria, norms and structural conditions that compel a child, or any other observable phenomenon with potential, to evolve according to its greatest probability. What

Concepts: Information Entropy, Negentropy, Noise

31

needs to be retained here is simply that the physical concept of potential, with its noble Aristotelian heritage, is by far not an anodyne synonym for the possible, when it comes to understanding the relation between probability and ‘freedom of choice’. It is thus important to stay alert when considering whether ‘information entropy’ must be understood as ‘potential information’. Physical potential implies that an event is more likely to occur, thus in fact reducing the number of possible events, as when one switches on a light circuit and the electricity is compelled by the physical potential to rush through the wire. Both potential and freedom of choice are manners of speaking about a possible event, yet the difference of inflection between potential, perceived as an option, as greater ‘freedom of choice’ and potential as greater probability of occurrence, hence reduction of choice, is not without consequence. In the state of maximal entropy, on the contrary, initial differences in energetic potential have equalized through interaction, until the system as a whole finally reaches a state of energetic equilibrium, where flows of matter or energy from one part of the system to another are highly improbable, at best random effects, because the micro-constituents of the system are no longer exposed to the tension of discrepancies between energy levels, no longer compelled by the physical potential that arises from these differences. Consequently, each state of the entropic system occurs with equal probability or, differently put, with the greatest ‘freedom of choice’. Coming back to Shannon’s formal mathematical definition of information, this means that to define ‘information entropy’ as potential information is to inverse it completely: potential, strictly speaking, would be a negentropic factor, negating entropy. In Boltzmann’s definition the algorithm [k] serves to indicate flows and displacements of thermal charges. Yet by leaving out the reference to this physical aspect, Shannon transforms Boltzmann’s mathematical expression of entropy into an ontologically arbitrary measure of probability. Shannon thereby unmoors probability from Boltzmann’s empirical measure of calorific conversion of energy and work related to thermal displacements in a physical system. Although Shannon himself applies this formula to the problem of electronic signal transmission, his concept of ‘information entropy’ and hence also of noise is now devoid of any ontological reference: it could inform us about the probability of occurrence of any phenomenon involving large numbers, be it the flow of signals, flows of people, of goods or unicorns – in short it is ontologically arbitrary.

32

An Epistemology of Noise

The abstraction of Shannon’s quantitative measure of information is undoubtedly what facilitated its translation with great ease into every imaginable field of research involving mass phenomena. It lends itself to statistical analysis, not only in communication technology, but also in economical or biological systems or any other domain. This gives Shannon’s concept of information an eminently analogical, if not paradigmatic function. Shannon’s definition is lower in theoretical constraints than Boltzmann’s formula, giving it greater polyvalence, but by the same token also increasing ambiguity of its interpretation: with respect to Boltzmann’s definition of entropy, Shannon’s concept of ‘information entropy’ is thus itself a prime example of what ‘information entropy’ does, when it increases the number of possible interpretations, namely increasing also uncertainty. The analogy between information and entropy, nevertheless, remains paradigmatic in the denomination of information as ‘information entropy’. The application of the term information to communication technology also reinforces this analogy with physical processes, because the engineer must deal with the effect of physical entropy in order to ensure the message is sent without loss due to perturbations, such as thermal noise, during the transmission of an acoustic or electric signal. The physical analogy thus persists, but as we have seen on a metaphorical rather than formal, mathematical level (Morange 2006). As a result the notion of physical potential remains, despite the obliteration in Shannon’s concept of the physical referent [k], a key feature of the concept of information. What persists is also a margin of ambiguity, when one speaks of ‘information entropy’ as the information one lacks, or as ‘potential information’, as the German physicist and philosopher Carl Friedrich von Weizsäcker does, using the term entropy here with specific reference to Shannon’s ‘information entropy’: Positive entropy is potential (or virtual) information. (Weizsäcker 1994, 167)

However, the idea of possibility which is implied in both physical ‘potential’ and the ‘virutal’, risks blurring a distinction that is perfectly clear to the physicist and much less clear in ordinary language. What von Weizsäcker means is that entropy, which denotes the number of possible states of a system, corresponds to virtual information and he specifies this in the parenthesis. Von Weizsäcker thus speaks of ‘potential’ information in order to make the idea of ‘information entropy’ more accessible. However, the virtual refers to the number of possible states, which increases with ‘information entropy’, while physical potential, as

Concepts: Information Entropy, Negentropy, Noise

33

we have seen places a constraint on the physical system and thus decreases the number of possible states by making one event more likely than another. Where the notion of information ‘potential’ is introduced, it is thus in fact reintroduced as an extrinsic criterion for the evaluation of ‘information entropy’, more specifically of its hoped-for use-value as information in the traditional sense of certainty and constraint. Better put, the idea of potential information introduces the idea of the capacity of ‘information’ to perform work, to make sense, which in turn is specific to the recipient of this information and the use s/he can make of it. What remains ambiguous and unspoken is the necessary conversion between the uncertainty that ‘information entropy’ introduces as ‘freedom of choice’, as under-determination, and the implied sense of potential information leading to negentropy, i.e. of increased certainty and constraint. Implied is that the actualization of potential information is equivalent with this conversion of uncertainty into certainty. And nothing could be further from certain than the spontaneous consolidation of uncertainty into certainty. For what this requires, is also that the nature of the boundary between information and noise changes, from being a limit that curtails the uncertainty of the ‘entropy of the message’ vis-à-vis the unlimited uncertainty of noise, to a border that opposes information and noise as certainty and uncertainty. There is thus continuous ambiguity at the level of conceptualization when the notion of physical entropy is transformed into the pure probability of ‘information entropy’, at once untethered from the physical paradigm, yet indelibly tied to it through metaphor and philosophical tradition. It is this ambiguity that constitutes ‘epistemological noise’ when Shannon’s concepts of information and noise are exported, alongside negentropy and often without distinction, to other domains, like biology, sociology and economics, where the physical paradigm risks becoming prematurely reductive.

V

Physical Concepts of Information and Informational Concepts of Physics

The controversial aspect of Shannon’s definition of information is that it is one of randomness and unpredictability, for which Shannon uses not only the mathematical expression but also the term entropy. Shannon in fact continued the line of enquiry into the mathematical treatment of signal transmission begun by H. Nyquist and R. V. L. Hartley at the Bell laboratories. He openly declared his debt to Norbert Wiener’s work, who in turn pointed out the innovation that Shannon’s ‘entropic ideas’ represented for information theory (Shannon and Weaver 1964, 3). As Weaver points out in his introduction to MTC, the notion of entropy was already associated with the notion of information in physics, notably in the work of L. Szilard, and had proven useful in quantum mechanics and particle physics, notably in the work of von Neumann (Neumann 1932, chap. V). Yet albeit being fundamental to the physical sciences and engineering, it is easy to see how the notion of entropy becomes a source of confusion when introduced into common language. There is of course no obvious reason why the concept of information should be introduced in this way into our understanding of physical processes, or why concepts from physics should pertain to our understanding of information1. The biophysicist Henri Atlan already takes this controversy into account, referring to Leon Brillouin’s 1959 Science and Information Theory, but also to the more general problem of the use of intuitive concepts in physics, such as ‘energy’, ‘force’ and ‘speed’, which has been discussed notably by Cornelius Castoriadis, G. Hirsch and J.-M. Levy-Leblond (Balibar, Lehoucq, and Lévy-Leblond 2005; Brillouin 2013; Hirsch 1976; Lévy-Leblond 1976). Outside of the purely scientific engagement with the notions of entropy and information, Shannon’s entropic definition of information also provoked and still provokes controversies as being excessively technical and alienating, if not

36

An Epistemology of Noise

contradictory (Capurro and Hjorland 2003). The term ‘information entropy’ evokes the paradoxical notion that information is reduced to disorder, if not chaos or, on the contrary, that entropy corresponds to the idea of homogeneity and un-differentiation, which is the opposite of what one would normally associate with the idea of a signal or message that stands out against the indifference of background noise. The mathematical formalization Shannon uses is, as we have seen, almost identical to the way in which Ludwig Boltzmann first formalized the statistical measure of entropy in a physical system, as expressing the average of all its possible microphysical configurations, occurring with equal probability under specified constraints. It is understandable that this notion of ‘information entropy’ is incompatible with what one ordinarily calls information, if ‘information entropy’ evokes simultaneously the ideas of disorder and of homogeneity, and which to boot becomes a measure of the information we lack: Dr. Shannon’s work roots back, as von Neumann has pointed out, to Boltzmann’s observation, in some of his work on statistical physics (1894), that entropy is related to ‘missing information’, inasmuch as it is related to the number of alternatives which remain possible to a physical system after all the macroscopically observable information concerning it has been recorded. (Shannon and Weaver 1964, 3, n. 1)

Yet how can the quantity of ‘information we lack’ correspond to the ‘quantity of information’ we receive? The natural answer to is to say that it is precisely the opposite that is the case, that information is the opposite of the ‘information entropy’, namely its negation, and to explain this with the minus sign that precedes the symbol Σ in the equation used to measure ‘information entropy’: H = – Σ pilog pi This definition of information as negation of entropy has become core to the now dominant definition of information as negentropy. But to accept this neologism, without enquiring into the theoretical conversion it implies, risks discarding too hastily the philosophical potential of Shannon’s ‘entropic ideas’, thereby risking to transform the concept of information into one of redundancy. To negate entropy, is to negate all possible alternatives, and hence to affirm an identity that cannot change. Another way of solving this dilemma is to say that the more improbable the occurrence of a particular sign is a priori, the more informative it is a posteriori (conversely, if its occurrence was certain a priori, then it would

Concepts: Information Entropy, Negentropy, Noise

37

bring no new information a posteriori) (Atlan 1979, 33). While this appears to be a good compromise, it leaves us with the abyssal question: how do we turn the unexpected, and hence that which we could not anticipate or know a priori, into something we know a posteriori? The difference between a priori and a posteriori is a little more complicated than a mere before and after the fact, if we accept that these terms have been irreversibly conditioned by Kant’s critical philosophy. The idea that we can turn the a priori unknown into what is known a posteriori implies an epistemological conversion that isn’t entirely straightforward. Let us recapitulate the idea by which the paradox of Shannon’s information entropy could be brought back into the fold: information entropy is what is unknown a priori, but known a posteriori and, crucially, the more unknown it is a priori (i.e. the more unexpected it is), the more knowledge it procures (i.e. the more it informs us in the traditional sense of the word information) a posteriori. Now, if the a priori is a critical term that designates the conditions of possibility of cognition, i.e. the concepts without which there is no coherent unified experience, and the a posteriori designates that which is experienced on the basis of these concepts, then I am not sure what such a conversion of a priori uncertainty into a posteriori certainty could mean. It could mean making the absence or indistinctness of concepts (or our uncertainty about the a priori conditions of thought) the prerequisite for our certainty about experience. In other words, it would mean a rejection of Kant’s critical legacy, and a return to dogmatic intuitionism, where experience, if not irrationalism, supplants reason. This, it appears to me, is not a solution to the paradoxical relation of information and uncertainty, but an even greater paradox. If, on the other hand, the terms a priori and a posteriori here do not refer to critical philosophical concepts, but are simply used as erudite terms for ‘before’ and ‘after’ the fact, then we are still faced with a difficult epistemological conversion, namely of the virtual (the purely possible) into the actual. In this case, the more uncertain we are about the virtual possibilities inherent in a situation, in other words, the more unexpected the evolution of this situation is, the more knowledge this transformation will have imparted on us once it has occurred. This inverse relation of the virtual and the actual may indeed provide a fruitful conceptual framework for thinking about information. However, what is certain, is that such a way of thinking about information is revolutionary by default, if by revolution we mean the radical and unexpected transformation of a situation, (and not a sudden reversal understood as a return to something pre-existing). Such an approach, whereby the maximal value of information is

38

An Epistemology of Noise

the most revolutionary, implies an epistemological attitude to information that could not be further from the idea of negentropy, if the latter is understood as the negation of alternatives.

VI

Information as Process Rather Than Content

The conversion of uncertainty into certainty is implicit and therefore taken for granted, when ‘information entropy’ is defined as potential information, or as information we ‘lack’. This conversion, however, cannot be the same as a simple actualization of a potential, neither in the metaphysical, nor in the physical sense. The problem of uncertainty and ‘freedom of choice’ and its transformation into the opposite, into information as certainty and constraint takes as its starting assumption what still needs to be explained, namely how knowledge constitutes itself in the face of contingency and what role uncertainty plays in the constitution of knowledge. What is taken for granted is thus the fundamentally dynamical problem of information at the heart of epistemology: information can only be understood as a process rather than a given, a factum or a datum. Intuitively the idea of quantifying information as ‘bits’ suggests a simple encounter of form and content, as if one could quantify a certain amount of information as when one measures how full a cup is. This risks obscuring one of the most important aspects of Shannon’s entropic concept of information, which quantifies not the individual signal or message, but its relation of probability with the set of all possible messages given particular constraints – such as for instance a string of letters in relation to a finite number of possible letters in an alphabet: a message can be composed of a selection of discrete symbols, which could be letters, words, musical tones or any imaginable other signal, each however belonging to a set of symbols or a spectrum within which there is a certain ‘freedom of choice’ in terms of probability. Each choice furthermore stands not on its own, but always in relation to previous choices having already occurred in a discrete or continuous transmission of information. The previous state is thus factored into the probability with which the next symbol is chosen as the most likely, in what is called the Markoff process. It is this progressive relation of probability, which turns out, as Weaver says, ‘to

40

An Epistemology of Noise

be exactly that which is known in thermodynamics as entropy’ (Shannon and Weaver 1964, 12). ‘Information Entropy’ is thus a measure of the probabilities involved in progressing through stages of selection, indicating the probabilities with which, at each stage, certain symbols will be chosen next. It is thus never, the individual message that is carrier of information, but its relation with the set of all possible messages under equivalent constraints, a relation that changes as the transmission progresses: The concept of information applies not to the individual messages (as the concept of meaning would), but rather to the situation as a whole, the unit information indicating that in this situation one has an amount of freedom of choice, in selecting a message, which it is convenient to regard as a standard or unit amount. (Shannon and Weaver 1964, 9)

When Weaver says the ‘unit of information is called a “bit”’, what is thereby quantified is not a signal or message, but a changing relation between the actual and the possible, within a given frame of constraints. To predict the actual symbol or even message on the basis of the set of all possible symbols or even messages, is to anticipate the relation between a set of n independent symbols and the probability of choice p1, p2, …, pn. It is this relation of probability that finds mathematical expression in Shannon’s formula: H = – Σ pilog pi The calculus of probability therefore measures how rich in entropy information is, in terms of the progressive relation between our ‘freedom of choice’ and our capacity to predict. This means that the quantity of information is never measured as content or amount of the transmitted message alone, but as a function of the relation between this message and all possible messages with equivalent constraints. Information is thus understood as a dynamical relation of probability that measures a process rather than a content. It is this progressive sequence of probability between the actual and the possible that becomes the raw material of communication, quantified in terms of ‘freedom of choice’ prior to any possible interpretation and evaluation of the message as being significant or not within a semantic context. The habitual sense, in which information is considered like a vessel, a carrier of a certain amount of signification, is thus transformed by Shannon into a measure of the relation between the set of all possibilities, allowing a certain ‘freedom of choice’ in terms of probability, and the probability of prediction based on already actualized choices:

Concepts: Information Entropy, Negentropy, Noise

41

The significant aspect is that the actual message is one selected from a set of possible messages. If the number of messages in the set is finite then this number […] can be regarded as a measure of the information produced when one message is chosen from the set, all choices being equally likely. (Shannon and Weaver 1964, 31)

For one, this implies that information is never a given, because it characterizes a progressive modulation of certainty and uncertainty. Information presupposes as essential the structural and operational synergy between context and individual message, as between the uncertainty of ‘freedom of choice’, and the progressive modulation of certainty during the evolution of the individual message. Information, then, is the progressive unfolding of this relation between uncertainty and certainty.

VII

To Think about Information as a Process of Individuation

The philosophical significance of Shannon’s shift of emphasis, from individual signal to process, can be appreciated if we look at it through the lens of a medieval problem that was once known as the problem of individuation. The idea of individuation must ring unfamiliar to the contemporary ear, evoking at best a vague impression of scholastic disputes, perhaps echoing faintly around the names of Aquinas, Scotus or Ockham. Readers of Leibnitz or Wolff will recognize its gradually effacing traces in early modern philosophy. However, with Descartes and the modern empiricists, the problem of individuation appears to slide into oblivion, and what we are left with, qualifying the object of experience, is the idea of ‘All Things, that exist, being Particulars … ’ (Locke 1975, 409 in Barber and Gracia 1994, 2). The individual has since become the starting point of critical reflection and even the axiom of any possible rationality, be it as the cogito or as the touchstone of empirical investigation. By the same token it has become the stronghold of what we now call information according to common sense. The individual, as object of experience, is what informs us on ourselves and on the world (constituting either a bundle of faculties or an aggregate of attributes). Even though science has proceeded to dissolve individuality all the way down to quantum fields, and has dismantled any residual faith in its intuitive givenness through the neurocognitive sciences, the individual has nevertheless ossified into a tenacious idea of personhood. The concept of the individual has congealed into a political and moral sine qua non. We like to flatter it, when we qualify the individual as a subject, paying no mind to the pejorative connotation of subjection, which implies that we are subject to other powers, and that we thereby glorify what the Ancients considered a passive substrate to an active principle. A narcissistic investment in the idea of the individual thus makes it difficult to render this notion unfamiliar once more, or even to recognize the

44

An Epistemology of Noise

impact it has on our way of thinking about the world, and about what can inform us and how. To question the legitimacy of the individual’s primordial role in the contemporary Zeitgeist is perhaps even threatening to some, as it touches the centre piece of contemporary humanism: the individual and its identitarian reclamations. Yet, what is left of the idea of humanity appears to be a fragmented, hedonistic individualism. It has the merit of keeping the economy alive with its voracious need to accessorize individuality and to soothe its fear of dissolution with consumption. However, by the same token, the idolatry of the individual also heralds the potential demise of humanity. Biologist Eugene F. Stoermer and atmospheric chemist Jozef Crutzen even proposed to call our current geological era the Anthropocene, indicating that the presence of humans on earth now has the power to catalyse a process of such magnitude that the planetary survival of all forms of life is put in doubt. (Crutzen 2002). If the question of individuation appeared to belong to the Middle Ages, it may yet acquire a new urgency in light of the consequences of today’s unbridled individualism. It is the singular merit of Gilbert Simondon (1924–1989), to have put the ossified concept of individuality back into motion, by reviving the question of individuation. An atom, a biological cell or, indeed, a person is no longer considered a given, either in the form of a monadic entity or as an always already constituted whole. Instead, whichever entity or term we call ‘individual’ is seen as the end product of a process of individuation, whose most final stage of individualization is but the exhaustion of its potential for further individuation. Simondon thus proceeds to put the metaphysician back on the school bench. He takes us on a theoretical tour de force through modern physics, biology and psychosocial theory – without neglecting a critical appraisal of information theory and cybernetics, within a wider philosophical analysis of technical reality. He postulates that any theoretical problem or even existential crisis can be said to reflect a field of tension such that, in analogy with an electromagnetic field, whatever discovery, concept or idea is inserted as a new element is at once seized by this field and polarized. The process of individuation is thus compelled by the potential that characterizes this field of tension. This impetus is comparable, yet subtly different from Aristotelian entelechy, because form no longer predetermines the outcome, but enters in a recurrent causality of form and matter. In an ambitious synthesis of Plato’s concept of form and Aristotle’s hylomorphism, which he updates with the scientific concept of the

Concepts: Information Entropy, Negentropy, Noise

45

‘field’, Simondon revives the formal power of the idea, but embeds it in a revised hylomorphic schema: the structuring or organizing principle of form enters a reciprocal, mutually determining relation with a field of tension. More simply put: what Plato considers a superior reality, the idea or form, is no longer aloof of matter, but seized in a hylomorphic relation. In turn also Aristotle’s hylomorphism is reformed. Form and matter are no longer abstractly linked, as active and passive principles. The field of tension that receives structuration, comparable to Aristotle’s matter, is itself active: it polarizes and affects the idea or form, as much as it is structured by it. There is thus an embeddedness of the formal power of ideas and concepts in an empirical field. The idea becomes constitutive of this field and its process of transformation, but is also polarized by it. Simondon’s account of the process of individuation thus not only comprises the emergence of form, i.e. the gradual or sudden structuration of a domain, but implies also the concurrent transformation of the field itself, whose preindividual state gives rise to a milieu associated with the process of individuation. Individuation co-evolves with its own milieu. Both the final individual and its associated milieu are thus seen as by-products of a same process of differentiation. Rather than being the first object of consideration, a given, the individual is thus what comes last. Conversely, the milieu is not what precedes individuation – in other words, it is not simply that to which the emerging individual adapts – but is itself a correlate of individuation (Simondon 2005a). Crucially, what qualifies the genesis of form, for Simondon, is information. Information is not an aspect of the individual alone, such that one could compare the content of a piece of information to the complexity of an individual entity (or signal). Rather, information is whatever catalyses a process of differentiation, comparable to the effect that a crystalline germ has on an oversaturated solution, but information also qualifies whatever modulates this process, amplifying or regulating it. In other words, information is both what triggers a process of differentiation and what acts as an organizing principle. Whatever catalyses and modulates the process of differentiation, as a resolution of tensions or the solution of a problem, can thus be qualified as information. Information is thus, for Simondon, what links the epistemological and the empirical aspects of individuation. It is the idea or concept (or form) that catalyzes a process of differentiation, that modulates this process and that, hence, informs empirical reality, which in turn polarizes the idea. However, the coupling of reason and experience is also subject to a formal analogy between thought processes and empirical processes, whereby

46

An Epistemology of Noise no norm, no system detached from its content can be defined: the individuation of thought alone can, by accomplishing itself, accompany the individuation of beings other than thought; it is thus not an immediate nor a mediate knowledge that we can have of individuation, but only a knowledge that is an operation parallel to the known operation; we cannot, in the habitual sense of the term, know individuation; we can only individuate, individuate ourselves, and individuate within ourselves; […] an analogy between two operations, which is a certain mode of communication. (ILFI, 36. Emphasis in the original; my translation)

Information thus becomes a key concept for individuation. The new inflections that the concept of information receives by way of information theory and cybernetics are present throughout Simondon’s two main works: his major doctoral theses entitled L’individuation à la lumière des notions de forme et d’information (published as a whole only in 2005 by Éditions Jérôme Millon (Paris) and as yet untranslated),2 and his secondary thesis, written in accordance with academic requirements at the time, entitled Du mode d’existence des objets techniques, which was published as early as 1958 in by Aubier, Flammarion, and whose official English translation has been published by Univocal Press only in 2017 (Simondon, Malaspina and Rogove 2017). It is in this key role given to a processual understanding of information that we find an affinity with Shannon’s mathematical approach to information as an evolving relation of probability. The philosophical magnitude of Shannon’s processual understanding of information can be gleaned by comparing it to Simondon’s metaphysics and epistemology of individuation: there is no ‘piece’ of information whose quantity could be determined in and of itself, any more than there are individuals that can be abstracted from a process of individuation without rendering them sterile and lifeless. The point that can be made here, without imposing too violent a reading on Simondon’s work, is that information is fundamentally misunderstood, if it is taken to characterize an entity (a piece or information, a message in and of itself) rather than a process, from which the transformation of context cannot be dissociated. Shannon and Simondon both operate a Copernican revolution, replacing the individual (being or message), traditionally at the centre of the attention, with a careful attention to the co-evolution of both individuation and its context or milieu. However, Simondon’s own critique of the insufficiency of any quantitative concept of information must be born in mind. It would be wrong to

Concepts: Information Entropy, Negentropy, Noise

47

suggest that Simondon’s theory of individuation lends itself to an appraisal of Shannon’s concepts of ‘information entropy’, and especially to the philosophical revaluation of the concept of noise, such as it is argued for here. Yet without seeking to harmonize the difference between Simondon’s concept of quality of information and Shannon’s ‘information entropy’, we can still find resources in Simondon’s theory of individuation which enable us to shed new light on Shannon’s conception of information as a progressive relation of probability. For one, Simondon enables us to grasp the philosophical enormity of a concept of information that puts the individual (message) last, and brings to the foreground a progressive relation of probability between the individual state of the message and the set of all possible messages under a given set of constraints. To think with Simondon, without thinking exclusively in Simondonian terms, thus helps to critically redress our understanding of information, and notably to reject the common conflation of information with ‘data’, understood as something given. However, Shannon’s definition of both information and noise as entropy, distinguishing only desirable from spurious uncertainty, no doubt strains the analogy with Simondon’s theory of individuation. For Simondon the role of information remains essentially one of differentiation and structuration. It is ultimately based on an opposition to entropy, understood, as Wiener and Brillouin do, as the final state of equilibrium or the death reached by a closed system. Although, technically speaking, Wiener’s concept of negentropy is based on the same mathematical model as Shannon’s, imposing an understanding of information as process rather than entity, in cybernetics it is always in the service of an already constituted and correctly functioning entity, a machine or an organism, that information is required to counteract entropy, notably through feedback processes. This is not to say that Simondon settles for Wiener’s definition of information as negation of entropy, any more than Shannon’s. Both fall short of providing a qualitative definition of information, such that it could encompass individuation. What Simondon finds lacking in the quantitative definition of information is the notion of potential, the tension that polarizes and thus gives a sense, if not a signification to information. Simondon rejects the idea that information, measured in bits, could in any way encompass what we must understand by the quality of information, which is what characterizes not only the capacity to inform or regulate reality, but, crucially, its capacity ‘to illuminate new domains’ (Simondon 2005, 549).

48

An Epistemology of Noise

Simondon does, however, incorporate a great number of aspects into his theory of individuation, which would appear to lend themselves to a revaluation of various aspects of noise, such as we are aiming at here, by relativizing its opposition to information. He has a subtle understanding of demodulations of structure, even of crisis, which he sees as a necessary reload of potential for novel structuration and reorganization. He mentions, for instance, processes of dis-adaptation in developmental psychology, as necessarily preceding reorganization at a higher level (Simondon 2005, 545). Several key terms in his philosophy express a deep appreciation for indeterminacy or even dissolution of structure or form. In his On the Mode of Existence of Technical Objects he values margins of indeterminacy in the functioning of so-called open machines, as necessary for their capacity to respond to input from the environment. In his lecture Communication et Information Simondon even delves into the motivating or bonding aspect of social noise (Simondon, Simondon and Chateau 2010; Simondon 2010). Furthermore, Simondon’s notion of a pre-individual state lends itself to an analogy with noise if we think of the pre-individual state of being as characterized by the equal probability of all possible states, and thus rich in the greatest possible ‘freedom of choice’. As such a primordial noise could now be thought, in light of Simondon’s concept of the pre-individual, as a positive ground for differentiation, in other words, as ground for the emergence of form and its transformation. This would mean that a primordial maximal uncertainty not only grounds but co-evolves with the process of individuation. Noise in the ordinary sense, understood as interference, would thus be the remaining uncertainty, tethered to the process of individuation by its associated milieu. What Simondon calls the ‘pre-individual’ state of being is thus not entirely alien to our idea of a subtler difference between ‘information entropy’ and noise entropy, both understood as a margin of ‘freedom of choice’ or uncertainty. However, even this limited analogy requires an important proviso. The analogy works only if both concepts, the ‘pre-individual’ and ‘maximum entropy’, are treated as regulative ideas, acknowledging that no situation is absolutely closed, in the sense that the absolute absolves from any relation to an outside. There are approximations of absolute closure of a system (for instance, in nanotechnology or quantum technology) that come at the cost of a technological apparatus and of mathematical constructions of great sophistication. However, no process of individuation and no process of information ever takes place on absolutely blank slate: nowhere in the empirical world is a closed system

Concepts: Information Entropy, Negentropy, Noise

49

realized in absolute terms, i.e. without having to take into account the role of the observer or the impermanence of this closure. Every system known to mankind is always already situated in a reality that is densely packed with pre-existing processes of individuation. To acknowledge this is to admit that empirical reality is always already a noisy mess of competing processes of individuation, involving dissolution of form and a wrangle of certainty and uncertainty. However, even if we acknowledge that no process of individuation is absolute and that multiple processes of individuation may compete noisily, it must also be acknowledged that Simondon explicitly stops short of an affirmation of noise, (or of phenomena involving de-differenciation, indeterminacy or even chance), as constitutive of information processes. Simondon compares for instance the tension of form, which he sees as a precondition for the quality of information, to social phenomena such as pre-revolutionary tensions. It is conceivable that in such situations, he says, a ‘thought coming from elsewhere’ (le fait qu’une idée tombe d’ailleurs) triggers a sudden structuration (ILFI, p. 550). Just as a chance correlation of molecules may set off the process of crystallization, so a ‘chance encounter’ may set of a revolutionary process. However, and this is crucial, ‘it is very difficult to admit that chance has a value of creation of good form’ (ibid.). This is because the quality of information is more than a fortuitous aggregation: its structuring effect must be more than just fleeting, it must sustain a structuring power, and sustain what in French is called sens and which we can only partially translate as both signification or direction. The quality of information mediates information’s power of structuration and the tension that characterizes a domain capable of receiving information. This mediation is quite literally the sense that information makes, the signification or organizing power it catalyses. A purely fortuitous process, in turn, would be subject to an equally fortuitous dissolution. In other words, what we could call a noise phenomenon, and which Simondon here characterizes as coincidence or chance (hazard), may act as a trigger for spontaneous structuration but – and this is the crux – the process of information itself remains a process of structuration for Simondon and ultimately a negation of entropy. While this appears to cut short any analogy between Shannon’s conceptual audacity and Simondon’s return to the problem of individuation, it must be born in mind that Simondon is already several steps ahead of the problem we are addressing here. His emphasis on the quality of information is already a problem of signification. For us, on the other hand, what is at stake is a rather limited

50

An Epistemology of Noise

problem that does not yet encompass the question of signification, but only the presence of uncertainty among its conditions of possibility. What Shannon enables us to think is not an absolute value of noise as novelty – which one could provocatively call ‘pure information’, if one were to attribute a maximal information value to maximal entropy. It is, rather, the fact that we can now think of information as a subtler difference than that between organization and chaos or sense and non-sense, a difference that takes place within the conceptual space of entropy, within the space of uncertainty: if information can be thought as qualified uncertainty, then noise too can be released from the theoretical exile of negation into which it was thrown. Noise can become possible information. In other words, unqualified uncertainty can be understood as one of the preconditions of qualified uncertainty and, hence, of information.

VIII

Redundancy and Necessity

Information entropy, understood as ‘freedom of choice’ or as equi-probability of events, is thus never absolute, especially when we consider that this term qualifies the quantity of information not merely as ‘entropy’, but as ‘entropy of the message’. ‘Entropy of the message’ designates a variability that is always already limited by the condition that the source of information continues to employ the same set of symbols and that this set is finite. It is a ‘relative entropy’ that implies a certain amount of redundancy, in other words of repetition (frequency) giving it a head-start on the margin of predictability. As Weaver puts it: If the relative entropy of a certain source is, say 0.8, this roughly means that this source is, in its choice of symbols to form a message, about 80 per cent as free as it could possibly be with these same symbols. (Shannon and Weaver 1964, 13)

If the relative entropy of a source (of continuous signal transmission, like for instance a radio transmission) is given a value of 0.8, then the remaining 0.2 corresponds to constraints that are placed upon the message, in other words, to what makes this message minimally predictable and hence, to what will be redundant within it: One minus the relative entropy is called the redundancy. (Shannon and Weaver 1964, 13)

Constraints on the entropy of the message can be, for instance, statistical rules governing the use of symbols, or the set of letters in an alphabet or syntactical rules. The predictable part of the message is what can be reconstructed and is therefore considered to be inessential to the novelty of the message, and in this sense ‘redundant’. It is what separates the ‘entropy of the message’ from complete randomness or noise. Interestingly Weaver goes so far as to call the redundant part of the message unnecessary, which appears to suggest that the message can still be a message without it:

52

An Epistemology of Noise [T]his fraction of the message [that] is in fact redundant in something close to the ordinary sense; that is to say, this fraction of the message is unnecessary (and hence repetitive or redundant) in the sense that if it were missing the message would still be essentially complete, or at least could be completed. (Shannon and Weaver 1964, 13)

Weaver is of course right in the sense, for instance, that most vowels can be left out of a typed message, without making it impossible to reconstruct the message. The journalistic convention of replacing letters in offensive words with the symbol * (i.e. f***) is indicative of the fact that what is redundant need not be reiterated. Redundancy is nothing other than the predictable part of a message. Weaver indeed goes on to note that language has a very high level of basic redundancy: It is most interesting to note that the redundancy of English is just about 50 per cent, so that about half of the letters or words we choose in writing or speaking are under our free choice, and about half (although we are not ordinarily aware of it) are really controlled by the statistical structure of the language. (Shannon and Weaver 1964, 13)

The conceptual presence of necessity here slips into Weaver’s expression nonchalantly, in the form of a negation: redundancy is the part of the message that is not necessary. However, to define redundancy as unnecessary, meaning inessential or accidental, could lead to a misunderstanding of far reaching theoretical consequences. While Weaver appears to say something obvious, the concept of necessity is one that cannot leave philosophical analysis of information and noise indifferent. Weaver’s way of putting it, namely that redundancy is the part of the message that is not necessary is potentially misleading, not least because his introduction seeks to lay the conceptual foundations for a new understanding of the broader theoretical relevance of MTC. It is a far from negligible slippage of logic to describe redundancy as unnecessary, because it shows and even performs that the necessary, that which cannot not be and which constrains ‘freedom of choice’, is what, as self-evident, can be left unsaid and hence un-thought. What does it mean for the redundant part of the message to be unnecessary? The necessary is, in simple terms, whatever is absolutely indispensable and hence of utmost importance (as for instance in the expression of the ‘bare necessities’ for survival). Necessity can also be understood as a constraint, such that its stringency or unavoidability is recognized in law even where a necessity contravenes the law, as in the expression ‘state of necessity’: the necessity to

Concepts: Information Entropy, Negentropy, Noise

53

safeguard the interests of a person may, before the law, result in the impunity of an incriminating act (dir. Jaen-Marie Pierrel et al. n.d.). Necessity thus designates what is required by a situation (material, practical, technical or vital necessities), but more fundamentally, in the philosophical tradition, that which cannot not be, or which cannot be otherwise. In other words, necessity is the mother of all philosophical concepts: a categorial, logical or metaphysical necessity is what reason posits as valid in any circumstance and whose contradiction is an impossibility. For reason, necessity is nothing less than the axiomatic starting point of rational thought. Everything else has been, since Greek Antiquity, attributed to the order of opinion, of mere phenomena or appearance. What is not necessary is contingent: either absolutely contingent or contingent upon a necessity that we may or may not know. The introduction of probability into reasoning is, therefore, a significant event in the history of thought. That something can be said to be 0.2 per cent certain and 0.8 per cent uncertain introduces the possibility of nuance and process: genesis and corruption are no longer excluded from the realm of reason. It is, in epistemological terms, the metaphorical equivalent of introducing colour into a black and white vision of truth. All the more reason to take note that necessity is what Pascal, one of the founding fathers of the calculus of probability, called a ‘state of constraint or restraint that annuls freedom of choice’ (dir. Jaen-Marie Pierrel et al. n.d.; Pascal and Guern 1987). Now, if redundancy is the part of the message that imposes a constraint, that reduces ‘freedom of choice’ in terms of the message’s probability, then it is hard to see how it could be unnecessary to the message. Redundancy is what in the message remains constant, what is stable and not subject to degradation through noise. Redundancy, in other words, is the very state qualified by the Latin root of necessity: non cedens, that which does not give in and which, in its regular form necée, is close to the idea of chastity: untouched by genesis and degradation. Is redundancy in the message not precisely that which remains untouched by ‘freedom of choice’, by entropic degradation of the message, by contingency, in short, by noise? Redundancy, without which ‘information entropy’ would be indistinguishable from noise, is thus not only necessary to the message, it is what, as self-evident, becomes the invisible or unthinkable a priori of information. The consequences of underestimating redundancy as unnecessary are far from trivial, if we acknowledge that every form of organization is based on constraints that introduce redundancy. Every system is informed by constraints that discriminate ‘freedom of choice’ according to given (hence redundant) criteria of pertinence.

54

An Epistemology of Noise

To euphemize redundancy as unnecessary thus risks losing from sight what founds every system: necessity, in other words that without which a system cannot be what it is. By becoming redundant in the sense of self-evident, necessity is what can remain unspoken and hence un-thought. Necessity, that which makes a system what it is, is thus also what most easily subtracts itself from critical analysis, as an a priori of this system, of this organization, of this way of functioning, that is always already taken for granted. In linguistic terms, the redundant part of language is what can be taken for granted, what can be ignored. But we must not overlook the fact that it can afford to appear unnecessary, it can afford to be ignored, only as long as it operates as the condition sine qua non of communication. Redundancy, contrary to Weaver’s claim, is necessary – and only therefore is it not necessary to reiterate – just as good manners do not need spelling out where they can be taken for granted, and just as the work of illegal immigrants can remain invisible to the public, while operating as a condition of possibility for an economy hungry for cheap labour, unregulated by workers’ rights. The question could be transposed onto the idea of work redundancy. It is especially relevant in view of the digital revolution, which exposes entire segments of professional activity to potential redundancy. The question, as with redundancy in language, is this: which a priori are being taken for granted? What goes without saying? A critical approach would be to seek the blind-spot in redundancy, in order to debunk mere preconceptions masked as necessity. When the necessary conditions of the information process, of communication more generally and of organization (including social or political) are what is redundant, then they are nothing less than the a priori of our way of thinking and acting. Yet if we fail to address the informational value of redundancy by minimizing it as ‘not necessary’, then it becomes increasingly difficult to ask: when is an a priori a necessity, a sine qua non of being thus, of thinking and of communicating thus? And when is it mere prejudice? The a priori restriction on the ‘freedom of choice’ in the message, is nothing less than the condition of possibility of communication, also because, without it, nothing would offset the uncertainty that is a correlate of the novelty of information. In other words, without redundancy the pure novelty [entropy] of information would be absolutely incomprehensible and equivalent with noise. It is only on the basis of redundancy that novelty demarcates itself from what is already certain. Redundancy is, furthermore, in this sense, also an essential concept for our understanding of physical entropy. For the measure of entropy in a physical system is a direct correlate of the knowledge we already

Concepts: Information Entropy, Negentropy, Noise

55

have of it and the knowledge we lack. The knowledge we have of a system, for instance of chemical rules of interaction between elements, acts as a constraint in epistemological terms: it reduces the entropy of the system. Without this knowledge the behaviour of the system is absolutely random to us. Conversely, what we call complexity is a correlate of low redundancy, in other words, of a low level of pre-knowledge about the system. By complexity we must understand the degree of indeterminacy of a system, rather than its level of structural complication. Biophysicist and philosopher Henri Atlan defines complexity as the measure of the observer’s ignorance as to the precise determination of a system. Greater complexity of information denotes greater uncertainty. What he calls the ‘maximum maximorum’ of ignorance is the state of greatest complexity. It corresponds to the most basic measure of information in Shannon’s sense (H), which informs us only about pure multiplicity, nothing but the number of elements in a system (H= log N). Atlan calls this the first, ‘trivial and maximal’ measure of complexity. It corresponds to the observer’s maximal ignorance of other factors, such as variety, frequency and other constraints. The second measure of information takes into account statistical distribution and frequency (H= Σ p log p). Its quantitative value is therefore smaller than the first, as its complexity is reduced. The third measure of information, finally, introduces redundancy through the addition of constraints [H = Hmax (1 – R)]. This corresponds to the least complex level of information, as determining factors carve away at the complexity of the pure multiplicity that characterized the first and maximal level of complexity. (Atlan, 1979, p. 80).

IX

Logic and Freedom of Choice

In logic, deductive redundancy is achieved when each proposition is tightly correlated with the propositions deduced from it. Step by step, each proposition becomes resonant with the others and no element can be modified without compromising the whole (Blanché 2009, 10). The philosopher of mathematics Robert Blanché thus describes the process of deductive thinking as propagating a structure of constraint where, like the emerging lattice structure of a crystal: [S]tep by step, a tight network is constituted where, directly or indirectly, all propositions communicate. (Blanché 2009, 9–10)

This is why one speaks of the deductive resonance of a mathematical theorem if, like a crystalline structure, it achieves a rock-solid correlation between each and all terms, where nothing is left to chance and no ambiguity can arise. The logical coherence of the whole can then be called isomorphic, like the lattice structure of a crystal. This resonance is a form of redundancy in the logical chain. It eliminates ambiguity, just as redundancy in the transmission of a message serves to reduce noise. It was long hoped that logic would provide a language without ambiguity, in other words without noise. If the elimination of ambiguity could ensure the truth of all statements in a logical chain, by eliminating ambiguity and hence error, then deductive redundancy was deemed necessary in order to stabilize philosophical discourse. If, however, all statements are not bound by necessity then they retain a trace of ambiguity. This is the case for unproven postulates, in other words, for all statements that are not yet proven axioms. A single postulate in a system of axioms becomes the gateway to a potentially radically diverging axiomatic system. A paradigm shift thus lies dormant in each postulate, in each trace of ambiguity and openness. This ambiguity and openness is what we will call noise, because it persists as a margin of uncertainty and freedom of choice, until it too

58

An Epistemology of Noise

is axiomatized. The fifth postulate in Euclid’s Elements long remained open to such uncertainty, because of the impossibility of proving it. According to this postulate, also called the ‘parallel postulate’: If a line segment intersects two straight lines forming two interior angles on the same side that sum to less than two right angles, then the two lines, if extended indefinitely, meet on that side on which the angles sum to less than two right angles.

The impossibility of proving this postulate despaired mathematicians for centuries. The uncertainty it represented revealed itself as a freedom of choice only when, in 1829 Nikolai Ivanovich Lobachevsky could prove that the fifth postulate does not hold true in a geometry of infinite dimensions. The search for the impossible proof of Euclid’s fifth postulate thus paved the way for the emergence of so-called hyperbolic or ‘non-Euclidian’ geometry. The classical three dimensions of Euclid’s geometry opened up to the infinite dimensions of Lobachevsky’s geometry, and as a consequence, also the space-time of classical physics could eventually open up to the possibility of special space-time relativity (Blanché 2009, 47). Eric Temple Bell called Lobachevsky the ‘Copernicus of all thought’, and saw in Lobachevsky’s work an incentive for mathematicians and scientists to ‘challenge other “axioms” or accepted “truths”’ (Bell 1986). For us, the ambiguity of Euclid’s fifth postulate can be thought as a borderline case, where uncertainty becomes freedom of choice, in other words, where noise becomes information. If Euclid’s fifth postulate represented a source of uncertainty for centuries, Lobachevsky transformed this uncertainty into freedom of choice, opening up three dimensions to infinite dimensions, and opening up the idea of mathematical truth to the emergence of new paradigms. In the early twentieth century, Georg Cantor set out to harmonize the domain of mathematics and to rescue the unity of mathematics in light of the plurality of geometries. In the attempt to axiomatize set theory, Cantor encountered foundational paradoxes that shook mathematics to its core no less than Lobachevsky’s achievements in geometry, and thereby irreversibly unmoored its theoretical capacities from its last anchorage in intuition. Mathematics and geometry were henceforth wide open to the prospect of infinite or transfinite sets of infinite numbers, chaperoned only by the paradox that the set of all stets cannot ground itself. In response to the crisis in the foundation of mathematics that was provoked by Cantor, logic, it was hoped, would reign mathematics in and prevent its speculative excesses.

Concepts: Information Entropy, Negentropy, Noise

59

Robert Blanché gives an account of the axiomatization of mathematics that is valuable in this regard, because it helps understand both the necessity and the absurdity of deductive redundancy, when taken to its extreme and when erected as the sole pillar of truth. Logic, it was hoped, would rescue mathematics from the paradoxes of axiomatic set theory, which appeared to have turned mathematics into a science where one never knows what it is that one is talking about, nor whether it’s true. […] By proposing to ground […] the entire edifice of mathematics on logic, Frege and Russell’s ‘logicism’ aimed further than returning to its principles: it intended bringing it to its term, reaching the rock, the ultimate foundation. (Blanché 2009, 70)

It is history, by now, that rather than eliminating the antinomies that had sprung up within axiomatic set theory, disagreement about the validity of logical principles were, in turn, to shake also the foundations of logic to their core, ultimately demolishing ‘the idea of an absolute, unique and universal logical legislation’. Blanché reconstructs how the axiomatization of logic finally led to the ‘disintegration of logic from within’, issuing forth into a pluralization of logics. Even if the intra-logical and axiomatic problems raised by a plurality of logics and of axiomatic systems could be set aside, what remains problematic is thus the ‘fit’ of logical redundancy and reason’s intrinsic complexity. Its formalized terms, albeit submitting to the utmost criteria of necessity, no longer represented anything but the mediation between ‘simple tautologies’. Although perfectly harmonized in a deductive redundancy, without what we may call the noise of ambiguity, they ‘say strictly nothing about the real, but […] for this reason, remain valid whatever content one applies’ (Blanché 2009, 71). In a different context the philosopher of biology Marjorie Greene struggled with the limitations of Logical Positivism in conceiving of the problems specific to biology. Its incapacity to deal with the imprecision of the empirical world eventually turned Greene away from Logical Positivism, in which she saw a sterility when it came to the life sciences: In the Anglophone tradition (which I derive partly from the Germano-Austrian tradition) that which one called the received view dominated until recently. I participated personally in Carnap’s seminar in Chicago during the year 1937– 1938. Having previously studied zoology, I was rapidly disappointed. It seemed impossible to treat the praxis of zoology with a purely extensional logic. I tried to explain this difficulty to Hempel, who was Carnap’s assistant in this seminar and

60

An Epistemology of Noise he replied: ‘We only say what we can explain with precision.’ […] twenty-five years later […] logical positivism was taught under its new less aggressive name of logical empiricism. It dealt with laws, theories, the deductive relation between theories and laws, the problem of confirmation etc. […] Today however, this old orthodoxy is, if not entirely buried, then in a – how shall I say – catatonic, vegetative state. (Grene 2007, 24–25)

Greene’s disappointment in articulating the deductive power of Logical Positivism with the open field of biological complexity is directly relevant to our interest in the question of ‘epistemological noise’, to the unavoidable ambiguity and distortions of concepts and theories, when we seek to crossfertilize fields of knowledge and praxis. It shows that ‘epistemological noise’ is not merely a deplorable side effect of interdisciplinary communication, or of the transdisciplinary formulation of scientific problems, but a positive requirement, without which specialized discourse becomes not only sterile, but worse, redundant.

X

Noise as Spurious Uncertainty

The question that has emerged from the previous sections is: how do we draw the line between constraint and ‘freedom of choice’? We cannot avoid complexity and ambiguity entirely without risking sterility of information, but we still need to impose a boundary between a level of complexity relevant for the formation of knowledge and infinite complexity, which consigns us to the power of oracles, not reason. Here lies the difficulty in distinguishing between the ‘entropic’ understanding of information and the entropy of noise. Yet this is precisely the question that is suspended when the engineer transmits a readymade message, regardless of whether it is the rambling telephone conversation of someone’s mother-in-law, an encrypted message, or Schönberg variations. The a posteriori evaluation of what is received, as either spurious or significant, literally doesn’t come into the equation. Any value judgement that discerns a message with the mark of distinction, namely that it is informative, is either pre-given, a priori, in the decision to transmit a message as information, or it occurs a posteriori, as the result of a process of evaluation and interpretation. The relatively restrictive scenario, where an already selected signal is transmitted as information has this particularity: the decision upon which we base the distinction between information and noise has already been made when the message is transmitted. The engineer treats the message as ready-made, but in order for information to be transmitted, in order for the entire technological apparatus of signal transmission to exist, a decision must have preceded transmission: namely the decision that information must be transmitted, and the decision that what is transmitted is information. Even if this is not a decision that concerns the communication engineer, it is a decision that needs to be made, repeatedly, every time a message is selected for transmission as information. However, the pre-established distinction between information and noise prevails only in the narrow technical context of communication, presupposing

62

An Epistemology of Noise

the decision has been made and information has been selected for transmission. Outside the fully specified function of transmission, the question ‘what counts as information?’ is wide open. When faced with new experiences, our ready made distinctions between information and noise, are more often than not contested. The redundancy of pre-established constraints on information’s ‘freedom of choice’ is only a temporary fix before the ultimately unpredictable course of events in our human, geological and cosmological history. The distinction between information and noise is stabilized by way of scientific methods that were once innovative and that become conventions and even traditions. Enshrined in the institutions of knowledge, almost iron-cast into the transmission of knowledge, methods become habits of thought, so deeply engrained that they inhabit the blind spot of redundancy, rendering indistinguishable the conditions of possibility of knowledge from the discovery of what we find informative. What we are left with is thus a tenuous, historically contingent line separating the entropy intentionally selected as information and the entropy spontaneously adding itself to information as noise. How to draw this line becomes an epistemological problem once the concepts of information and noise are taken outside the context of electronic signal transmission, as when they are translated into molecular biology or systems analysis more generally – and a fortiori when they enter general discourse. This is perhaps why Shannon’s audacious levelling of information and uncertainty is attenuated by Weaver, as a way of curtailing the radical consequences of aligning information with uncertainty, revealing a reluctance perhaps to let go of the clear-cut distinction between information and noise. Weaver indeed acknowledges the bond between information and uncertainty, but only to rescue what he can of the ordinary understanding of information, by expelling noise as ‘spurious’ uncertainty: It is generally true that when there is noise, the received signal exhibits greater information – or better, the received signal is selected out of a more varied set than is the transmitted signal. […] Uncertainty which arises by virtue of freedom of choice on the part of the sender is desirable uncertainty. Uncertainty which arises because of errors or because of the influence of noise is […] spurious and undesirable.

Weaver’s assumption therefore is that noise tells us nothing new, because there is no telling at its origin, no intention: what noise tells us is ‘spurious’ because accidental. Yet, as we have seen above, noise cannot be accidental in the sense that it is less necessary than ‘information entropy’, since both are and remain

Concepts: Information Entropy, Negentropy, Noise

63

a measure of uncertainty whose positive quantity is negatively correlated with redundancy. Noise can be spurious in Weaver’s sense only if it is accidental visà-vis intention Intention, however, pertains only to semantic communication, where some form of consciousness can be presumed. But information theoretical concepts of information and noise have proven their relevance in fields that far outstrip problems of intentional communication. Consider for a moment just how broad the field of conceptual relevance is for the concepts of information and noise: if we accept that the very concept of system relies on communication between its parts and the whole, then all systems can be thought of in terms of processes of information. All empirical systems imply communication, which can be understood as ‘causal relation’, or with Ross Ashby as relations of ‘non-independence’ between its parts and the whole (Ashby 1962) and even more cautiously as merely statistical correlation. From the quantum level to the cosmic level, all correlations of events can thus be formulated more or less loosely in terms of information processes. Yet if we accept this premise, which vastly broadens the concepts of information and noise beyond the scope of verbal or technical communication, then intention can be considered to be relevant to only a small number of in principle infinitely many possible systems that operate through processes of information. Conscious intentional communication, which we perhaps too hastily attribute to human beings as a mark of distinction, becomes a limited domain, the only domain where the distinction between desirable and ‘spurious’ uncertainty pertains. We may have to concede that the centrality of human communication, understood as a semantic and culturally saturated information system is, at least in principle, neither the first system in which information processes occur, nor necessarily the most efficient. Systematic communication, understood broadly as non-independence between parts and the whole, can be assumed to have preceded our cognitive faculties to communicate, via physical, chemical, evolutionary processes of information. Even if we concede that we can have knowledge of any such system only in the cultural domain and via semantic mediation, that the very concepts of system, information and noise are nothing but a mental construct, we cannot account for mental constructs without assuming that there is genesis and transformation of form that has preceded, as a condition of possibility, the forms of cognition and the processes of information we call our own. How apt a handle the concepts of system, information and noise provide, as

64

An Epistemology of Noise

mental constructs, to think any such processes in themselves, is another matter altogether. When Weaver narrows the conceptual reach of Shannon’s ‘entropic ideas’, by consigning Shannon’s concept of information to intentional communication, he thus generalizes the pre-given distinction between information and noise. And yet, intentionality is not a concept comprehensive enough to encompass the difference between information and noise on the large spectrum of theoretical and experimental domains in which Shannon’s concepts are relevant. The distinction between intentionally and accidentally transmitted ‘entropy’, between information and noise, is precisely what is not pre-given in most of the contexts into which the information-theoretical concepts have been translated. Most areas of scientific research, on the contrary, explicitly set out to redefine the boundary between relevant and irrelevant phenomena, between uncertainty productive of new scientific insights, and uncertainty that is a mere distraction or loss of focus. The very core of scientific endeavour is to establish what, in the vast contingency of interrelated empirical events, is relevant as information and what, on the contrary, stands in the way of the identification of phenomena relevant for enquiry. The difference between information and noise is thus rarely ready-made, and always provisional in the context of research. The difficult elaboration as to what, in a research question, must be isolated as a relevant phenomenon, and what, on the contrary must be discarded as irrelevant is ultimately a methodological decision. It is always threatened with the potential error that what is excluded as noise may lengthen the path of erring towards understanding. Even acquired certainties are subject to this rule, whereby history unravels scientifically held beliefs as ‘epistemological obstacles’, as Bachelard would say, often prompted by phenomena of perturbation or errors that reveal a flaw in those certainties, and thereby turn information (in the sense of acquired certainty) into noise and noise into information (in the sense of a productive uncertainty). Weaver’s idea of noise as by definition ‘spurious’ uncertainty thus risks unnecessarily limiting the relevance of Shannon’s open definition of information as ‘information entropy’, by artificially hardening a boundary between the measure of entropy that is apportioned to ‘information entropy’ and the measure of entropy that is considered to be an excess, i.e. noise.

XI

Negentropy

There is a peculiar tension between on the one hand Shannon’s very minimal definition of information as pure probability, shaved even of Boltzmann’s reference to physical processes (k), and on the other hand of the cluster of concepts relating to purpose and organization that characterize negentropy and even Weaver’s reliance on intention. Peculiar because it opposes different epistemic frameworks for the conceptualization of information and noise, that don’t reduce well to mere opposition, such as the idea of the simple negation of entropy suggests. The notion of negentropy in fact implies a multilayered concept of information. No longer a pure concept of probability, theories of organization and order are tacitly aligned with the concept of negentropy. The question this concatenation raises is the following: how can the negation of entropy increase the complexity of information, which we commonly associate with organized forms of life? The neologism negentropy first established itself first in light of Erwin Schrödinger’s conception of life as that which distinguishes itself from the proneness of physical processes to entropic dissipation of energy and death (Schrödinger 1945). Leon Brillouin’s seminal article ‘Life, Thermodynamics and Cybernetics’ (Brillouin 1949) has further cemented this and Norbert Wiener’s use of the term information as a measure of prediction of future behaviour in the domain of control and communications systems. Information here answers the question, reiterated by Brillouin: [W]hen we possess a certain number of data about the behaviour of a system in the past, how much can we predict of the behaviour of that system in the future? (Brillouin 1949, 554)

This conception is now cemented in current usage with Brillouin’s neologism negentropy. The degree of organization of a system thus comes to represent its quantity of information, which contrary to Shannon, stands in inverse relation to entropy. This leads Brillouin to make observations about the relation of

66

An Epistemology of Noise

information to time and memory, which we cannot pursue here, but which would be of great interest to every Bergsonian or Deleuzian. As we have seen earlier, Warren Weaver also signals this link between organization and the negation of entropy in his introduction to MTC, but on the contrary explicitly states that, in light of Shannon’s entropic ideas, the state of organization is one where ‘information […] is low’: Thus for a communication source one can say, just as he would also say it of a thermodynamic ensemble, ‘This situation is highly organized, it is not characterized by a large degree of randomness or of choice – that is to say, the information (or the entropy) is low’. (Shannon and Weaver 1964, 13)

Weaver and Brillouin’s shared assessment of the role of organization as constraint of ‘freedom of choice’, in other words, as negation of entropy, must not lead us to overlook that two very distinct or rather opposed definitions of information are nevertheless at stake depending on whether we define information as negentropy or, with Shannon, as ‘information entropy’. In Science and Information Theory Brillouin makes this difference explicit: Entropy is usually described as measuring the amount of disorder in a physical system. A more precise statement is that entropy measures the lack of information about the actual structure of the system. This lack of information introduces the possibility of a great variety of microscopically distinct structures, which we are, in practice, unable to distinguish from one another. Since any one of these different microstructures can actually be realized at any given time, the lack of information corresponds to actual disorder in the hidden degrees of freedom. […] The origin of our modern ideas about entropy and information can be found in an old paper by Szilard […]. The connection between entropy and information was rediscovered by Shannon, but he defined entropy with a sign just opposite to that of the standard thermodynamic definition. Hence what Shannon calls entropy of information actually represents negentropy. […] To obtain agreement with our conventions, reverse the sign and read negentropy. (Brillouin 1949. Emphasis added)

What does Brillouin mean, when he says that Shannon defines entropy with a sign opposite to the standard thermodynamic definition? When Brillouin says ‘he [Shannon] defined entropy with a sign just opposite to that of the standard thermodynamic definition’ then presumably he does not mean Boltzmann’s definition (S = −K Σ pi log pi.), which as we have seen is preceded by the same minus sign as Shannon’s definition (H = −Σ pi log pi). By standard

Concepts: Information Entropy, Negentropy, Noise

67

thermodynamic definition Brillouin must mean Wiener’s definition of information as negation of entropy. To follow Brillouin we must ‘reverse the sign and read negentropy’, in other words, the definition of information as negentropy is based essentially on the negation of a negation. Not surprisingly this convoluted relation between information and entropy lends itself to a certain amount of confusion when the terms information and noise enter the mainstream. The concept of negentropy is in fact not rarely justified with the minus sign that precedes the symbol for entropy [Σ] in the mathematical expression for ‘information entropy’ [H = − Σ pi log pi]. This apparent subtraction of the value of entropy from the quantity of information [H] appears to chime with Wiener’s statement that information is simply the negative of entropy or disorder: entropy consequently can be said to decrease as information [H] increases. In other words, the idea is, the smaller the quantity of entropy, the more reliably we can predict. Raymond Ruyer, in La cybernétique et l’origine de l’information, explains this point of view: Entropy goes in the direction of the most probable states; information, with the opposite sign, is thus an ‘anti-probability’ or, to use Edington’s expression, an ‘anti-contingency’ [anti-hasard] […[ the formula that expresses it is exactly the formula of entropy, the logarithm of equal probability, but with the contrary sign. Information is negative entropy. (Ruyer 1954, 114–15)

What results is an apparent paradox that has riddled the interpretation of Shannon’s use of the term information ever since. How are we to interpret the minus sign that precedes Σ, in both Boltzmann’s and Shannon’s formalization of ‘information entropy’? Warren Weaver demystifies this difficulty for nonmathematicians in his introductory essay to MTC, by spelling out the role of this minus sign: Do not worry about the minus sign. Any probability is a number less than or equal to one, and the logarithms of numbers less than one are themselves negative. Thus the minus sign is necessary in order that H be in fact positive. (Shannon and Weaver 1964, 15, n. 5)

The minus sign in Shannon’s mathematical expression of ‘information entropy’ thus expresses in positive quantitative terms the entropy, hence ‘freedom of choice’ of a message. Clearing up any potential confusion, we can now say that the minus sign in the standard mathematical definition of entropy, and in Shannon’s definition of ‘information entropy’, expresses the positive value of entropy or ‘information entropy’. It does not express the negation of entropy or negentropy.

68

An Epistemology of Noise

It is the reversal of this minus sign, which reduces ‘freedom of choice’ in terms of probability, and hence ‘entropy’. The minus sign whereby any probability below 1 is transformed into a positive number, meaning the entropy of the message, is what must be reversed, in order to align the positive quantity of information not with entropy but with the negation of entropy. Hidden in this reversal of a reversal is thus a major redefinition of the concept of information, but also a subtle shifting of ground from information as pure probability to information as index of organization (still understood straightforwardly as negation of entropy) and from organization to order (as the negation of disorder, but also as instruction or function). Where Shannon freed the concept of information from the reference to physical potential, as we have seen earlier with reference to the algorithm ‘k’, Brillouin reintroduces the idea of physical potential as a precondition for the organization of vital processes. Information is therefore put under the helm of Schrödinger’s nascent theory of emergent biological organization, which is tacitly aligned with the idea of purpose that underlies Wiener’s cybernetic definition of information. In short, negentropy becomes an index of organization, which in turn becomes tethered through cybernetics to the telos of predictable functioning in manmade systems. Brillouin rightly recognized that highly organized systems delay the entropic dissipation of energy. In this sense it is true that Shannon’s notion of ‘information entropy’ cannot suffice on its own to qualify the informational content we associate with highly organized systems and that redundancy, (understood as frequency or constraint which reduces entropy), must be taken into account as one of the conditions of possibility of organization. It can be agreed that what qualifies highly organized systems is the capacity to sustain conditions far from equilibrium. However, by reversing Shannon’s ‘entropic ideas’ of information, negentropy cannot help but become nothing but a degree of predictability. Paradoxically, the greatest measure of information then is what is most predictable and consequently what tells us nothing new at all. Equating information with negentropy means that the maximum of information is, ultimately, what is completely redundant. Negentropy, by negating entropy, renders equivalent the idea of ‘information entropy’ and noise: both negated as a mere measure of entropy. It thereby surrenders novelty, unpredictability and hence also complexity of information to the imperative of certainty at any cost. Yet what is lost when Shannon’s definition of ‘information entropy’ is simply negated, is whatever occurs with a lower probability than the redundant message. As a result, the idea of growing information complexity makes no sense in terms of negentropy. The unintended

Concepts: Information Entropy, Negentropy, Noise

69

consequence of this negation is that, strictly speaking, information and indeed organization can learn nothing new. They cannot become more complex, but only less complex. There is thus a paradox inherent in Norbert Wiener and Claude Shannon’s radically divergent evaluations of the same mathematical formulation of information entropy. And yet, it is important to stress that the mathematical formulation based on Boltzmann’s definition of entropy ultimately remains unchanged. What changes is only the discursive, conceptual and theoretical conversion of entropy into its negation, of ‘freedom of choice’ into constraint, of pure probability into conceptions of order and disorder. The fidelity to the mathematical formula shows that it is the need to interpret and translate mathematical concepts into general discourse that becomes a source of what we can call ‘epistemological noise’. The multiple conceptualizations of information, and implicitly of noise, that have sprung forth from cybernetics and information theory across the natural sciences, but also the humanities, thereby come to confront different ideals of information and organization: on the one hand an ideal of information as ‘freedom of choice’ and on the other hand an ideal of organization as increase of constraint. While ‘freedom of choice’ must take uncertainty in its stride, the functional constraint of the cybernetic information model must ensure control through noise cancelling feedback. Where Shannon’s concept of information risks becoming indistinguishable from noise, Wiener’s cybernetic definition risks producing a radically conservative form of information that can learn nothing new and can only preserve what is already there. The conceptualization of information, and as a consequence of noise, is caught in this field of tension. Yet a satisfactory definition of either information or noise cannot be reduced to either of the two, entropy or its negation, except at the cost of becoming self-contradictory when taken to its extreme as a maximum value, i.e. by becoming either pure noise or mere redundancy.

XII

Complexity on the Basis of Noise

Neither absolute uncertainty nor complete redundancy, on their own, thus suffice for a notion of information that does what the word information says, which is to inform. A viable concept of information must satisfy the criterion of genesis of form, whether we understand form in Platonic terms as idea, or in morphological terms as the emergence of a topological difference. Gilbert Simondon recognizes this tension when he defines information as an activity and an irradiation, the capacity to enlighten new domains. (Simondon 2005b, 541)

A concept of information that satisfies the requirement for this capacity to sustain a process of genesis of form and transformation of diverse domains, must be capable of sustaining the conceptual tension between the two opposed terms. Wiener and Brillouin’s negation of entropy does not suffice to characterize highly organized systems, if by ‘highly organized systems’ we mean those systems that maintain themselves far from equilibrium, like a biological organism or an ecosystem functioning on the basis of homeostasis, i.e. of self-regulation of an internal milieu. Yet it is precisely to this understanding of self-regulation that the cybernetic model extends itself. The negation of entropy, paradoxically, substitutes one form of equilibrium for another, negating entropic equilibrium by substituting it with structural equilibrium. To be far from equilibrium, however, means that a system maintains itself far not only from entropic equilibrium, but also far from the equilibrium of structural redundancy. As a consequence neither the concept of negentropy, nor that of ‘information entropy’ can account for a median term on their own, i.e. for a dynamical equilibrium or metastability. We have already seen that the state of maximal entropy is one in which all exchanges between unequal energy levels have taken place, all energy potential has been exhausted, until no energy potential further propels the system to behave or move in any particular way. However, also its negation, for instance

72

An Epistemology of Noise

the lattice structure of a crystal is (macroscopically at least) at rest when all possible molecular reactions have resulted in stable molecular concatenations. The process of crystallization is one example whereby energetic equilibrium is attained through structural equilibrium, by means of a process of molecular interaction that diffuses entropy in the form of heat into the environment. The crystalline structure itself has more or less exhausted the chemical and physical potential for transformation that was present in the saturated solution from which it emerged. It remains at rest, unless it still contains potential for further crystallization if catalysed, for instance, by a change in temperature. Consequently, the process of crystallization, too, reaches energetic equilibrium, as the final (relative) absence of entropy falls short of a self-perpetuating dynamic or metastability, an instead results in the rigor mortis of a system with energetically spent structures. Paradoxically, the ultimate conclusion of a concept of information based on the negation of entropy would be that a twokilogram crystal contains more information than the average human brain, on the grounds that it more effectively negates entropy. While it certainly makes sense to qualify an organized system as rich in information according to its self-regulating capacity, the simple negation of entropy fails to do justice to the generative and transformative qualities we associate with information. In both forms of equilibrium, structural and entropic, no physical potential is left to propel transformation further (Atlan 1979, 31; Tonnelat 1996). The difference between structural and entropic equilibrium is that one form of energetic equilibrium is bound up in structure, while the other allows for the random movement of free particles. Neither form of equilibrium on its own, however, can be said to retain the potential for the genesis or transformation of form. As a consequence, if by information we seek to characterize processes of organization capable of sustaining themselves in a changeable environment, if we agree that information is what qualifies organized systems that are both far from equilibrium and capable of growing complexity, then neither ‘information entropy’ nor its negation can suffice on their own to account for the genesis of organized form and its dynamical and transformative potential. It becomes clear that neither the maximal state of entropy nor that of structural equilibrium can fully encompass the notion of information, whose maximal value would in both cases correspond to the loss of potential for the genesis and transformation of form. Fully redundant ‘information’ and maximum uncertainty of ‘information entropy’ alike fail to maintain the potential to inform. The dialectic of negation between ‘information entropy’ and negentropy thus leads us to a dead end. What characterizes the resilience of a dynamical and adaptable

Concepts: Information Entropy, Negentropy, Noise

73

system with organizational capacity thus cannot be the mere negation of entropy, which would ultimately mean structural equilibrium rather than being far from equilibrium. It must be, instead, the persistence of a metastable state. Metastability is the dynamical suspension of a system between two forms of equilibrium, between entropic dispersion and structural inertia. It was Gilbert Simondon’s merit to have introduced the concept of metastability to the philosophical corpus, by making it the cornerstone of his theory of individuation. Insofar as our definition of information and noise must answer this tension between both forms of equilibrium, it is not only the physical systems we can describe as rich or poor in either ‘information entropy’ or negentropy, but the very concept of information itself must be characterized by metastability. Neither entropy nor its opposite, redundancy, can act as the seat of information: if information can inform only when it is far from equilibrium, then it cannot rest in either of the two forms of equilibrium. The process of information must be, like the act of walking, a controlled way of falling or, as Henri Atlan formulates it, a succession of ‘recuperated disorganizations’: On can conceive of the evolution of organized systems or of the phenomenon of self-organization more generally as a process of increasing structural and functional complexity, as a result of a succession of recuperated disorganizations, followed each time by re-organization with a higher level of variety and a lower level of redundancy. (Atlan 1979, 49)

Information, in other words, must be sought in a cycle that expends and reloads its potential for transformation, through repeated cycles of acquisition and loss of equilibrium: it must fluctuate between the two extremes, between entropic dispersion and structural rigidity, between uncertainty and certainty – without succumbing to the temptation to seek rest in either of them. As a consequence, the role of entropy in systems with potentially complex organization cannot be dismissed as nothing but noise. Entropy cannot be eliminated without undermining the very principle of organization as that which resists both entropic dispersion and the rigor mortis of structural rigidity. Nor can entropy constitute, on its own, a viable concept of information that can encompass ‘freedom of choice’ beyond its purely mathematical value of probability. The line we draw between information and noise thus becomes the point of highest tension: it is here that information can arise as genesis of form, as new idea, and as transformation of forms of knowledge. Consequently, if the general concept of information is to express the idea of a sustained process of genesis and transformation of form, in a formal analogy

74

An Epistemology of Noise

with physical processes, then it must include a form of disparity that is in some way comparable to the disparity between energy levels in a physical system. But how do we translate this understanding of tension created by disparity of energy levels in a physical system, into a more general concept of disparity or tension of an epistemological kind? The epistemological tension between ‘information entropy’ and negentropy could be defined as the tension between on the one hand an a priori uncertainty, which gives us a measure of ‘freedom of choice’ and, on the other hand, the empirical need to predict, decide and act, which requires a reduction of this ‘freedom of choice’. What is needed is a concept of information that places a relative uncertainty in the context of existing knowledge and problems that constitute such a field of tension, in other words, within which the antagonism between a priori uncertainty and a posteriori reasons to believe creates the potential to propel a process of information further. Our conceptualization of noise in turn makes sense only when introduced into our understanding of processes that must maintain themselves far from equilibrium, and hence far from both absolute uncertainty and complete redundancy. The need to address this tension as the very core of the concept of information helps us to devise new criteria also for our understanding of the relation between information and noise. Just as the sustained genesis and transformation of form, which we call information, cannot be reduced to the mere negation of entropy, so the role of noise, in other words of unintentional or accidental increase of entropy, cannot be discarded offhand as having no informational value. It calls for no less than a methodological revival of the docte ignorance. At stake, however, is no longer the Cusean wisdom of knowing the finitude of knowledge before God, but the knowledge we gain from the mathematical specification of our uncertainty. Noise can thus be said to have a positive epistemic value, if it is understood to specify the uncertainty in which we are about the unfolding of a system. Noise, as a result, is no longer an entirely negative nor an entirely unspecified form of ignorance, but on the contrary is what specifies complexity and informs on what remains to be known. Information and noise in turn become more refined concepts, than the conflation of information with data, signification or knowledge suggests, as for instance in the definition of information as ‘facts provided or learned about something or someone’, as in the Oxford Dictionary (‘Information’ 2014a). The Merriam-Webster dictionary further emphasizes this epistemological aspect of information as ‘the communication or reception of knowledge’ (‘Information’ 2014b). These definitions suppose an equivalence of information and knowledge, which is relayed also in the Penguin

Concepts: Information Entropy, Negentropy, Noise

75

dictionary, where information and knowledge form a whole, composed of facts, data, signification and even ideas: Information/[…]/noun 1a knowledge obtained from investigation, study, or instruction […] b facts that represents data […] esp. regarded as significant […] 2 communication or reception of facts or ideas […].

The participation of noise in a positive aspect of uncertainty, closely welded to the concept of ‘information entropy’, thus no longer indicates only lack of data, loss of signification and the margin of error. Uncertainty instead becomes a measurable function of information as possible knowledge, stratified according to entropy of the message and spontaneous noise in the channel of communication. Knowledge in turn, is not necessarily obtained only by elimination of uncertainty and noise, but by cultivating noise as both reservoir and dynamo for learning, allowing for the dual role of information, as both increasing ‘freedom of choice’ through the novelty of ‘information entropy’ and as reducing uncertainty (information as negentropy) through a process of analysis and interpretation. The idea that ‘the creation of information can only occur on the basis of noise’ was in fact already commonplace in research on artificial intelligence (Atlan 1979, 63–64; Cowan 1965; Neumann 1956). Henri Atlan has been amongst the first to argue explicitly for a theory of ‘complexity on the basis of noise’ from the early 1970s. In order to adapt Shannon’s concept of information to biophysics, Atlan refers to Ross Ashby’s theory of self-organizing systems (Ashby 1962), in particular his ‘law of requisite variety’. For Ashby a ‘function’ denotes a correlation and in this sense ‘communication’ between subsystems. Whether an organism can match the number of its possible states to the alea of the environment, depends on the variety of its subsystems. While noise in the ‘channel of communication’ between subsystems is detrimental to a particular function, making the correlation it establishes insecure, it can still be viewed positively, as an aspect of increased variety of the system as a whole. Noise may, Atlan argues, ultimately contribute to what Ashby defined as the ‘requisite variety’ of the system’s possible states. By increasing the system’s number of possible states, its ‘freedom of choice’ in probabilistic terms, noise can thus be said to contribute positively to the resilience of a complex or hyper complex system in an unpredictable environment – provided it does not precipitate the system’s break down. Any positive understanding of noise as a factor of increased variety must therefore respect the difference between noise affecting the channel of communication, and noise that is transversal to the set of all channels of communication, constituting the system as a whole, which it may affect positively as increased variety. This is where it begins to make

76

An Epistemology of Noise

sense to think both Shannon’s entropic conception of information and Wiener’s negation of entropy alongside one another. Insofar as any system relies on the stable communication of its subsystems, noise at the level of the subsystem is always going to be detrimental to its predictable functioning. The very concept of organization is, if we follow Ashby, premised on a mode of communication, understood as a more or less stable set of correlations between subsystems. If the level of noise in each subsystem is such that signal transmission breaks down, then the correlation between subsystems no longer pertains, making them ‘independent’ from each other. The more independent individual subsystems are, the more vulnerable the system as a whole is to a break down. Functional constraint indeed negates ‘entropy’ in the sense that Schrödinger, Brillouin and Wiener, but also Shannon and Weaver saw and clearly argued: organization requires redundancy. Let us not forget that, despite our emphasis on the conceptual importance of Shannon’s ‘entropic’ definition of information, he is remembered first and foremost for his contributions to the high fidelity of signal transmissions without noise. Yet while noise can never add information at the level of the individual channels of communication, where information is essentially negentropic, in the system as a whole it can be seen to increase the variety of possible states, and consequently improve its capacity to respond to a changing environment, allowing for a greater number of possible states. A system that is constituted by a hyper-large number of subsystems is less likely to break down, if only some subsystems are affected by noise. Systems such as those which von Neumann called ‘extremely highly complicated’ because of the hyper large numbers of components and connections, and which Edgar Morin baptized ‘hyper complex systems’ (Atlan 1979, 48), can thus afford a high degree of noise without collapsing. The quantity of information considered at the level of the organism as ‘information entropy’ (variety) can thus be seen to increase on the basis of noise, in Shannon’s sense of ‘freedom of choice’, rather than in Wiener and Brillouin’s sense of negentropy. Noise between subsystems, although detrimental to their channel of communication, may thus even be considered as creating information (as ‘information entropy’), in other words, as increasing variety in Shannon’s sense of increasing ‘freedom of choice’ at the level of the system as a whole. Despite the now dominant definition of information as negation of entropy (negentropy), Shannon’s positive value of ‘information entropy’ as a measure of uncertainty has thus increasingly fed into various theories of ‘complexity on the basis of noise’: in game theory, computing, artificial intelligence, but also

Concepts: Information Entropy, Negentropy, Noise

77

in theories of physical and biological complexity. Just as physical entropy is no longer understood only as loss of available energy, as it was in classical mechanics, but has become a core concept in our understanding of non-classical physics and mechanics, so the role of noise is set to continue to enjoy re-evaluation in light of the ‘information-entropy’ of complex systems.

XIII

The Astigmatism of Intuition

The positive appreciation of the informational value of entropy now belongs to the mathematical dexterity, without which empirical reality can no longer be understood beneath and beyond the threshold of intuition. The counter-intuitive nature of mathematical reasoning, however, also leads to the pulverization of the idea of simplicity: the apparently simple becomes maximally uncertain. The analytical method of the modern sciences, based on the Cartesian method of analysis, which consists in progressing from the simple to the more complex, here it encounters a fundamental difficulty, as apparent simplicity reveals itself to be an epistemological stumbling block, rather than constituting than the rock of foundation of knowledge. Not only does the apparent homogeneity of the entropic mix reveal uncertainty and unpredictability, but even the atom, the god-father concept of all individualism, the simplest building block of empirical reality since Democritus, is pulverized into complexity by quantum physics. Noise and uncertainty are no longer the prerogative of (large or hyper-large) aggregates, but even of simple elements: simplicity itself becomes complex. The conversion of highly formalized concepts of information and noise to those which are less formalized, and therefore more speculative, is not aided by the fact that the mathematized sciences have for a long time, and certainly since Lobachevsky, Boltzmann and Cantor, become accustomed to proceed in a counterintuitive way, to trust in mathematics where the senses can no longer follow. As a result, one of the difficulties in correlating noise with complexity is the common perception of entropy as the homogeneous murmur of ‘white noise’. Apparently bland and predictable, noise becomes a blanket background against which a signal appears to stand out as what is least predictable. The common perception of entropy is one of lacking difference, of a homogeneous and therefore presumed simple and predictable state. If we are to see how entropy is related to low probability and high uncertainty, it is therefore indispensable to deal with this tension between counterintuitive and intuitive modes of reasoning as a veritable clash of cultures.

80

An Epistemology of Noise

If we want to understand what is at stake in the definition of information as negation of entropy, or negentropy, we must tackle this paradox, whereby the greatest novelty of information, that which is maximally unpredictable, corresponding to ‘information entropy’ in Shannon’s sense, appears to be incompatible with the image we may have of entropy as a state that is of no interest at all, homogeneous and unchanging like a fully diffused drop of ink in water. This image of maximum entropy as a bland homogeneity from which nothing new can arise, implies that nothing can surprise us: the entropic system no longer has any physical potential to further transform itself. From this point of view maximum entropy would appear to be perfectly redundant. Following this logic one could thus say, arguing against Shannon’s concept of ‘information entropy’, that it is entropy that tells us nothing new, that entropy is redundant like the image of a homogeneous and undifferentiated whirr of ‘white noise’. Nothing interesting will happen, nothing will surprise us in a closed system that has reached the state of entropic equilibrium. Raymond Ruyer insists on this point and also Gilbert Simondon argues in this direction (Ruyer 1954, 114; Simondon 2005, 541). What is thereby taken for granted are two things: firstly, that information can only be defined as the negation of entropy, as Wiener did in the context of self-regulating systems with feedback; and secondly, in Simondon’s case, that the concept of information negates entropy, insofar as information corresponds to the novelty of what may still happen within a system that has not yet reached entropic equilibrium. Schrödinger’s conception of life as that which delays the spontaneously entropic dissipation of physical processes, reinforced by Brillouin’s neologism negentropy, has led to a conception in the humanities whereby information is what occurs against all probability, meaning against the greater likelihood with which physical processes are said to evolve spontaneously towards the complete entropic dissipation of energy. This is a well disseminated view that appears to feed on the philosophical and literary imaginary of an infinitely unpredictable world, imagined as an open system, where anything can happen, so long as it has not exhausted itself and reached its entropic death. This idea of an open, complex and changeable world is contrasted with the image ‘white noise’ in a closed system, symbol of the decomposition of all possibility. What this understanding of information as negation of entropy implies, however, is that the second law of thermodynamics, which stipulates that every closed system invariably evolves towards maximum entropy, is in fact contingent rather than necessary: it can be resisted. This, however, requires that we leave physics behind and consider the second law of thermodynamics as a mere

Concepts: Information Entropy, Negentropy, Noise

81

image, where the necessity of a natural law is no different from the contingency of probability: a law that is no law because it does not always apply with the same vigour, but is contingent upon the particular type of empirical reality which may submit to it or resist it. It makes no sense to speak of the probability with which a system evolves towards entropy, unless there is at least one possibility that it will not – and in this case either the second law of thermodynamics is no law at all, or information is a miracle. It is thus difficult to maintain that the second law of thermodynamics is contingent rather than necessary, while holding onto it as a core element in the definition of information. What is certain, however, is that the second law of thermodynamics applies to closed systems. Open systems, on the other hand, which are called metastable when they maintain themselves far from equilibrium, must answer a different set of theoretical requirements, to which the second law of thermodynamics still applies, but requiring explicit theoretical conversion. Consequently, the idea that information corresponds to the negation of entropy, commonly understood as what occurs against all probability, cannot be applied without reservation to the notion of information in an open system. This conversion is in fact the object and originality of theories of so-called ‘self ’organization, such as those initiated by Manfred Eigen’s work on the origin of life in chemical cycles and hyper-cycles, Ilia Prigogine’s work on dissipative structures and complexity or Katzir-Katchalsky work on chemico-diffusional coupling in biological flow-structures (Atlan 1979, 99–128; Atlan and KatzirKatchalsky 1973; Eigen 1971; Prigogine and Stengers 2009). These approaches, however, are not representative of an interpretation of the notion of information as negation of entropy, but on the contrary render possible theories that consider the positive role of entropy in organized systems with increasing complexity and indeed theories like Atlan’s ‘complexity on the basis of noise’. To associate information (negentropy) with unpredictability insofar as it negates entropy thus relies on the conflation of multiple levels of analysis: a system that has reached a state of energetic equilibrium, hence of entropy, will indeed have no predictable transformation going on at the molar, visible level. But this apparent ‘death’ of the system, which is no longer compelled by physical potential to transform itself, belies the lack of constraint that physical potential exercises on a system: the compelling nature of physical potential is, as we have seen earlier, what makes the system’s transformation more rather than less predictable. The higher the potential, i.e. the energy differential, the more predictably a system evolves. This is why we tell children not to play with electrical sockets. What remains maximally unpredictable, on the contrary, is

82

An Epistemology of Noise

the movement, position, speed and direction of microparticles in the entropic mix, like the whirring atoms in a gas canister. The failure to see complexity and unpredictability in the apparently simple homogeneity of the entropic mix comes from the fact that the macroscopic, molar level of observation tacitly steps into the foreground, pushing the analysis of the molecular and statistical level into the background. The macroscopic presentation of a system in the state of maximal entropy is indeed often characterized by homogeneity, as in the example of the diffused drop of ink. Although strictly speaking, the term entropy refers to the homogeneity of energy levels, which does not necessarily preclude macroscopic heterogeneity, as is the case of an entropic mixture of water and oil, which is not morphologically homogeneous. In any case, it is also true that in a maximally entropic closed system no dynamics are likely to arise spontaneously (even if localized and transitory random oscillations exist). The entropic mix appears completely redundant, like the homogenous murmur of white noise. It appears to harbour nothing of interest, no informational potential can be detected in it: no directional movement, no flows, no noticeable difference. Viewed like this, it is indeed entropy that appears redundant and predictable. This apparent simplicity, however, diverts attention from the complexity of its microphysical presentation. The absence of constraints, which distinguishes entropy from redundancy, and which entails equal probability of all its possible states, is still what characterizes entropy in mathematical terms as the greatest possible ‘freedom of choice’. In other words, a misunderstanding can arise if we conflate the molar and the molecular level and a fortiori if we conflate the role of entropy in a closed and in an open system. Norbert Wiener’s definition of information as the negation of entropy does not deny this, it merely performs a reversal of Shannon’s concept of ‘information entropy’: information is no longer understood as what increases the ‘freedom of choice’ in terms of probability, but as its negation. Wiener’s functional, machine oriented concept of information as the reduction of ‘freedom of choice’ and hence as an increase in predictability simply means that information as negentropy is what is ultimately redundant. To say otherwise is to induce a curious short circuit, whereby the negation of entropy, negentropy, would result in a concept of information that expresses not frequency, repetition, or deductive redundancy, but randomness – transforming the trajectory of a self-directing missile into the erratic loop of a swallow. There is thus a certain amount of confusion, when the concept of information as negentropy is seen as increasing ‘freedom of choice’ and unpredictability vis-à-vis the state of the apparent redundancy of entropy or ‘white noise’.

Concepts: Information Entropy, Negentropy, Noise

83

Negentropy has as its main purpose the reduction of unpredictability, notably in the cybernetic context, where the requirement is the prediction for instance of the trajectory of a self-directing missile, which certainly cannot tolerate any increase in unpredictability. When we narrow down the options, decrease the number of possible states of a system and thus increase probability of accurate prediction, we generate ‘information’ in the sense of negentropy, i.e. a greater prediction. If we now go on to say that the negation of entropy serves to reduce the margin of error of prediction, by narrowing down the number of possible scenarios, then what Shannon calls ‘information entropy’, in opposition to negentropy, becomes even less clearly distinct from the concept of noise, if noise like ‘information entropy’ is what makes reality less predictable. Both ‘information entropy’ and noise must be understood as what increases uncertainty. ‘Information entropy’ and noise thus stand together in opposition to what the negation of entropy designates as information, which is the accuracy of prediction and negation of possible alternative scenarios. Entropy, finally, corresponds to the unpredictability of the micro-complexions of the system, which is frequently represented as the random collision of particles. The apparent simplicity of entropic homogeneity merely masks an indiscernible complexity, which is revealed only at the microscopic level, where there is increased uncertainty, if one were to predict the movement of its microconstitutive elements. Equal probability thus corresponds, despite the apparent simplicity of ‘white noise’, to the greatest possible uncertainty in which the observer is as to the prediction of the micro-complexions of a physical system. Despite its apparent simplicity this uncertainty corresponds to the greatest ‘freedom of choice’ and to what Shannon calls ‘information entropy’. Apparent simplicity thus merely masks its indiscernible complexity at the microscopic level. When entropy is seen as ‘simple’ rather than complex, in accordance with the appearance of ‘white noise’, it is thus a case of phenomenological astigmatism.

XIV

The Path of Despair

The two dominant interpretations of the term ‘information’, having taken hold in the natural and human sciences, carry significantly different implications: one values information as a specific form of uncertainty, implying increased ‘freedom of choice’ (‘information entropy’), and the other negates the latter in favour of prediction (negentropy). Shannon’s ‘entropic ideas’, whereby the information containing the greatest degree of novelty is the most difficult to predict, and Wiener’s cybernetic definition of information in control theory, where the objective is reliability and prediction, appear incompatible yet are more often than not diffused across scientific discourse without clear demarcation. Yet both are based on the same mathematical expression of entropy. Precisely because they offer a purely quantitative value, their definition of information is not determined by epistemic, aesthetic or moral values: the engineer does not judge the quality of the information s/he transmits – whether it is a redundant phone call of your mother-in-law or a transmission of Schönberg’s Variations for Orchestra, only the quality of transmission. Nevertheless, Shannon’s entropic definition of information opens a chasm of uncertainty at the heart of our understanding of information. It allows for a positive epistemic appreciation of uncertainty (‘information entropy’), whose generalization beyond the technical problem of signal transmission enables a renewal of the Socratic maxim, ‘I know that I know nothing’. Such knowledge of our uncertainty, however, is no longer an intuition, but measures the ratio and hence limitation of what we can predict with the greatest possible precision. Throughout its history philosophy has been prone to a perpetual high-jacking of certainty. This high-jacking of certainty is perhaps not as different in nature to Shannon’s conceptual audacity as the ‘merely’ technological status of his concept of information suggests. Philosophy indeed never rests on its laurels for long. Radical doubt and new concepts make established certainties redundant, or recover and transform

86

An Epistemology of Noise

defunct certainties that were left for dead. Whatever certainties philosophy thus acquires or rehabilitates in its own right (rather than, say, mathematical proof or mystical faith), they are by default exposed to critique’s systematically erosive force of doubt and renewal. From Greek scepticism to Hume, from De Cues’s docte ignorance to Pascal’s wager, from Descartes’s tabula rasa to Kant’s epistemic humility and from Schelling’s speculation about the absence of ground to Hegel’s ‘path of despair’, the cycles of liquidation of illegitimately held grounds for certainty are inseparable from the production of new, reconfigured foundations. Wittgenstein’s paradoxes, Feyerabend’s radical critique of method and Popper’s falsificationism are not the end of the tether for philosophy, but the flowering of its many cultures of systematic uncertainty. So-called ‘continental’ philosophy, from post-structuralism to post-modernism, has been ridiculed for preferring the deconstruction of our certainties to the plain language of constructive or analytic objectives of philosophy. Yet both the ‘continental’ and analytical traditions display a radical distrust of certainties, which they see as exposing reason to the risk of naivety, if not ideology. Both the philosophers associated with ‘continental’ philosophy and those associated with the analytical tradition thus belong together in the abstract tableau of philosophy’s great etching of imprudent convictions. Looking at the diverse philosophical cultures of uncertainty, one might be tempted to see in philosophy the vocation of overcoming dogma with ever more recalcitrant cultures of uncertainty. If philosophy can be said to inform us, in a way that participates in the natural and the human sciences, then it is fair to say that Shannon’s entropic conception of information is better suited to qualify such information as specified uncertainty than the traditional correlation of information with certainty or even the negation of uncertainty implied in negentropy. However, every self-respecting philosopher must surely object that the discipline of philosophy is, on the contrary, to put the systematic use of reason in the service of solid foundations for certainty. What is the vocation of philosophy, if not to surpass the unreliability of the senses, to overcome the changeability of opinion and to reveal the unfoundedness of doxa? In other words, what else is the vocation of philosophy, than to dispel uncertainty – just as Shannon puts mathematics at the service of a noise-free channel of communication? From this point of view, it would seem justified to say that philosophy doubts in the service of certainty and its ultimate victory over uncertainty. Is not the objective of philosophy to reduce the margin of error in reason in order to arrive at an ultimate victory of reason over empirical contingency? To make philosophy

Concepts: Information Entropy, Negentropy, Noise

87

duty-bound towards certainty, however, would be to neglect what drives philosophy, the divine mania of Platonic inspiration, the esprit de finesse with which Pascal complements the reason of the geometer, Kant’s acknowledgement of the vertigo of the philosopher before the question of ground, Schelling’s notion of ground as anti-idea and even Hegel’s path of despair. Michel Foucault’s famous expression that ‘error is at the root of what makes human thought and its history’ here reveals its true relevance for the conceptualizations of noise, which is the normative relation between reason and contingency (Foucault 1989, 22). Foucault’s appraisal of ‘error’ in his introductory statement to George Canguilhem’s 1966 edition of The Normal and the Pathological stands in apparent contrast with the admiration he proclaims for a philosopher who imparted the utmost exigency of rigour to an entire generation of philosophers in France, for whom rigour was elevated to an ethos of philosophy. But the counter-intuitive consequences of this rigour produced an offence to common sense that contributed to the apparently unsurmountable distance between French and Anglo-Saxon philosophy. The distance has since been caricatured as opposing the presumed literary and imprecise bent of so-called ‘continental’ philosophy, to Anglo-Saxon philosophy, which in turn is downplayed by many a ‘continental’ philosopher as wallowing in pseudo-scientific rationality. What should in itself arouse suspicion about the geographical labeling is that much of the so-called ‘Anglo-Saxon’ philosophy pledges critical fidelity to a moment of great conceptual inventiveness taking place in Europe at the turn of the twentieth century around the Vienna Circle. Continental philosophy is deeply entwined with this critical moment in the history of philosophy; some of its most crucial developments during the 1960s, like for instance the journal Cahiers pour l’analyse, created an intense dialogue, notably with Wittgenstein’s thought and also via translations, like philosopher of logic Claude Imbert’s translation of the logician Gottlob Frege into French (‘Cahiers Pour l’Analyse (An Electronic Edition)’ 2017; Hallward 2012a, 2012b). French epistemology and post-structuralism’s interest in the ruptures, bifurcations and unforeseeable turns in the historical unfolding of scientific theories has, in turn, reached well beyond France, and beyond the theory of knowledge, into the Anglo-Saxon humanities, taking back alleys through the fields of literary, cultural and political theory and aesthetics. And while its ramifications into the Anglo-Saxon theory of knowledge are less evident, they are no less significant, if we consider for instance the importance attributed to Thomas Kuhn’s theory of scientific revolutions, who in turn pays tribute

88

An Epistemology of Noise

to French epistemology. To this day the lack of translation of key works of what we now call ‘French epistemology’ – such as those of the philosophers of mathematics, Jean Cavailles or until very recently Albert Lautman, into English – testifies to the obstacles that must still be overcome before these two approaches can dialogue in earnest. Nevertheless, it can’t be denied that at a time when the Vienna Circle made it the objective of philosophy to eliminate ambiguity from discourse, espousing the formalization of language through logic and an emphasis of scientific empiricism, an alternative approach to the philosophy and the history of the sciences emerged in France. Without further distorting or exaggerating the divergence between ‘analytical’ philosophy and French epistemology, it can be acknowledged that analytical philosophy made it its objective to stabilize philosophical discourse and its history by emulating scientific methodology and emphasizing logic, while French epistemology espoused the unpredictable historical unfolding of mathematics and of scientific theories more generally. Its emphasis on the historical singularity of the ‘event’ in thought opened philosophy to an understanding of uncertainty as constitutive of the question of the foundation of truth. Jean Cavaillès thus enquired into the consequences for philosophy implied by Georg Cantor’s unmooring of axiomatic set theory from intuition. Even the Kantian proposition of an unchangeable structure of our transcendental a priori found itself shaken in Cavaillès analysis, and thrown into the historical becoming of mathematics (Cavaillès 2000). Gaston Bachelard, in turn, called for a philosophy of complexity in light of developments in quantum physics and chemistry (Bachelard 2003), and George Canguilhem’s revaluation of the concept of error in his study on The Normal and the Pathological (Canguilhem 1991) was driven by the realization that reason is not only subject to norms, but is itself a source of the norms of reason: this normativity of reason accounts also for the unpredictability of science’s historical unfolding, revealing its necessity only retrospectively. From a different perspective to that of Logical Positivism, French epistemology thereby reopens the question of ground each time reason is caught in the act of generating new norms, thereby outgrowing and trespassing the limits of established norms: Almost all progress in algebra has come from the desire to accomplish forbidden operations (negative, rational, imaginary numbers etc.). (Atlan 1979, 229; Canguilhem 1993; Collectif 1979, 508)

Concepts: Information Entropy, Negentropy, Noise

89

This quotation is a curious case of palimpsest, where different texts and different authors transpire and appear to speak in one voice. Georges Canguilhem here quotes the mathematician René Thom, a Fields Medal winner known for catastrophe theory. What is interesting is that Canguilhem cites this passage from Henri Atlan, who himself cites René Thom’s intervention at the Colloque de Royaumont in 1975, in his book Entre la fumée et le crystal, 1979. It is the quotation of a quotation through which transpires a philosophy that values non-linear thought positively as a vector for scientific striving, as an impulse for reason’s normativity, be it in different fields of theoretical enquiry. It is significant that Canguilhem here cites Thom in a conference paper entitled ‘Le cerveau et la pensée’ (Canguilhem 1993), a now-famous attack on reductionist theories of the brain and thought, which Canguilhem associates with cybernetics and a market-driven obsession with computers. For Canguilhem one name is emblematic for the cybernetic ‘ironmongery’ of thought: [L]et it suffice to cite a name: that of Leonid Pliouchtch, and an emblem: that of I.B.M. (Canguilhem 1993, 11)

It may thus appear particularly ill-fated to seek in Canguilhem’s thought a motive to unpack the philosophical relevance of Shannon’s ‘entropic ideas’. Shannon’s quantitative definition of information has not unfrequently been blamed, alongside Wiener, for the reduction of thought to the mere calculating power that stands accused in Canguilhem’s paper. The reduction of thought to what Canguilhem calls disparagingly (quoting Thom), the ‘ironmongery’ of thought,3 could indeed be associated with the successful application of Shannon’s information theory to informatics alongside Wiener’s cybernetics. To correlate the idea of noise with Canguilhem and Foucault’s revaluation of error as the ‘origin and history of thought’ could thus appear nothing short of heretical. The now common notion of ‘noise pollution’ furthermore evokes the idea that noise is the acoustic waste generated by industrial development, rather than the late flowering of the Enlightenment. Rather than pointing to the origin of thought, which is traditionally sought in the peace and quiet of contemplation, nothing could seem less conducive to human thought and its history than noise, and nothing more readily the source of litigation in increasingly densely populated urban spaces than noise (Berglund and Lindvall 1995; Bijsterveld 2001, 2003). Even Pascal, who may be credited with having invented the calculus of probability in games of chance, which later informed the concept of noise, saw in noisy entertainment and distraction an avoidance of thought (Pascal 1966, 71–73). Schopenhauer even defined noise as ‘the most

90

An Epistemology of Noise

impertinent of all interruptions, since it interrupts and even shatters our own thoughts’: even a great mind is powerless if dispersed by noise, like a scattered army or a splintered diamond it loses its power and incisiveness (Schopenhauer 1851, 517). And yet, if two ways of reading ‘information entropy’ are possible, then the fundamental ambiguity in Shannon’s definition of information for which we have argued thus far, namely the ambiguity between ‘information entropy’ and noise, enables us to take Canguilhem and Thom’s defence of thought’s normativity as an argument for the defence of Shannon’s ‘entropic ideas’. The cultural significance of the ‘freedom of choice’ inherent in Shannon’s entropic ideas can then be measured against Wilhelm von Humboldt’s words, quoted by Canguilhem in this same text, according to which [i]t [language] must therefore make infinite use of finite means (Sie [die Sprache] muss daher von endlichen Mitteln einen unendlichen Gebrauch machen …). (Canguilhem 1993, 26; von Humboldt 1903, VII, 98–99; 1836, 106)

Notes 1 I thank Emmanuel Picavet for his insightful comments regarding this point. 2 Only the second part of the main thesis was published in 1964 as L’individu et sa genèse physico-biologique by Presses universitaires de France, while the first and second parts were published together by Aubier as L’individu et sa genèse physicobiologique and L’individuation psychique et collective. 3 ‘The traders of electronic ironmongery [or junk; French quincaillerie] would like to make us believe that with the distribution of computers a new era will open up for scientific thought and for humanity’ (Atlan 1979, 228–29).

Part Two

Empirical Noise

I

On the Transduction of the Concept of Noise

The new understanding of noise, brought about by the mathematical approach to information, has transformed modern society beyond recognition, both from non-classical mechanics and quantum physics to molecular biology and across the natural and human sciences wherever statistical reasoning holds sway. New communication media have sprung forth, which have in turn generated new flows of information but also new structures of information distribution, and therefore new dimensions for noise, ambiguity and error. Shannon’s ‘information entropy’, but also Wiener’s negation of entropy, in this sense carry forward the leap in modernization that had been accelerated during the nineteenth century by mathematical physicists, notably by James Clerk Maxwell’s theory of electromagnetic radiation and Boltzmann’s statistical expression of entropy. The increasing energy efficiency of machines, capable of harnessing entropy, together with the emergence of the electric grid are the direct result of the molecular understanding of entropy in electrodynamics, which in turn helped lay the conceptual foundations of what Floridi has called ‘the information revolution’ (Floridi 2014) which ushered us from an industrial to a post-industrial modernity. Noise no longer characterizes only entropic processes related to mechanical work, but increasingly conditions information networks, and even, if differently, the co-emergence of cognitive labour, characterized by information overload and even the ‘mental state of noise’. Boltzmann’s statistical expression of entropy and Shannon’s generalization of the concept of entropy as a problem of probability and statistics thus represent not only the relay of a profound conceptual innovation, but also a leap in technological innovation. The scientific and technological mastery of noise implies novel scientific views, provoking what Thomas Kuhn called paradigm shifts: the shift not only from classical to non-classical mechanics, but also from

94

An Epistemology of Noise

the logic of transformation of energy (for instance, mechanical to electrical) to a logic of information, amplification and control through feedback. How can we understand this transfer of the idea of noise from mechanics to information, and from its antithetical relation with work in classical mechanics to self-regulating systems with noise, bearing in mind also that the general notion of noise is derived from an aesthetic and moral connotation of acoustic events? Gilbert Simondon, once more, provides a key for understanding this conversion of noise from its aesthetic/moral connotations of sound to mechanics, and from the industrial era of mechanical work to the post-industrial era of regulative information networks and cognitive labour: the concept of transduction. The term ‘transduction’ commonly describes the conversion of one form of energy into another (for instance, kinetic into electrical energy) or the transmission of a signal from one system into another (as when a signal is ‘transduced’ across the membranes of a biological cell). Simondon uses the term transduction to describe the process of individuation, using the example of a crystal that grows in all directions, by transducing its structure such that ‘each molecular layer already constituted serves as a base for the layer in the process of formation’ (Simondon 2005, 33). He also establishes an analogy between transduction in empirical reality and transduction occurring in the domain cognition, which consists in ‘following being in its genesis, in carrying out the genesis of thought’ in analogy with ‘the genesis of the object’ (Simondon 2005, 34). In other words, the domain of cognition, and the fields of knowledge it establishes, is said to come about through a process of transduction analogous to the processes of transduction observed in empirical reality. In fact, it is this process of conceptual transduction that enables Simondon to apply the concept of individuation to a great variety of empirical domains, from quantum physics to psychosocial reality. Simondon’s epistemological appropriation of the idea of transduction is helpful if we want to understand the transformation of this cluster of concepts (work-order-entropy-information-regulation-organization) around noise. The transduction of the concept of noise would imply not only the transformation of one domain, for instance, of information theory, but also the conceptual transduction from one domain to another. Applied to the domain of knowledge, with its increasingly specialized and diverse areas of expertise, this means not only that transduction is how an area of knowledge is progressively structured according to guiding principles or concepts, but that the structuration of one field of knowledge also transduces its guiding principles, concepts or problems, across academic divisions and institutional boundaries, into other fields of knowledge.

Empirical Noise

95

When Simondon says, ‘by transduction, we mean a physical biological, mental, or social operation, though which an activity propagates from point to point within a domain’, we must therefore hear also the transduction a concept or idea developed in one domain of knowledge into others. It is in this sense that we can say that that Boltzmann’s statistical concept of entropy has been transduced by Shannon, Wiener and other mathematicians, into the statistical concepts of noise and information, which we now find across the natural and human sciences and even in the arts. To be sure the idea of transduction of concepts from one domain of knowledge to another, like that of the concept of noise, is an appropriation of Simondon’s own appropriation of the term. It is taylored here to help us to think about epistemological resonance and epistemological noise. The effect of Boltzmann’s statistical understanding of entropy can now be understood in its significance beyond that of innovation in the specific field of thermodynamics and mechanical engineering. The profound consequences it had as a conceptual innovation illuminated the domain of new communication technologies and progressed from there to all other domains of knowledge, thereby transducing the concept of entropy from mechanical and thermodynamic domain to the concepts of information and noise in communication technology and beyond. The transduction of the concept of noise is at once conceptual and technological, because it goes hand in hand with the material recording and transmitting information. Conceptual transduction requires more than just translation of a term (noise), from acoustics to thermodynamics, or from statistics, say, to molecular biology. There is an irreducibility between these domains. There is a lack of one-to-one correspondence between the diverse uses of the term noise. In other words, they are not isomorphic as two sets of axioms could be. The consequences of thinking in terms of noise as a result differ from domain to domain. As a consequence, the transduction of the concept of noise could be said to scatter, more than it resonates. It hits obstacles, as reason struggles to find a universal key of conversion from one domain of knowledge to another. In other words, there is no transduction in the epistemological sense without noise. This noise increases ‘freedom of choice’ in the form of ever greater complexity of the field of knowledge as a whole. In this sense the transduction of concepts is irreversible, like entropic dispersion, which means that it does not lead to a reductive one-to-one correspondence of concepts between one field and another.

96

An Epistemology of Noise

As a result, any concept and any problem that becomes relevant across theoretical or experimental boundaries must accept a certain metaphorical warping, unless a strictly reductive logic is applied. Indeed it is doubtful that many domains of knowledge currently submit to a reductive logic. It seems, rather, that many reductive theories today simply postulate that the bridging between levels of complexity is merely a question of time and a matter of filling in the gaps. Yet we have seen how in an axiomatic system even one postulate can become a gateway to a radically different axiomatic system, as was the case with Euclid’s unproven fifth postulate and the subsequent development of nonEuclidian geometry with infinite dimensions. For any conceptual transfer to occur without noise, and without metaphorical distortion, both fields would therefore have to be fully axiomatized, any divergence arising from mere postulates fully accounted for, in all its amplitude. However, to accept metaphorical warping, and to understand that concepts are often transduced not only in the form of abstract formalization but also in the form of mental images, must not mean to accept the intrusion of concepts coming from other fields of knowledge uncritically or without precision. On the contrary, the currently germinal relation between the conceptual creativity in the sciences and in the arts could provide a new opportunity to enable scientific discourse to engage in metaphor and images of thought critically and creatively, but most of all consciously. In other words, if we want to understand how the concepts of noise and entropy have contributed to restructuring the epistemic field as a whole, transductively and with the epistemological noise arising from this transduction, then the necessary metaphorical latitude must be engaged with actively rather than passively, creatively and critically rather than through denial or convention.

II

Accidental Information, Predictable Noise

To think about the scattering that necessarily accompanies any transduction of concepts in the form of epistemological noise, no doubt implies a return to Weaver’s distinction between information and noise as the difference between desirable uncertainty (information entropy) and accidental, hence spurious uncertainty (noise, or just entropy). And yet a problem immediately arises, when the difference between information and noise is reduced to the difference between intention and spontaneity or accident, because this difference is almost never as clear-cut in vivo, as it is in the situation of the engineer, who transmits a ready-made message. Research, for instance, occasionally happens upon the spontaneously informative as an instance of serendipity. The element of accidental discovery implies a shifting of the boundary between the informative and noise, which must not remain unthought. Moreover, in statistics it is not only serendipity (which every researcher must hope for), but the systematic use of randomness that requires us to place the accidental at the heart of the information process. As Andrew Gelman and Cosma Rohilla Shalizi put it in ‘Philosophy and the practice of Bayesian statistics’: [t]he statistician begins with a model that stochastically generates all the data […]. (Gelman and Rohilla Shalizi 2013, 8)

In other words, data must be generated randomly, even if the goal is to arrive at an understanding of the systematic relation between variables, between variables and their parameters, and between variables and ‘noisy aspects of the data’, meaning its ‘contingent, accidental, irreproducible’ (my emphasis) aspects. Bear in mind that randomness is a mathematical problem, and that the word ‘stochastic’ was introduced in early probability theory from the Greek, meaning to conjecture, to take aim or guess. The definition of noise is thus only slightly distinct from stochastically generated data insofar as the stochastic implies taking aim, while noise implies missing the target. The accidental thus partakes

98

An Epistemology of Noise

in the way a research question is formulated against the backdrop of uncertainty, in the design of a statistical model or scientific experiment, and in the testing of models or theories. The distinction between information and noise is a clear task only for the telecommunications engineer: here the problem is only that of fidelity of a message, after its selection has taken place and only once the decision has been made to transmit this message, whatever it may be, as information. The difference between information and noise derives from this decision alone, as it does from the selection of one scientific model over another. It is this selection that designates something, anything as far as information. It is only on the basis of an always already made selection that the telecommunications engineer or any other statistician can root out any increase in the message’s entropy during transmission, which can then be discarded as accidental, i.e. as noise. What remains in the dark, if we consider only the problem of noise in the channel of communication, is thus the part that the accidental may play in the decision to select and designate something, in principle anything, as information. The origin of noise may be an external perturbation (afferent noise) or noise may generated by the thermal agitation of the transmission itself (as efferent noise). The engineer is consigned the task of ensuring the safe passage of information through the channel of communication, warding off afferent noise and keeping information free from corruption through efferent noise. There is noise generated by the transmission of information, either produced by the thermal vibration of atoms in the conductors (Johnson-Nyquist noise), or the noise already intrinsic to the means of measuring of discrete electric charges, or of photons in optical devices (shot noise or Poisson process). In other words, information is always already thrown into the noise of its own presence, but also that of objects surrounding it (not least the engineer), of earth and even of celestial bodies and black holes, each emitting a frequency spectrum characteristic of its own electromagnetic radiation – its black body radiation (so called not because it is black, but because the eye cannot perceive colour in low intensities of thermal radiation, unless the temperature is elevated). All of the above forms of noise participate in a pool of so-called Gaussian noise, meaning that this ‘general’ noise, is characterized statistically by its ‘normal’ distribution, also represented by a so-called bell curve of the average of a large number of random Gaussian processes. It is worth dwelling on this slightly technical point, lest we miss the paradox it implies for our understanding of noise. Noise (accidental, non-necessary, increasing uncertainty and ‘freedom of choice’ in probabilistic terms) is also normal and ubiquitous, often subject to

Empirical Noise

99

a highly predictable statistical form, called a normal distribution. So predictable, in fact, that telecommunications can mimic its uniform distribution across the frequency band, and use what is called additive white Gaussian noise (AWGN) as a basic noise model for the underlying behaviour of a system, so as to later distinguish phenomena like fading, interference, dispersion or other chaotic or unpredictable (non-linear) events. What at first presented itself as non-reproducible stochasticity, for the communications engineer noise on the contrary becomes characteristic for the underlying behaviour of a system. The least we can say is that the relation between noise, stochasticity, the accidental and the predictable is far from simple. In a course entitled ‘Uncertainty in Measurement: Noise and How to Deal with It’, Peter Scott, Emeritus Associate Professor at the University of Santa Cruz Physics Department, explains the difference between noise, understood in the empirical sciences as a statistical term, meaning random error in the sense of statistical uncertainty, and as a systematic error in measurement, such as it can arise, for instance, from a mis-calibration of a metre or from some physical effect not taken into account. The idea of noise in physics applies when a quantity measured in an experiment involves a large number of random processes that can only be approximated statistically. The pressure of a gas, for instance, results from summing the random motions of a very large number of molecules. These random motions are likely to be distributed normally, according to the Central Limit Theorem, which states that if we have a number of random variables, say u, v, w, …, and if we form a new variable z that is the sum of these (z = u + v + w + · · ·), then as the number of such variables becomes large, z will be distributed normally, i.e., described by a normal distribution, regardless of how the individual variables u, v, w, … are distributed. (http://physics.ucsc.edu/~drip/133/ch2.pdf)

Almost all measurements that are normally distributed around some average value (over 99 per cent) will lie within three standard deviations of the mean, giving this normally distributed from of noise a great advantage of predictability. However, Scot warns that not every data point falls within the classic bell curve and that ‘many are non-normal – too far away from the mean – A voltage spike or a vibration caused by a truck passing by, for instance […]’ (http://physics. ucsc.edu/~drip/133/ch2.pdf ). Such extreme, non-normal values are no doubt closer to our intuitive understanding of noise in general discourse, derived from the experience of an unwanted or startling sound. What is clear, however, is that we have a

100

An Epistemology of Noise

relatively solid basic conception of noise in physics, tied to the normal statistical distribution of random events (represented by the characteristic Bell curve), which does not exclude the definition of other forms of noise that can be defined as special cases, obeying different mathematical parameters than the normal or Gaussian distribution. In short, information is lifted out of an (in principle infinite) pool of accidental frequencies, of which some are ‘normal’ and others, consequently anomalous. The fundamental problem in the definition of noise will thus be, at what point anomalous or extreme values in a statistical distribution call for a change of perspective (or model) from an approach that privileges the normal distribution of noise and thus de-emphasizes the importance of extreme values as rare and insignificant, to an approach that calls into question a given model, by emphasizing the importance of values that by far exceed the norm. The history of science shows how the paradigm of Newtonian physics buckled, amongst other things, under anomalies in the measurement of Mercury’s orbit, the smallest and innermost planet of the solar system. It was only when these anomalies could no longer be contained in a normal distribution of errors (and hence disregarded) that the possibility of an alternative paradigm could arise, eventually enabling the emergence of relativity theory. The power of the dominant Newtonian model to discard the relevance of extreme values in favour of the normal distribution of the ‘law of errors’ revealed itself to have become an epistemological obstacle, delaying the realization that a profound change in theory was called for. Thomas Kuhn later theorized the subsequent change of perspective as a paradigm shift, which he defined as a revolution of consensus in the scientific community. Statistics, as a theoretical and applied discipline, has since created a highly complex arsenal of mathematical approaches, including the combination of models and their systematic testing, ranging from Bayesian data analysis to complex stochastic models and simulation-based model checking. The problem is that the promiscuity between this highly specialized discipline and the hunger for statistical ‘evidence’ in general discourse risks a continual return, in general discourse, to the epistemological obstacle physicists faced when the highly coherent Newtonian system had to be overhauled. In other words, the public appetite for the ‘empirical evidence’ accorded to statistics risks falling prey to an underestimation of the need for testing and revision of models. As a result, information is deemed acquired, and noise can be more easily discarded as meaninglessness. As Andrew Gelman and Cosma Rhilla Shalizi put it:

Empirical Noise

101

All models will have errors of approximation. Statistical models, however, typically assert that their errors of approximation will be unsystematic and patternless – noise. Testing this can be valuable in revising the model. (Gelman and Rohilla Shalizi 2013, 20–21)

Gelman and Rohilla Shalizi argue that the central task of statistical analysis, namely the search for consequential errors by means of ‘severe’ testing, is not only a problem specific to improving statistics, but a philosophical problem. The necessary testing and re-calibrating of statistical models leads to a new understanding, notably of Bayesian statistics, which can no longer be seen as a merely inductive form of reasoning (where the inflow of new data continuously alters the distribution of probability), but where data and concepts are in the constant dialogue of a hypothetico-deductive mode of reasoning: where a hypothesized model makes certain probabilistic assumptions, from which other probabilistic implications follow deductively, via a detour of testing and recalibration. In other words, conceptualization of a model, deductions and their recurrent critical analysis, are at least as important as the incoming flow of data. For us, the philosophical relevance of the necessary testing of models is closely related to the problem noise. Noise, as a result, becomes a highly articulated concept in the context of statistics, with its own frame of debate. Information, in turn, remains much more difficult to define. The criterion of intention must be disencumbered from bias and loosened up, so as to allow for the accidental as a possible reason to shift the intentional framework. The distinction between what is accidental (noise) and what is intentional (information, data) thereby becomes the Gordian knot of a hypothesis and its recalibration: whereby the information of a poorly calibrated model can degenerate into noise, while what was a priori discarded as noise may become information. As a consequence, even the residual or negative definition of information (negentropy), which implies that information is simply what is left after noise is discarded, becomes a highly restless and fragile proposition.

III

Ready-Made Information

The distinction between what is accidental and what is intentional, which still plays a part in Weaver’s definition of noise as ‘spurious uncertainty’ (as opposed to ‘information entropy’ as desirable uncertainty), thus appears to be clear-cut, at least in principle, only in a highly formalized empirical setting, and only provided that the framework of its theoretical conventions and its fitness for purpose are accepted as a starting point. In turn, the telecommunications engineer deals with the distinction between ‘information entropy’ and noise as the consequence of a ready-made decision (the selection of a message to be transmitted as information). The principal task is to ensure safe passage of information – whatever its content – through noise: neither the selection of something to be transmitted as information, nor its evaluation upon receipt as informative, comes into the equation. Shannon explicitly defers the problem of the decision, according to which information is evaluated as informative and hence distinguished form noise. By separating the probability of its occurrence from the purpose of transmission and from its subsequent evaluation, Shannon has earned great criticism for his definition of ‘information entropy’. The first criticisms levied against this purely quantitative definition of information was, consequently, that it fails to take into account the heuristic criteria by which we qualify something as informative. For us, this suspension of the heuristic question is, on the contrary, the greatest epistemological advantage of Shannon’s definition of information. It is epistemological in the sense that – taken outside of the technical problem of noise in the channel of communication – it allows us to think about the shifting boundary between information and noise in the emergence of knowledge and its transformation. While most of what we ordinarily call noise in the statistical or scientific sense can be categorized as relatively predictable, according to its ‘normal’ statistical distribution, some events, as Scott warned, cannot: extreme or ‘non-

104

An Epistemology of Noise

normal’ values, chaotic or non-linear events can become highly informative, if we consider only the probability of their occurrence, untethering this form of extreme noise from pre-established criteria of pertinence that would discard it as irrelevant, because too far from the norm. By deferring the choice upon which the distinction between information and noise is based, Shannon renders a hiatus explicit that is tacitly smoothed over, when information is viewed as already endowed with a signification, as an already given datum or, in the plural, data. The artist Marcel Duchamp revealed this implicit conflation between what is simply given and what is already selected as meaningful, by radically questioning what is acceptable as a work of art. By presenting an open-entry sculpture exhibition in 1917 with a urinal signed with the pseudonym R. Mutt., Duchamp created a situation that rendered the implicit assumption of criteria of pertinence explicit. His riposte to the removal of the piece from the exhibition and to the indignation it caused, is here reproduced in his own words, as a brief question and answer that points to the issue that interests us when thinking about the difference between information and noise, namely the issue of freedom of choice in the artistic process, and in its reception: They say any artist paying six dollars may exhibit. Mr Richard Mutt sent in a fountain. Without discussion this article disappeared and never was exhibited. What were the grounds for refusing Mr Mutt’s fountain: 1 Some contended it was immoral, vulgar. 2 Others, it was plagiarism, a plain piece of plumbing. Now Mr Mutt’s fountain is not immoral, that is absurd, no more than a bathtub is immoral. It is a fixture that you see every day in plumbers’ show windows. Whether Mr Mutt with his own hands made the fountain or not has no importance. He CHOSE it. He took an ordinary article of life, placed it so that its useful significance disappeared under the new title and point of view – created a new thought for that object. As for plumbing, that is absurd. The only works of art America has given are her plumbing and her bridges. (Harrison and Wood 1999, 248)

By separating out the act of selection from the attribution of signification, Duchamp demonstrates the normativity of the artist – insofar as he can set new norms – and allows these to compete with the normative role of the art institution – insofar as the latter has a mostly conservative function of enforcing and reproducing existing norms. Duchamp thus diverts the attention from the

Empirical Noise

105

content and form of the work of art as datum, to the normativity that is in play in the production/selection, presentation and reception of a work of art. Duchamp’s ready-made thus enables us to question the problem of normativity which, albeit in a scientific and technological context that could not seem more remote from Duchamp’s concerns, also Shannon enables us to single out: what is transmitted as ‘information entropy’ is transmitted as raw material, isolated from meaning and value; both the choice that triggers the transmission of something as information and its evaluation upon reception imply a normative process. Only once these two normative stages in the information cycle are isolated can they become apparent as problems in their own right. Shannon thus illuminates what otherwise remains obscure in the emergence of something new as significant or otherwise informative: namely the normative process that must precede and conclude the distinction between information and noise. Shannon’s definition of ‘information entropy’ as ‘freedom of choice’ thus carries conceptual, theoretical and more generally cultural relevance well beyond the question of the purely mathematical evaluation of probability. What appears to offend common sense in Shannon’s definition of information is the brutality with which ‘information entropy’ is presented, like Marcel Duchamp’s urinal, as a brute fact, unprocessed by interpretation, denuded of signification, and of its greatest value when at its most unpredictable. Yet, by suspending the question of interpretation and evaluation, Shannon, like Duchamp, also leaves open the possibility that the constraints of interpretation may change, that what may be, under current interpretational constraints, discarded as useless, may on the contrary become highly relevant, if different rules of interpretation are applied. This normative aspect of the selection and evaluation of a contingent fact can now be posed as a condition of possibility of signification. Shannon’s conceptual audacity was to treat information as a raw fact, in all its ontological and epistemological nudity. Devoid of Boltzmann’s reference to physical reality, but also of Wiener’s utilitarian injunctions regarding organization or purpose, Shannon’s ‘entropic ideas’ help us to rethink information as a pure event of which we know nothing but the improbability of its occurrence. What information philosopher Luciano Floridi has called the most profound epistemic upheaval since the invention of the Gutenberg Press, must therefore be considered not only in light of the profound impact of new communication technologies (NTCs) facilitated by Shannon’s contribution, but also in light of its truly philosophical audacity (Floridi 2014).

106

An Epistemology of Noise

In light of this normativity, which characterizes the boundary between information and noise, it becomes an imperative to rethink the epistemic value also of noise. As virtual ‘freedom of choice’, understood as a speculative ‘maximum’ of uncertainty, noise is that against which the profile of relative information is drawn as a ‘relative’ uncertainty. The difference between information and noise thereby loses the aspect of its Manichean dualism, characterized by mental images of light and dark, rational and irrational, order and disorder, life and death. The relation between information and noise instead becomes a normative line drawn through the contingency of the empirical world, but also through the contingency of our mental constructs. If a cut persists, it is thus no longer pre-established and guaranteed by a moral fault line, but animated by what philosopher of mathematics, Jean Cavaillès, called ‘successive fractures of independence’. Whatever imposes itself as the ‘new’ norm for thought thereby claims for itself whatever imperial profile of the norm preceded it, yet one it is bound to concede to whatever will come as the next fracture. In a logic not unfamiliar to the logic of the artistic avant-garde, Cavaillès sees the rhythms of reason’s unfolding as cadenced according to these successive fractures of independence, [which] each time detach from what precedes it the imperial profile of what comes afterwards, necessarily and in order to surpass it. (Cavaillès 2000, 42)

In light of this conflictual and dynamical rhythm, also the epistemological line of fracture between information and noise is mobilized, becomes impermanent – in the empirical sciences even more so than in mathematics. Yet, as we have seen with regard to Lobachevsky or Cantor, even the history of mathematics reveals a continuously shifting ground of the certainties of established forms of knowledge. Shannon’s very open definition of both information and noise as a pure measure of uncertainty, thus enables us to think of epistemological normativity, meaning the emergence and perseverance of new norms of knowledge, as the act of separating general uncertainty into two kinds: uncertainty we estimate to be of potential use and uncertainty which we deem to be ‘spurious’, whose pursuit would be ‘futile’. Shannon thereby enables us to ask, as Duchamp did: on what grounds do we cut the empirical manifold in two, into what counts as information (or art) and what can be discounted as noise? While the engineer is not asked to assess the message s/he transmits, the scientist, the statistician and even the financial trader must, like the artist, continuously perform a normative

Empirical Noise

107

act in order to draw and redraw the boundary between relevant and irrelevant empirical contingency, testing and recalibrating running assumptions. Only on this condition can ‘a new thought be created’, to use Duchamp’s expression, for a reality that is never entirely divested of uncertainty.

IV

Cosmic Background Radiation1

The background noise of the unfolding universe can perhaps rightly be called the archetype of noise. As a consequence of its discovery the classical idea of a cosmos in perpetual equilibrium, alongside the Pythagorean idea of a harmonious ‘music of the spheres’, no longer provides the stable epistemic space within which Ancient Greek philosophy hoped to detach itself from myth and religion. The tranquil firmament for our ideas about the world was irreversibly lost by the chance encounter of isotropic rays during the 1960s. Arno Penzias and Robert Wilson’s discovery of isotropic rays, and their identification as background noise of the universe, roused the classical idea of cosmic order and equilibrium from its slumber with the ebbing murmur of a catastrophic origin of the universe. Cosmic noise henceforth becomes a reminiscence, of Proustian proportions, of a swarming ‘microgenesis of cosmogenesis’. In its light, or rather in its soundscape, the classical idea of cosmic equilibrium and our conceptions of order, in nature and in the structures of our understanding, radically changes. The perpetuity of cosmic order is henceforth thrown into a hypothetical scenario of ongoing genesis and perpetual metamorphosis: the stability of the cosmos itself is bent into the meta-stability of cosmic becoming (Morin 2008, 77). Its accidental discovery in 1978, initially as a persistent noise in signal transmission, serves here to exemplify the transition from the empirical problem of distinguishing ‘information entropy’ form noise, to the epistemological problem of the shifting boundary between the relatively uncertain and the absolutely uncertain. For, what at first presented itself to radio astronomers Arno Penzias and Robert Wilson as unwanted noise, interfering with the reception of chosen signals, subsequently revealed itself to be one of the most valuable sources of information in modern astronomy. The two scientists pursued the source of an irreducible radio noise that eventually paved the way for modern theories of the evolution of the universe:

110

An Epistemology of Noise

The cosmic microwave background radiation, considered a relic of the explosion at the beginning of the universe some 18 billion years ago, is one of the most powerful aids in determining these features of the universe. (Wilson 1978, 463)

The information potential of noise was of course not recognized at first and was, to begin with, treated only as perturbation of transmission, as nothing but noise. The task therefore was to identify the source of noise in order to eliminate it from the desired signal transmission. Yet this particular persistent background noise could not be attributed to any of the known sources of noise. What subsequently revealed itself to be of utmost importance for astronomy, for our understanding of the universe and our place in it, was thus at first an element of the greatest possible uncertainty: not a known source of noise, but pure noise. Penzias and Wilson’s persistence and openness to noise was rewarded with a Nobel Prize. This leads to an understanding of noise of literally cosmic proportions: everything that is not selected, is in principle considered as noise, the earth’s atmosphere, no less than the thermal radiation of all things and people around us, the noise of our activity, of our observation interfering with what is being observed. Rather than being generic, noise is thus always already stratified according to its different identifiable sources. Very rarely do we encounter a noise as pure noise: To measure the intensity of an extra-terrestrial radio source with a radio telescope, one must distinguish the source from local noise sources – noise from the radiometer, noise from the ground, noise from the earth’s atmosphere, and noise from the structure of the antenna itself. (Wilson 1978, 466)

In other words, what we consider to be information must be carved out from noise, which is expected to add nothing but ‘spurious’ uncertainty. But noise, in the process, must be attributed to a source or at least correlated with a known variable, and thus dismembered, dissected into to identifiable and nonidentifiable sources of noise: contingent though it may be, noise thereby specifies itself as a source of information about the empirical world. The analysis of what at first appeared as one mingled background noise and its attribution to different sources thus leads to an understanding of noise that is stratified, separating out the predictable sources of perturbation of signal transmission, which have already been incorporated into knowledge, from noise arising from an unknown source. In this case, the signal interference that could be attributed to no known source, that represented the highest

Empirical Noise

111

possible degree of uncertainty, is what became the object of greatest interest and the source of the most innovative information about the history of the universe. In La Méthode, Edgar Morin underlines the momentous transformation this implies for the classical conception of order and of the universe as the cradle of life, for Kepler’s ‘steady state’ and Newton and Laplace’s mechanical clockwork universe. His fast-paced summary of the profound epistemological upheaval during the past century and a half contrasts the ‘scenario’ of cosmogenesis with the classical conception of cosmic order: the hypothesis at the origin of the universe is that of a photon cloud dilating at a counterintuitive temperature of K, before granulating into electrons, neutrons, protons as it cools: collisions at still formidable temperatures force the nucleo-synthesis of deuterium, helium and hydrogen. Cosmic expansion is henceforth wrought with gravitational dynamics that put the expansive cloud under tension and even fissure it under the pressure of regional selfamplifying density, from which a ‘schismatic morphogenetic’ process ensues: the cloud ‘cracks, dislocates’ into proto-galaxies. These proto-galaxies in turn fissure and break up into gravitational assemblages, accelerating the localized growth of density to the point where the collision of particles provoke a nuclear chain reaction: in a titanic explosion, contained only by the equally titanic force of gravitational pull, a star is lit: a cog in the gravitational clockwork of the galaxy, industriously producing heavy matter from atoms that are forged in the kiln of the star, before it finally implodes or explodes (Morin 2008, 77). From the first principle of thermodynamics, which postulates the indestructible nature of energy, i.e. that there can be no loss of energy, to the second principle of thermodynamics that subjects the conversions of energy from mechanical to electrical or chemical energy to calorific degradation, the emphasis shifts from perpetuity of energy to the irreversible loss of the universe’s potential for transformation. Implied is a dialectic of order and disorder that now enfolds the classical idea of perpetual cosmic order in the idea of its becoming: from its chaotic beginning to its inevitable entropic death, based on Clausius’ assumption that the universe can be seen as a closed system vowed to entropic equalization of energy levels. The eschatological scenario of entropic death of the cosmos, however, runs into an aporia: the observable genesis of order. Galaxies form and even life emerges against the predicted entropic dissipation of energy. The negentropic capacity of the cosmos to pull itself together at all levels, from gravitational

112

An Epistemology of Noise

order to biological organization, defies the idea of the cosmos as a closed system vowed to entropic death. The pendulum swings back, from the catastrophic scenario of original chaos and inevitable entropic exhaustion, to order as the negation of entropy, implying the marginalization of disorder as background noise. The loss of the classical ideal of the perpetual cosmic machine is eventually compensated, not only by life but by our understanding of negentropic processes, and epistemologically, by the mastery of entropy in statistical mechanics. The maximization of thermal efficiency in manmade machines reverses the reversal of order that the understanding of entropy had first brought about, and re-installs the reign of order, this time as the mastery of entropy. A refined understanding of metastable systems eventually enables the articulation of entropy and the emergence of structure. And quantum physics, finally, provide the mathematical formalization of an uncertainty relation constitutive of matter itself. Cyberneticians and those working on Shannon’s information theoretical concepts begin to place entropy and noise at the heart of emerging theories of complexity, amongst which, to name but a few: the physicist and cybernetician Heinz von Foerster who discovers the principle of order from noise, based on the recognition of initial constraints; the mathematician, physicist and computer scientist John von Neumann, who introduces an understanding of self-reproducing ‘natural’ automata, functioning with disorder, and of course Henri Atlan, who incorporates Shannon’s definition of information in light of Ashby’s law of requisite variety into a theory of biological complexity on the basis of noise. Disorder, quantum uncertainty, entropic diffusion of heat are no longer seen as the mere negation of order, but interweave constraints and determinism with the indelible singularity of evental conditions. The emergence of structure can now be thought via the theory of metastable systems such as Prigogine studied them. The phenomenon of high molecular cooperativity under the effect of entropy, for instance, comes to explain the emergence of order from entropy in the form of hexagonal convection cells, observed by Bernard and generalized by Prigogine in his theory of systems far from equilibrium. René Thom’s catastrophe theory, in turn, provides a deterministic mathematical framework for thinking catastrophic bifurcations in dynamical systems, aligning mathematical rationality with the non-linear, and resulting in the conceptual monstrosity born from the necessarily unpredictable: deterministic chaos.

Empirical Noise

113

The classical opposition of order and disorder thus enters into a dialogue, an increasingly intimate cooperation of both notions. There is in fact a dialectical torsion of order and disorder, which, as Edgar Morin shows in La Méthode (Morin 2008), corresponds to the irreversible transformation and complexification of these notions, to a refinement of an initially coarse opposition between order and disorder and to the increasing interpenetration of both.

V

Noise in the Gap between Narratives

Morin’s epic of cosmogenesis and of the parallel complexification of our notions of order and disorder resembles, by Morin’s own account, an ‘adventure story with chance, suspense and drama’ (Morin 2008, 76). It is not this narrative of astronomic proportions, however, that is the item of interest for Morin, but the transformation of concepts and theories, forced by the breakdown of the classical notion of order. In one essential way Morin’s dilemma of drawing the line between science and fiction is thus analogous to the one faced by the writer Hans Magnus Enzensberger in The Short Summer of Anarchy (Enzensberger 1977), which relays Spanish revolutionary Buenaventura Durruti’s last summer. Enzensberger draws on a myriad of sources and of causal explanations, whose inequalities and tensions are left to create fissures in the dominant narrative. The peculiarity of Enzensberger’s account of Durruti’s controversy-surrounded death, is that he creates a discursive fugue, where historical sources, documents, interviews and accounts of Durruti’s contemporaries appear to repeat a common theme, threading together the same events, yet always from different viewpoints, resonating together to produce a riotous yet confluent discourse. Enzensberger’s reliance of a great array of sources, however, does not serve the purpose of consolidating historical facts by synchronizing the sources. He even derides the anxiety of fellow historians who fear a degeneration of historical discourse from historical fact to adventure novel. The peculiarity of Enzensberger’s approach to fiction is analogous to Morin’s Méthode, in that the seriousness of the historical account lies not where diverse narratives classically converge into a powerful discourse, but in revealing the friction that bristles between these narratives: The contradictoriness of forms only announces the fissures that run through the material itself. The reconstruction resembles a puzzle, whose pieces do not join seamlessly. It is on the joints of this picture that one must dwell. Perhaps it is in

116

An Epistemology of Noise

these that the truth lays in view of which, unbeknown to the narrators, there is narration. (Enzensberger 1977, 14)

History as a magisterial discourse, on the contrary, even the history of the universe, as Morin shows, is a tune that emerges from synchronizing the heterogeneous sources and discourses. The historian, no less than the scientist or statistician, is expected to normalize and synchronize the singularity of its sources. The authoritative tone with which a final narrative is traditionally presented as objective implies that we must pass a blind eye on the self-interest of each narrator, who uses, selects and shapes the raw material of empirical contingency into a streamlined informational flow. To this Enzensberger objects: What he finds is not mere ‘material’, unintentionally dumped, in pure objectivity, untouched by human hands. On the contrary, everything that you see here has gone through many hands, shows signs of use. (Enzensberger 1977, 16)2

The consolidation of a myriad of narratives into one discursive flow thus masks the murmur of inconsistency, the loss of precision that, like entropy, makes every process, also the discursive process, irreversible. Noise is relegated to the margins of scientific discourse, dispensable and finally cut from the narrative as mere error or imprecision. As a consequence, we can say that the form and hence limits of every scientific narrative or model are drawn against the backdrop of noise, cut out and lifted from noise with the surgical precision of a theoretical prism. Yet by cutting this narrative or model out from the empirical manifold, knowledge generates an excess that is then discarded as noise: a left over discarded on the other side of this epistemological cut through uncertainty. As mere off-cut, noise becomes the refuse, generated by the process of information itself – that which falls by the wayside, and whose negative connotation as epistemological refuse recalls the Germans call Abfall, and the moral connotation of apostate.3 This implicit moral connotation is what perhaps explains the anxiety of seeing scientific discourse contaminated with fiction. The excess that comes to haunt information as alien, as threat and as eliminable presence – epitomized by the metaphor of noise as parasite in the channel of communication – in fact reveals itself to be a function of the selection itself. If this selection establishes a norm, by drawing a critical line through empirical contingency, then noise is that which does not conform to the norm, implying a certain threat of subversion of the norm. To quote Jean Cavaillès once more:

Empirical Noise

117

[T]he empirical manifests its essential fragility in the unpredictability of its characters, in the illegality of sorts that it harbours […]. (Cavaillès 2000, 19)

Experience, but also the inferences we draw and the deliberate testing of our models, compel us to constantly redraw the line that separates information from noise, to reposition the cut according to which we select useful uncertainty as information and de-select ‘spurious’ uncertainty as noise. In the process we may be required to attribute value and accept as legitimate what was previously devalued and considered illegitimate. Analysing how the history of the natural sciences has been jolted into action more than once by phenomena that were previously discarded as marginal perturbation, how many an experimental perturbation has led to the recasting of scientific theories, Bachelard concluded that the very idea of perturbation […] will have to be eliminated eventually. One won’t speak any more of simple laws that are perturbed, but of complex and organic laws that are sometimes touched by certain viscosities, certain effacements. The previous simple law becomes a simple example, a mutilated truth, an unfinished image, a sketch copied onto a chalk board.

As the distinction between information and noise reveals itself to be subject to historical change, so the line we draw between information and noise becomes, like an artist’s sketch, a fine overlay, a vibrant impression transpiring the gist of the movement of thought itself. As the territories of our knowledge are redrawn according to new lines that divide information from noise, and the boundaries between theories shift or disappear, also the rules according to which we distinguish information from noise are challenged by experience or concept, and rewritten.

VI

Noise in Finance

Finance provides a setting that demonstrates, like few others, how valuable it can be to look at the metaphorical latitude involved, when the idea of noise is converted from a scientific concept to a schema of thought. Here the idea of noise is generalized in relation to the investment and regulation of financial transactions – without forgetting its origins in the common understanding of noise as unwanted sound or disturbance. The financial krach of 2007 and 2008 revealed itself to be tied to a semantic, rather than just financial speculation. The sophistication of financial products, and of mathematical models of prediction permitted a new esotericism to arise in financial discourse, deferring critical analysis of the internal dynamics of the financial sector to the presumed expertise of those in charge. And yet, the subprime crisis erupted apparently without warning and even in direct contrast with official prognostics, such as those famous last words of the International Monetary Fund (IMF), according to which ‘global growth should remain vigorous in 2007 and 2008’. The most common definition of noise in finance is that of unexplained price and volume fluctuations in general equity trading. It can be described by statistical models that are specific to time scales that vary from monthly or longer time scales, (often modelled as a diffusion process with a drift), to a ‘tick-by-tick’ resolution, following every up- or downtick in the price or volume of a market or security price, (unpredictable beyond a few minutes).4 The time scale is essential to understanding the dynamics of trading. As Eisler, Kertesz and Lillo argue, ‘the qualitative picture changes dramatically when one moves down to the resolution of individual transactions’. As a consequence, for example, the notorious highfrequency trading strategies, remain a challenge for mathematical modelling despite their notoriety (Eisler, Kertesz and Lillo 2007). Noise and its conceptual amplification into stock market volatility is in part blamed on such automatically executed logarithmic program trades, which are often designed as a hedge, in order to counterbalance the risk of potential losses

120

An Epistemology of Noise

due to market weakness. The algorithm triggers the selling or buying of large quantities of securities, when a predetermined price is reached. As targets are hit, the liquidation of large volumes precipitates further price falls, which may in turn trigger other stock liquidations, doing so at the precise moment in which the same programs also stop buying. This phenomenon was largely blamed for the 1987 crash, when stocks dropped by 22 per cent in a single day. Due to the risk associated with algorithmically induced market volatility, exchanges now limit the time window for program trades. The second characteristic commonly associated with noise in the stock market is irrationality. The so-called noise trader is defined variously as an amateurish investor, the gambling type, whose decisions are based on feelings rather than either fundamental or technical analysis. In other words, the noise trader responds to price fluctuations with the gambler’s instinct. Rather than basing decisions on knowledge of the presumed intrinsic value of a security, he/ she ignores the fundamental macroeconomic factors, such as general economy and industry-specific conditions, and microeconomic factors, such as company management and financial soundness. Simply put, noise trader designates someone who apparently ignores the fundamental data necessary in order to assess whether a security’s market price is over- or undervalued. The noise trader is generally believed to also lack the technical knowledge, which is associated with the purely mathematical forecasting of a security’s stock market volume and price movements, (often evaluated in and of itself on the basis of recognizable patterns, without taking fundamentals into account). This portrait of the noise trader as ignorant and purely impulse driven, however, is in some need of adjustment, as Alex Preda’s analysis in Noise: Living and Trading in Electronic Finance reveals: increasingly tech-savvy and mathematically literate traders are participating in computerized trading (Alex Preda 2017). In any case, the image that persists of the noise trader is that s/he is incapable of distinguishing patterns and trends from random fluctuations or noise. Unlike the institutional investor behind much program trading of large stocks, it is the large number of individual noise traders that is seen to contribute to market volatility, by introducing irrationality and accidentally amplifying random fluctuations into trends. The noise trader is thus not only subject to the alea of economic factors and random market fluctuations, but also seen to play a role in its amplification into a trend. As a consequence, the noise trader stands out, vis-à-vis the technical specialist, or the economist handling economic fundamentals, when consecutive financial crises are blamed on reckless speculative behaviour.

Empirical Noise

121

Blaming the noise trader for agitation of the financial markets, in fact, goes back to the origins of mathematical finance in nineteenth-century moral philosophy. Jules Regnault’s Calcul des Chances et Philosophie de la Bourse, published in 1863, first put the question of the financial markets’ morality on a purely mathematical ground (Regnault 1863; Walter and de Pracontal 2009). In it, Regnault opposes the natural and just self-regulation of the markets to anomalous market dynamics, blamed on the irresponsible behaviour of speculators agitating the markets and perverting the natural course of events. One could say, to use von Foerster’s expression, that Regnault paints the financial markets as the spontaneous emergence of order from noise, in which the healthy and just self-interest of the sound investor, who helps to build the economy, is opposed to the spurious speculator’s fall into personal and collective ruin, which is at the same time a moral fall from grace: The stock market is the temple of modern society: it is here that all the great interests of an eminently positive and industrial century are destined to converge; but the stock exchange is also the official sanctuary of gambling, where fortunes and existences founder. (Regnault 1863, 1)

Regnault goes on to oppose the ‘real’ causes of market variations in offer and demand to the ‘sterile movements’ of agitation that result from pure speculation, characterizing this difference by a strongly phrased moral contrast: There are thus two kinds or varieties of speculators: one […] a real parasite on genuine speculation […] is based on ignorance, cupidity, satisfaction of brutal appetites, all passions that engender and characterise gambling; it is shameful and a disgrace. The other […] has the talent to create, edify, transform, having as his only goal the utility for the common good; he corrects the exaggerated movements, which blind trust or senseless panic produce in the stock exchange, offers credit and maintains a constant equilibrium between the diverse values according to the utility of their products […] he cannot be praised and encouraged enough by all governments, for he is the true source of public credit. (Regnault 1863, 102)

With these words, Regnault sediments a moral dichotomy in thinking about finance, opposing on the one hand the legitimate, informed and rational investor, who is responsible for the normal function of the financial markets and contributes to financing the economy, to the agitator who acts in pure selfinterest, and whose unsound speculations causes the perturbations responsible for the abnormal functioning of the markets and ensuing financial crises.

122

An Epistemology of Noise

Regnault’s argument is that, rather than chastising speculators for the immorality of their gambling behaviour, it is far more effective to demonstrate that the random distribution of wins and losses eventually cancels out any advantage that the speculator may have obtained over time. In other words, the chaotic variability of a great number individual speculator’s decisions eventually convergences around a statistical average, which also represents the best approximation of ‘true market value’, while more extreme values peter out in proportion to their distance from the average. Although each one of the speculators’ decisions can be considered to be independent from one another, like the toss of a coin, future values always distribute according to the Gaussian curve, irrespective of the past. What good is it to blame individuals for their reckless behaviour, if all you have to do is prove them mathematically that they can’t win? Even though the scope and depth of probability theory and statistics in finance has changed dramatically since the nineteenth century, Regnault’s confidence in the normal distribution of wins and losses of pure speculation appears to have held up remarkably well: despite the fact that private computers have given millions of amateur traders the possibility of participating directly in trading, less than 2 per cent stand to make consistent profit (Preda 2017). Although the blame is still squarely placed on the immorality of the spurious speculator, Regnault in fact opens an avenue for thinking about market volatility in structural terms, by proposing a mathematical solution to what he perceives to be a moral problem. Probability, at this very moment, acquires the status of a diagnostic and prognostic tool for the financial markets, which the mathematician Louis Bachelier (Bachelier 1900) will further elaborate in 1900, in his doctoral thesis Théorie de la spéculation. As Christian Walter and Michel de Pracontal remark in Le Virus B, crise financière et mathématiques, Bachelier effectively lays the mathematical foundations for Regnault’s philosophy of finance, arriving at the mathematical expression of a particular case of the martingale, where probabilities are independent from each other and their variance within an interval of time follows a Gaussian curve. The mathematical properties Bachelier discovers, Walter and de Pracontal note, are equivalent with those of Brownian motion, which Norbert Wiener will later further define as a continuous time stochastic process. It is also during the period of this codevelopment of theory and technology that mathematical finance becomes a field in its own right. After a lukewarm reception in France and a period of neglect, Bachelier’s work was rediscovered during the 1950s by a group of future Nobel laureates in the United States. Paul Samuelson, professor at MIT, saw in Bachelier’s work the basis

Empirical Noise

123

for a solution to the problem of estimating the price of options, which obtained its full mathematical expression in 1973 in the Black-Scholes model. No longer a mere means of counting and tracking, financial mathematics from this point onwards becomes a factory of conceptual engines driving the financial markets. Brownian motion, Walter and de Pracontal argue, will eventually become the pithy mental image for fluctuations in the stock exchange (Walter and de Pracontal 2009, 44), and on which the now ubiquitous idea of noise is based, imagined as a form of static murmur in the regular distribution of variance. Markowitz 1952 model of portfolio choice, later completed by James Tobin, and the Black-Scholes option price model, developed in 1973 – all diversify and further sediment the idea of regular random distribution as a dominant paradigm (Walter and de Pracontal 2009, 44). Regnault’s argument that the normal functioning of the markets counterbalances the agitation of the speculators thus enters the mainstream of mathematical finance vial Bachelier’s work, and continues to have contemporary relevance also in relation to recent thinking about financial information and noise. Paul C. Tetlock, now Professor of Finance at Columbia University, indeed argued in 2006 that, although noise traders (conceived as agents with hedging motives or irrational reasons to trade) reduce informational efficiency in a securities market, their secondary effect is that rational agents counteract this increased uncertainty by trading ‘more aggressively on their existing information’ and by acquiring better information: For these reasons, two of the most widely used models in finance, Grossman and Stiglitz (1980) and Kyle (1985), predict that an increase in noise trading will not harm informational efficiency. In fact, if one allows informed traders to acquire costly information, the Kyle (1985) model unambiguously predicts that an increase in noise trading leads to an improvement in informational efficiency. (Tetlock 2006)

The moral condemnation of speculation thus persists, obliquely, in the opposition between information and noise, and in the opposition between genuine investors and noise traders. But it does so in a relativized form, where the reality no longer seems to be so clear-cut. Not only does information now emerge as a reaction to uncertainty, if not noise, but going one step further, Elena Esposito argues that [t]here is no difference between ‘objective information’ which changes expectations about fundamental variables, and pure noise, introduced by specific agents (noise traders) for purely speculative purposes. (Esposito 2011, 81)

124

An Epistemology of Noise

In fact, Esposito goes so far as to say that the very idea of perfect and uniformly distributed information is unrealistic, and that the explicit recognition of imperfect information is a better starting point for understanding financial systems. Underlying her and Tetlock’s revaluation of information and noise in finance is the idea of the financial markets as an information system that, like Wiener’s cybernetic system, has the capacity to self-regulate and compensate noise through feedback. The point Esposito makes, however, goes one step further: namely that there is also another factor. Alongside the traditional idea of noise as accidental perturbation and of information as compensation of noise, this other factor is the market’s capacity to observe itself. If information efficiency in finance is to be thought along the lines of a self-regulating system, it is thus better understood as a self-observing system, in analogy with a Second Order cybernetic system: The market provides a framework in which the operators can recognize themselves and their inclinations. This paradoxically contradicts the hypothesis of market efficiency because it is instead subject to a ‘“dynamic imbalance” that is neither efficient nor rational, and can be exploited in a non-random way’. (Esposito 2011, 66–67)

The persistence of an apparently clear-cut opposition between information and noise, as indeed between genuine investors and noise traders, thus grows increasingly out of sync with the insights afforded by Second Order cybernetics – whose relevance is even more acute in the technical sense, when one considers the impact of growing availability of new information technologies to individuals who want to trade. Alex Preda, interviewed by Mike O’Hara and Ricky Treadwell, in the HFT Review (High Frequency and Algorithmic Trading) observes that the so-called day traders, who emerged during the internet boom and were dismissed as noise traders, have not only gained greater access to the computerized trading of electronic markets (accounting for about 5 per cent of Forex’s daily value of $4.7 trillion), but are also increasingly educated in the natural sciences, mathematics or engineering, capable of producing, testing and using their own mathematical models. In other words, not only do so-called noise traders participate actively in producing and shaping financial information, they also participate in the strategic complexity of a self-observing market, whose agents can capitalize on the discovery of a competitor’s strategies and trading patterns, and by dissimulating their own:

Empirical Noise

125

I remember speaking with a screen trader not too long ago who was trading futures spreads, and he said he could always tell when the algos and the bots really start kicking in, because there were certain identifiable patterns of activity. He shaped his trading around that, to either step back or to trade in a certain way to counter what the bots were doing. (Preda, O’Hara, and Treadwell 2013)

The intricacy of mutual recognition, adaptation and self-dissimulation in the interaction between humans and algorithms, between individual and institutional investors, not only complicates the distinction between information and noise, it also complicates the question who is observing, and who or what is being observed? To continue blaming consecutive financial crises throughout the twentieth century on the ignorance and immorality of the speculator or noise trader, thus becomes inaccurate from today’s perspective, when hedging, program trades, institutional investors and individual punters all contribute to a complex informational eco-system. It is also blatantly ineffective when moral condemnation regains preeminence in the aftermath of a krach. Although the urge to blame and condemn was undoubtedly laid in the cradle of mathematical finance by Regnault’s moral philosophy, it is also he who argued, already in 1863, that moral condemnation has indeed very little effect on behaviour, in comparison to mathematical proof. In words that bring home the futility of the moral outcry after the subprime crisis of 2007, Regnault insisted that [m]orality […] to date, has never failed attacking the abuse of speculation […] (yet) it is not by abstract and odious declamations […] (and) empty words that one can hope to reform bad instincts […]. (Regnault 1863, 1–2)

Regnault clearly has no qualms over the designation of bad instincts, but his observation of the inefficacy of moral condemnation has lost nothing of its actuality today. In fact, the public outcry after the financial crisis in 2007 and 2008 led to a major public debate, calling for tighter regulation of the finance sector, the curbing of immoral bonuses, and an unequivocal condemnation of unbridled greed and speculation. Yet the moral imperative, so strongly felt in the aftermath of the crisis, has failed in adequately addressing and hence extirpating the root cause of exposure to financial crashes. Just this month the Bank of England was compelled to warn that the financial sector is yet again precipitating itself on a dangerous slope of easy credit, recalling the crisis of 2007 (Elliott 2017).

126

An Epistemology of Noise

In fact, it was already as soon as 2013 that a Financial Times headline warned of ‘Boom-era credit deals poised for comeback’ (Alloway and Mackenzie 2013). Despite this warning, Christopher Thompson remarks two years later in a Financial Times article that ‘Global CDO volumes have totalled $100bn in 2014, two-thirds more than the year before’, while also ‘Global volumes of synthetic collateralized debt obligations [CDOs] roughly doubled last year’ (Thompson 2015). Some will recall what role CDOs and swaps played in catalysing the financial crisis of 2007 and 2008, alongside the bonanza of unsecured debt.5 If we want to understand the inefficacy of moral indignation, it is worth drawing a rough sketch of the plasticity with which financial information branches out from these CDOs into increasingly protracted feedback loops, where financial information becomes opaque and viscous. Synthetic CDOs are essentially those structural financial instruments, which enable banks and investors to reduce their exposure to the risk of unpaid loans by selling the entitlement to repayment as a security – while at the same time investing in hedge funds that counterbalance the risk of such credit defaults, by betting precisely on the defaulting of these loans. Synthetic CDOs are famously divided into ‘tranches’ with varying degrees of risk and seniority. In order to promote their most senior tranches, which offer the lowest returns, banks can make them more attractive by creating a so-called ‘leveraged super senior’ tranche, which allows investors to pay only a fraction of its total value. Risk is thereby taken off the bank’s books, while in fact continuing to expose the same bank to the possibility of losses, should the market value of these products decline, such that contract clauses require investors to provide ‘billions of dollars’ of collateral (assets guaranteeing a counter value to the loan) or, indeed, ‘walk away’ (Alloway and Mackenzie 2013). A shadow banking system emerges from this market in synthetic CDOs, requiring the creation of commercial conduits for the increasingly protracted chains of debt obligations, so-called special purpose vehicles (SPV). SPVs are commercial entities set up by banks or other lending institutions, in order to provide short-term financing for companies or fund investments. Alongside set-up and running fees, these conduits profit from selling short term by generating funds with which to buy and then sell financial assets such as MBS (mortgage backed securities), CMOs (collateralized mortgage obligations) or CDOs (collateralized debt obligations).6 The gap between short-term borrowing costs and returns on long-term investments in debt derivatives (‘securities arbitrage’), in turn, opens up a niche, allowing hedge funds and banks alike to raise cash from so-called asset-backed commercial papers (ABCP), which can

Empirical Noise

127

be used to buy higher-yielding securities (‘What Are Conduits, SIVs and SIVLites?’ 2017). As banks are required to hold a capital proportional to the loans and mortgages they issue, selling their entitlement to mortgage repayment as a mortgagebacked security (MBS) to such conduits means these debt obligations are taken off the banks’ own balance sheets, leaving more capital available as leverage to provide new loans and mortgages to customers. In the meanwhile, Hedge funds wishing to increase their own returns can then borrow money from banks. As Lisa Abramowicz puts it: ‘banks are lending money to hedge funds to invest in derivatives that guarantee losses on loans held by banks’ (Abramowicz 2017). Not only does the sold-off risk not disappear, in Abramowicz’s words these individually crafted transactions look ‘a lot like the synthetic collateralized debt obligations made infamous amid the 2008 financial meltdown’ (Abramowicz 2017). This multiplying, shuffling and sampling of debt and promise, one could argue, leads to increasing information opacity and viscosity: it is not only difficult to see through, but also difficult to dissect the ramifications of liability. Risk assessment and market analysis is, of course, always already vulnerable to what one might call the ‘afferent noise’ of unforeseeable circumstances, both in the form political and macro-economic uncertainty, but also in the form of historical firsts, so-called ‘black swan’ events that cannot be foreseen on the basis of past experience. But the obliqueness and viscosity of financial information becomes a form of efferent noise, meaning a self-generated uncertainty, adding itself to the basic contingency of pricing on the trading floor (Ayache 1771; Taleb 2008). Although the mirage created by this vanishing act of exposure to debt is not what is usually meant by noise in finance, synthetic CDOs, together with the shadow banking system they generate, nevertheless add a dimension of uncertainty that is not essentially distinguishable from noise and which predictive models must strive to take into account. The unexplained price and volume fluctuations in general equity trading we commonly associate with the term ‘noise’, thus floats on a deeper uncertainty, carried by currents of ever more oblique channels of financial information that pulsate according to the cadence of the up-ticks and down-ticks of trading positions on the market – an information uncertainty, or to use Shannon’s term ‘information entropy’, whose complexity rivals that of noise, (even without taking fraud, cybercrime and fiscal crime into account). Blaming the immorality of speculation not only blatantly fails to bring about change through moral persuasion. What is worse is that it fails to understand

128

An Epistemology of Noise

the way in which these synthetic derivatives are tied to the double bind of tighter regulation and the demand to make more money available for economies blighted with austerity. The requirement for banks to increase their capital and reduce exposure to risk in fact means that regulators are compelled to sign off deals that enable banks to sell on the risk tied to the loans they issue to hedge funds, which in turn enables them to have more capital available to lend (Abramowicz 2017). Far from cowing under the moral pressure to reform the financing culture that led to the financial crisis, one could argue that regulators are in fact morally compelled to enable the ever more sophisticated distribution of risk, in order to generate liquidity and enable banks to fulfil their moral obligation of lending and investing in the economy. You may ask: what happens to these loans and derivatives of loans, and derivatives of derivatives, in a financial world that considers itself enlightened by the catharsis of the financial krach of 2007 and 2008? The answer is that the biggest investor base in European synthetic securitizations, after Hedge funds, are pension funds. This means that banks, regulators and politicians alike face a nearing ‘silver tsunami’, meaning an ageing population with exponentially rising care and health care costs (‘The Silver Tsunami | The Economist’ 2017), with the financial tools that can transform the debt of the young (from student loans, car loans, credit cards, mortgages, bonds) into liquidity for the old. Taken as one, the derivatives market is thus a veritable will to the future, both a will to power in quasi Nietzschean terms, in that it assumes the role of creator of wealth (worth $553 trillion in over the counter synthetic CDOs in 2015), and a legally binding pledge not unlike a testament, taking the form of futures, options and swaps. It is thus both a logos of the future, by means of mathematical models that underlie the pledged structuration of the future, and a nomos, procuring entitlement to a future carved up into varying tranches of risk, ranging from minimal uncertainty with low returns to high risk with potentially great returns. The immediate future is thereby transformed into a financial legacy that will be fulfilled after the contractual life cycle of each security. Each security, in turn, is a ratified, legally binding promise, not unlike a testament in which an imminent future will be inherited in tranches of risk. What effectively amounts to a monetarization of the future in fact constitutes a new paradigm – almost a New Testament – for the financial markets. Although investment is by definition an investment in the future, the expectation of future value could until recently still be grounded in its relation with the ‘real economy’. Real here refers to an economy tethered to an actualized set of relations of capital and productive forces. The derivatives market, in turn, capitalizes on the

Empirical Noise

129

virtuality of the future, whereby the actual becomes abstract, in the sense that the link between the virtual and the actual is no longer intuitive in so-called ‘complex’ financial products. In contradistinction with the New Testament, the fulfilment of this promise is no longer postponed to the afterlife, but to an always imminent future, continuously scattering value according specific temporal and legal modalities (futures, options, swaps, mortgage- or asset-based securities, etc.). What is more, the owned share of this promised future must mature before, not after the day of reckoning, or it will be void. The financial crisis of 2007 was such a reckoning, in which the promise of the future temporarily collapsed, as the noise of stock market volatility turned into a global krach. The bible’s New Testament supplants the old paradigm of the law with one of redemption through Christ’s ultimate sacrifice of himself. Every new financial crisis, in turn, reveals the stark choice between perdition and redemption in the form of public sacrifice. The uncertainty that could previously be blamed on stock market volatility triggered by so-called noise traders, thus increasingly resembles a flight into the future, ecstatic with the virtual wealth of futures, options and swaps, such that the German noun Rausch [ecstasy, intoxication, jouissance], is better suited than the technical term for noise [the verb Rauschen] (‘Global Economic Prospects | Data’ 2017). Regnault’s conviction that moral blame is ineffective in producing lasting change in behaviour has become inaudible before the colossal, almost sublime proportions of the risk of financial krachs. Yet what has persisted are the basic tenets of moral philosophy, which subtend the opposition of normal and abnormal functioning of the market, and the opposition of financial information and noise. The apparent paradox is that the moral condemnation of irrational speculation, inaugurated by Regnault’s moral philosophy, appears to go hand in hand with the mathematical models used, with their underlying assumption that the law of normal distribution stabilizes this virtual monument to the future, and ultimately cancels out the distribution of risk in the securities market. There is, in fact, a correlation between this moral dichotomy, opposing noise and information, and the reliance on the law of normal distribution as a paradigm for mathematical finance. This is one of the main arguments in Christian Walter and Michel de Pracontal book Le virus B – crise financière et mathématiques. The dichotomy of normal versus irrational financial markets is, the authors argue, precisely what perpetuates the repeated failures of predicting

130

An Epistemology of Noise

the financial crisis of 1987, the Asian krach in 1997, the dot-com bubble and more recently the 2007 subprime crisis, which unraveled in a global financial meltdown in 2008. In fact, Walter and de Pracontal argue that the dominant methods used for mathematical modelling in finance today, are still premised on this distinction between normal and irrational markets. In an interview with Le Monde, 28 March 2008, they quote Nicole El Karoui, founder of the influential masters in ‘Probability and finance’ at the Pierre et Marie Curie University, as saying that the current crisis is not a crisis in mathematics, because the probabilistic models used in finance to assess risk are ‘made to function in ordinary situations’ and not ‘“in periods of overheating, of bubbles,” during which “behaviour is no longer rational”’ (Walter and de Pracontal 2009, 16). And yet, alternative mathematical models exist and risk predictions have been made, yet could not be heard. Nouriel Roubini’s now famous assessment of the risk of an impending global crisis was not even included in the IMF report in April 2007 and his warning was discarded as ‘absurd pessimism’ (Walter and de Pracontal 2009, 17; ‘Transcript of IMF Seminar – The Risk of a U.S. Hard Landing and Implications for the Global Economy and Financial Markets’ 2007). In the words of Nobel laureate Paul Krugman, influent voices, in recognized publications like the Wall Street Journal, Forbes and National Review, ridiculed the ‘doom-sayers’. Alternative approaches to the analysis of risk were brushed aside as undue pessimism at a time of an unprecedented bonanza of debt, fuelled by the growth of the subprime market, whose value had increased from $35 billion in 1994 to $600 billion in 2006. The extent of risk was instead considered to be regional and, in the opinion of Ben Bernanke, then chairman of the Federal Reserve, limited to the possibility of slowed growth in the United States. What followed was the krach, which took down three of the five dominating banks on Wall Street (Lehman Brothers, Bear Stearns and Merrill Lynch) and consigned the other two (Goldman Sachs, Morgan Stanley) and the insurance company AIG to the most formidable state intervention in living memory, engulfing the global financial markets, shaking the economy of many countries to the core and provoking the worst and most widespread recession since the Great Depression. A krach is an event of such magnitude that it breaks the cyclical logic of economic rise and fall, acting instead as a tabula rasa. In other words, krach amplifies the very idea of noise in finance to the level of an unheard-of crisis. Charles Kindleberger and Robert Aliber for instance used the term already in 1978 in their book Manias, Panics and Crashes (now in its sixth edition), to describe the Austrian krach of 1873 (Kindleberger and Aliber 2011, 151).

Empirical Noise

131

Although the term ‘Börsenkrach’ is commonly used in German to translate financial crash, it is, strictly speaking, only the verb zusammenkrachen that signifies a crash in the sense of a crashing together of two or more elements, while the verb einkrachen signifies a collapse. While a crash designates a certain type of event, such as a collision or collapse (here the collapse of value attributed to financial products and shares traded in the stock exchange), a Krach in fact qualifies the subjective dimension of a traumatic noise associated with such an event. In German to have a Krach also means experiencing a relationship breakdown, a dispute with high levels of animosity – not a crash then, but a noisy agitation unravelling chaotically. Krach thus links the evocation of a rousing sound (or alarming realization) that is in itself potentially traumatic, with the idea of imminent, highly amplified discord: it would be wise to remember, when we use the word krach to qualify a major financial crisis, that the origin of the more common German word for noise, ‘Lärm’, like the English ‘alarm’, comes from the Italian call for arms: all’arme! Krach is one of several German words for noise – alongside Lärm (which designates acoustic nuisance), and Rauschen (which designates the murmur of common natural phenomena, like rustling leaves in a tree or the regular rumbling of rolling waves, but also noise in the technical sense of static noise or entropy). It is also the most emphatic word for noise, an onomatopoeia that not only designates, but expresses a crashing noise. Closer to the Dutch word for power [kracht] than either the English noise [etym. nausea] or the German technical term for noise [Rauschen – rustling or murmur], Krach designates the traumatic confrontation with a violently unpredictable reality, acoustic or otherwise, and its catastrophic unravelling. In short: there can be subtle noise, but not a subtle Krach. Describing the global repercussions of the so-called ‘subprime crisis’ that rocked the financial markets in 2007 and 2008 as a krach, as many commentators have done, thus adds one more dimension to our understanding of noise in finance. It emphasizes not merely a greater, more powerful and more unpredictable form of statistical variation as chaotic, non-linear, it shifts our attention to a qualitative dimension of trauma associated with the unpredictable. A global krach means not just more uncertainty with regard to the probability of estimating or speculating on stock value (or its collapse), but the experience of a traumatic implosion of the whole set of assumptions upon which the world-wide financial system is based. Rather than considering only the technical definition of noise, we must thus look slightly askew at the question of the mathematical abstraction of noise:

132

An Epistemology of Noise

we must ask ourselves how unburdened the idea of noise is by its sublimation into pure probability, both of the physical paradigm of entropy (implying ideas of loss of potential and of capacity to work), and of noise’s first and foremost socio-aesthetic, moral and political aspects. Rather than sounding out only the depth of noise’s mathematical formalization, a transdisciplinary approach to the conceptualization of noise in finance thus faces the task of analysing its multiple facets, from the anxiolytic (artificially calming) murmur of Brownian motion to the alarming screech of non-linear, chaotic processes of a crisis, and from rumour to krach.

VII

Statistics: The Discipline of the Prince

The dictionary defines information as ‘facts provided or learned about something or someone’ (‘Information’ 2014a). Some dictionaries, like the Merriem Webster, further consolidate this epistemological aspect of information as ‘the communication or reception of knowledge’ (‘Information’ 2014b). What these definitions presuppose, is an equivalence of information and knowledge, which is relayed also in the Penguin dictionary, where information and knowledge form a whole, composed of facts, data, signification and even ideas: Information/[…]/noun 1a knowledge obtained from investigation, study, or instruction […] b facts that represents data […] esp. regarded as significant […] 2 communication or reception of facts or ideas […]. (Allen 2001)

Given as a contextual reinforcement to this definition is a quotation by anthropologist Margaret Mead, which adds a moral emphasis to the idea that information should lead to the reduction of uncertainty by means of accuracy: I was brought up to believe that the only thing worth doing was to add to the amount of accurate information in the world – Margaret Mead. (Allen 2001)

Information thus requires trust in the obtainability of accurate facts and data, indeed information is presented by the Penguin dictionary as ‘B facts or data […] 2 the communication or reception of facts that represents data’ (Allen 2001). The dictionary definition thus builds a moral caveat into the definition of information, which is that of trusted facts and data, in the form of accuracy. And this moral requirement that information accurately reflects or indeed is ‘facts and data’ comes into its own in the juridical aspect of the verb ‘to inform’, which confers upon it also the power of: 3 a formal accusation presented to a magistrate. (Allen 2001)

Information, conflating accuracy/facts/data/knowledge, thus acquires a moral and juridical signification, whereby accuracy becomes the basis for legitimacy,

134

An Epistemology of Noise

grounding the just exercise of reason and power and safeguarding of law and order: Inform/in’fawm/[…] → verb intrans 1 to give information or knowledge. 2 (usu + against/on) to give the police or other authorities information about a criminal or crime. (Allen 2001)

Facts and data are thus considered to be a given that can be obtained and provided as accurate knowledge or suspected as false. The question what motivates our search and our suspicions, and by what means empirical reality becomes a given thus does not arise in the definition of information. It does not arise, because if facts and data are given, then their emergence is not a question of becoming, and the uncertainty preceding the emergence of facts and data does not come into the epistemological equation of information. Information as facts and data points to form and its accurate transmission, not to the emergence of form from the formless and their mutual transformation. The common definition of information thus has little room for uncertainty as partaking positively in what we call information. Uncertainty falls on the side of noise and excess together with inaccuracy, illegitimacy and illegality. Consequently, it is not surprising that in the dictionary definition of information also novelty, which necessarily implies a degree of uncertainty, is the least prominent feature and mentioned last. In fact, a single word for the relation between information and novelty suffices: c news. (Allen 2001)

‘News’ is thus, in the context of this definition of information, expected to serve a purpose, to consolidate knowledge, reinforce order and legitimate the exercise of the law. If these dictionary definitions accurately reflect the understanding of information in ordinary language, then information must be understood as facts and data that serve the consolidation of knowledge and law and order, hence consolidating established power through established knowledge, by reducing uncertainty, disorder and subversion. What follows is that noise is not, in fact, a straightforward opposite of information in the ordinary sense, since the opposite corresponding to the first and thus predominant definition of information as ‘knowledge obtained’ would have to be ignorance or uncertainty, rather than noise, which the dictionary defines first and foremost as ‘loud, confused, discordant sound, e.g. of shouting; din’. Noise is only secondarily defined in relation to the concept of information, and only in the context of the output of a computer, as

Empirical Noise

135

4. Irrelevant or meaningless information occurring with desired information in the output of a computer. (Allen 2000, 945)

The etymology of noise also prioritizes this socio-juridical rather than epistemic tension between information and noise: Middle English via Old French noise strife, quarrel, noise. (Allen 2000, 945)

On the basis of these different levels of definitions of information and noise, it is worth asking ourselves how it is that noise becomes the opposite of information, in other words how the ordinary connotations of noise as improper and immoral behaviour becomes the opposite of knowledge. To answer this question, it is not enough to simply state that the information/ noise couple is specific to computer science, since we have seen that two very distinct connotations cohabit in Wiener and Shannon’s definitions of information. Values thus come into play, even in computer science, communication technology and cybernetics more generally. The question that the discrepancy between Shannon and Wiener’s definition of information enables us to ask is: where do these values come from and how do they arrive at having such a determining impact on technoscientific concepts? It is worth looking for the answer to this question in the institutions of knowledge, where the activity that produces information, i.e. investigation, study or instruction, is consolidated under the authority to exclude uncertainty, and to expel noisy individuals and all those unreceptive to the communication or reception of accurate facts or data. Only thus can we understand how cultural and scientific conceptualizations do not constitute two separate spheres. To bring cultural and scientific conceptualizations of information and noise up to speed with each other may help us understand the problems inherent in the conceptualization of noise, and the function of both information and noise more generally in relation to established knowledge and power. The opposition between Wiener’s understanding of information as negation of entropy and Shannon’s definition of information as ‘freedom of choice’ is not the question of a mere point of view, but a problem deeply rooted in our need for prediction and control, but also of discovery of novelty. Let us briefly revisit how these roots intertwine in the history of the calculus of probability and statistics, from which we derive our contemporary understanding of information and noise. The historical origin of statistics indeed reveals how the objective of maintaining stability of power through knowledge becomes a culturally determining factor for the definition of information and noise.

136

An Epistemology of Noise

Statistics provide the data and facts that have become generally synonymous with information, for instance when we obtain news about the rate of unemployment or price indexes. It is also the discipline that has enabled Boltzmann’s non-classical definition of entropy and, consequently oriented Wiener and Shannon’s definitions of information and noise. This discipline comes, as Alain Desrosières reminds us (Desrosieres 2006), from the German Statistik. In its historical origin, the relation between knowledge and power is thus not only explicit, but fundamental and moreover foundational, since statistics first designated the ‘science of the state’, and was institutionalized as such during the seventeenth century by Hermann Conrig (1606–1681) who called it the: ‘nomenclature of knowledges necessary to the Prince’ (Desrosieres 2006). As such statistics were first of all the sovereign’s privative domain of knowledge. It is only from the 1830s that the domain of statistics was opened up to be accessed by ‘enlightened men’ (Desrosieres 2006, 1019). Before being reduced to the numeric data we now associate with statistics, the discipline of the prince was broad ranging, comprising history, law, political sciences, economy and geography, all classified according to Aristotle’s logic into material, formal, final and of efficient causes. It is when John Gaunt (1620– 1674) developed Conrig’s system of taxonomy into a new method of ‘political arithmetic’ that the mathematical foundations were laid for what we now call statistics. Gaunt’s technique of transforming parish registers of baptisms, weddings and deaths into facts and data relevant to the exercise of power, was subsequently systematized by William Petty (1623–1687) as the beginning of demographics. The knowledge statistics are now able to produce thus arises, first of all, as a by-product of state administration, detailing not only the number of births, deaths and marriages, but also of the number of crimes, epidemics, commerce, schooling, prison and hospital admissions. What we call statistics today thus emerges from this history of state administration, from the correlation it establishes between the knowledge and power. Yet this perspective would be one-sided, if one therefore overlooked that contemporary statistics also emerge from the merging of two very different, if not opposed forms of knowledge, by combining the ‘political arithmetic’ of statistics with the calculus of probability. Desrosières shows how these two very different disciplines intertwine historically and how great an epistemological obstacle had to be overcome in order to arrive at today’s hybrid of statistics and probability, which articulates the need to know with the mathematical measure of doubt (Desrosieres 2006). It is this tension that persists in the apparently

Empirical Noise

137

paradoxical definitions Shannon and Wiener give of information. And it is in the space of doubt that is characterized by this tension that we can find reasons for the new conceptualizations of noise as a culture of doubt. The historical origin of the calculus of probability in games of chance dates back to the sixteenth century with Gerolamo Cardano. It receives its mathematical foundation in the seventeenth century with the work of Fermat and Pascal, and Bernoulli and Laplace in the eighteenth and nineteenth centuries7 (Desrosieres 2006; Snell 2012, ii). Bernoulli’s 1713 law of great numbers demonstrates that a priori probability will be confirmed a posteriori, when events are repeated a sufficient number of times in experience. The ‘normal distribution’ of Bernoulli’s law of large numbers is also called the ‘law of possibilities’. Bernoulli thereby provides the mathematical basis for the ‘hypothetico-deductive’ method of probability and statistical frequency, laying the ground for the ‘experimentalinductive’ method of statistics. The ‘law of possibilities’ is subsequently interpreted by Laplace as an adequate representation of ‘errors in measurement in astronomy’ and re-baptized the ‘law of errors’. Amongst statisticians of the period, however, this new alliance between the purely theoretical calculus of probability and empirical, inductive evidence is by far not accepted as self-evident. The calculus of probability, as a purely formal, abstract theoretical method of a priori determination of a problem first appears too ‘subjective’ to statisticians, for whom the ‘frequentist’ observation of statistical frequency of an event provides the needed empirical a posteriori grounding required for reliable information. This distrust on the part of the ‘frequentists’ persists until the end of the nineteenth century, generating a material, frequentist ‘avalanche of statistics’ well into the 1930s (Desrosieres 2006, 1019). For two centuries the emerging discipline of modern statistics is thus taught between on the one hand the ‘subjectivism’ of the calculus of probability, which provides an a priori ‘measure of uncertainty’, and on the other hand the certainty provided a posteriori by the statistical frequency of events. Twentieth-century statistics thus result not only historically, but also epistemologically from the hybridization of the discipline of doubt and the discipline of the prince, combining the known uncertainty of a priori probability with the need to know, satisfied by the a posteriori frequency of events and the taxonomy and enumeration that formed the ‘nomenclature of knowledge necessary to the Prince’. This crossroads, where the measure of doubt meets the need to know, consolidating the law of large numbers with William Petty’s ‘political arithmetic’ (1623–1687), becomes the birthplace of modern demographics.

VIII

The Man without Qualities

The Belgian astronomer and mathematician Adolphe Quetelet (1796–1874) marked nineteenth-century statistics not only with institutional reforms, but also with his moral interpretation of the ‘law of errors’. If the average of imperfect measurements of a star give us a true idea of the real star, he argued, then the average of the empirical variation of the height of military conscripts, for instance, would give us a true idea of the ‘normal man’. Just as beyond the distribution of observed positions of a star there was a real star, so the Gaussian distribution of the sizes of military conscripts was the indication of a truth comparable to the reality of the star. For Quetelet the ‘normal’ distribution of heights of conscripts was proof of a constant cause, comparable to the real star that transpired through the average of astronomic measurement. The statistical average thus drew the portrait of man as a Gaussian ideal: the statistical average henceforth becomes the ideal form, the true idea (Desrosieres 2006, 1019). Not only the height of conscripts can now be measured against an ideal average or norm, but any contingent event. The average number of crimes or suicides for certain populations, can henceforth point to a real ‘moral propensity’, just as the average of fluctuations in the stock exchange becomes indicative of true market value in Jules Regnault’s philosophy of finance. The singularities we now associate with the idea of noise as ‘unexplained variation’ fail to constitute an ‘event’ in the eye of the nineteenth-century statistician, as they are subsumed and diffused into the necessary unfolding of the statistical tide of large numbers. So too the singularity of free will, blends itself into the propensities of the masses. What emerges is the ‘man without qualities’, whose singularity is washed away by the flow of great numbers, subjecting the individuality of human agency to the same laws as statistical physics: ‘Statistical physics’ were constructed upon the idea that erratic movements of microscopic particles, defined in a probabilistic manner, could result in macroscopic regularities, just as the average man was relatively stable and

140

An Epistemology of Noise

predictable, while individuals were volatile and unpredictable. A macroscopic determinism was compatible with radical uncertainty at the level of elementary particles. (Desrosieres 2006, 1020)

Francis Galton’s (1822–1911) concept of distribution eventually supplants Quetelet’s notion of statistical causality with that of partial causality, sought in the correlation between variables, in the co-occurrence and simultaneity of traits8 (Desrosieres 2006, 1021). Karl Pearson further formalizes this ‘table of contingencies’ during the 1880s and eventually renounces the concept of causality altogether: causality is no longer deduced even from strong correlation, nor from regression towards a normal average (‘regression toward the mean’), as one variable cannot be said to be oriented towards another. Pearson will thus reject causality as a purely nominal concept: reality henceforth resides in the subjective correlation, or as Hume would say, in a habit of perception. Even scientific laws, in this light, become summaries of such habits, codified routines of perception. Twentieth-century statistics inherits Pearson’s inferential use of statistics, which it increasingly uses together with probabilistic models to prove or test a hypothesis. The calculus of probability is thereby made to operate alongside statistical inference, articulating the conditions of a priori uncertainty with a posteriori certainty, bringing together the need to doubt and the need to know. The difference between the calculus of probability and the analysis of statistical frequency, in Desrosieres’ words, is now that between the ‘indeterminacy of the world itself ’ and the ‘protocols of observation’. The merging of these two distinct methodological aspects has also been called ‘the taming of chance’ (Desrosieres 2006, 1018; Hacking 1990). Yet despite inheriting Pearson’s epistemic humility, statistics have become the bedrock of public persuasion. The rhetoric of contemporary news media and government reports alike, and even the dictionary definition of information, call upon an almost infantile trust in facts and data – in other words in data as facts. These are represented not only as indicative of real causes, but often also imply Quetelet’s ideal of the statistical norm as indicative of truth and moral inclination. The epistemic humility of a priori uncertainty, imposed by the calculus of probability, no less than by Pearson’s inferential pragmatism, is thereby effaced in the common perception of statistical results as ready-made facts and data, which are taken to be indicative of causal determination. In the process, the function of a priori doubt and uncertainty is supplanted by a quasi-religious trust in the accuracy of the facts and data we derive from large numbers. The more data, the more truth, hence big data.

Empirical Noise

141

And yet, we have seen from Eisler, Kertesz and Lillo’s analysis, how important for instance time scales are for the statistical analysis of stock market variations, that unpredictability and dynamics are highly dependent on the degree of temporal resolution, and that, as a consequence, ‘the qualitative picture changes dramatically when one moves down to the resolution of individual transactions’ (Eisler, Kertesz, and Lillo 2007). This being just one of many parameters that make up the conceptual framework of statistical analysis, it should be evident how much knowledge is required before statistical data, and the appreciation of its accuracy, is no longer just given, but acquired. Although Quetelet’s interpretation of the statistical norm as revealing an ‘ideal cause’ eventually faded from statistical theory, it nevertheless left a mark on what we now perceive as normal or abnormal. Today’s dominant idea of standardized beautify is a sorry reminder of the soft power of Quetelet’s statistical idealism. But also the implicit moral connotations that still linger in our concepts of information and noise, as what is conform or divergent from expectation, as what is true in essence as opposed to corrupted and distorted by experience, still pay tribute to Quetelet’s numerical interpretation of the Platonic ideal. As Desrosières points out, the literature on the history of statistics is divided between distinct approaches that cover either the sociological analysis of its concepts and methods (Porter 1986), the history of institutions (Anderson 1988; Coll. INSEE 1987), the history of its mathematical formalisms, or the philosophical implications of this new discipline of knowledge (Benzecri 1982; Stigler 1986). This division of labour in the attempt to understand the full breath of the significance of statistics and its history gives an appreciation of the numerous fields of research that ought to be taken into account, if we wanted to understand the now dominant conflation of information with both data and knowledge. Our trusting reliance on statistical of data, consolidated in the dictionary definition of information, appears to owe a non-negligible debt to the conflation of these fields. The epistemic, legal and moral implications of statistics, acknowledged by the division of labour in its historical, sociological, formal and philosophical analysis, remain implicit in the notion of information and must be read between the lines of today’s dictionaries. Information is conflated with data, data is conflated with facts, both together become the medium for informing on or against. As data and facts become the bricks and mortar of the reinforcement of law and order, information is affected by a feeling of an ought to, in other words of a moral imperative, implicit in

142

An Epistemology of Noise

the act of informing. It is thus necessary to remain vigilant of the conflation of information and data especially in light of today’s culture of socially networked personal confessions, paired with the means for statistical data mining and hyper surveillance, which become all the more sinister when information is treated as a given, when data are treated as facts, and when information effectively eclipses uncertainty. We have the carelessness of rhetorical persuasion to thank for, if the era of ‘post-truth’ can fall back on the brandishing of statistics. It is worth remembering that what is a given at the origin of the modern discipline of statistics, from which we derive our notions of data and information, is not information, but power. What is given as the origin of Statistik is indeed the God given power of the Sovereign, consolidated by a form of knowledge and administration that becomes the discipline of the prince. This originally political power of information is distributed, anonymized and generalized throughout the rise of state administration during the nineteenth century. Statistical data is opened up for the first time to public administration and the enlightened public. The foundational role of power, as a consequence, is pulverized into a disinterested scientific discipline: in other words, the power of information acts incognito. Conversely when we now speak of noise in statistics, of error in signal transmission or in the logical chain of reason, we must beware of connotations that are as much cultural and political as they are technoscientific. The connotations implicit in the words used in scientific discourse act as a murmur of moral and political values. The value judgement levied against noise, when defined as ‘parasitic’ on information, thus expresses not only a measure of uncertainty, but also an implicit threat of insubordination to the norm.

IX

Noise Abatement: The Dawn of Noise

Of course, we no longer mean acoustic noise, when we speak about noise in statistical terms, be it in the stock exchange or in molecular biology. And yet, despite the complexity of thinking about noise in mathematical terms, the term noise, with its imagined inverted commas, has become increasingly current because it benefits also from its intuitive appeal. The common experience of acoustic noise remains strongly suggestive, even if we know that what is meant is not, for instance in finance, the din of the trading floor. The ‘special’ understanding of noise may no longer be reducible to the ordinary sense of noise as acoustic nuisance, but its empirical relevance nevertheless remains associated with the intuitive experience of acoustic noise. Noise appears to be the acoustic waste generated by industrial development, a side effect of the mastery of thermal noise in machines. The latter entailed the sudden presence of acoustic noise from machines, automobiles and, increasingly, the growth of entertainment and communication technology for an increasingly stimulus hungry modernity. While the general notion of acoustic noise may be at the origin of the noise metaphor in information theory, cybernetics, but also in finance and the empirical sciences more generally, ‘noise pollution’, in turn now appears as a side effect of technological innovation and industrialization. The two noise discourses, statistical and acoustic, henceforth develop side by side, yet surprisingly without much dialogue: while the humanities take stock of the effects of acoustic and visual noise on society, Science and Technology studies discover, alongside the visual design of new technologies, also the economic incentive for designing the sound new technology makes. Engines purr, domestic appliances bring a panoply of signalling sounds into our homes, while the ideal operation of an extractor hood or vacuum cleaner is subject to decibel ratings. Transport, industry, machines, but also the presence of technology in our homes makes itself heard, producing a cacophony without orchestration.

144

An Epistemology of Noise

An author whose analysis of the condensation of the scientific, technological, aesthetic, moral and political dimensions of noise is, in my view, exemplary is Karen Bijsterveld. Although noise has an impressive bibliography at its heals in historical, cultural and even sound studies, Bijsterveld’s articles ‘The Diabolical Symphony of the Mechanical Age: Technology and Symbolism of Sound in European and North American Noise Abatement Campaigns 1900–1940’ and ‘City of Din: Decibels, Noise and Neighbours in the Netherlands, 1910–1980’ provide a remarkably rich entry point to the theoretical polymorphism of noise, which will help us to contextualize the contemporary problem of ‘noise pollution’ and of its quantification. Technology and science are not only what enables us to discern the formal acoustic properties of the sounds we call noise, but it is the very progress of science and technology that has irreversibly altered the soundscape of modern society. All the more surprising, Karin Bijsterveld notes, that there is what one could call radio silence on the topic of noise between the cultural theory of noise on the one hand, and Science and Technology studies on the other (Bijsterveld 2001). It might be surprising that such a ‘radio silence’ could long persist between academic disciplines that are implicated, respectively, in assessing the sounds of technological innovation and the sociocultural transformation of our soundscape. While cultural theory takes into account the symbolic values and sociopolitical significance of noise or silence, Science and Technology studies focus on functionality and design of new technologies, until recently privileging visual characteristics over acoustic design and more generally privileging issues of sound engineering over an understanding of the wider acoustic resonance of modern technologies. Yet the absence of dialogue between Science and Technology Studies and the cultural analysis of the affective and symbolic dimensions of noise can only be detrimental to both. It is a silence that should be broken, since the sound of technology not only tunes our sonic environment, but has also been a highly controversial aspect of technology loaded with symbolic significance. (Bijsterveld 2001, 37)

The affective and symbolic charge, associated with the noise of modern technology, co-determines desirability or rejection of innovations. Road and air traffic noise; car alarms and ambulances; the ubiquity of mobile phones ringing; fragments of private conversations dissipating into public space; call waiting lines with digitalized Mozart tunes; shopping malls resounding with a cacophony of ‘ambient’ music: if the noise of industrialization could arouse

Empirical Noise

145

futuristic excitation at the beginning of the twentieth century, the postmodern soundscape has become a delirious imposition. Noise becomes an interdisciplinary paradox: while the noises of modern life require orchestration, so does the knowledge about noise coming from the humanities on the one hand and science and technology on the other. As the enjoyment and rejection of the sounds of new technologies divides opinion, the apparently straightforward problem of noise pollution becomes a sociopolitical problem of power: the power to judge what is noise and what is not and the power to regulate noise. The divisiveness of value judgements regarding the noises of modern technologies accompanies the history of noise abatement right from its beginning. Technological novelties such as the telephone, radio and gramophone are greeted with enthusiasm about modernity, but also raise concern about the deafening presence of new technologies in social life: [People] wanted to listen to their radio, but some nearby electric machinery interfered with proper reception. They wanted to talk to each other, but people shouting into their telephones prevented them from doing so. […] radio was third on the list of the most reported sources of annoyance [while] the gramophone came up for discussion in Dutch local politics as early as 1913. (Bijsterveld 2003, 175–78)

In response to the perceived intrusion of the noise of modern technologies, arises the need to identify, assess and quantify the perceived rise in noise nuisance in the urban space. Audiometers (previously developed for hearing tests), are converted into noise meters and put to use by public law enforcement in London, Chicago and New York as early as 1926. Bijsterveld’s historical case study of the Netherlands brings us the delectable story of the collaboration between a scientist and an Amsterdam police chief, in developing the first portable noise measuring device, the ‘silent witness’ or ‘Silenta’. The Silenta will become the first legally valid means for the measurement of noise, recognized by the Dutch Supreme Court in 1939 (Bijsterveld 2003, 185). A ‘Silence Brigade’ is formed in the police force, and ‘four policemen equipped with a motorcycle with sidecar’ are attributed the task of ‘the hunt for decibels’ in the pursuit of traffic noise offenders. Almost like a comical staging of the scientific problem of measuring noise, the attempt to measure noise is immediately confronted with the noise arising from the method of measurement itself: the roaring of the police officers’ motorbikes hunting down narrow streets of Amsterdam and the silenta’s airstream sensitivity

146

An Epistemology of Noise

condemn the exercise to a first wave of public scepticism, as initial enthusiasm in the press quickly yields to disappointment (Bijsterveld 2003, 186). The tide of public opinion sways against the upholders of civil and acoustic order, whose scientifically objective noise assessment is ridiculed in newspaper articles, such as the following extract of an article in De Telegraaf’s of 7 July 1937: You snatch the phone and dial our silence dictator, Police Chief Bakker, and explain the case to him […] Hardly three minutes later, you can hear the silence brigade’s motorcycle rush into your street […] With them, they have professor Zwikker’s sound meter and indeed, true enough – this is too much needless noise. Yet whereas the noise little John produces exceeds the limit by 0.12, little Peter, his twin brother, exceeds the maximum limit by no less than 8.65cB. What is the brigade to do in such a case? Arrest little Peter and take him with them […]? (Bijsterveld 2003, 187)

In the process, it is not only the image of the Silenta’s scientific accuracy that is tarnished. More importantly, it is the moral high-ground associated with purported scientific objectivity that is now subject to ridicule, as the Silenta comes to represent not progressive technology but a reactionary attitude to modernity: a symbol not of technological innovation, but of conservative reticence. As the scientific challenge of measuring noise in view of its regulation increasingly meets the rise of leisure occupations, associated with loud sound, so the technological problem becomes a moral and aesthetic debate over the attempt to curb noise – which soon takes on the political dimension of selfdetermination. The right to protect one’s acoustic space from the noise of others is confronted with the claim to the right of one’s own acoustic presence. The debate dividing Dutch opinion on noise in fact dates back to the beginning of the century, when new technologies such as the radio and the gramophone are blamed across Europe and the United Sates for the increase in noise. Mass media stand accused and yet are called upon to instil acoustic manners by teaching a ‘noise etiquette’ through public education – requiring the radio for instance to advise listeners to turn down the volume at a certain time of the evening. Combatting noise becomes a problem of redressing the moral decadence associated with modernity. As Bijsterveld notes, [M]aking noise was thought of as barbarian, uncivilized, anti-intellectual and disruptive behaviour – in short, as a lack of self-control. (Bijsterveld 2003, 175)

In a backlash against the apparent moral high-ground of the assault on noise, political representatives of the working classes will in turn assert the right of workers to listen to popular music or to their favourite radio shows on their

Empirical Noise

147

new gramophones and radios, notably at night, because the long working hours in factories prevent daytime leisure and justify night-time recreation. The low cost of the gramophone makes it the ‘“musical instrument” of the lower classes’, whereby not only the decibels it emitts, but also the popular music played on it and the behaviour associated with it are seen as objectionable by the middle class: One did not hear gramophones, a critic claimed, in the city’s upper-class districts. (Bijsterveld 2003, 178)

A political rift thus arises at the heart of the noise dispute, which takes noise from the problem of quantification, via a detour of aesthetic and moral judgement, to that of political arbitration. Leftist council members rebuke the conservative accusation of acoustic ‘debauchery’ as being elitist, pointing out that that the workers have a right to a ‘sound culture’ of their own, and that ‘one could be equally bothered by lady singers, trombonists […] “maltreated” pianos [and the] “miserable lamentations” of concertinas’ (Bijsterveld 2003, 179). Interestingly, a first victory is won by the conservatives, on the basis of the difference between mechanical sources of music and acoustical music practice, offering the perhaps surprising argument of the latter’s moderation due to human fatigue: [A]fter an hour’s practice flutists, oboists and so on […] feel a need to do something else, like smoking a cigar, talking to someone or devoting themselves to some other study. They get tired and stop playing and therefore they cause les hindrance, nuisance, harassment and irritation. (Bijsterveld 2003; ‘MA Rotterdam, Proceedings City Council Rotterdam, Session 26 February 1914, 64–65’, n.d.)

The argument succeeds in passing an ordinance in Rotterdam with a vote won by 24 to 13, followed by an amendment giving the police the right to penetrate the private property of the noise offender against his will, when accompanied by a high official. A decade later this ordinance is adopted by the councils of The Hague, Leiden, Breda, Utrecht, Amsterdam, Maastricht and Groningen, countering the invasive nature of the noises and new technologies with the invasive power of policing and regulation. The conservatives ban the ‘stupid, machinelike’ playing of records and radios for ‘hours and hours without really listening’ (Bijsterveld 2003, 180). Yet the conservative reaction to noise appears increasingly reactionary in the eye of progressive members of the public and ‘the grounds on which the loud playing of gramophones and radios could be legally punished became narrower’. The continuing resistance of left liberals,

148

An Epistemology of Noise

social democrats and communists eventually leads to the obsolescence of neighbourhood noise regulations in general (Bijsterveld 2003, 180). Noise in the process becomes more than acoustic excess, it becomes the soundtrack of an emancipatory confrontation with dominant norms. Today’s porous regulation regarding neighbourhood noise, noted by the WHO report on noise pollution (‘WHO Guidelines for Community Noise’ 1999), is a direct result of this cultural and political divisiveness of the definition of noise.

X

Noise Pollution

The historical shifting of the boundary between what is considered a noise offence and what is considered a legitimate acoustic presence and civil right generates not only new social and political norms, but also creates a new culpability of transgression. Noise becomes the excess generated by the acceptance of a new norm. What is at stake here is not only the changing soundscape of modern society, but the divisive and decisive power of norms. While neighbourhood related noise fails to generate consensus, growing awareness of the toxicity of ‘noise pollution’ means that the quantitative measure of noise remains an urgent topic. Acoustic noise, of course, is not a novel concern, tied only to modern technology, but one of the most ancient recorded objects of legal dispute. The Romans already imposed legislation regarding noise of ironed wheels hitting stone paved roads at night. It is with the advent of modern transport, however, that noise from lorries, diesel engines, aircraft, trains and industrial means of production becomes an indelible and increasingly toxic feature of industrialization and urbanization, recognized by the World Health Organization (WHO) as a serious cause of concern for public health: The extent of the noise problem is large. In the European Union countries about 40% of the population are exposed to road traffic noise with an equivalent sound pressure level exceeding 55dB(A) daytime and 20% are exposed to levels exceeding 65 dB(A). Taking all exposure to transportation noise together about half of the European Union citizens are estimated to live in zones which do not ensure acoustical comfort to residents. More than 30% are exposed at night to equivalent sound pressure levels exceeding 55 dB(A) which are disturbing to sleep. The noise pollution problem is also severe in cities of developing countries and caused mainly by traffic. (Berglund and Lindvall 1995, iii)

A side effect of economic development, noise thus presents itself as a luxury problem of developed countries: the noise of economic success. This view, however,

150

An Epistemology of Noise

risks underestimating the impact of noise on public health, an underestimation reflected by the lack of decisive action at the level of local, national and international policy. Noise, the report argues, must be recognized as an issue of unsustainable development. Defined in terms of its toxicity as a pollutant, ‘noise pollution’ is comparable in its toxicity with chemical substances, subject to cumulative adverse effects on health, not only in the present, but on future generations. In order to shift the lethargy in the perception of the noise problem amongst policy makes, the WHO report pitched the public health risk of noise in the only terms it thinks will rouse attention: namely in terms of the economic impact of noise pollution. Intensive urbanization of large cities in particular raises the problem of noise as a threat to productivity. The reduction of hearing range and even loss of hearing, due to noise, is an obvious example. But also other psychophysiological variables of the city dweller are affected by noise, threatening notably the performance of urban white collar workers: A major step forward in raising the awareness of both the public and of decision makers is the recommendation to concentrate more research and development on variables which have monetary consequences. This means that research should consider not only dose-response relationships between sound levels, but also politically relevant variables, such as noise-induced social handicap; reduced productivity; decreased performance in learning; workplace and school absenteeism; increased drug use; and accidents. (Berglund and Lindvall 1995, xviii)

Noise impairs the learning process of school-children, especially in the phase of speech acquisition, burdens communication in communal spaces with poor architectural design of reverberation and even the increase of aggressive behaviour in ‘predisposed’ individuals is considered a correlate of noise pollution. Also aesthetic considerations come into play, concerning for instance the report’s recommendation to preserve the acoustic serenity of conservation areas (Berglund and Lindvall 1995, vii). The report thus places the urgency of a quantitative measure of noise within a complex framework of its adverse effects, from the medical, as in hearing loss, to the sociopolitical, touching on issues of sociability, all of which become more relevant, it seems, in economic terms of productivity rather than wellbeing. The question is, how can such a complex set of variables be pinned down to a quantitative measure of noise? Is a quantitative measure of the ‘complex pattern of sound waves’ that makes up the urban soundscape adequate as a basis for policy making and accountability? Critics have been quick to warn that a purely quantitative measure of noise fails to address the complexity of the problem of

Empirical Noise

151

noise, unless it can take into account the competing interests in making and curtailing noise (Hydaralli 2012). While this sociopolitical aspect remains a controversial dimension of the problem of noise, as we have seen with Bijsterveld, the WHO report also stands out with a highly stratified methodology, taking into account not only loudness, but frequency, time variables, context and hearing range, differentiated enough to encompass a wide scope of environmental noise, comprising noise from road and rail traffic, air traffic, industries, construction, but also neighbourhood noise arising from the catering sector, from live or recorded music, from sports events, playgrounds, car parks and even domestic animals, such as barking dogs. Not only outdoor, but also indoor noise is listed amongst the sources of noise pollution; loud, but also anodyne noises are taken into consideration, such as the noise of ventilation systems, of office machines and home appliances, loudspeakers and headphones and architectural reverberation properties of buildings. Frequencies are measured alongside sound pressure and also the ‘signal-tonoise’ ratio is factored in, in order to assess impact on communication, basic tasks and performance, including sleep. Context- and task-related specificity also means taking into account the progression of noise over time: does it have a sudden onset with startling effect, or does it weigh continuously on daily activity, like traffic noise? Is its intensity stable or variable? A very quiet or low decibel noise and certain types of frequencies that barely register in the measurement of traffic noise, become highly significant in the signal-to-noise ratio, when they impede a conversation, disturb sleep or impair cognitive effort. Noise affects not only individuals but groups, for instance when poor architectural design results in reverberation in public buildings, impairing learning in schools or convalescence and even medical staff ’s attention to signals coming from monitoring devices in hospitals. Logarithms for ‘frequency weighting’ adapt the ‘combined sound energy’ over a period of time, according to different parameters. The logarithm LAeq, T, for instance, measures the energy average of continuous noise, such as road traffic, so-called A-weighted sound, over a specific period of time, weighting lower frequencies less than mid- and high frequencies, which are more noticeable in the human hearing. The type of noise onset and its intensity are measured against maximum noise level (LAmax) and sound exposure level (SEL). Frequencies, i.e. vibrations per second, measured in Hertz (Hz), as well as the loudness of sound pressure levels, measured in Decibels, are measured against the range of human hearing, situated between 20 and 20,000 Hz for unimpaired younger listeners.

152

An Epistemology of Noise

This logarithmic scaling takes into account not only context as a time variable, like night-time or daytime sounds, but also accounts for signal-to-noise ratio and degree of habituation in order to assess its impact on activity, be it sleep, communication or cognitive work load (Berglund and Lindvall 1995, vii). The quantification of noise is thus already armed with a highly sophisticated methodology, measuring additive sound events against a maximum tolerable to human hearing, and stratifying noise according to context and occupation. In other words, this method provides quantitative pointers for the qualitative tipping point at which our normal experience of the acoustic environment becomes toxic and is suffered rather than experienced. The quantification of noise is thus not a simple metric of loudness, but a tailoring of method to context as well as task specificities. While limitations of these measuring methods and their articulation are acknowledged in the report, both the economy and practical advantages of a standardized approach are argued to outweigh these. The difficulty that arises for us from the quantitative measure of noise is thus not the lack of complexity of this measure. It is clear that a methodology underlies this approach that can take many variables into account. As a result, however, also the definition of noise scatters over multiple contexts and acoustic characteristics. Noise is not necessarily loud, but undesirable across a large array of factors, involving acoustic properties, time, disposition and activity. This leads to a definition of noise as audible frequencies and sound pressure, whose toxicity must be classified according to both intrinsic and extrinsic conditions. As a result, the measure of ‘noise pollution’ leads to a taxonomy which, rather than giving a unified definition of the concept of noise, on the contrary pulverizes its object: there is not noise, but an open ended series of definitions of noise corresponding to an open ended series of contexts and tasks. The starkest contrast is to be found between the definition of noise in a purely statistical form of acoustic analysis on the one hand, and the analysis and taxonomy of ‘noise pollution’ on the other. In purely acoustic terms, noise has a very simple definition: Noise is an aleatory or irregular wave. In acoustics, therefore, noise is not an unpleasant or un-aesthetic sound, but a continuous signal in which no particular frequency can be distinguished. […] What we call white noise is a noise containing all the frequencies at the same intensity. (Volcler and Volk 2013, 10)

Empirical Noise

153

The concept of noise pollution, far from being simple, thus becomes polymorphous, moulding itself to different criteria and types of experience. As a consequence, as the method of measuring noise pollution increases in mathematical specificity, the concept of noise loses the simplicity of both its original intuitive appeal and of its purely statistical acoustic definition. Yet was it not either this intuitive appeal, or the trust in its mathematical formalization, which we presumed to underlie the ease with which the idea of noise has invested theoretical and experimental fields? What had appeared to cast the great conceptual arch of noise, from unwanted sound to statistical variation and even stock market volatility, was the simple idea of the undesirable, articulated in so many analogies of acoustic perturbation. What does the quantitative measure of ‘noise pollution’ have in common with the quantitative definition of noise in information theory and cybernetics? The obvious point in common is that both propose a quantitative measure of noise using statistical analysis. But ‘noise pollution’ and noise in the channel of communication are not just two different objects of scientific investigation. ‘Noise pollution’ measures and classifies noise as an object of experience (acoustic noise in the audible range of human hearing), according to its different logarithmic weightings. Information theory, as we have seen in Part One, measures both noise and information as a relation of probability. The two quantitative measures of ‘noise pollution’ and of ‘noise in the channel of communication’ thus share a statistical base, but they differ, apart from their domains of application, also in their epistemological priority. The definition of ‘noise pollution’ gives epistemological priority to the object (the phenomenon of noise with its various properties), which constitutes one term in a relation of perception, (the other term being that of individual or collective perception). The definition of noise in information theory, on the other hand, gives epistemological priority to the relation of probability between a signal and the set of all equally possible signals given certain constraints. Information and noise are constituted by this relation of probability. ‘Noise pollution’, too, pays attention to relation, by emphasizing context and task specificity – but it still measures the amplitude, frequency and duration of something that constitutes, and is categorized and weighted as, an object of perception: noise. The very term, ‘noise pollution’, designates a substance, a pollutant. The quantity designated as entropy in information theory, on the other hand, measures the relation of probability of an event, not its ontological characteristics. This degree of probability can also be expressed in terms of frequencies, amplitude or duration, but it is the relation with a set of virtual events of equal probability that

154

An Epistemology of Noise

characterizes both noise and information in information theory. The difference between ‘noise pollution’ and noise in information theory is thus a subtle shift of emphasis from object to relation, easily overlooked in favour of objective ‘similarities’ and the analogy with perturbation. The difference is thus not merely that both approaches deal with a different set of phenomena. Information theory does not define the object it measures, because it is precisely the ontologically arbitrary definition of a relation of probability from which Shannon’s definition of information and noise derives its wide applicability. This precisely is the problem posed to the definition of noise by the difference between the quantitative measure of noise (as sound) and the quantitative measure of information and noise (as a relation of probability). One measures noise as object of perception, refracting the definition of this object over an in principle infinite taxonomy of types of acoustic properties, with different weightings according to context and task. The other measures neither object nor content, but the relation of probability between our power to predict the actual occurrence of an event on the basis of what came before and the number of virtual possible alternatives occurring with equal probability within equivalent constraints. The difference between ‘noise pollution’ and noise in the channel of communication thus comes down to the difference between a theory of objects of perception, which constitute the terms of a relation between individual and environment, and a theory of relation that constitutes its terms as probable or improbable. This difference may seem irrelevant at this point, but it acquires all its importance if we recall Bijsterveld’s historical analysis, where noise becomes the function of a conflictual relation that crystallizes the sociopolitical terms of the relation, both in the form of opposing political groups and in the form of their respective soundscapes.

XI

Toxic, Viral, Parasitic

The prime concern for the WHO’s definition of noise’s toxicity as a pollutant is its impact on health. The most obvious toxic effect of noise is, of course, the impact of high decibels on the hearing apparatus. What surprises are the figures, which reveal the gravity and the commonality of the problem: Worldwide, noise-induced hearing impairment is the most prevalent irreversible occupational hazard and it is estimated that 120 million people worldwide have disabling hearing difficulties. (Berglund and Lindvall 1995, viii)

Hearing impairment is defined here as the increase in threshold at which sound becomes audible. Also tinnitus, a permanent ringing sound, may be triggered by high frequencies between 3,000 and 6,000 Hz, but even by frequencies as low as 2,000 Hz when exposure takes place over a prolonged period of time (Laeq, 8h). The consequences of hearing impairment are not only medical, however, but also and eminently, social. Impairment in the ability to understand speech, even at a reduction of hearing as small as 10 bB, becomes a ‘severe social handicap’ affecting communication and learning processes that involve language acquisition or recognition of speech. Not only exposure to extreme levels of sound, affecting the hearing apparatus, but even subtle noises that interfere with speech understanding contribute towards social handicap induced by hearing deficit, delaying language acquisition and impairing communication. Consequently, it is not only deafening noises, but even the slightest reverberation in poorly planned architecture of public spaces, especially schools and hospitals, where signal to noise ratios matter most, that have toxic noise effects, and where affected speech discrimination increases cognitive load, making ‘speech perception difficult and straining’ (Berglund and Lindvall 1995, ix). Guidelines concerning the control of noise pollution therefore concern not only exposure to decibels above 140 dB, to prevent damage to the hearing apparatus, but also the ‘signal-to-noise’ ratio in buildings with reverberation. The difference between speech level and surrounding noise should be 15 dB(A),

156

An Epistemology of Noise

but there is also reverberation time, which is desirable to be below 0.6 second, even in a quiet environment, and especially where vulnerable subgroups are concerned, such as children in the process of language and reading acquisition or those not yet familiar with the language spoken. Also hospitals, where the perception of signals from monitoring devices is vital and where delayed recovery of patients is correlated with disturbed rest, present an area where the signal-to-noise ratio is of the greatest importance (‘WHO Guidelines for Community Noise’ 1999, 11). What the impact of noise on learning highlights, is the effect noise has, beyond damage to the hearing apparatus and beyond the mere annoyance or disturbance that it ordinarily evokes, on the process of cognition and the wider impact on communication and participation in the social fabric. This aspect of toxicity de-qualifies any relativistic approach, that would have the idea of unwanted sound become a personal point of view. Nor can noise be reduced to a merely quantitative variable, correlated with impaired performance of school children and cognitive load of white collar workers exposed to multiple sources of noise – comparable perhaps to having to carry a surplus weight while working. Noise instead acquires an eminently psychosocial importance that pertains to hearing, but also to the way in which the social fabric can tear at the slightest lowering of threshold of sensitivity to sound. Although findings on the relation between environmental noise and mental health effects were as yet inconclusive at the time of the cited report, the WHO guidelines nevertheless stress the need to further investigate the scientific basis for understanding the relation between noise and mental health. The insufficiency of scientific evidence for the impact of noise on mental health is criticized as demonstrating lack of interest in the problem at the level of government, and this despite the availability of data concerning the measurable use of drugs such as tranquilizers, sleeping pills and hospital admission rates. The lack of scientific dedication to the matter is all the more disappointing, according the report, as statistical correlation between occupational noise and the development of mental disorders, such as neurosis, but also acceleration and intensification of latent mental illness, already exists (Berglund and Lindvall 1995, x). There is, of course, an established field of research into the psychology of perception of acoustic noise, which the WHO report does not mention for obvious reasons. However, this field has been fruitful less in cementing scientific proof of the correlation between ‘noise pollution’ and mental health problems, than in developing technological means aimed at the deliberate use of noise as deterrent, non-lethal and even lethal weapon.

Empirical Noise

157

By presenting noise as a toxic, but accidental by-product of industrial development and urban density, the WHO report in fact casts a blind eye on the development of technologies for the deliberate use of noise. These technologies aim at exercising a new form of control and at producing a more acceptable show of power than the now unpopular visibility of violent militarization. Research into the psychology of perception and the physiological and psychological effects of noise has indeed become a valuable resource for the development of military weapons to be deployed in situations where blood-shed is not politically acceptable, and of commercially available deterrent devices, aimed at situations where the visibility of such measures would be detrimental to the perception of a brand. What is striking that the scientific research and technological advances in the field of noise weapons appears to have no productive overlap with available scientific data on the correlation between noise and mental health. It is, on the contrary, characterized by the absence of public information and lack of access to independent experts. In Jürgen Altmann’s words: Acoustic weapons are under research and development in a few countries. Advertised as one type of non-lethal weapon, they are said to immediately incapacitate opponents while avoiding permanent physical damage. Reliable information on specifications or effects is scarce, however. (Altmann 2001)

Such technologies range from the extremely high power sound pressure, deployed with intent to injure, disorient, incapacitate and even kill, to the barely noticeable, nauseating effect of low and almost inaudible frequencies, used as a deterrent of loitering behaviour. The Long Range Acoustic Device [LRAD] (LRAD 2015), for instance, can be used to transmit warning messages at long distances, but can also be used with the intent of transmitting pain-inducing and harmful sounds in a 30° beam at 2.5 kHz. Its applications range from the deterrence of wildlife from industrial facilities to counter-piracy maritime law enforcement and crowd dispersion at public demonstrations and events. The LRAD is now widely used by governments as an effective means of control, even within their own territory and against their own population (Thomas 2012). In the commercial sector noise is not only used as alarm, but as a subtle deterrent device based on the diffusion of high frequencies audible only to young people, and aimed at dispersing loitering, anti-social behaviour and vandalism near shops (BBC 2008; Campbell 2008). Juliette Volcler’s Extremely Loud, Sound as Weapon (Volcler and Volk 2013) gives a thoroughly researched account of the military and commercial history of sonic weapons, spanning a period starting with the use of loudspeakers on

158

An Epistemology of Noise

the battlefields of the Second World War, used with the intent of deception and psychological abuse, through to the development of deafening infrasonic grenades during the 1960s and 1970s, and including the most recent developments emphasizing ‘non-lethal’ technologies as a means of bypassing public opinion on torture in the ‘war on terror’. ‘Sound cannons’ have seen deployment as crowd control and dispersal devices, in both Wall Street and Gaza, and were on standby also during 2012 London Olympics. Military deployment of sound, has long been made use of in fittingly called theatres of operation; notoriously in Iraq during 2003 and 2004, where trucks with loudspeakers broadcast ‘harassment operations’ that consisted in playing hard rock, heavy metal and rap for several days and nights on maximum volume. The dramatic siege of Fallujah in 2004 (during which the United States admitted also to using white phosphorous), the so-called ‘clash of cultures’ also took the form of an acoustic battle: US loudspeakers battled for dominance of the urban soundscape by broadcasting high volume AC/DC and Guns N’Roses titles, in response to which the mullah’s broadcast chants of Al-lahu Akbar and Arabic music. US military spokesman, Ben Abel compared ‘these harassment missions’ in urban settings with the disorienting and confusing effect of a ‘smoke bomb’ (Volcler and Volk 2013, 104). Also CIA interrogation techniques are known to have long relied on sound in so called ‘no-touch torture’, relying on ‘the capacity of sound and music to destroy subjectivity’ (Volcler and Volk 2013, 104; Cusick 2006; ‘White Phosphorus: Weapon on the Edge’ 2005). But as Volcler points out, using Axel Schafer’s expression: ‘hearing is touching form a distance’. Sound is a mechanical vibration, whereby a pulse is transmitted thanks to the oscillation of molecules or atoms, whose intensity is measured in watts per square meter, whose pressure is measured in pascals and whose amplitude is measured in decibels (dB), reaching the human pain threshold at about 140 dB, while respiratory problems become severe at 150 cB and frequencies between 50 and 100 Hz and inaudible, infrasonic, sounds (below 20 Hz) are deemed to be potentially fatal from 174 dB (Volcler and Volk 2013, 8, 14, 28). This, certainly, is where the analogy between acoustic and statistical noise breaks down, if the latter relies, as Weaver does, on the distinction between the intentional (information) and the accidental (noise). Wherever extremely loud or uncomfortable audible noise is used as a threat or even weapon, used with intent to injure or kill, the message is certainly clear: noise becomes a perverse form of information, in the sense of an imperative signal, while its incapacitating properties retain the characteristics of noise, not information.

Empirical Noise

159

As we have seen, Volcler specifies that sound, even when used with the intent to disturb or harm, is not equivalent with the acoustic definition of noise, which is confined to an aleatory wave with equal distribution of frequencies. In the specific context of defining the toxicity of noise, however, it still makes sense to speak of noise rather than using the generic term sound. Consequently, reference to Volcler’s insights into the military and commercial use of sound with intent to harm will feed directly into our problem of conceptualizing noise. The dominant idea of noise as audible disturbance, in the meantime, presents us with a serious drawback, if it neglects the non-audible range of acoustic events. It thereby fails to answer the challenge posed by the sophisticated use of noise in the defense industry and commercial security, but also by the experimental use made of noise in contemporary film, music and art. Although the difference between audible and non-audible noise may appear to be little more than a difference of degree or intensity, the consequences for the conceptualization of noise are worth considering. For what is lost, when the core conceptualization of noise is limited to the auditory range of acoustic events, is the full breadth of the physical phenomenon, comprising sound pressure levels and frequencies above and below hearing range, whose impact on health and cognitive performance is well known, relating sub-base frequencies that fall in the barely perceptible range of audible frequencies, as well as infrasound below 20 Hz, with states of anxiety and nausea, if not physical harm. Laura Wilson, for instance, analyses the strategic use of low frequencies in avant-garde cinema, with the intention of causing physical and emotional unease, provoking in the viewer what Wilson calls a ‘physical spectatorship’. The conscious processes of perception and cognition and their cultural coordinates are intentionally subverted by inaudible noise, imposing an involuntary physiological response that can be disconnected from to the visual spectatorship of the witnessed scene. Wilson gives the example of a rape scene in the film Irreversible by Gaspar Noé, 2002, where noise in the range of barely audible frequencies constitutes an onslaught on the cultural dominance of vision and hence, implicitly, on the role of the voyeur (P. Wilson 2012). The profound effect on the viewer of barely audible noise is here used deliberately as a critical and subversive technique, in order to high-jack the cultural dominance of vision, and of the male gaze in particular – meaning here not the male gaze as a form of self-perception of dominance in male spectators, or their potential identification with the rapist, but more generally the dominance of vision associated with a sense of power and control, which is culturally

160

An Epistemology of Noise

codified as a tacit assumption that this power is one of male dominance over the object of his gaze. The strategic use of non-audible noise is thus a way of extending the viewing experience beyond the predominance of seeing, but also of subverting conscious and culturally formatted perception processes. This subversion of conscious processes of perception in the service of creative practices, gives us a wider angle on the problem of noise than that associated typically with the perception of unwanted audible sound, extending the latter to the full spectrum of soundpressure levels and frequencies and their effects on us. Also the critical limit between conscious perception of phenomena and the pre-conscious substrate of perception is called into question and rendered evident by its strategic artistic manipulation. However, the anxiogenic effect of barely audible and inaudible noise features not only in artistic practices – as a means of critique of our faculties and cultural codes of perception – but also in audio-visual media developed for mass consumption more generally. Artists, the defence industry and commercial interests in mass media alike, compete in the use of non- or barely audible noise, with awareness of its potential to override the rational and culturally codified criteria of perception. By targeting the preconscious substrate of perception, and manipulating the physiological and affective dispositions of those exposed, audible and inaudible noise becomes effective in bypassing the cognitive functions not only of individuals, but of groups and potentially of entire populations. Steve Goodman’s Sonic Warfare – Sound, Affect and the Ecology of Fear in fact argues for the idea of a ‘sonic ecology’, less in terms of noise pollution, than with respect to what he sees as a wholesale assault on perception, by means of technological mechanisms of fear production. In addition to the already mentioned use of noise as weapon or threat, and even artistic subversion of traditional codes of perception, Goodman draws our attention to the urgent need for a critical understanding of the way in which acoustic ambience in general is being manipulated – ranging from branding experiences to the induction of a general sense of unease, and even of fear or dread, notably by mobilizing the periphery of auditory perception or what he calls the ‘unsound’ of vibrational environments. Interestingly, he speaks of a transduction to describe the propagation of affective tonalities that modulate collective dispositions of fear and anxiety, and thereby potentially ready the ground for the reception of ideologies (Goodman 2012, xx):

Empirical Noise

161

As film sound designers know only too well, certain frequencies can produce an affective tonality of fear in which the body is left poised in anticipation, expectant of incoming events: every pore listens for the future. Just think of the uneasy listening of atonal or discordant sound, or the sense of dread induced by low- frequency drones. […] Unlike an emotional state, affective tonality […] envelops a subject […] short- circuiting […] attention and consciousness […] (Goodman 2012, xx, 34).

Goodman sees in the interplay of analogue and digital technologies the articulation of vibrational substance and information, a ‘sensual mathematics’, evolving into an ‘ecology of code and vibration’ (Goodman 2012, xix). At stake in the notion of a sound ecology is thus not the question of ‘noise pollution’ as the unregulated production of sounds in an industrial and post-industrial world, but rather the purposeful manipulation of audible and inaudible acoustic properties, the deliberate creation or distortion of ambience, with the intent of overriding rational cogency and conscious perception. Dean Lockwood comments on Goodman’s notion of sound ecology by describing it as ‘an affective capture of the materiality of the body [via a] sonic ecology of fear in which we are collectively controlled, kept on edge, by the cultivation of a chronic affect of dread’. (Lockwood 2012, 74)

We may or may not agree with Goodman and Lockwood on the shift from the evidence of a weaponized noise industry to their much more general hypothesis that appears to include a conspiracy to orchestrate large-scale affect-modulation through the targeted manipulation of our soundscape. However, what emerges clearly, is that noise is no longer only a question of reason, of ratio, in the sense of calculation, (as the statistical measure of noise suggest), but also points to a reality that engages pre-conscious levels of perception, which in turn may become the object of targeted manipulation of perception’s affective disposition. The well-worn metaphor of noise as parasite in communication technology here finds its almost physiological counterpoint. Goodman indeed insists on the epidemiological metaphor to describe the contagion of affect as a viral process that affects groups or populations. He describes the use, for instance, of jingling noises in advertising and its evolution into corporate sonic branding, notably through earworms, as an ‘affectively contagious radiation of sonic events through the networks of cybernetic capitalism’ (Goodman 2012, xix). It is no longer clear whether this manipulation of ambience and perceptive disposition counts as sound, noise or even information, (be it in the form of veiled imperatives to buy and consume, like earworms). What this stretching

162

An Epistemology of Noise

of the objective parameters of noise beyond the audible means for us, is that whatever noise is, it can no longer be understood merely as something added to the field of perception – be it intentionally or not. It is perception itself, rather than that which is perceived, that is subject to contamination. Although no clear definition of noise emerges from these new parameters, thinking about audible and non-audible sound in the context of intentional manipulation of the soundscape does have one important consequence. It means that our focus must change from what is perceived, to the act of perception. This act of perception is what is vulnerable – either to mechanical damage of the hearing apparatus or to the short-circuiting of conscious processes of selection and of pre-conscious physiological and affective dispositions. The conceptualization of noise is thus no longer limited to the classical philosophical problem of determining what we can understand of the reality of noise ‘in itself ’ or even ‘for us’. It is irreversibly contaminated by a political problem, which is the possibility of deliberate distortion of our critical faculties through noise. Goodman’s proposition to think in terms of a ‘sound ecology’, his recognition that sound and what he calls ‘unsound’ is correlated with individual and collective affective disposition, has implications that cannot be contained in the mere quantification of acoustic phenomena as audible or even inaudible noise. Noise is no longer merely a question of defining tolerable ranges of audible frequencies and sound pressure levels, of finding optimum signal-to-noise ratios to measure the effect of noise on communication pathways. It becomes an issue of vital self-determination, be it at the psychological level of affect, or, as we have seen with Bijsterveld, at the level of sociopolitical cohesion.

Notes 1 A part of this chapter is drawn from a previously published article (Malaspina 2014). 2 ‘[…] was er vorfindet, ist kein bloßes “Material”, absichtslos vor ihn hingekippt, in reiner Objektivität, untouched by human hands. Im Gegenteil. Alles was hier steht, ist durch viele Hände gegangen, zeigt Spuren des Gebrauchs.’ 3 apostate, mid-14c., from apostenai ‘to defect’, literally ‘to stand off ’, from apo- ‘away from’ (see apo-) + stenai ‘to stand’, (ww.dictionary.reference.com/browse/apostate). 4 A security is a tradeable financial asset that entitles either to a share or stock ownership in a publicly traded company, or the share of a debt repayment, (which

Empirical Noise

5

6

7 8

163

can take the form of a bond that entitles the owner to interest and repayment of a mid- to long-term loan), or a derivative such as a future, which commits to buy or sell an asset at a certain price at a certain date in the future (and thus rises or falls in value as real prices of the commodity fluctuate), or an option which entitles to the same right as a future, but without obligation to buy or sell. CDOs are derivatives of securities, so-called, because they derive their value from an underlying asset, such as entitlement to the future repayment on auto, credit card or mortgage loans (MBS, mortgage-backed security) or corporate and business debt (ABCP asset-backed commercial paper). Another type of derivative is a swap, where one asset (i.e. a debt, currency or interest rate) can be swapped for another, for instance to insure against the default of asset- or mortgage-backed securities (such as MBS or ABCP). Mortgage-backed securities (MBS) are a guarantee of repayment of a loan that can be sold as a financial asset, and from which collateralized mortgage obligations (CMOs) can be derived, which essentially bundle a pool of mortgage-backed securities (organized by date of maturity and level of risk, different principal balances, interest rates, maturity dates and rise of repayment defaults). As borrowers repay the mortgages that act as collateral on these securities, principal and interest payments are paid to investors based on CMO terms. The value of these financial assets fluctuates with interest rate changes, refinancing and foreclosure rates as well as house prices. CMOs are a subcategory of more general collateralized debt obligations (CDO) including mortgages, bonds and loans, but also CDOs derived from CDOs called CDO2. ‘Probability theory, Encyclopaedia Britannica’. Britannica.com. The height of parents, for instance, henceforth explains but does not determine the height of children: the average height of sons of a same father is a growing linear function of the father’s height, but the dispersion around this average is independent of the father’s height, whereby dispersion of heights of all sons is equal to that of all fathers (Desrosieres 2006).

Part Three

The ‘Mental State of Noise’

I

The Crossroads: Mathematical, Technical, Empirical and Subjective Noise

The problem of noise now reaches a theoretical crossroads, characterized by a dynamic correlation of scientific formalization, technological regulation, psychosocial convention and, not least, the act of perception (understood as an act of self-determination of criteria of pertinence, vulnerable to affective dispositions). This conceptual crossroads is relevant for anyone who wants to understand the transdisciplinary appeal of noise. It is what accounts for the ease with which the concept of noise has conquered general and specialized discourse in a great array of theoretical and experimental domains. It requires us to conjugate artificial and natural variables of noise, its digital and analogue dimensions, its rational and affective parameters, its conscious and pre-conscious perception, no less than individual and collective dispositions. The relation between cognition and the real it attempts to cognize constitutes the speculative dimension not only of the experience of noise, but of any scientific concept of noise. The problem raised by Goodman’s notion of ‘sonic ecology’ is that of recognizing within the process of cognition also the possibility of inclinations, of affective dispositions and hence also of the possibility that cognition is ‘contaminated’ by a pre-cognitive ground of experience. When noise is thought in epidemiological terms and presented as parasitic upon the conscious processes of perception of its host or, importantly, host population, then what we are potentially dealing with is a bio-politics of noise in the Foucauldian sense. The critical problem is thus not to determine noise ‘in itself ’ or even ‘for us’, but emancipation: because it concerns the power of judgement and the power of control over its pre-cognitive ground. We have seen that failing to expand the conceptualization of noise beyond the audible phenomenon means that the concept of noise is unable to encompass the full breath of the experience of noise, including its strategic use where noise below hearing range is employed as artistic device or, in the case of commercial

168

An Epistemology of Noise

and military applications, as a deterrent or weapon. What the use of noise below the audible range emphasizes, is the continuing relevance of the etymological twists and turns that have led to the contemporary notion of noise, which leads us via the Latin term nausea, back to the Greek ν α υ σ ί α [nausia]. It is this aspect that is most relevant for us to pursue, because it leads to the core of the conceptualization of noise. If noise can be argued to affect cognitive and pre-cognitive processes, then the conceptualization of noise touches on more than the quantitative measure of sound volume or frequency, no matter how sophisticated, on more than statistical analysis, and certainly on more than mere aesthetic appreciation or personal taste. Noise becomes a philosophical problem, when it has to be factored into the conditions of possibility of cognition itself. For it is hard to see how the conditions of possibility of rational thought can be engaged with, without enquiring also into the distortion of cognition. It is thus not only the impact of noise on cognition, as extraneous factor, but the role of noise within the process of cognition that is at stake. It requires, in other words, that we think of noise not as object of thought but as a variable within the process of thought. This correlation between noise and cognition, between noise as distortion of information and noise as a factor of the distortion of cognition, emerges as an important aspect of the conceptualization of noise. Any philosophical enquiry into rationality, human agency and collective self-determination must therefore arrive at an understanding also of the state of indecision and confusion associated with noise – a state to which information and knowledge are temporary and always fragile solutions. Any epistemological enquiry into the nature of knowledge, finally, must contend with the role of noise as lived ambiguity, indecision and error.

II

Internal Chaos, Terror and Confusion

Flectere si nequeo superos, acheronta movebo.

(If I cannot bend the higher powers, I will move the infernal regions.) Aeneid, Virgil In 1986 two colleagues at the Massachusetts Mental Health Center wrote an article entitled ‘The Concept of Noise’, pressing ahead where the WHO will years later still deplore lack of scientific investigation into the relation between noise and mental health (Sands and Ratey 1986). In it, then clinical instructor in psychology Steven Sands and Assistant Professor in psychiatry John Ratey, both in the Harvard Medical School’s Department of Psychiatry, set out to add a new concept to clinical assessment: that of the ‘mental state of noise’. The authors hoped to show that the concept of noise is pertinent not only in diagnostic and therapeutic terms, but as an epistemological category for research into mental health. In practical terms the article sets out to articulate a psychiatric concept of noise. It provides a starting point and incentive for research into the concept of noise in view of therapeutic measures of psychiatric conditions, and more generally as a conceptual contribution to medical epistemology. The idea of noise is thereby brought to the foreground of attention and theorized as a hitherto unrecognized common denominator of psychiatric illnesses. The authors suggest that nosology, or classification of psychiatric illnesses according to sets of symptoms, may change profoundly once the role of noise is conceptualized and taken into account as transversal to most mental illnesses. The article begins with the author’s proposition of a general definition of noise. Today, thirty years after the article was written, many a white-collar worker, with high cognitive load and at risk of ‘burn out’ (Schaufeli, Leiter and Maslach 2009), may recognize some aspects of the ‘mental state of noise’, whose definition is here still reserved for very ill psychiatric patients:

170

An Epistemology of Noise

Noise is a term we are using to describe a complex and distressing aspect of the bodily and cognitive experience of many very ill psychiatric patients. By ‘noise’ we mean an internally experienced state of crowding and confusion created by a variety of stimuli, the quantity, intensity and unpredictability of which make it difficult for individuals so afflicted to tolerate and organize their experience. Attempts to do so may only add to confusion and psychotic phenomena. (Sands and Ratey 1986, 290)

In order to arrive at a concept of noise relevant to the domain of psychiatry, the authors draw concentric circles around the initial definition quoted above, situating the concept of noise within ever wider limits to the theoretical context. The issue of the psychiatric aspects of noise is thereby raised in the contexts of developmental and evolutionary psychology, at the level of learning and memory, as well as affect and social relations. The authors’ references span as wide as urban noise studies, psycho-analytic considerations and even art theory. Towards the end of the article we come to understand the extent to which the authors’ interest in noise is guided by research into to the pharmacological containment of the effects of the ‘mental state of noise’, when placed alongside other therapeutic measures. This is the perspective that, in practical terms, orients their need for a more rigorous understanding of noise. The pharmacological control of noise is what finally articulates the problem of noise in light of the amplification of the response to ‘the mental state of noise’, via feedback in the nervous system. Referring to the nervous system as a self-regulating system with feedback here appears to place the idea of the nervous system’s homeostasis, (i.e. of selfregulation as the ‘internal milieu’ of a living organism), into the logical framework of the cybernetic theory. Cybernetics theorizes machines with self-regulation through feedback and extends findings from control theory in mechanics beyond the machine paradigm to the analysis of living systems with homeostasis. Yet despite the apparent closeness, noise is never defined by the authors in cybernetic or information theoretical terms. While the understanding of noise is clearly extended beyond the common understanding of acoustic noise, and is couched explicitly in more general terms of systems, there is only one mention of Norbert Wiener, who is at once acknowledged and dismissed by emphasizing the ‘different’ role noise plays in psychiatry. This difference, however, is stated, but not rendered explicit. Sands and Ratey’s perspective on the pharmacological containment of the ‘mental state of noise’ nevertheless testifies to an implicit cybernetic inclination towards the problem of noise in psychiatry, as noise ultimately becomes a problem of control of

The ‘Mental State of Noise’

171

physiological feedback and amplification of arousal in the nervous system. What we can glean from this fleeting, if uneasy, reference to cybernetics is the malaise towards a reductive mechanistic paradigm that has, thirty years after the publication of this article, become common currency in cognitive neurosciences (Wiese and Metzinger 2017).1 Cybernetic theory, before effectively gaining the allure of a paradigm that explicitly structures research, has thus gone through a phase of latency, where its claims are not yet accepted as authoritative, but where the theoretical field is readied through a semantic expansion, a subtle change in discourse. Sands and Ratey thus chart multiple dimensions of the conceptualization of noise, but stop short of an explicit reliance on cybernetic theory. Their malaise towards cybernetic reduction is perhaps symptomatic of the tension that arises from this multiplicity. What could at first give the impression of eclecticism, accidentally reveals the real philosophical stakes of the problem of noise: the non-reductive articulation of biological, psychosocial and pharmacological, but also technocultural perspectives. In the process this article becomes an opportunity for us to hinge the philosophical understanding concept of noise, on problematizing this theoretical field of tension. It is the difficulties the authors face in conceptualizing noise, difficulties arising from the plurality of perspectives they engage, that will inform us as much as what they achieve in this paper. These difficulties become symptomatic for all other attempts to conceptualize noise, insofar as they are situated at a critical junction. The conceptualization of noise here touches upon the core of its epistemological relevance. As it oscillates symptomatically between the perspectives of object and subject, it presents itself alternately as object of cognition and as a factor in the process of cognition. The difficulty of stabilizing the perspective on noise reveals something essential about the subject of the theory of knowledge. The experience of ‘self ’ and of control over one’s own experience is seen in a new light only when it is in crisis. It is when the sense of self and of control breaks down that some fundamental cognitive and pre-cognitive factors step into the foreground as a problem, while they operate silently in the background of cognition during ‘normal’ functioning. If René Leriche could famously say that ‘health is the silence of the organs’, Sands and Ratey draw our attention to the philosophical implications of the fact that mental health is the silence of our embodied cognitive faculties. As a consequence, their article helps contribute a new aspect to our understanding of the preconditions for rational discourse – and a fortiori for the possibility of rational discourse without error or ambiguity, in other words ‘without noise’.

172

An Epistemology of Noise

It is thus important to situate the relevance of Sand and Ratey’s article in view of this epistemological question of noise. Even if the new understanding they propose of noise as a mental state dramatizes the embodiment of perception, it is not a phenomenology of the acoustic experience of noise that we seek to enrich here with recourse the psychiatric notion of noise. Nor is it in view of an existential philosophy of noise that we call upon Sands and Ratey’s article, such that the individual becomes the core reference of the concept of noise or even of ‘epistemological noise’. However, this crisis will help us to think about normativity more generally, including in its collective dimension, at the edge of reason. The relevance of Sand and Ratey’s proposition for thinking about noise as a mental state is that of highlighting a blind spot in the modern theory of knowledge, which existentialism was not alone to address: our modern assumptions about rationality are built on the Cartesian presupposition of a coherent self. Kant helped us to specify this presupposition, by showing that rationality rests on the universal structure of apperception, preformatted by a set of unchanging transcendental a priori. Philosophy’s critical and self-critical method has long called the subject of rationality into question. It was notably in response to Hume’s scepticism that Kant sought to stabilize the foundations of rational thought. Marx, Freud and Nietzsche have each taken a brick out of the monolithic edifice of the classical subject of reason, followed by the structuralist analysis of linguistics, of anthropology and of Lacanian psychoanalysis, which, together with postcolonial theory, revealed the subject of modern reason to be a constituted subject, before being a subject constitutive of reason and of the world it thinks. These traditions of critique have arrived at an anti-humanist understanding of the subject – anti-humanist not because against the human, but against classical humanism, whose hybris loomed large against the background of the traumas of the world wars, of Stalinism and colonialization (Alliez 2017). Analytical philosophers like Wittgenstein and Feyerabend have rattled at the scaffolding of rationality no less effectively – only to find that the presuppositions of rational discourse and the legitimacy of consensus are far from set in stone. Our reading of Sands and Ratey’s article thus serves the purpose of revealing, through the tensions and contradictions that arise within it, some of the fault lines and paradoxes that the concept of noise reveals within the constitution of the rational subject. Far from dealing with noise as mere object of perception, we will see that noise is also what un-conditions the capacity to discern and evaluate the object of perception. Not just a state of confusion and indecision, noise is

The ‘Mental State of Noise’

173

also amplified by the panicked attempt to redraw the boundaries of the sense of self. To regain the sense of self as first object of cognition thereby becomes the precondition to reasserting its relation with other objects of cognition. At stake, in other words, are not the noises we perceive, but the noise of cognition constituting itself, against the always looming crisis of its dissolution. The mental state of noise is first defined by Sands and Ratey as ‘internal chaos’, even as ‘inner confusion and terror’. The feeling of being overwhelmed is correlated by the authors with psychiatric symptoms that range from a lowered level of adaptation, to various forms of psychopathology: perceptual distortions, impulsive actions, impaired functioning and increased physiological stress are only some of the coordinates of the mental state of noise. Sands and Ratey begin their analysis by listing classic studies in evolutionary and developmental psychology that position noise as a fundamental correlate of avoidance behaviour and threat. A series of empirical studies in evolutionary theory of perception indeed correlates the effect of bright lights and startling, loud noises, with avoidance mechanism observed even in unicellular organisms. The conceptualization of noise is thus placed on a par with basic evolutionary concepts, in the sense that all organisms, including unicellular ones, are said to display withdrawal behaviour in response to intense stimulation by light or sound (Sands and Ratey 1986, 290; Schneirla 1959). The avoidance of noise is thus posited as a basic, even evolutionary aspect of perception (Schneirla), which is then linked by the authors with theories of perception and developmental psychology of human infants (Watson and Rayner 1920). The negative reaction to noise is similarly posited as a basic developmental disposition. Sands and Ratey mention J. B. Watson and R. Raynor’s observation that human infants respond with fear to ‘loud noise, unpredictable shock, removal of familiar forms of support’ (referring to Watson’s controversial experiments with ‘little Albert’ (Harris 1979)). Starting with the rejection of startling loud sounds as a basic feature of the experience of noise, Sands and Ratey move on to refer also to D. C. Glass and J. E. Singer’s theory of stimulus-overload from the environment. Unpredictable or uncontrollable stimuli are identified in the context of urban noise as a social stressor, associated with the experience of ‘internal chaos’ and correlated with consequences ranging from personal distortions, impulsive actions, impaired functioning and increased physiological stress (Glass and Singer 1972; Sands and Ratey 1986). It is, however, not only external stimuli coming from the ‘outside world’ that come into the equation of ‘the mental state of noise’. The crux appears to be

174

An Epistemology of Noise

that the ‘mental state of noise’ arises as a correlation between environmental stimuli, which act as extrinsic stress factors, and the individual’s own, intrinsic capacity to deal with the complexity of cognitive tasks. Sands and Ratey here cite Norman’s 1968 study on the ability to ‘filter’ information for pertinence: while criteria of pertinence vary in relation to momentary interest (one’s own name, for instance, tends to remain pertinent at all times), individuals vulnerable to the ‘mental state of noise’ appear to have in common that the critical threshold of attention, which sets the criteria of perception, is inoperative or impaired. This critical threshold of perception is what cuts through the continuum of sense data. Where it appears to be lacking or defective in some individuals, they become vulnerable to mental disorders. It is, in Sands and Ratey’s words, as if patients vulnerable to the chaos of overloading that we are calling noise are always wide open. (Sands and Ratey 1986)

The image that is evoked here is one of a lacking boundary that would otherwise protect the subject from the excess of external stimuli. As if the vulnerable subject was a fortress whose drawbridge is always ‘wide open’. To a certain extent we could infer, from the use Sands and Ratey make of this notion of threshold, that such a filter of perception is cognition’s own pre-critical boundary, filtering stimuli according to pre-conscious a priori. These pre-conscious a priori act as a perceptive firewall, separating out not only superfluous stimuli, but thereby ensuring the very condition of perception: a stable sense of self. The excessive openness, defined as vulnerability to noise, is subsequently associated by Sands and Ratey also with the individual’s diminished confidence in the ability to organize his or her experience adequately. Sands and Ratey refer again to Glass and Singer’s studies, in which the stressful effect of external stimuli is found to be proportional to self-perception as ‘helpless’. The ‘mental state of noise’ is thus posited, in the first instance, not only as the result of the ‘crowding’ of sensations due a failed critical threshold of perception, but as noise in the sense that it is amplified by the fear of losing this critical boundary, in other words of losing control. The fear of disintegration of the sense of self is thus also a consequence of the inability to impose a critical limit. Loss of confidence furthermore implies the threat of losing a reliable sense of self. Loss of boundary, loss of a defined sense of self, loss of control, thereby emerge as a chain reaction amplified by the loss of confidence that leads to the heightened sense of vulnerability that characterizes the ‘mental state of noise’. However, it is not only the correlation of external stimuli and internal disposition, which becomes relevant to understanding the ‘mental state of

The ‘Mental State of Noise’

175

noise’, but the changes of external stimuli over time and the changes of internal disposition over time conjointly modulate the experience of ‘the mental state of noise’, potentially progressing from confusion and anxiety to what Sands and Ratey will identify as the ‘catastrophic reaction’. The range of behaviours Sands and Ratey associate with this ‘catastrophic reaction’ to the ‘mental state of noise’ may take the form of a wide variety of behaviours and psychodynamic processes, spanning from ‘boisterousness to fainting and passive weakness’, from ‘internal and social withdrawal to catatonia’ and ‘stereotypies’. Significantly, also ‘excessive orderliness’ is listed alongside other behaviours, as a form of behavioural and cognitive withdrawal from noise. In defining these behaviours as a ‘catastrophic reaction’, Sands and Ratey refer to early-twentieth-century neuropsychiatrist Kurt Goldstein. Observing soldiers returning from the First World War, having suffered brain lesions or shell shock, Goldstein noted that even ordinary situations were experienced as ‘catastrophic situations’. Brain damage and other types of physical and psychological trauma were seen as increasing the vulnerability, not only to noises commonly perceived as excessive acoustic stimulus, but even to ordinary situations. Goldstein paid attention also to the patient’s feeling of inadequacy, resulting from the radically decreased ability to deal with normal and ordinary experiences. Together with the feeling of being defenceless against the stimuli coming into the central nervous system, this feeling of inadequacy appeared to contribute to the anxiety of impending catastrophe: As soon as an excitation is felt that emanates from an objectively dangerous situation, a catastrophic reaction occurs immediately, all other adequate utilization of excitation is excluded and the ill individual appears completely closed to the world. (Goldstein 1983, 37)

Goldstein’s clinical observations led him to correlate this perceived threat of disintegration, not only of familiar patterns of experience, but of the individual’s sense of self, with the individual’s adoption of a ‘rigid attitude’, intended to stave off the state of mental confusion. This rigidity, in turn, was seen to aggravate the catastrophic experience, provoking ‘disorderly, disharmonious, defective performances’ (Goldstein 1948; Sands and Ratey 1986, 291). Sands and Ratey’s conceptualization of the ‘mental state of noise’ is thus referring to a mental state more complex than the idea of confusion and overstimulation appears to suggest, more multifaceted also than today’s attention to ‘information overload’. One of the factors of this complexity is the correlation of extrinsic and intrinsic noise factors. This correlation leads to a

176

An Epistemology of Noise

concept of noise that must encompass the nexus between both. This nexus, moreover, cannot be reduced to that between two states, but must encompass the correlation between two forms of duration: between the change over time of extrinsic noise factors (for instance, the difference between sudden onset, progressive increase or intermittence) and of intrinsic noise factors (progressive openness, disintegration of critical pre-cognitive thresholds of attention, erosion of boundary, of sense of self, progressive loss of control and loss of confidence). What makes this correlation more complex is that also a tipping point must be taken into consideration, where the correlation between stimuli and internal disposition abruptly lead to a catastrophic reaction, which rather than ending or even alleviating the ‘mental state of noise’ in fact amplifies it like feedback between a microphone and a speaker. What is required, therefore, if we understand the correlation between these factors as a correlation of processes with a duration in time, is to see the ‘mental state of noise’ as subject to a variable of variables, making the concept of noise subject to a function of functions, rather than simply a form of excess of stimuli, whose modulation is either extrinsic or intrinsic or a simple correlation of these two factors. Its complexity can thus be reduced neither to the unpredictable or intense nature of stimuli, or to an individual’s internal disposition, nor to a mere summation of both. The experience of noise must instead be understood as a stochastic effect of correlation between extrinsic and intrinsic variables, as a complex function of functions that conditions or deconditions our experience, rather than being reducible simply to an anxiogenic state of over-excitation, overstimulation or confusion. We could question in passing to what extent this subject, the subject of noise, is based tacitly on sociopolitical assumptions about male presuppositions of ‘self ’. Let us pause, for argument’s sake, on a sociopolitical aspect that is often neglected in scientific discourse. Is not, at least in evolutionary terms, the extremely low threshold of pertinence of a mother’s perception of stimuli the very condition of survival of an infant, soliciting response to the infant’s every minute face expression, to every change in the infant’s behaviour even during the mother’s sleep? Is not the perpetual attention of a mother, presumed always mentally available for her offspring, taken by psychologists and psycho-analysts to be the ground for the latter’s satisfactory development and insertion into society? A society that, in many parts of the world, politically structures childcare so that mothers must refrain from full insertion into the labour market and stay continuously available for their offspring, while men can afford a more selective attention threshold? Is not the very definition of a woman’s ‘self ’ premised on

The ‘Mental State of Noise’

177

being always ‘wide open’ to her children’s needs, to her partner and parent’s needs, to her community’s needs? It would seem then, that the mental state of noise is perhaps not a neutral criterion for the assessment of mental health in general, but that it is an implicitly gendered one, whose sociopolitical aspects are ignored as a matter of discursive convenience: the mental state of noise is perhaps a normal state of mind for many women who assume tasks with a high cognitive load, while assuming also the traditional role associated with child rearing. It becomes pathological or abnormal, when it impedes the traditionally male privilege of focus. The loss of confidence and of a stable sense of self may indeed be a more common experience for many women raising a family in modern patriarchal societies than is often aknowledged – indeed perhaps more so as career expectations increase the tension between the requirement for openness and the requirement for focus. Openness is thus more likely to be correlated with pathological aspects of the ‘mental state of noise’ when the prerequisite for survival in a capitalist economy based on cognitive labour is threatened: namely male focus and confidence. However, it is perhaps not only a question of gender, but also more generally one of care. Men and women alike can be subject to a requirement of ‘openness’ experienced as excessive in an economy with high cognitive load, notably in the medical profession. The latter appears to put also men on a spectrum of required ‘openness’ to a multitude of solicitations and others’ needs, comparable to what is traditionally the case for women. The prevalence of mental health issues associated with the medical profession, unsurprisingly, appears to fit some of the criteria for the ‘mental state of noise’, as indicated by a recent survey published in the British Medical Journal: ‘A recent Medical Protection survey of over 600 members revealed that 85% have experienced mental health issues, with stress, anxiety, low self-esteem, and depression being among the most common complaints’ (‘BMJ Careers – Doctors’ Own Mental Health Issues’ 2017).2

III

The Vicious Whir of Sensations

Sands and Ratey go on to compare the ‘mental state of noise’ in the adult to the ‘vicious whir’ of sensations supposed to be experienced by the infant, due to an initial lack of cognitive differentiation. Following Piaget, Sands and Ratey draw attention to the infant’s lack of cognitive differentiation, as a result of which neither stable representation, nor continuous memory can as yet serve as grounds for a stable sense of self. The low degree of cognitive differentiation and hence low capacity for integration of experience’s ‘component parts’ are deemed to expose the infant to an experience that can be compared, in Sands and Ratey’s view, to the excessive ‘openness’ of those who suffer from the ‘mental state of noise’. Both the infant and the mentally ill are thus compared in their vulnerability to a feeling of being overwhelmed and powerless in the face of unpredictable and uncontrollable stimuli. Sands and Ratey cite Phyllis Greenacre’s 1952 study on trauma, growth and personality in the earliest period of life, in which she defines the infant’s helplessness in coping with what she calls the ‘vicious whir’ of sensations, as the result of the immaturity of the sensorial and cognitive systems. Greenacre calls this basic form of bodily and psychic distress in the infant ‘pre-anxiety tensions’, which are compensated only by a ‘holding environment’ that facilitates the integration of early experience. This ‘holding environment’ in turn would act as a foundation for future organizational capacity (Greenacre 1952; Sands and Ratey 1986, 292). Greenacre and Piaget are thus called upon by Sands and Ratey, in order to compare the adult, in whom this differentiation is impaired or lost in the ‘mental state of noise’, to the infant’s initial deficit in cognitive differentiation. Unstable representation and the defaulting sense of self are thereby posited as affecting the infant and the mentally ill in analogous fashion, as both are seen as ‘trapped in experiential noise’, because the criteria for continuous representation, for stable memory and hence for a consolidated sense of self are inoperative (Sands and Ratey 1986, 295).

180

An Epistemology of Noise

Sands and Ratey cite, alongside Piaget and Phyllis, also an author contemporary to their work, whose observations suggested that schizophrenia can be associated with a process of ‘de-differentiation’ of discourse (Frosch 1983; Sands and Ratey 1986, 292). Now, how this research into the de-differentiation of discourse holds up to more recent understandings of schizophrenia is less important for us, than the fact that Frosch’s hypothesis serves to consolidate Sand and Ratey’s hypothesis, namely that the ‘mental state of noise’ can be associated with a form of regression, not only on the level of cognitive and sensorial development, but also on the level of discourse’s regression the infant’s undifferentiated stage of linguistic development. Sands and Ratey thereby place the entire logic of the ‘mental state of noise’ under the authority of Freud’s theory of regression: the return to the ‘vicious whir’ of sensations of the infant, to the vulnerability to chaos and tensions characteristic of an immature stage of neuropsychological development, and finally in terms of the de-differentiation of discourse. Suggesting that psychosis is related to the dedifferentiation of discourse, they also see in this process a possible explanation for Schachter and Arieti’s clinical observation that psychosis is associated with heightened ‘evaluative needs’. The idea is that, in response to the dissolution of the threshold of perception, of the sense of self and its discourse, the psychotic reacts with a ‘defensive searching for a “name” or label’ for their experience. For Schachter and Arieti it is the ‘basic need to understand one’s experience’, precipitated and amplified by anxiety, that provokes psychotic delirium as a ‘premature flight into meaning’: Schachter (1959) suggested that when the body is in a state of hyper-arousal, there arises in the patient […] evaluative needs – in other words, a defensive searching for a ‘name’ or cognitive label to help him explain and understand his bodily feelings [leading to the] premature flight into meaning [as] a source for psychotic explanations for sensations and associated affects that cannot be recognized and integrated. (Lemaine 1960; Sands and Ratey 1986, 292)

The idea of closure of discourse, through a premature precipitation of meaning, also recalls Goldstein’s observation of the catastrophic reaction, in the form of a rigid ‘methodical character’ and obsession with order, the attempt to create a focus and thereby stabilize the cognitive process. Yet despite attempting to stave off the ‘mental state of noise’ through the closure of discourse and methodical rigidity, the ‘catastrophic’ reaction fails to alleviate the tension and the catastrophic attempt to organize experience ‘may only add to confusion and psychotic phenomena’ (Sands and Ratey 1986, 290).

IV

Keat’s Negative Capability

Sands and Ratey go on to contrast the flexibility that, according to Piaget, is essential to the child’s development with the ‘catastrophic reaction’. Indeed, they will emphasize openness as an ‘essential human trait’, by referring to art historian Meyer Shapiro. Shapiro was known to have taken great interest in psychoanalysis and for having interpreted the conservative rejection of modern art as indicative of a loss of this ‘essential human trait’ of openness. He interpreted the conservative rejection of abstract modern art a classicist ‘reliance on details, for lack of a grasp of the gestalt’, which he compared to some pathologies resulting from brain lesions. Shapiro saw in the conservative ‘need for sameness’ the sign of an ‘impoverishment of an essential human trait’ (Sands and Ratey 1986, 292). Sands and Ratey thus make a more general statement about the catastrophic reaction to the ‘mental state of noise’ than the initial clinical context for the contextualization of noise appeared to suggest. Extending the psychiatric conceptualization of noise in some way to a conservative cultural disposition, Sands and Ratey go on to compare the child’s healthy attitude to what the poet John Keats called the ‘negative capability’, by which Keats meant the ability to approach new, strange or confusing material without prematurely resorting to an armour of pre-set attitudes or behaviours. (Keats 1958; Sands and Ratey 1986, 292)

This definition of a ‘negative capability’ as the opposite of both the conservative reaction to abstract modern art and the catastrophic reaction, however, also raises a problem in Sands and Ratey’s line of reasoning. Is there not an ambivalence between this emphasis on openness as an ‘essential human trait’ and the negative aspect of excessive openness, which first characterized the internal disposition and vulnerability to the ‘mental state of noise’?

182

An Epistemology of Noise

This ambivalence, in our view, is of the same nature as the one we encountered in part one of this thesis, when we found that ‘information entropy’ can be evaluated positively as ‘freedom of choice’ that augments the quantity of information while also augmenting uncertainty, or negatively, when information is on the contrary defined as negation of entropy, as negentropy, by pitting information as reduction of uncertainty against both ‘information entropy’ and noise. Much is at stake in this comparison because negentropy is also that which, according to Brillouin, characterizes the degree of organization of living beings. If one were to speak of a negative capability in the sense of negentropy, it would be the capability of negation of entropy, the capability in other words, of imposing constraints and critical limits and thereby negating openness as the threat of organization’s liquefaction. Did Sands and Ratey not at first suggest that the organization of discourse relies on the capability of imposing a threshold of attention, in order to ensure stable structures of representation, of memory and, as a consequence, of the sense of self? Keats ‘negative capability’ on the contrary implies not this form of negation, but the negation of this negation. It is worth going back to Keats’ letters in order to reveal just how radical the consequences of this ‘negative capability’ would be, in the context of Sands and Ratey’s article, if fully taken on board. The ‘negative capability’ is not only the ‘absence of an armour of pre-set attitudes and behaviours’, but very much the negation of a sense of ‘self ’. It is the capability of ‘being in uncertainties, mysteries, doubts’, of accepting ‘half-knowledge’ serenely because one ‘trusts in the heart’s perceptions’, negating only the closure and concreteness of an ‘irritable’ attitude that ‘reaches after fact and reason’ (Keats 1958, 193–94). In other words, Keats’ negative capability essentially negates the flight into meaning and the closure of the sense of self. If we give Sands and Ratey’s reference to Keats’ ‘negative capability’ its full weight in the definition of an ‘essential human trait’, then we must ask ourselves: what does it imply if the essential trait is essentially negative? The ‘negative capability’, for Keats, is far more radical than a tame liberal motto of refraining from preconceptions. Keats goes one essential step further than being merely ‘open minded’, he takes one step further into the abyss of reason – the ‘negative capability’ is essentially the courage of allowing the representative structures of one’s own ‘self ’ to dissolve. In his 1818 letter to Woodhouse, Keats opposes the poet to the virtuous philosopher, affirming that ‘[w]hat shocks the virtuous philosopher, delights the chameleon Poet’. The poetical character, according to Keats, relishes that he ‘has no self ’, ‘has no identity’ not even the ‘the egotistical sublime; which is a thing per se and stands alone’.

The ‘Mental State of Noise’

183

If the reference to Keats’ ‘negative capability’ is to have any weight, then the analogy between the ‘negative capability’, and the openness that Piaget sees as essential in the child’s development, needs a critical proviso: for, in contradistinction to the insouciance of a child at play, the ‘negative capability’ of the poet implies knowingly taking the risk of losing one’s mind, the voluntary regression of the sense of self to a dedifferentiated state: When I am in a room with People if I ever am free from speculating on creations of my own brain, then not myself goes home to myself: but the identity of everyone in the room begins so to press upon me that I am in a very little time annihilated […] I will assay to reach to as high a summit in Poetry as the nerve bestowed upon me will suffer.

There is thus a knowing risk involved in the negative capability which, when we read Keats, is more radical than that of merely overcoming preconceptions, because it makes strategic use of a process of de-differentiation of identity, which alone makes poetry possible according to Keats. While the child may be open by default, it happens upon an already structured environment (the ‘holding environment’ mentioned earlier in the article as foundation for the integration of experience). The poet, on the contrary, seeks out the absence of structure and pre-conception, where society is already structured and closed. Not only the dissolution of a sense of self is at stake in the negative capability, ‘not one word I ever utter can be taken for granted as an opinion growing out of my identical nature’, but also the capability of others to tolerate the poet’s withdrawal from identity (Keats 1818). It is not only the poet’s own anxiety, but that of his addressee, which is the object of Keats’s letter on the ‘negative capability’: I feel your anxiety, good opinion and friendliness in the highest degree. (Keats 1818)

The negation implied in Keats’ ‘negative capability’ in the service of artistic creation is thus not negation of contingency, as that implied in the notion of negentropy, but negation of the negation of contingency. It is negation not only of the individual’s own negentropic needs for organization and stability, but negation also of the negentropic needs of others for stable identities and thus stable intersubjective relations. If Keats’ negative capability truly represents the ‘essential’ human trait for Sands and Ratey, then we must acknowledge that it stands in contradiction with the individual’s ‘pertinence filter’ which Sands and Ratey cite in the beginning of their article as requisite protection of the individual from the excessive openness of the ‘mental state of noise’. The partitioning, regulation and judgement of

184

An Epistemology of Noise

incoming stimuli can only occur on the basis of pre-conscious criteria of pertinence, without which the individual would be flooded with stimuli, unable to discern and judge what constitutes information or noise. If we interpret this pre-conscious filter of pertinence as a ‘fire-wall’ of perception, protecting healthy individuals from stimulus overload and experiential ‘chaos’, then defining Keats’ ‘negative capability’ as an ‘essential human trait’ means that it is an essential human trait to risk one’s mental sanity. While Keats clearly thought that this opposes the poet to the philosopher, we cannot help but recall here George Canguilhem’s well-known vindication of the freedom of thought: The norm in matters of the human psyche is the vindication and use of freedom, as power of revision and of institution of norms, an assumption that implies, normally, the risk of madness. (Canguilhem 2000, 217)

The paradox at the heart of Sands and Ratey’s article on the ‘mental state of noise’ is that it refers at once to the negative function of a pre-conscious selective threshold of attention and to the negation of any such pre-established armour of perception, encapsulated in Keats’ ‘negative capability’ as an ‘essential human trait’. The ‘essential human trait’ thus appears to run counter to the evolutionary and developmental avoidance of noise. It implies, on the contrary, a voluntary vulnerability to the ‘mental state of noise’. By referring to art history and poetry, Sands and Ratey appear to get more than they bargained for, making the ‘mental state of noise’ a problem of culture that exceeds the evaluative and therapeutic objectives of the clinical context with which they set out. We are, at this point, in the paradoxical situation that openness, flexibility and abstaining from the critical and pre-critical faculty of negating contingency constitutes an ‘essential human trait’, that this openness is both requisite for cognitive assimilation during the normal development of the child and a fortiori for creativity in the adult, in the form or the ‘negative capability’. Yet on the other hand this openness implies also the absence of a critical faculty of negation (which we could call a negentropic function in the sense of organization, certainty and information), which is no less essential to mental health. What qualified as a description of the mentally ill, ‘always wide open; anything seems to satisfy their pertinence filter and they are thus prone for flooding’, thus becomes the prerequisite for the child’s normal development according to Piaget, for cultural progressiveness according to Shapiro, and

The ‘Mental State of Noise’

185

for poetic creation according to Keats – and even for philosophical thought, if the inclusion of Canguilhem is granted (Sands and Ratey 1986, 291). If one were to take this paradox to its natural conclusion, then one would have to say, with Canguilhem, that in order to maintain one’s health one has to risk one’s health.

V

Closure to Noise and the Paradox of the Declining Life

What remains to be answered, then, is the difference between a declining life and a thriving life. We have seen that openness of the as yet not fully differentiated nervous system in the infant presumes an external ‘holding structure’ that facilitates integration of experience’s component parts in the infant, alleviating the ‘pre-anxiety’ tensions caused by the ‘vicious whir of sensations’. A certain degree of closure thus appears to be a precondition the infant’s cognitive development. It is also what, in the form of a pre-conscious threshold of attention, keeps at bay a ‘catastrophic reaction’ to the ‘mental state of noise’ in the adult. There is a tension between the requirement for both openness and closure, and also a risk inherent to both openness, which may provoke the ‘catastrophic reaction’ and closure, which risks impeding the openness required by development and culture. This problem complicates Sands and Ratey’s suggestion that the ‘mental state of noise’ can be thought in Freudian terms as a regression. It is not stated clearly whether this regression is meant to explain the similarity between the excessive openness of the child and of the psychotic experiencing the ‘mental state of noise’, or the similarity between the behaviours that are implied to characterize the immaturity of the child (such as shouting, or withdrawal behaviour) and the behaviour that characterizes the ‘catastrophic reaction’. Yet what is clear, is that the infant’s ‘vicious whir of sensations’ is different from the excessive openness in the ‘mental state of noise’, first of all because the child is thriving on this openness, while for the mentally ill it is associated with a decline in health. There is thus a flaw in the comparison between the mentally ill and the ‘pre-anxiety tensions’ of the infant and, as a consequence, in the idea of regression: even if the anxiety provoked by the ‘mental state of noise’ in the mentally ill is comparable to the ‘pre-anxiety tensions’ Phyllis observed

188

An Epistemology of Noise

in the infant, the mentally ill cannot be said to regress to the infant’s normal disposition to thrive and develop. The tension between openness and closure, between progression and regression, between extroversion (i.e. excessive openness) and introversion (i.e. withdrawal) thus constitutes a fundamental difficulty that Sands and Ratey face in the definition of the ‘mental state of noise’. This tension is dramatized by the proneness to switch catastrophically from excessive openness to excessive closure. The authors’ acknowledge this difficulty to a certain extent. After comparing the mental state of noise to the infant’s ‘vicious whir of sensations’, Sands and Ratey refer once more to Piaget, in order to highlight the difference between the ‘catastrophic reaction’ and the ‘flexibility essential to adaptive functioning’ in the normal development of the child (Piaget 1999). The ambiguity appears to lie in the status of openness in the infant and child, and in the adult experiencing a ‘mental state of noise’. According to Piaget reflex and sporadic imitation serve the purpose of initiating a learning process in the infant, by acting as a structural basis for the consolidation of experience, upon which the child will be able to build the capacity for symbolization and representation: after the first year of life sensorimotor schemas begin, according to Piaget, to structure memory, which in turn enables the transition to the formation of conceptual schemas, enabling the assimilation of new stimuli and the accommodation of new experiences and new behaviours. Piaget’s schema of systematic imitation, formation of sensorimotor schemas, consolidation of memory and the formation of conceptual schemas, thus provides several strata of what we could call structural redundancy. If we rephrase Piaget’s notion of imitation, by interpreting it as a function of redundancy informing the child’s development (redundancy being achieved in the channel of communication by repeating a message or signal and used to counter-act noise in the channel of communication), then the child’s openness to imitate serves essentially a negentropic function. Imitation indeed appears to constitute an information process that, like negentropy, reduces ‘freedom of choice’. By laying the necessary structural conditions for sensory-motor schemas, for the consolidation of memory and the formation of conceptual schemas, imitation, like redundancy in the channel of communication, would appear to make experience more predictable, in other words, it enables the child to learn. In order to overcome the ‘vicious whir’ of sensations and create a basis for the capacity for symbolization and representation, the infant could be said to reduce the ‘information entropy’ of its initial sensorial openness through the repetitive structure of imitation – decreasing uncertainty, by increasing redundancy.

The ‘Mental State of Noise’

189

What we could call the maximum entropy of the ‘vicious whir’ of the infant’s undifferentiated sensations is thus to be reduced, in order to generate a stable basis for a growing interaction with the environment. Imitation can thus be understood as introducing a structural redundancy that transforms the pure contingency, or ‘maximum entropy’ of the ‘vicious whir’ of sensations into the ‘relative entropy’ of what we can now call potential information (Sands and Ratey 1986, 294). Yet, if such is the open disposition of the child, open in other words to imitate and acquire structural redundancy in the process of learning, then the poet could not be more different to the child, at least insofar as Sands and Ratey refer to Piaget and Keats. The ‘negative capability’ that characterizes the poet’s openness is not a ‘default’ openness (it is not predisposed to growing structural redundancy and organizational constraint of the initial ‘vicious whir’ of sensations), but on the contrary constitutes a knowing and willed undoing of the certainties that have arisen on the basis, at least initially, of learning through imitation and repetition. Implied in the reference to the Freudian notion of regression was the idea that the experience of the ‘mental state of noise’ affects the arrow of time, that it bends the progression, the natural development of the subject, backwards: the experience of ‘the mental state of noise’ would be like a time boomerang returning the adult to the ‘vicious whir’ experienced by the infant. The catastrophic reaction, in turn, would be a regression to coping mechanisms comparable with those of a child. Logically, the opposite of a catastrophic reaction would then be the idea of progress in healthy mental development: yet, if regression is the return to a maximally un-differentiated state of openness, then progress would logically consist in progressively closure. Sands and Ratey’s reliance on the Freudian concept of regression thus raises a problem analogous to the one we encountered when comparing Shannon’s definition of information as ‘information entropy’ and Wieners negation of the latter as negentropy: ‘information entropy’ becomes indistinguishable in principle from noise, while negentropy ultimately means redundancy. It is not clear, whether Sands and Ratey saw the contradiction in their concepts of openness, that arises between the openness of infant, the openness that serves as a basis to explain the vulnerability to the ‘mental state of noise’ in the adult, the openness of the child according to Piaget, and the wilful openness of the poet.

VI

The Catastophic Reaction to Noise

The importance of Goldstein’s concept of the catastrophic reaction is evident in the central position it is given in Sands and Ratey’s article. Beyond its importance in psychiatry, however, Goldstein’s concept of the catastrophic reaction also has a role to play in the philosophical conceptualization of noise which we undertake here, because it is relevant not only for Sands and Ratey’s interpretation of noise, but also for a particularly important moment in the history of twentieth-century philosophy. It is well known how important Goldstein’s influence was on a book that has become a corner stone of French epistemology: George Canguilhem’s The Normal and the Pathological (Canguilhem 2009). Canguilhem’s turn towards the philosophy of medicine has played a significant role in the unfolding of contemporary French philosophy when, during the 1960s, it intersected with two other trajectories of contemporary thought, the formal engagement with systems and their structure and an emancipatory philosophy of politics. Canguilhem rendered evident the intersection between, on the one hand, the value judgement inherent in the concepts of health and disease, and on the other hand the disinterested search for scientific truth in the concepts of the normal and the pathological. His analysis resonated with the philosophical tension between the need to act and the pursuit of a formal engagement with a structural understanding of systems that would erupt in 1968, amidst of a number of highly original philosophical projects. In what follows we will develop some of the implications of Goldstein’s notion of the ‘catastrophic reaction’. This serves the double purpose of deepening our understanding of its role in Sands and Ratey’s concept of the ‘mental state of noise’, but also of drawing out some of the philosophical consequences for a more general view on the epistemological relevance of noise. Goldstein’s ‘catastrophic reaction’ will serve as a basis notably to approach Canguilhem’s concept of normativity, whose meaning is here not restricted to the power of the law, but extended to the question of the source of our norms of thinking and living. In

192

An Epistemology of Noise

the context of the philosophy of medicine, normativity, for Canguilhem, is first of all the individual’s capacity to organize and, more importantly, reorganize experience after a traumatic event or lesion. Rehabilitation therefore is not merely a question of therapeutic or pharmacological reduction of noise, but of facilitating the reassertion of subjectivity via the normative power to shape one’s own experience. At stake in Canguilhem’s concept of normativity is therefore not merely the power of existing norms, and hence not the individual’s return to a previously normal state, but normativity understood as the individual’s reassertion of his or her power to act, judge and decide, in other words, the power to generate new norms in answer to life’s contingent events. Grounded not in existing norms but in the individual or even the organism’s capacity to invent new norms in response to contingency, this normativity presents itself not as a function of the understanding or reason alone, but as a gamble, of which Canguilhem famously said: [L]ife gambles against growing entropy. (Canguilhem 1991)

The concept of normativity can thus be understood as a covert way to pose the problem of what makes an individual a subject, capable of judging, deciding according to norms invented by him/herself – covert, because Canguilhem guarded himself well from producing a theory of the ‘subject’ in a time when existentialism and phenomenology occupied this philosophical terrain. Another reason why there is no abstract theory of the subject in Canguilhem’s work is that while the concept of a subject is linked to this normativity, the notion of the subject itself is not reducible, for Canguilhem, to that of a conscious self, as it is for existentialism and phenomenology. Nor can the subject be generalized beyond the conscious self to an abstract notion, such as Bergson’s élan vital. The subject instead remains a difficult and open question in Canguilhem’s work (Badiou 1993). The philosophical consequences we will draw from Goldstein’s concept of the ‘catastrophic reaction’ will address Canguilhem’s concept of normativity, which in turn will enable us to broaden the philosophical approach to noise, by taking into account once more Shannon and Weaver’s definition of ‘information entropy’ as freedom of choice, but also Sands and Ratey’s attempt to conceptualize noise as a mental state. It is in this broader constellation that Goldstein’s concept of the ‘catastrophic reaction’ and Canguilhem’s concept of normativity become important, because they allow us to ask anew the question, in extremis, at the edge of reason: how do we draw the line between information and noise?

The ‘Mental State of Noise’

193

Although the topic of Sands and Ratey’s article and indeed of Goldstein’s concept of the ‘catastrophic reaction’ is clinical, focussing in particular on mental health issues that deserve to be studied with due care for their singularity and without rash generalizations, it is difficult to overlook the resonance with current political developments. The return of isolationist politics on the global stage, accompanied by a wave of xenophobia, and the concurrent politicization of extremist, conservative interpretations of religion, cannot help but resonate darkly with Goldstein’s concept of the ‘catastrophic reaction’. While every care must be taken not to generalize where careful attention to the specificity of a problem is required, it would also be wrong to completely ignore that George Canguilhem’s thesis on the Normal and the Pathological, was written while he was an active member of the resistance against fascism; that the profound innovation Goldstein brought to the rehabilitation of the mentally ill, came in response to a generation of young men returning shell-shocked from the First World War; and that his Logic of the Organism was written in 1934, when Goldstein was forced to give up his position as clinical director of psychiatry in Königsberg and flee Germany, after being arrested and imprisoned for being a Jew. Not only their implicit rejection of the conservative impulse, at the level of both concept analysis and methodology, but also the extreme circumstances in which Goldstein and Canguilhem wrote these important works, testify to the ethical and political relevance of their thinking about normativity in the context of pathology.

VII

Anxiety

In defining the ‘mental state of noise’ in terms of a ‘catastrophic reaction’ Sands and Ratey place Kurt Goldstein at the core of their definition of the ‘mental state of noise’. Here, we will go back briefly to Goldstein’s Logic of the Organism in order to complement Sands and Ratey’s interpretation of the catastrophic reaction, in particular by zeroing in on the opposition between noise and order (Goldstein 1948; Sands and Ratey 1986, 290). As Sands and Ratey note, the ‘catastrophic reaction’ consists in an attempt to establish an atmosphere of ‘sameness’, in order to compensate for the difficulty of controlling confusing internal affective and cognitive states, frequently achieved via ‘excessive orderliness’, social withdrawal, or recourse to ‘stereotypes’ (Sands and Ratey 1986, 291). Goldstein indeed observed in the context of mental illness that a ‘methodical character’ tied to the ‘fanatic need for order’ (Goldstein 1983, 39) serves the purpose for the affected individual, ‘to maintain himself in a situation he is able to master’ (Goldstein 1983, 37). He found the excessively methodical enforcement of order to be symptomatic of a conservative attitude that is, in his now famous expression, the phenomenon of a ‘declining life’: The instinct for [self-]preservation can appear like an essential trait of the organism, even though in reality, the tendency towards preservation is a phenomenon of disease, of a life ‘in decline’. (Goldstein 1983, 355. Emphasis in the original.)3

Goldstein thus saw in the catastrophic reaction a desperate attempt by the individual to re-organize experience, by drawing attention around a radically reduced focus of attention. He interpreted the ‘catastrophic’ reaction as a retreat to a more constricted form of organization, whose sole purpose becomes the preservation of self: In order to escape catastrophic situations, the sick who suffer cerebral lesions have a particularly characteristic means: it is their methodical character. These

196

An Epistemology of Noise

individuals have a truly fanatic need of order. The wardrobes of cerebrally traumatized I have observed during many years were exemplary models of a certain type of determined order. […] If one places a messy heap of objects before such an individual one soon observes that he, if he notices them, places them immediately methodically one next to the other, sometimes even making little heaps of those which, for him belong to a same category. (Goldstein 1983, 39. Emphasis in the original.)

The avoidance of stressful situations, by narrowing the focus of attention around a single event or situation, however, invariably fails to have the desired effect. The ‘catastrophic reaction’ in fact lowers the capability of adaptation, rather than enabling the individual to adapt to the new situation. Even if a release of tension is afforded by the ‘catastrophic reaction’, it ultimately augments the feeling of impending catastrophe. Rather than reducing it, the ‘catastrophic reaction’ thus amplifies the perception of a catastrophic situation. Goldstein took his assessment of the rigidity of the conservative impulse a significant step further, by comparing the spontaneous closure that characterizes the ‘methodical character’, the isolation from the exterior world and obsessive orderliness, with a state of death-like paralysis, which he compared with the ‘simulated death’ of certain animals, characterized by paralysis and fixation of the source of danger: This isolation from the exterior world […] must probably be considered as analogous to what we call simulated death of animals. (Goldstein 1983, 39)

The animal’s rigid fixation on the source of threat of imminent death, in the human situation, is a fixation on that which is at once frightening and inconceivable. In order to fully characterize the ‘catastrophic reaction’ Goldstein therefore insists on a subtle nuance: the ‘catastrophic reaction’ is reducible to neither fear nor instinct, it is of the order of anxiety. The ‘catastrophic reaction’ – and this will be essential to our more general epistemological conceptualization of noise – does not arise from fear of something, but from anxiety: it is the fear of something inconceivable. What characterizes the ‘catastrophic reaction’ according to Goldstein is thus not the fear of disorder, strictly speaking. He sees compulsive order as the response to an anxiety aroused by an experience that has become inconceivable: [What] causes him to cling obstinately to the order which is adequate to him and which appears to us, normal beings, as an abnormally primitive order, abnormally rigid and forced, [is not] the fear of objective disorder, which the ill

The ‘Mental State of Noise’

197

person naturally doesn’t feel objectively as such, but [of that] which he feels as anxiety. (Goldstein 1983, 40–41)

When introducing the distinction between fear and anxiety, as between fear of something and fear of the inconceivable, Goldstein puts the catastrophic reaction into an explicitly philosophical context: ‘I am thinking of Pascal, Kierkegaard, Heidegger’ (Goldstein 1983, 250). An existentialist orientation in Goldstein’s thought could be further pursued here, by referring notably to Sartre’s La nausée. But what we seek in Goldstein’s concept of the ‘catastrophic reaction’, in the broader context of our argument about the epistemological aspect of noise, is his attention to anxiety as the fear of the inconceivable. It is less an existentialist état d’âme, than his definition of disorder that is of the highest relevance to our considerations about noise: What does disorder mean in this case? It goes without saying that objective disorder does not exist any more than objective order. Disorder means an arrangement such that it imposes no single, determined perspective, nor a unique mode of utilization, but allows several or even many. Total disorder (if it were possible) would, however, impose nothing, but freedom of choice. (Goldstein 1983, 39)

The catastrophic reaction is thus characterized by the rigid negation of all contingency. It is what we could call a catastrophically negentropic attitude. Its stability is in fact rigidity, a state of control petrified in an inflexible structure. The only certainty it avails is the avoidance of all ambiguity and complexity. Disorder for Goldstein means, as entropy did for Shannon and Weaver, ‘freedom of choice’. Such ‘freedom of choice’ becomes inconceivable to the individual struck by the devastating effect of contingency, traumatized or injured, because openness requires of the individual the association of contingency with the freedom of choosing amongst more than one possible perspective, of acting and reacting in more than one possible way. This pluri-valence of an open situation results precisely in what ‘the catastrophic reaction’ seeks to avoid, because of the uncertainty it entails. Goldstein’s insightful notion of anxiety as fear of the inconceivable, as inconceivable multiplicity of choice, now illuminates for us Shannon’s conceptual feat in defining information as ‘information entropy’, as that which occurs with the lowest probability and whose prediction is thus characterized by the greatest uncertainty, but also by the greatest ‘freedom of choice’ in probabilistic terms. Goldstein’s concept of the ‘catastrophic reaction’ resonates so compellingly

198

An Epistemology of Noise

beyond the realm of the individual’s psychopathology to the level of collective attitudes, that it enables us to fully appreciate the weight of conservative preconceptions that persist against Shannon’s audacious interpretation of ‘information entropy’ as ‘freedom of choice’. This definition of information was so vehemently rejected, said to have no relevance beyond the sphere of electronic signal transmission, because it implied disorder, uncertainty and thus stoked anxiety about loss of control. Shannon and Weaver’s definition of information as ‘information entropy’ appeared to sacrifice too much of what we ordinarily associate with information (certainty, knowledge, facts and data) to an intimate relation of information with contingency (Capurro and Hjorland 2003; Janich 2006). The difference between information and noise in Shannon and Weaver’s definition of ‘information entropy’ is not, as we have seen in the first part, a difference between order and disorder, but the finest of lines separating the contingency of both ‘information entropy’ and noise, a line drawn by the intention alone with which a certain ‘entropy of information’ is selected and transmitted as a message, against the backdrop of an accidental entropy that is discarded as noise. By analogy we can infer the fundamental difference between the intentional openness and ‘freedom of choice’, embraced by the poet and the philosopher, and the accidental and in this sense excessive openness suffered as the ‘mental state of noise’. Novelty and uncertainty, which for Shannon go hand in hand with a high content of information, require of the individual the capacity to assert uncertainty as a choice rather than be swamped by it in the mental state of noise as confusion and indecision. Noise is thus not an object of perception, but that which swamps it. In this respect we could adopt Goldstein’s perspective on anxiety and say that noise is, like disorder, an inconceivable freedom of choice.

VIII

Order

Entropy is often defined as molecular disorder; noise in molecular biology implies the idea of variation from the norm, or of disorderly DNA transcription or disorderly cell proliferation. Noise in communication technology evokes the disorderly transmission of signals, and noise litigation in the social context evokes most emphatically the idea of disorder: here noise signals most explicitly behaviour that disrupts the social order and defies orders given by the authorities. Intuitive as it may be, the notion of order, which is frequently opposed to noise as disorder or chaos, is not all that obvious. It is therefore worth dwelling on the interdisciplinary circulation of the concept of noise and its cortege of concepts, order and disorder, even if these considerations appear to take us too far away from the problem of the mental state of noise. Although we were here dealing with the excessive orderliness of the individual’s catastrophic reaction to noise, and therefore with a behaviour defined as pathological, the relevance of the concepts of order and disorder appears to extend beyond the psychiatric definition of the ‘catastrophic reaction’, to epistemological and methodological considerations at the level of scientific and philosophical discourse. Recall, for instance, the assessment the philosopher of biology, Marjory Greene, gave of the rigidity of Logical Positivism, of its incapacity to deal with the imprecision of the empirical world. This rigidity led, in her view, to the ‘catatonic, vegetative state’ of Logical Positivism in contemporary philosophy. The question is not how accurate a description Greene’s is of Logical Positivism, which is not the object of study here. What is striking rather, is that the passions run high, even in scientific and philosophical discourse, when the requirement to engage with the uncertainty and imprecision of novel perspectives clashes with an authoritative call to order that is experienced as intellectually repressive. The first thing to point out is that the concepts of order and disorder are, scientifically speaking, no less ambiguous than the concept of noise. In

200

An Epistemology of Noise

mathematical terms it makes no sense to equate even chaos with disorder, since mathematical chaos, for instance as treated it in catastrophe theory by René Thom, is deterministic – determined and thus orderly in its necessary unfolding, even if its graphic representation is baffling and unpredictable. For this reason alone it would be incorrect to equate entropy with disorder or with molecular chaos: because the idea of disorder wrongly implies that physical reality favours one state over another. It would be wrong also because entropy is a concept belonging to the field of probability and is thus inherently not deterministic and thus cannot be said to contravene a deterministic order – and how can order be anything other than deterministic? The difference between probability and determinism is, in fact, a line of fracture that runs through the history of the sciences, between classical and nonclassical thermodynamics, classical mechanics and quantum mechanics. This fracture is exemplified by the dispute between René Thom and Ilya Prigogine, recounted here by Rainer Zimmermann in ‘Order and disorder – on the recent dispute about determinism between Thom and Prigogine’: Besides Prigogine, Thom accuses other authors such as Monod, Morin, Atlan and Serres of attributing an inordinate and thus inappropriate significance to notions such as contingency, noise, oscillations: ‘[…] all make contingency responsible for either the organization of the world or the emergence of life and thought […]’. Thom sees this as an ‘anti-scientific attitude par excellence’, a ‘mental state of confusion’, which one may forgive authors in the human sciences, but never in the natural sciences. (Zimmermann 1988)

Boltzmann’s statistical formalization of entropy thus signals a radical departure from classical physics and mechanical reversibility, which plunges physics into what we may opportunistically call a mental state of noise, and which Heisenberg established as an uncertainty relation in quantum physics. Even in physics it is therefore difficult to give an intrinsic reason why one should qualify one state of a system as more orderly than another – according to which intrinsic physical criteria? How does the value of a physical state as order or its disqualification as disorder insert itself into the indifference of physical states, which simply are what they are and which unfold, by definition, in perfect accordance with the laws of physics? Unless one asserts with Hume and Pearson that there are no laws of nature, only habits of perception – but in this case also the notion of disorder no longer has any relevance, because disorder simply becomes that to which we are not habituated.

The ‘Mental State of Noise’

201

It is not that the power of prediction has lessened with the introduction of the statistical method in physics, on the contrary, physical processes with entropy have only been mastered on account of probability and statistics. What has happened is that the idea of indeterminacy has gained prominence with the introduction of probability and statistics: the impossibility of an observer’s perfect knowledge of initial conditions of a deterministic chaotic system, calls for the probabilistic method, which in turn introduces the sliver of in-determinacy at the microscopic level of obersvation into an overall highly performing method of prediction. Even if mass phenomena are statistically mastered and allow for an impressive power of prediction, this sliver of indeterminacy in turn introduces an irreducible unpredictability and with it the irreversibility of any probabilistic process with increasing entropy. This sliver of irreducible indeterminacy in the probabilistic process thus severs, irreversibly, probability from determinism. Yet if nothing, at least in physical reality, can be said to be fully determined by reason, on what grounds can an idea of order be erected against which we can measure disorder? In biology the idea of molecular disorder, associated with the notion of noise and entropy, preserves an implicit determinism that lingers with genetic theory, despite the softening of the concept of teleology, which Jacques Monod, in his Chance and Necessity, already replaced with the term teleonomy (Monod 1973). The now common, if still controversial term, teleonomy, implies the rule-bound unfolding of molecular mechanisms, such as that of the structural modification of proteins in relation to their environment, or the role of nucleic acids in regulating the formation of proteins (Moulin 2006, 1073). Rather than reducing the observable regularity of biological mechanisms to a purposive logic, teleonomy reduces teleology to the acknowledgement of apparent norms in biological mechanisms, without inferring a telos or purpose. The neologism teleonomy was introduced to replace the metaphysical notion of teleology, with the assumption that biological organisms act in a rule-bound way and submit to norms, even though these may be the chance result of evolution. As the philosopher of biology, Michel Morange, notes, the idea of an implacable order in the form of a ‘genetic program’ was proposed independently by Ernst Mayr, Francois Jacob and Jacques Monod, inaugurating the golden age of genetic theory, where [l]ife was considered to be the possession of genetic information and of a genetic code allowing this information to be translated. (Morange 2005, 432)

This reduction of life to a predetermined order, even if the latter is the child of evolution’s contingency, provoked the biologist Stanley Shostak to declare that

202

An Epistemology of Noise

the reduction of life to the sum of the properties of the macromolecules present in organisms, was nothing other than ‘the death of life’. Yet if, like Monod, we accept the premise that the order imposed by the genetic ‘program’ is itself the result of evolutionary contingency, or to say it with more emphasis, that the very concept of order in biology is grounded in contingency, then noise, as maximum information entropy or maximum in terms of evolution’s ‘freedom of choice’, acquires the status an altogether fundamental role for the very concept of order in biology, as that which both precedes and exceeds order, as that from which order arises and into which it founders when pathology, eco-systemic transformations or geological catastrophes reassert contingency over predictability. There still lingers in the idea of noise in molecular biology the idea that it is a source of disorder or deviance from the orderly and regular unfolding of biological processes, which in turn points, in the concept of order, to an implicit function of purpose. Even if teleology is ruled out as a metaphysical concept, order and purpose appear to remain close cousins. They are related, in conceptual terms, to the ideas of use-value and work, which we had previously seen in relation to information (as opposed to the ‘spurious uncertainty’ of noise), and of physical entropy, which according to Carnot’s classical definition, is energy that is no longer available to perform work. To oppose order to noise thus continues to imply that disorder, uncertainty and error are the negation of a rule, of purpose, use-value or work. In this light the entire enterprise of rational and scientific discourse, in so far as it relies on the idea of order both in discourse and in the empirical world, must be understood as the assertion of the power to impose order and purpose where confusion, error and uncertainty loom large. This power, perhaps, is nothing other than the more or less successful attempt to exercise control, temporarily, in the midst of the fundamentally contingent unfolding of events that accounts for time’s historical irreversibility and for the future’s irreducibility.

IX

Control

Noise, beyond the reference to unwanted sound, thus reveals itself to be conceptually polymorphous because it has never been about types, classes or measures of phenomena that qualify noise as a particular type of disturbance, but about the relation between contingency and control. Contingency and control, especially loss of control, means that in various domains of theoretical investigation and practical application, very different types of phenomena are at stake. Sands and Ratey explicitly distance their use of the term ‘noise’ from Wiener’s ‘different’ understanding of noise in cybernetics. Without wanting to diminishing the reasons for insisting on this difference we are nevertheless obliged to address the link between their own approach to the ‘mental state of noise’ and the idea of control in cybernetics, when Sands and Ratey emphasize their ‘theoretical’ formulation of the ‘catastrophic reaction’, by phrasing it as a ‘closing down of the system’s circuitry’ (Sands and Ratey 1986, 296). Their decision to formulate noise as a trans-nosological concept appears to expresses, indirectly, a generalized diffusion of the cybernetic notion of ‘control’ to scientific discourse, even where the cybernetic paradigm is not adopted explicitly. It is not only the wording in this key moment of the article, but also the objective of Sands and Ratey’s conceptualization of the ‘mental state of noise’ that links back to the cybernetic notion of control: for it is, ultimately, in view of a pharmacological approach to feedback and amplification mechanisms in the nervous system, that the notion of noise is developed here. Sands and Ratey’s objective is ultimately to elucidate the potential benefit of certain drugs in alleviating the ‘mental state of noise’, in combination with psychological training of awareness of bodily sensations associated with it. This becomes clear in the conclusion of their article, where they refer to the effects of cardiac drugs on autonomic reactions. The attempt to conceptualize noise in the context of psychiatry is thus finally put in the service of illuminating the role

204

An Epistemology of Noise

of the involuntary nervous system in the ‘mental state of noise’, acting below the level of consciousness and controlling visceral functions including heart rate, respiratory rate, perspiration and arousal of the nervous system. It is at this preconscious level that the authors propose to intervene with the use of cardiac drugs such as beta-blockers and clonidine, in view of reducing what the authors call the ‘reverberating circuit’ of emotional agitation and autonomic reactions. The objective is thus to control the feedback mechanisms that occur between the cognitive aspects of confusion, anxiety and agitation and the involuntary physiological response to stress, which together characterize the ‘mental state of noise’ and the ‘catastrophic reaction’. This approach, Sands and Ratey argue, is relevant for the treatment of all psychiatric illnesses, as wide ranging as brain-damage, mania, mental retardation, schizophrenia and autism, but also in treating the effects of antipsychotic neuroleptic drugs, such as restlessness (Sands and Ratey 1986, 293). Drugs such as Beta-blockers and clonidine are discussed as acting on the ‘mental state of noise’ by regulating feedback mechanisms between the central and the peripheral nervous systems. So what are these drugs and what role do they play in saving the ‘system’s circuitry’ from breaking down? Clonidine notably acts on the ‘central down-regulation’ of the locus ceruleous, in other words, on the part of the brainstem involved in the synthesis of the hormone noradrenaline. Noradrenaline is a neurotransmitter involved not only in physiological responses to stress and panic, but also in the state of vigilant concentration. By lowering sympathetic outflow, this drug lowers the levels of stress hormones in the autonomic nervous system, whose function is both to maintain homeostasis, but also to mobilize the body in a ‘fight or flight’ response. Beta-blockers, in turn, are drugs that act on the peripheral nervous system, by blocking the reception of the stress hormone noradrenaline in the sympathetic nervous system. Both drugs modulate the physiological response to stress. They are found, according to Sand and Ratey, to ally anxiety, impulsivity and psychic and bodily disorganization. The soothing effect of beta-blockers in relaxing striate musculature (comprising both voluntary muscles and cardiac muscles), reduces anxiety response in a way the authors compare to the ‘holding’ environment that alleviates the ‘vicious whir’ of sensations in the infant, described by Greenacre (Greenacre 1952). These drugs, proposed to combat the ‘mental state of noise’, can be said without too much exaggeration to act in an analogous way to the control of noise in cybernetics: both approaches have in common that information and noise are not dealt with at the level of signification or conscious processes, but

The ‘Mental State of Noise’

205

by controlling the amplification of undesired perturbations through feedback mechanisms. This is indeed how Sands and Ratey argue for the use of beneficial effects of these drugs, which ‘interrupt the reverberating circuit of emotional agitation, cognitive disarray and peripheral arousal’. What underlies the conceptualization of noise is thus ultimately a cybernetic logic of control, driving the complex dynamic functions involved in the experience of the ‘mental state of noise’ and its alleviation. At stake is thus the positive amplification of noise in the catastrophic reaction, and its control through negative feedback. In other words, what is at stake is the barely acknowledged yet thereby even more crucial, because subterraneous relation with cybernetics. The authors do acknowledge the cybernetic conceptualization of noise in passing (‘The concept of noise, a term fist applied by Wiener […]’), but what is more, they describe the ‘catastrophic reaction’ in terms that leave no doubt about the technical paradigm (‘concreteness and closed circuitry of the system’), even while they distance their own use of the term noise, by specifying that the term is used by Wiener ‘in a somewhat different way to the discussion of mental processes […]’. The theory of cybernetics is, of course, itself from the start closely related to the physiological concept of homeostasis, the self-regulation of an internal milieu, which cybernetics analyses in analogy with man-made systems, i.e. machines with self-regulation through feedback mechanisms. The ‘circuitry’ of man-made systems, of machines, thereby acquires a paradigmatic status. Our understanding of machines henceforth encompasses our understanding of natural and biological systems, because of the greater certainty and the possibilities they offer for accurate mathematical analysis, in comparison to a living organism, whose internal regulations and relations with the environment are still too complex to provide an equally stable theoretical framework. This is not to say that the convergence of physiology and cybernetics around the term ‘noise’ expresses an explicit adherence to either cybernetics, secondorder cybernetics, or even Shannon’s conceptual framework for thinking about information and noise. Indeed, neither the mathematics of Wiener’s cybernetics nor those of Shannon’s information theoretical algorithms are, strictly speaking, taken on board or even mentioned as an explicit theoretical framework when the term ‘noise’ is used. It is rather the oblique reference to theories of contingency or control that gives this and many other conceptions of noise their quotations marks, allowing noise to become a transdisciplinary passe-partout. Sands and Ratey’s ambivalent acknowledgement of the technical paradigm in relation to their approach to the ‘mental state of noise’ is understandable, insofar

206

An Epistemology of Noise

as the psychiatry and the neurosciences of the 1980s were perhaps still closer to the humanities and psycho-dynamic approaches than today’s wholehearted cognitivist mechanization of the human mind and its metaphorical reduction of the brain to the model of a super computer. The theoretical influence of cybernetics on Sands and Ratey’s approach to the ‘mental state of noise’ is not intended here as an argument for the idea that noise is reducible to ‘crossed wires’ in the brain. The computer, no matter how sophisticated and even superior to the brain in its calculating power, memory and speed, does not suffer the noise that interferes with its optimal functioning, it makes no value judgements, and it does not fear the loss of control. The computer applies norms of optimal operation, it does not invent these norms – optimal conditions are ‘optimal’ for the user and completely indifferent to the computer. In this sense one could perhaps say that, ironically, the computer is superior not only because of its superior combinatorial capacity, but simply because of its indifference to man’s needs. Even though the computer submits to rules and norms that may indeed be designed to allow room for the ‘learning’ of new decisions, the computer does not make judgements that ground these decisions. It can adjust and diversify, but it cannot know the difference between adjustment and a just act according to selfgenerated norms. And yet, something is lost, if this already pregnant cybernetic link remains obscured, something that can illuminate differently not only the idea of a ‘mental state of noise’, but also the concept of control. Sand and Ratey’s highly original attempt to define the ‘mental state of noise’ allows us to think about control as a fine-tuning and adjustment of complex dynamical variables, rather than the simple imposition of a norm through pharmacological control. Although this is hardly news to the engineer or mathematician, it introduces a nuance that is lacking wherever cybernetics become the excuse for reductive mechanization in general discourse, and wherever control assumes a dark connotation, as in the context of political and commercial abuse of civil liberties through new vulnerabilities that come with the ubiquity of new communication technologies.

X

The Helmsman Metaphor: Kybernetes

A nuance to the reductive analogy between the computer and the brain can be found, perhaps surprisingly, if we go back to cybernetics’ origins in control theory. Here we find not only the metaphorical link between control and power in the political sense of governing, but also its intrinsic limits as a mathematical paradigm. Contemporary modelizations of complex systems owe much to the field of control theory that emerged during the 1950s. Since then the idea has taken hold that such modelizations offer a sufficient explanatory basis for complex systems, such as the brain, when combined with statistical data form empirical observations. Yet while there can be no doubt of the technical utility of control theory for mechanical engineering, and indeed for our growing understanding of complex systems, including the neurological basis for cognition (‘Bluebrain | EPFL’ 2013), the question remains: what are the limits of control theory as a conceptual framework for our understanding of the control of powers, be they mechanical, psychosocial or downright political? Where does the legitimacy of the mathematical model end, and where does the ideologically motivated capturing and manipulation of processes of cognition, and of flows of information and wealth begin? This requires us to ask what we mean, when we speak of control in the context of cybernetics and what conceptual conversions are required for the concept of control to be exported to non-mathematical contexts. Although archaeological evidence suggests that the invention of control mechanisms dates back as far as irrigation systems in Mesopotamia, it is not until the 1950s that control theory became a theoretical field in its own right. It now extends from human or automatic control of mechanical devices, for instance through measuring devices such as thermostats turning refrigerators on or off, to remote controls or servomechanisms and motor governors. It deals with individual systems or with the coordination of devices at a large scale, such as a power plant or traffic control. More recent developments see the importance

208

An Epistemology of Noise

of control theory in bio-technology, for instance in the development of artificial organs or nerve-controlled prosthetics and artificial intelligence through pattern recognition and speech recognition (‘Control Theory | Mathematics’ 2017a). In his 1961 Cybernetics, Control and Communication in the Animal and the Machine, Wiener refers to the origins of cybernetics in the formal mathematical analysis of control mechanisms, by referring to James Clerk Maxwell’s 1868 On Governors (Maxwell 1868): We have decided to call the entire field of control and communication theory, whether in the machine or in the animal, by the name of Cybernetics, which we form from the Greek χυβερνήτης [kybernetes] or steersman. In choosing this term we wish to recognize that the first significant paper on feedback mechanisms is an article on governors, which was published by Clerk Maxwell in 1863,4 and that governor is derived from a Latin corruption of χυβερνήτης. We also wish to refer to the fact that the steering of engines of a ship is indeed one of the earliest and best-developed forms of feed-back mechanisms. (Wiener 1961, 11–12)

A governor is a device that measures and regulates the speed of a machine or engine. It is the basis for the invention of the servo mechanism, initially devised for the steering of big ships. The governor makes the steering engine of a ship independent of its load, by using readings from the steering wheel to regulate steam pressure valves offset from the tiller (the lever used to turn the rudder of the ship); the tiller’s movement to one side progressively closes some pressure valves, while opening others, so as to increase the force that produces the rotation (torque). Any force opposing the motion of the tiller keeps up the admission of steam, thereby increasing the push of the tiller to the desired position. However, if feedback is too brusque the rudder overshoots, leading to feedback in the other direction, which in turn amplifies into a ‘wild oscillation or hunting’ in the steering mechanism. Although Maxwell’s work predates by almost a century the emergence of control theory as a field of research in its own right, it lays the foundations for a fundamental problem tackled by control theory and cybernetics. And by providing a mathematical theory of ‘self-oscillation’ it paves the way for the understanding of phenomena of feedback, amplification and overcompensation in increasingly complex dynamical processes. Control thus means, first of all, the calibration of powers that vastly exceed the steering capacity itself. This means that the governor not only informs a greater power, but more importantly, that it calibrates the control action through continual feedback, so as to avoid wild oscillation with chaotic consequences. The idea of the governor thus represents more than a technical concept. It stands for a paradigm of self-regulation that can be extended to natural and human

The ‘Mental State of Noise’

209

forms of organization, and can even be said to imply an ethics of self-regulation for human forms of organization or governing. However, despite its vast possibilities of application, it must not be forgotten that control theory is first of all a consolidation of two classical branches of mathematics, the calculus of variations and the theory of differential equations. Differential equations help correlate a system’s measurable changes and rates of change. By providing a function that expresses dependences amongst these variables, it allows prediction of the behaviour of the system within given constraints. The calculus of variations in turn helps understand phenomena like elasticity, vibrations or electromagnetic theory. Finding the least surface area for a given volume so that it encounters minimum resistance during its trajectory is a typical aerodynamic problem solved through the calculus of variations (‘Calculus of Variations | Mathematics’ 2013). Control theory lends itself to any empirical domain, on condition that its parameters are understood comprehensively enough to be subject to exhaustive mathematical analysis. This means that whatever the system under consideration, it must afford precise mathematical description both of its internal dynamical state and of external influences affecting any control intervention, in every possible circumstance. It is not only the precise mathematical description of the system itself that is required, however, but also the mathematical definition of purpose, in terms of control criteria. Together with the environmental conditions and disturbances, these criteria constitute the margin of ‘optimization’ for control. The future behaviour of the system can then be deduced from knowledge of both the present state and future inputs on the basis of the so-called ‘control law’. The latter is the rule that defines the relation between variables of the state, in other words, it is the function that determines the control action to be taken. This control action may indeed take the form of feedback or feedforward methods, but the ‘control law’ is in itself a more general concept about the functioning of a system in its environment, than feedback (‘Control Theory | Mathematics’ 2017b). Since not every component of the dynamical state of a system (or state vector) can be measured simultaneously with instantaneous exactness, statistical prediction and filtering theory step in to determine the control law with the obtained estimate state vector. Cybernetics, and in its wake complex systems theory, thus emerged from the historical coincidence of new methods of statistical calculation and new means of computerized data compilation and powers of calculation. Although it was by then possible to accurately represent the state-vector equations of those physical systems in the natural world which

210

An Epistemology of Noise

experience only small deviations from the steady state behaviour, the question arose whether control theory could be applied also to non-linear models, in other words, to systems in which small changes in input can result in large deviations. Depending on the mathematical method, these systems are called stochastic or chaotic. The crucial point, however, is that whatever knowledge control theory affords us regarding empirical systems with greater or lesser complexity, its most critical conceptual contribution is the precise mathematical definition of the limits of controllability. This inherently critical understanding of the conditions and limits of control theory in the mathematical sense is forgotten at great cost. For this reason control theory undoubtedly also lends itself to spontaneous ideological distortion. The historical origins of statistics in the ‘discipline of the state’ indelibly links statistics with the question of power: the power to collect information and the power to predict and thereby influence the course of events. The word ‘control’ is itself so highly charged with political connotations that it cannot help but blur the boundaries between mathematics and common language, when the model of control theory is reintroduced into common discourse – distorting the critical limit between abstract concepts and concrete interests, between the need to know and the need to act. Without a clear understanding of its inherent critical limit, control theory thus becomes prey to its ideological appropriation for a control society without noise. Nevertheless, if we are to think freely about the conceptual potential inherent in this inventive convergence of diverse branches of mathematics in control theory, without either succumbing to, or ignoring, the sinister link between mechanical and political control, we may wish to recall George Cuvier’s ‘Report on the Progress of the Sciences and Mathematics’, presented to the French government in 1808. In this plea for parliament’s support of the sciences, Cuvier addresses the government by drawing the cautionary tale of ‘ordinary men’ who see in science only their immediate advantage. Supporting the sciences only when they can grasp their practical applications, they fail to understand the principles of science, like the ‘vulgar’ who fails to appreciate a work of art. And yet every proposition in science, Cuvier insists, is the germ of ‘a thousand common inventions’ that, in turn, affect the very bases upon which not only the state, but also the political relations between nations rest: [F]eudal anarchy would perhaps still subsist, if canon powder hadn’t changed the art of war; the two worlds would still be separate without the compass; and no one can predict what today’s relations would become, if one could supply colonial food-stocks with indigenous plants. (‘cpa9.17.cuvier.pdf ’ 1968)

The ‘Mental State of Noise’

211

What would the post-Second World War world look like, without the development of computer science, and what would the Cold War have been, without the contributions made by cybernetics to the development of selfdirecting missiles? We may well ask ourselves what scientific principle could more profoundly affect government, the relation between nations, and indeed the relation between mankind and nature, than the mathematical harnessing of uncertainty, the control of noise? But to fail to understand the sprawling, practical fecundity of a single valid mathematical principle, Cuvier warns, is also to fail to understand that a mathematical principle’s power lies within the intrinsic understanding of its limitations: In the mathematical sciences […] a single well stated and precisely measured fact becomes a principle, a starting point; the rest is the labour of calculation: but the limits of calculation are also those of the sciences. The theory of moral concerns and of their powers stops here and more abruptly still before the continual and incomprehensible mobility of the heart that ceaselessly defies every rule and every prediction […]. (Cahiers Pour l’analyse N° 9/Genealogie Des Sciences 1968; ‘Cpa9.17.Cuvier.Pdf ’ 1968; Cuvier 1837)

We would do well, then, to heed Cuvier’s warning when overextending the powers we attribute to control theory, by imagining a society controlled by data, if not by artificial intelligence, in which the human becomes superfluous, replaced not only as tool-bearer, but as critical intelligence. For the mathematical specification of uncertainty in control theory also implies, by definition and by design, that the limits of controllability are clear and that uncertainty, in the form of variation, feedback and perturbation, is stratified: carefully distinguishing between uncertainty that can be harnessed and predicted mathematically, uncertainty that can be proven mathematically to be unpredictable, and uncertainty on which we have no scientific handle. This knowledge of limits and stratification of uncertainty provides a conceptual dexterity that is lost, when general discourse is all too eager to soak up the ideas of control and prediction, hurried to promote the idea that extremely complex and poorly defined systems, like economics or the brain, are ‘controllable’ in the sense of causal explanations and control mechanisms. We need to remember that control theory itself – and by extension the forms of reasoning based on its mathematical paradigm, from cybernetics to complex systems theory – imposes a critical check on the political appropriation of scientific discourse for its own means, provided it can be articulated and heard in public discourse.

XI

The Helmsman in Plato’s Alcibiades Dialogue

Nevertheless, it is also clear that the power of prediction, afforded by control theory, is a hybrid of political power and scientific inventiveness – of the need to know and the need to act. When Wiener adopts the governor as the emblem of cybernetics, linking it to the image of the helmsman (Gr. χυβερνήτης, kybernetes), he in fact rekindles a classical philosophical analogy between the art of navigation and the political art of governing. The cybernetic metaphor of the ‘helmsman’ thereby cements a classical link between the art of navigation and the art of governing in the contemporary technical paradigm, dating back to the Platonic dialogues. Plato’s Alcibiades dialogue is crucial in this regard, not only because of the importance given to the helmsman as metaphor for the art of governing, but also because of the injunction it makes against ‘the most sickening’ aspect of poor governance, which is the ignorance of one’s own ignorance. (How could we fail to mention here that the etymology of noise leads back to the nautical field, i.e. nausea, or sea-sickness, completing the metaphorical analogy between noise and uncertainty.) In this dialogue Socrates is the confidante and adviser to the inexperienced Alcibiades, who expects his imminent ascension to political power on the grounds of his noble heritage.5 Socrates’ advice, however, concerns not how best to govern and control the populace. Instead, he insists that governance requires self-control of a particular kind: only the knowledge of one’s ignorance of unintended consequences can act as guarantor for good governance (Plato 1997, 557–95). Through the dialectical method of question and answer, Socrates leads Alcibiades to demonstrate his own ignorance to himself. At the risk of displeasing the future ruler, Socrates indeed forces Alcibiades to acknowledge that his is the worst of all forms of ignorance: ignorance of his own ignorance and, without holding back, condemns this as the most sickening of all forms of ignorance. Carefully calibrating the risk of too brusque a ‘feedback’ in the dialogue with Alcibiades’, Socrates thus treads a tightrope between the need

214

An Epistemology of Noise

to achieve an effect powerful enough to change the direction of Alcibiades’ thinking, and the risk of a catastrophic unravelling of the situation for Socrates. In this powerfully phrased condemnation of privilege as an excuse for ignorance in political governance, Socrates speaks truth to power, in the sense that Foucault would celebrate in his later lectures at the College de France. Here Foucault defines the role of the teacher or master, by distinguishing the modalities of speaking the truth (parrhesia, speaking the truth freely and without reserve) from rhetoric, as the modality of speaking in order to convince without regard for truth. In words that could not be more poignant in our own times of ‘post-truth’ elections and bogus referendums, which make even post-modern cynicism look naive, Foucault’s raises the question how an individual becomes a subject, by asking in which manner the individual constitutes itself in the act of speaking true, and how it is constituted by others as a subject holding a true discourse […]. (Foucault 2011)6

Rather than the creation or revelation of truth, (alethurgy), the praxis of parrhesia is thus a modality of speaking freely, an alethurgical form. It distinguishes itself form rhetoric, but is even more different from flattery, because to speak freely, to hold nothing of the truth back, implies the risk provoking a violent reaction. For Foucault, Socrates is the parrhesiast who pursues this task ‘even when he is threatened with death’ and to his last breath (Foucault 2011, 36). It is thus in this Foucauldian sense, as an act of parrhesia, as an act of courage, that Socrates uses the metaphor of the helmsman to hold up to the young and powerful Alcibiades the mirror of ‘the most disgraceful sort of stupidity’, which at once not only proves him unfit to govern but also gives Alcibiades the opportunity to acquire the awareness necessary for his incumbent responsibilities: Socrates: Well, if you were sailing a ship, would you be out there wondering whether to put the helm to port or starboard, and wavering because you didn’t know? […] Alcibiades: I’d leave it to the skipper. Socrates: So you don’t waver about what you don’t know if in fact you know that you don’t know. […] Don’t you realize that the errors in our conduct are caused by this kind of ignorance, of thinking that we know when we don’t know? […] the sorts of people who don’t think they know how to do things make no mistakes in life, because they leave those things to other people. […] Well, who are the ones making the mistakes? Surely not the ones who know? […] Since it’s not those who know, and it’s not those who don’t know and know they don’t

The ‘Mental State of Noise’

215

know, is there anyone left except those who don’t know but think they do know? […] So this is the ignorance that causes bad things; this is the most disgraceful sort of stupidity […]. (Plato 1997, 574–75)

Alcibiades is unfit to govern, not because he is young and lacks knowledge, but because he cannot conceive that there may be unintended consequences of his actions. His arrogance is the fruit not of lack of knowledge, but of his ignorance of this lack of knowledge. The risk of wavering, of putting the helm to port or starboard too abruptly, of underestimating the powers he modulates through the use of the tiller thus stands for Alcibiades’s lack of knowledge of the unintended consequences of his actions. Alcibiades lacks knowledge of the limits of his own competence: not only will he fail to predict the risk of unintended consequences of his action, his arrogance relies on the fact that he cannot even fathom this risk, in contradistinction with Socrates, who uses the dialectical method as ‘feedback’ in full knowledge of the risk he incurs that Alcibiades may rise to power and take revenge for the humiliation. Alcibiades lacks self-control, because the self-control Socrates speaks of comes from both the knowledge of one’s own power and the knowledge of one’s own ignorance. What Socrates leads him to understand is that Alcibiades wants to be master, when he is only the product of power: his opinion sways one way or another according to persuasion. Unlike the well-adjusted action of the helmsman, Alcibiades’ opinions are driven passively, subject to the powers that prevail, rather than being the expression of a subject, asserting itself in the face of contingency. In other words, the information Alcibiades needs in order to govern is both knowledge of the art of governance, of what is the good and just conduct, and knowledge of contingency, in the form of knowledge of his own ignorance. Wiener’s return to the metaphorical relation between the governor and the helmsman allows us to compare Alcibiades’ wavering and confusion as an inexperienced political governor, to the wild oscillation or hunting of the ill-adjusted tiller. Also Sand and Ratey’s definition of the ‘catastrophic reaction’, whereby any attempt to overcome the ‘mental state of noise’ ‘only adds to confusion’, can be understood, like the overshooting of the rudder, as an overcompensation, causing amplification of turbulence and ultimately breakdown. By analogy, noise becomes a political, ethical and epistemological problem in equal measure: lack of knowledge of one’s ignorance makes noise an epistemological problem of the docte ignorance; power without the knowledge of unintended consequences makes noise a political problem of good governance; and, the relation between contingency and self-governance, finally, makes noise an ethical problem of the just conduct.

216

An Epistemology of Noise

Taken together, the mathematical requirements of a critical definition of the limits of controllability, and the cautionary role of the helmsman metaphor in the Platonic dialogue, allow us to bring a nuance to either the purely utilitarian definition of control or the entirely negative association of control with the abuse of power. The notion of control is insufficiently understood when it is seen exclusively as leading to a totalitarian exercise of power – even though there is no doubt that the knowledge of control theory is put to efficient use in mechanizing human decisions and rationalizing human action with the objective of efficiency and frictionless exercise of power. Yet, what is lost, if the cybernetic concept of control is reduced to the idea of totalitarian domination without noise, is the nexus between knowledge and ignorance, in the form of systematic epistemic humility: the knowledge of noise, of non-linear dynamics and of the mathematical limits of control. The important status of noise in cybernetics thus allows us to add a nuance to the cybernetic paradigm, often overlooked when the focus is – rightly – on the confluence of technological, political and commercial control in an information intensive and globally networked economy (Mersch 2013). The point is not to soften the critique of cybernetic ‘control’ society, but to think freely, in order to rethink control as self-control in the form of epistemic humility before unintended consequences of actions. In turn, cybernetics itself is an incomplete paradigm, if it is concerned only with adjusting the trajectory of a self-directing missile and fails to consider the problem of the just act, as a problem involving contingency, ignorance and unintended consequences. In light of the cybernetic notion of control and interpreted in view of the Alcibiades dialogue as self-control through knowledge of one’s own ignorance, the problem of noise thus acquires an ethico-political dimension that is inseparable from the epistemological problem it poses. It requires us to distinguish the manifold objects of the definition of noise (acoustic, cognitive, biological, thermodynamic noise, etc.) from the normative power to regulate complex systems endowed with recursivity or feedback. The individual trapped in the ‘mental state of noise’ is, like the protagonist of a drama, bereft of certainties: incapacitated to judge the unpredictable nature of stimuli as either useful (information) or potentially harmful (noise) noise becomes equivalent with an undecidable state. Noise here becomes an unthinkable freedom of choice. The individual is nevertheless required to assert a decision, and to assert it on the grounds of nothing but this necessity to either appropriate this overwhelming experience as his or her own or face break down.

The ‘Mental State of Noise’

217

The catastrophic reaction to this state of indecision is the compulsive imposition of order, closure and the precipitation into meaning. For the ‘chameleon poet’, like Keats, but also for the resistance philosopher, like Canguilhem, and by default for every scientist working at the edge of what is known and can be proven, the willful confrontation with uncertainty becomes the occasion for a self-grounding of art, science or reason. What the individual has lost in the ‘mental state of noise’ are the physiological a priori of the healthy nervous system, which filter incoming stimuli according to pre-conscious criteria of pertinence, but also according to the adaptability of a posteriori rules which experience had consolidated over time as pre-set attitudes. Such pre-set attitudes feed into our cultural preconceptions and even into our epistemic a priori, notably the sense of duration and the sense of self. However, when these are powerless in the face of contingency, life calls for new norms of living. Reason too, calls for new a priori when its paradigms are called into question by radical uncertainty. This was the case with Lobachevsky’s answer to the impossibility of proving Euclid’s fifth postulate, which occasioned the invention of non-classical geometry with infinite dimensions. This recasting of geometry continued to throw waves in the epistemic field, precipitating the crisis of foundation of mathematics and the pluralization of logic. In this sense, noise, understood as maximum uncertainty, is what calls forth and hence precedes the normativity of reason, i.e. the judgement according to which uncertainty is valued as informative or discarded as spurious. This is why we can now interpret the problem of noise, its illegitimate and therefore abnormal character, in light of Canguilhem’s enigmatic dictum about the priority of the abnormal: Consequently it is not paradoxical to say that the abnormal, while logically second, is existentially first. (Canguilhem, 1991, p. 243)

We can now think of noise in terms of a fundamental epistemological contingency, a state of suspension or indecision, from which reason emancipates itself with acts of self-grounding. As the groundlessness that necessarily precedes our own rational self-grounding, noise is is no longer marginal to philosophical discourse, no longer reducible to mere error. Noise can, instead, be thought as as a fundamental philosophical problem: as the groundlessness that necessarily precedes reason’s act of self-grounding. What is at stake with the question of noise, is ultimately a vital and epistemological normativity, an emancipatory act of self-grounding, that is conditioned by no ready-made control law, grounded in nothing but itself.

218

An Epistemology of Noise

A cybernetics worthy of its noble philosophical heritage, a cybernetics of the just act, would thus not only be a cybernetics capable of speaking truth to power. The cybernetics of a just act, capable of harnessing powers far greater than itself, is one that cultivates the knowledge of its own ignorance, rather than boasting with the mechanical reduction of complexity; one that cultivates control as self-control and fine-tuning of doubt, rather than as an excuse for the hubris of domination. The transdisciplinary proliferation of concepts of noise, far from merely indicating error and perturbation everywhere, may be the index of a renewed and more confident scientific, artistic and philosophical culture of qualified uncertainty.

Notes 1 Thanks to Inigo Wilkins for drawing my attention to this. 2 On the ongoing cultural and economic shift away from a tradition of deep focus towards the requirement of polyfocal ‘hyper attention’, see Katherine Hayles, ‘Hyper and Deep Attention: The Generational Divide in Cognitive Modes’, Profession, pp. 187–199. 3 L’instinct de conservation peut apparaître comme un trait essentiel de l’organisme, bien qu’en réalité, la tendance à la conservation soit un phénomène de maladie, de vie ‘qui décline’ (Goldstein 1983, 355). 4 (Maxwell 1868). 5 Although the Alcibiades dialogue was traditionally attributed to Plato, D. S. Hutchison argues in his introduction to the dialogue that similarity with later Academic doctrine and the simplicity of the dialogue point to an Academic philosopher writing in the 350s or just afterwards and that it displays similarity notably with the Aristotelian idea of the Magna Moralia that self-knowledge is best gained through a philosophical friendship in which we see ourselves, as in a mirror (Plato 1997, 558). 6 My translation (Foucault 2009; Foucault and Davidson 2012).

Bibliography Abramowicz, Lisa. 2017. ‘A Synthetic CDO by Any Other Name Is Still Risky’. Bloomberg Gadfly. Accessed 18 July 2017. https://www.bloomberg.com/gadfly/ articles/2017-02-03/a-synthetic-cdo-by-any-other-name-is-still-risky Allen, Robert, ed. 2000. The New Penguin English Dictionary. 1 edition. London and New York: Penguin Books Ltd. Allen, Robert. 2001. ‘Information’. In The New Penguin English Dictionary. London: Penguin Group. Alliez, Eric. 2010. ‘What Is – or What Is Not – Contemporary French Philosophy, Today?’ Radical Philosophy. Accessed 31 May 2017. https://www.radicalphilosophy. com/article/what-is-%e2%80%93-or-what-is-not-%e2%80%93-contemporaryfrench-philosophy-today Alloway, Tracy, and Michael Mackenzie. 2013. ‘Boom-Era Credit Deals Poised for Comeback’. Financial Times. Accessed 18 July 2017. https://www.ft.com/ content/1d2d33d8-5ba5-11e3-848e-00144feabdc0 Altmann, Jürgen. 2001. ‘Acoustic Weapons – A Prospective Assessment’. Science & Gobal Security 9: 165–234. Anderson, M. 1988. The American Census. A Social History. New Haven: Yale University Press. Ashby, Ross. 1962. ‘Principles of the Self-Organizing System’. In Principles of SelfOrganization: Transactions of the University of Illinois Symposium, edited by H. Von Foerster and G. W. Zopf, 255–78. London: Pergamon Press. http://csis.pace .edu/~marchese/CS396x/Computing/Ashby.pdf Atlan, H, and A. Katzir-Katchalsky. 1973. ‘Tellegen’s Theorem for Bond Graphs. Its Relevance to Chemical Networks’. Currents in Modern Biology 5 (2): 55–65. Atlan, Henri. 1979. Entre Le Cristal et La Fumée. Paris: Seuil. Attali, Jacques. 1985. Noise: Political Economy of Music. Minneapolis: University Of Minnesota Press. Ayache, Elie. 2015. The Medium of Contingency: An Inverse View of the Market by Elie Ayache. London: Palgrave Macmillan. Bachelard, Gaston. 2003. Le Nouvel Esprit Scientifique. Paris: Presses Universitaires de France – PUF. Bachelier, L., 1900. Théorie de la spéculation. Annales Scientifiques E.N.S. 17, 21–86. Accessed 29 January 2018. http://archive.numdam.org/article/ ASENS_1900_3_17__21_0.pdf

220

Bibliography

Badiou, Alain. 1993. ‘Y-a-Til Une Théorie Du Sujet Chez Georges Canguilhem?’ In Georges Canguilhem. Philosophe, Historien Des Sciences, 295–304. Paris: Albin Michel. Badiou, Alain. 2005. Being and Event. London: Continuum. Balibar, Françoise, Roland Lehoucq, and Jean-Marc Lévy-Leblond. 2005. Qu’est-Ce Que La Matière ? Paris: Editions le Pommier. Barber, Kenneth F. and Jorge J.E Gracia. ed. 1994. Individuation and Identity in Early Modern Philosophy, Descartes to Kant. Albany : State University of New York Press. Bawden, David, and Lyn Robinson. 2009. ‘The Dark Side of Information: Overload, Anxiety and Other Paradoxes and Pathologies’. Journal of Information Science 35 (2): 180–91. doi:10.1177/0165551508095781. BBC. 2008. ‘Company Defends Anti-Teen Device’. BBC: Company Defends Anti-Teen Device. Accessed 12 February. http://news.bbc.co.uk/1/hi/wales/7240306.stm Bell, E. 1986. Men of Mathematics. 1st Touchstone edition. New York: Simon & Schuster. Benzecri, J. P. 1982. Histoire et Préhistoire de l’analyse Des Données. Paris: Dunod. Berglund, Brigitta, and Thomas Lindvall, eds. 1995. ‘Guidelines for Community Noise. Document Prepared for WHO’. Archives of the Centre for Sesory Research. Stockholm University and Karolinska Institute 2 (1): iii–140. Bijsterveld, Karin. 2001. ‘The Diabolical Symphony of the Mechanical Age: Technology and Symbolism of Sound in European and North American Noise Abatement Campaigns 1900–1940’. Social Studies of Science 31 (1): 37–70. Bijsterveld, Karin. 2003. ‘City of Din: Decibels, Noise and Neighbors in the Netherlands, 1910–1980’. OSIRIS 18: 173–93. Blanché, Robert. 2009. L’axiomatique. Paris: Presses Universitaires de France – PUF. ‘Bluebrain | EPFL’. 2013. Accessed 11 March. http://bluebrain.epfl.ch/ ‘BMJ Careers – Doctors’ Own Mental Health Issues’. 2017. Accessed 31 May. http:// careers.bmj.com/careers/advice/Doctors%E2%80%99_own_mental_health_issues Boutin, Aimee. 2015. City of Noise: Sound and Nineteenth-Century Paris. Urbana: University of Illinois Press. Brassier, Ray. 2007. ‘Multitudes Web – 22. Genre Is Obsolete’. Genre Is Obsolete. Accessed 7 March 2013. http://multitudes.samizdat.net/Genre-is-Obsolete Brillouin, Leon. 1949. ‘Life, Thermodynamics and Cybernetics’. American Scientist 37: 554–68. Brillouin, Leon. 2013. Science and Information Theory: Second Edition. 2nd revised edition. Mineola: Dover Publications Inc. Bunz, Mercedes. 2012. Die stille Revolution. Berlin: Suhrkamp Verlag Gmbh. Cage, John, Dieter Daniels, and Hartware-Medien-Kunst-Verein (Dortmund). 2012. Sounds like Silence: John Cage 4'33' Silence Today: 1912–1952–2012. Leipzig: Spector Books. ‘Cahiers Pour l’Analyse (An Electronic Edition)’. 2017. Accessed 29 May. http://cahiers. kingston.ac.uk/ Cahiers Pour l’analyse N° 9/Genealogie Des Sciences. 1968. Seuil/travaux du cercle d’epistemologie.

Bibliography

221

‘Calculus of Variations | Mathematics’, 2013. Encyclopedia Britannica. Accessed 9 June 2017. https://www.britannica.com/topic/calculus-of-variations-mathematics Campbell, Sarah. 2008. ‘Now Crime Gadget Can Annoy Us All’. BBC. Accessed 1 January 2015 sec. UK. http://news.bbc.co.uk/1/hi/uk/7759818.stm Canguilhem, Georges. 1991. The Normal and the Pathological. Translated by Carolyn R. Fawcett. New edition. New York: Zone Books. Canguilhem, Georges. 1993. ‘Le Cerveau et La Pensée’. In Georges Canguilhem. Philosophe, Historien Des Sciences. 1st edn., 11–36. Paris: Albin Michel. Canguilhem, Georges. 2000. A Vital Rationalist: Selected Writings from Georges Canguilhem. New edn. New York: Zone Books. Canguilhem, Georges. 2009. Le Normal et Le Pathologique. Presses Universitaires de France - PUF. Capurro, Rafael, and Birger Hjorland. 2003. ‘The Concept of Information’. Annual Review of Information Science and Technology 37: 343–411. Castellengo, Michele. 1994. ‘Les formes sonores’. In Les sciences de la forme aujourd’hui, edited by Yves Bouligand, 127–39. Paris: Ed. du Seuil. Cavaillès, Jean. 2000. Sur La Logique et La Théorie de La Science. Paris: Vrin. Coll. INSEE. 1987. Pour Une Histoire de La Statistique. Edited by J. Affichard. Vol. 2. Paris: Economica. Collectif. 1979. Colloque de Royaumont: Theories Du Langage, Theories de l’apprantissage. Paris: Seuil. ‘Control Theory | Mathematics’. 2017a. Encyclopedia Britannica. Accessed 8 June 2017. https://www.britannica.com/topic/control-theory-mathematics Cowan, J. D. 1965. ‘The Problem of Organismic Reliability’. Progress in Brain Research 17: 9–63. ‘Cpa9.17.Cuvier.Pdf ’. 1968. Accessed 10 June 2017. http://cahiers.kingston.ac.uk/pdf/ cpa9.17.cuvier.pdf Crutzen, P.J., 2002. Geology of mankind [WWW Document]. Nature. https://doi .org/10.1038/415023a Cummings, Dominic. 2017. ‘Dominic Cummings: How the Brexit Referendum Was Won’, 2017. The Spectator. Accessed 1 May 2017. https://blogs.spectator. co.uk/2017/01/dominic-cummings-brexit-referendum-won/ Cusick, Suzanne. 2006. ‘Music as Torture’. TRANS - Revista Transcultural de Música Transcultural Music Review, no. 10. http://www.sibetrans.com/trans/articulo/152 /music-as-torture-music-as-weapon Cuvier, M. Le Baron G. 1837. Histoire Des Progrès Des Sciences Naturelles : Depuis 1789 Jusqu’à ce Jour. Bruxelles: Société Belge de Librairie. Accessed 14 February 2018. https://archive.org/details/histoiredesprog02cuvi. Desrosieres, Alain. 2006. ‘Statistique’. In Dictionnaire d’histoire et philosohie des sciences, 4th edition, 1008–24. Paris: puf. Diaz Nafria, Jose Maria. 2010. ‘What Is Information? A Multidimensional Concern’. Triple C 8 (1): 77–108.

222

Bibliography

Dodig Crnkovic, Gordana, and Wolfgang Hofkirchner. 2011. ‘Floridi’s “Open Problems in Philosophy of Information” Ten Years Later’. Information Open access (2): 327–59. doi:10.3390/info2020327. Eigen, M. 1971. ‘Selforganization of Matter and the Evolution of Biological Macromolecules’. Die Naturwissenschaften 58 (10): 465–523. Eisler, Zoltan, Janos Kertesz, and Fabrizio Lillo. 2007. ‘The Limit Order Book on Different Time Scales’. Proc. SPIE 6601 Conference Volume 6601 (Noise and Stochastics in Complex Systems and Finance): 66010G. Elliott, Larry. 2017. ‘Bank of England Warns of Complacency over Big Rise in Personal Debt’. The Guardian. Accessed 29 July 2017, sec. Business. http://www.theguardian .com/business/2017/jul/24/bank-of-england-household-debt-bank-credit-card-carloans Enzensberger, Hans Magnus. 1977. Der kurze Sommer der Anarchie: Buenaventura Durrutis Leben und Tod. Roman. 19th edition. Frankfurt am Main: Suhrkamp Verlag. Esposito, Elena. 2011. The Future of Futures: The Time of Money in Financing and Society. Cheltenham and Northampton: Edward Elgar Publishing Ltd. Floridi, Luciano. 2002. Philosophy and Computing: An Introduction. 1st edition. London: Routledge. Floridi, Luciano. 2010. Information. Oxford and New York: Oxford University Press. Floridi, Luciano. 2014. The Fourth Revolution: How the Infosphere Is Reshaping Human Reality. 1st edition. Oxford: Oxford University Press. Foucault, Michel. 1989. ‘Introduction’. In The Normal and the Pathological. Translated by Carolyn R. Fawcett, 7–24. New York: Zone Books. Foucault, Michel. 2009. Le gouvernement de soi et des autres : Tome 2, Le courage de la vérité – Cours au Collège de France. Paris: Seuil. Foucault, Michel. 2011. Der Mut zur Wahrheit: Die Regierung des Selbst und der anderen II. Vorlesungen am Collège de France 1983/84. Translated by Jürgen Schröder. Berlin: Suhrkamp Verlag Ag. Foucault, Michel. 2012. The Courage of Truth: The Government of Self and Others II: 1983–1984. Translated by Graham Burchell. Edited by Arnold I. Davidson. Reprint edition. New York: Palgrave Macmillan; Picador USA. Frosch, John. 1983. The Psychotic Process. New York : International Universities Press. http://archive.org/details/psychoticprocess00frosrich Gelman, Andrew, and Cosma Rohilla Shalizi. 2013. ‘Philosophy and the Practice of Bayesian Statistics’. British Journal of Mathematical and Statistical Psychology 66: 8–38. Gibson, Kirsten, and Ian Biddle, eds. 2016. Cultural Histories of Noise, Sound and Listening in Europe, 1300–1918. Burlington: Routledge. Glass, David C., and Jerome E. Singer. 1972. Urban Stress: Experiments on Noise and Social Stressors. New York: Academic Press. ‘Global Economic Prospects | Data’. 2017. Accessed 19 July 2017. http://data.worldbank. org/data-catalog/global-economic-prospects

Bibliography

223

Goldstein, Kurt. 1948. Aftereffects of Brain Injuries in War: Their Evaluation and Treatment; the Application of Psychologic Methods in the Clinic. New York: Grune & Stratton. Goldstein, Kurt. 1983. Structure de l’organisme. Paris: Gallimard. Goodman, Steve. 2012. Sonic Warfare: Sound, Affect, and the Ecology of Fear. Cambridge, MA: The MIT Press. Greenacre, Phyllis. 1952. Trauma, Growth, and Personality. New York: Norton. Grene, M. 2007. ‘La vie des sciences et les sciences de la vie’. In Conceptions de La Science : Hier, Aujourd’hui, Demain. Hommage à Marjorie Grene, edited by J. Gayon and R. M. Burian, 23–36. Paris: Ousia. Grossman, Sanford J. and Joseph Stiglitz. 1980. ‘On the Impossibility of Informationally Efficient Markets’. American Economic Review 70: 393–408. Hacking, Ian. 1990. The Taming of Chance. Cambridge: Cambridge University Press. Hainge, Greg. 2012. Noise Matters: Towards an Ontology of Noise. London: Continuum Publishing Corporation. Hallward, Peter and Knox Peden ed. 2012a. Concept and Form, Volume 1. London and Brooklyn: Verso. Hallward, Peter and Knox Peden ed. 2012b. Concept and Form, Volume 2. London and Brooklyn: Verso. Harris, Ben. 1979. ‘Whatever Happened to Little Albert?’ American Psychologist 34 (2): 151–60. doi:10.1037/0003-066X.34.2.151. Harrison, Charles, and Paul Wood, eds. 1999. Art in Theory, 1900–1990. An Anthology of Changing Ideas. 1st edition. Oxford and Cambridge, MA: Blackwell. Hegarty, Paul. 2007. Noise/Music: A History. New York: Continuum. Herbert, Martin 2016. Tell Them I Said No. Berlin: Sternberg Press. Hirsch, G. 1976. ‘Langage et Pensée Mathématiques’. In Langage et Pensée Mathématique, 42–62. Luxembourg: Actes, Colloque International. Centre Universitaire de Luxembourg. Hydaralli, Saeed. 2012. ‘Noise: An Inquiry into Its Formal Properties’. In Reverberations, edited by Michael Goddard, Benjamin Halligan, Paul Hegarty, 221–32. London: Continuum. ‘Information’. 2014a. Oxford Dictionaries. Oxford University Press. http://www .oxforddictionaries.com ‘Information’. 2014b. Merriam-Webster. London: Merriam-Webster, Encyclopedia Brittanica Company. http://www.merriam-webster.com/ Jaen-Marie Pierrel (first) dir., William eds. del-Mancino, Laurent Gobert, Jean-Marc Humbert, and Etienne Petitjean. n.d. ‘Nécessité’. Centre National de Ressources Textuelles et Lexicales (CNRTL). http://www.cnrtl.fr/definition/nécessité Janich, Peter. 2006. Was ist Information?: Kritik einer Legende. Frankfurt am Main: Suhrkamp. Judah, Hettie. 2017. ‘Don’t Look Now: The Artists Who Turn Their Backs on the World’. The Guardian. Accessed 25 June 2017, sec. Art and design. https://www.theguardian. com/artanddesign/2017/apr/20/tell-them-i-said-no-artists-who-turn-backs-onworld-martin-herbert

224

Bibliography

Keats, John. 1818. ‘Letter to Richard Woodhouse’. Accessed 10 August 2014. http://www .john-keats.com/briefe/271018.htm Keats, John. 1958. The Letters of John Keats, 1814–1821. Cambridge, MA: Harvard University Press. Kindleberger, Charles P., and Robert Z. Aliber. 2011. Manias, Panics and Crashes: A History of Financial Crises. 6th edition. New York: Palgrave Macmillan. Kyle, Albert S. 1985. ‘Continuous Auctions and Insider Trading’. Econometrica 53: 1315–36. LaBelle, Brandon. 2006. Background Noise: Perspectives on Sound Art. 1st edition. London: Continuum. Larousse, Éditions. 2017. ‘Définitions: Entropie – Dictionnaire de Français Larousse’. Accessed 10 May 2017. http://www.larousse.fr/dictionnaires/francais/entropie/30093 Lemaine, Jean-Marie. 1960. ‘Schachter Stanley, The Psychology of Affiliation’. Revue française de sociologie 1 (4): 499–500. Lévy-Leblond, Jean-Marc. 1976. ‘Us et Abus de Langage: Mathématique, Didactique, Physique …’ In Langage et Pensée Mathématique. Luxembourg: Actes, Colloque International. Centre Universitaire de Luxembourg. Locke, John. 1975. An Essay Concerning Human Understanding, ed. P. H. Nidditch. Oxford: Clarendon Press. III, 27.3, 409. Lockwood, Dean. 2012. ‘Lovecraft’s Weird Ecology of Noise’. In Reverberations: The Philosophy, Aesthetics and Politics of Noise, edited by Paul Hegarty, Michael Goddard, and Benjamin Halligan, 73–83. London: Continuum Publishing Corporation. ‘MA Rotterdam, Proceedings City Council Rotterdam, Session 26 February 1914, 64–65’. in. Malaspina, Cecile. 2012a. ‘The Noise Paradigm’. In Reverberations: The Philosophy, Aesthetics and Politics of Noise, edited by Michael Goddard, Benjamin Halligan and Paul Hegarty, 58–72. London: Continuum. Malaspina, Cecile. 2014. ‘Epistemic noise’. Systema: Connecting Matter, Life, Culture and Technology 2 (1): 48–58. Manson, Phillip. 2014. The Philosophical Implications of Information Overload. Seattle: CreateSpace Independent Publishing Platform. Marcel Ducham p. 1917. ‘The Richard Mutt Case’, The Blind Man. New York. Maxwell, J. C. 1868. ‘On Governors’. Proceedings of the Royal Society 100: 270–83. Mersch, Dieter. 2013. Ordo Ab Chao - Order from Noise. Zürich: Diaphanes Verlag. Monod, Jacques. 1973. Le Hasard et La Necessite. Essai Sur La Philosophie Naturelle de La Biologie Moderne. Paris: POINTS. Morange, Michel. 2005. ‘What Is Life? A New Look at an Old Question’. In Logic, Methodology and Philosophy of Science, edited by Petr Valdes-Villanueva Hajek, Dag Luis Westerstahl, 431–40. King’s College Publications. Morange, Michel. 2006. ‘Information’. In Dictionnaire d’histoire et Philosohie Des Sciences, edited by Dominique Lecourt et al, 618–21. Paris: Puf. Morin, Edgar. 2008. La méthode : Coffret en 2 volumes. Paris: Seuil.

Bibliography

225

Moulin, Anne Marie. 2006. ‘Téléologie’. In Dictionnaire d’histoire et Philosohie Des Sciences, 4th edition, edited by Dominique Lecourt, 1072–74. Paris: puf. Neumann, J. von. 1932. Mathematische Grundlagen Der Quantenmechanik. Berlin: Springer Verlag. Neumann, J. von. 1956. ‘Probabilistic Logics and the Synthesis of Reliable Organisms from Unreliable Components’. In Automata Studies, edited by C. E. Shannon and J. McCarth, 43–98. Princeton: Princeton University Press. Nunes, Mark, ed. 2012. Error: Glitch, Noise, and Jam in New Media Cultures. New York: Continuum Publishing Corporation. Pascal, Blaise. 1966. Pensées. Paris: Bordas. Pascal, Blaise, and Michel Le Guern. 1987. Les Provinciales. Paris: Gallimard. Piaget, Jean. 1999. Play, Dreams and Imitation in Childhood. London: Routledge. Plato. 1997. ‘Alcibiades’. In Plato Complete Works, edited by J. M. Cooper, translated by D. S. Hutchison, 556–95. Indianapolis and Cambridge: Hackett Publishing Co, Inc. Porter, T. 1986. The Rise of Statistical Thinking. Princeton: Princeton University Press. Preda, Alex, Mike O’Hara, and Ricky Treadwell. 2013. ‘Humans vs. Robots: Man Fights Back’. HFT Review, High Frequency and Algorithmic Trading. Accessed 30 July. https://c.mql5.com/forextsd/forum/109/alex_preda-hftr-final.pdf Preda, Alex. 2017. Noise: Living and Trading in Electronic Finance. Chicago: University of Chicago Press. Prigogine, Ilya, and Isabelle Stengers. 2009. Entre Le Temps et l’éternité. Paris: Flammarion. Regnault, Jules. 1863. Calcul des chances et philosophie de la bourse. Paris: Mallet-Bachelier [et] Castel. Accessed 29 July 2017. http://archive.org/details/calculdeschances00regn Ruyer, Raymond. 1954. La Cybernétique et l’origine de l’information. Paris: Flammarion. Sands, S, and J. J. Ratey. 1986. ‘The Concept of Noise’. Psychiatry 49 (4): 290–97. Schachter, S. 1959. The Psychology of Affiliation. Stanford: Stanford University Press. Schafer, R. Murray. 1994. Soundscape: Our Sonic Environment and the Tuning of the World. Original edition. Rochester: Destiny Books. Schaufeli, Wilmer B., Michael P. Leiter, and Christina Maslach. 2009. ‘Burnout: 35 Years of Research and Practice’. Career Development International 14 (3): 204–20. doi:10.1108/13620430910966406. Schneirla, T. 1959. ‘An Evolutionary and Developmental Theory of Biphasic Processes Underlying Approach and Withdrawal’. In Nebraska Symposium on Motivation, 1959, edited by M. R. Jones, 1–42. Oxford: University of Nebraska Press. Schopenhauer. 1851. ‘Über Lärm Und Geräusch’. In Parerga Und Paralipomena, 2, 517–19. Berlin: A. W. Hain. Schrödinger, Erwin. 1945. What Is Life? The Physical Aspect of the Living Cell. Cambridge: Cambridge University Press. Schumpeter, L. 2010. ‘The Silver Tsunami | The Economist’. Accessed 18 July 2017. http://www.economist.com/node/15450864 Shannon, Claude E., and Warren Weaver. 1964. The Mathematical Theory of Communication. Urbana: University of Illinois Press.

226

Bibliography

Shenk, David. 1997. Data Smog: Surviving the Information Glut. San Francisco: HarperSanFrancisco. Simondon, Gilbert. 2005. L’individuation à La Lumière Des Notions de Forme et d’information. Grenoble: Editions Jérôme Millon. Simondon, and JohnGilbert, trans. Cecile M. and John R. 2017. On the Mode of Existence of Technical Objects. Minneapolis: University of Minnesota Press. Simondon, G., N. Simondon, J.-Y. Chateau, 2010. Communication et information: Cours et conférences. Editions de la Transparence. Smith, Mark M., ed. 2004. Hearing History: A Reader. Athens: University of Georgia Press. Smith, Steven. 2002. ‘Statistics, Probability and Noise’. In Digital Signal Processing: A Practical Guide for Engineers and Scientists. 3rd Revised edition, 11–34. Amsterdam and Boston: Newnes. Snell, James Laurie. 2012. Introduction to Probability. Providence: American Mathematical Soc. Stigler, S. 1986. The History of Statistics. The Measurement of Uncertainty Before 1900. Cambridge, MA: Harvard University Press. Taleb, Nassim Nicholas. 2008. The Black Swan: The Impact of the Highly Improbable. London: Penguin. Tamir, Eshel (ed.) 2005. Defense Update, International Online Defense Magazine. ‘Long Range Acoustic Device - LRAD’. Accessed 1 January 2005. http://www.defenseupdate.com/products/l/LRAD.htm. LRAD Corporation website. http://www.lradx. com/site/content/view/15/110/ Tetlock, Paul C. 2006. ‘Does Noise Trading Affect Securities Market Efficiency?’ http:// www.columbia.edu/~pt2238/papers/Tetlock_Noise_and_Efficiency_09_06.pdf Thomas, Gavin. 2012. ‘Sonic Device Deployed for Games’. BBC, Accessed 1 January 2015, sec. London. http://www.bbc.co.uk/news/uk-england-london-18042528 Thompson, Christopher. 2015. ‘Synthetic CDO Volumes Double amid Hunt for Yield’. Financial Times. Accessed 18 July 2017. https://www.ft.com/content/cb8d80d8-a32311e4-9c06-00144feab7de Tonnelat. 1996. Thermodynamique et Biologie. Paris: Maloine. ‘Transcript of IMF Seminar – The Risk of a U.S. Hard Landing and Implications for the Global Economy and Financial Markets’. 2007. Accessed 15 July 2017. http://www. imf.org/en/News/Articles/2015/09/28/04/54/tr070913z Voegelin, Salome. 2010. Listening to Noise and Silence: Toward a Philosophy of Sound Art. New York: Continnuum-3PL. Volcler, Juliette, and Carol Volk. 2013. Extremely Loud : Sound as a Weapon. New York: THE NEW PRESS. Von Humboldt, W., 1836. Über die Verschiedenheit des menschlichen Sprachbaues und ihren Einfluss auf die geistige Entwickelung des Menschengeschlechts. Berlin: Königliche Akademie der Wissenschaften. Von Humboldt, Wilhelm, and Albert Leitzmann. 1903. Gesammelte Schriften. B. Behr. Accessed 18 December 2014. http://archive.org/details/gesammelteschri03berlgoog

Bibliography

227

Walter, Christian, and Michel de Pracontal. 2009. Le Virus B, Crise Financière et Mathématiques. Paris: Seuil. Watkins, Calvert, ed. 2011. The American Heritage Dictionary of Indo-European Roots. 3rd edition. Houghton and Mifflin: Harcourt Publishing Company Inc. Watson, John B., and Rosalie Rayner. 1920. ‘Conditioned Emotional Reactions’. Journal of Experimental Psychology 3 (1): 1–14. Weizsäcker, Carl F. von. 1994. Aufbau Der Physik. München: DTV Deutscher Taschenbuch Verlag. ‘What Are Conduits, SIVs and SIV-Lites?’ 2017. Accessed 16 July 2017. http:// uk.reuters.com/article/factbox-conduits-sivs-idUKNOA13480720070911 ‘White Phosphorus: Weapon on the Edge’. 2005. Accessed 16 November 2005. http:// news.bbc.co.uk/1/hi/world/americas/4442988.stm ‘WHO Guidelines for Community Noise’. 1999. Geneva: World Health Organization. Accessed 17 June 2014. http://www.who.int/docstore/peh/noise/Comnoise-1.pdf Wiener, Norbert. 1961. Cybernetics; or, Control and Communication in the Animal and the Machine. New York: The MIT Press. Wiese, Wanja, and Thomas K. Metzinger. 2017. Vanilla PP for Philosophers: A Primer on Predictive Processing. Philosophy and Predictive Processing. Frankfurt am Main: MIND Group. http://predictive-mind.net/papers/vanilla-pp-for-philosophers-a -primer-on-predictive-processing/getAbstract. Wilkins, Inigo. Forthcoming. Irreversible Noise. Falmouth: Urbanomic. Wilson, Paul. 2012. ‘Physical Spectatorship: Noise and Rape in Irreversible’. In Reverberations: The Philosophy, Aesthetics and Politics of Noise, edited by Benjamin Halligan, Paul Hegarty, and Michael Goddard, 122. London: Continuum. Wilson, Robert W. 1978. ‘The Cosmic Microwave Background Radiation’. Accessed 6 August 2014. www.nobelprize.org http://www.nobelprize.org/nobel_prizes/physics/ laureates/1978/wilson-lecture.pdf Winograd, S., and J. D. Cowan, 1963. Reliable Computation in the Presence of Noise. Massachusetts: The MIT Press. Zimmermann, Rainer. 1988. ‘Ordnung Und Unordnung - Zum Neueren Determinismusstreit Zwischen Thom Und Prigogine’. Lendemains, 50: 60–73.

Index ABCP. See asset-backed commercial papers (ABCP) Abramowicz, Lisa 127–8 absurd pessimism 130 accidental entropy 198 accidental information 97–101 acoustic noise 143, 149, 153, 156, 170 additive white Gaussian noise (AWGN) 99 afferent noise 98, 127 Alcibiades 213–18 lack of knowledge 215 wavering and confusion 215 alethurgy 214 Aliber, Robert 130 alternative mathematical models 130 Altmann, Jurgen 157 ambiguity, margin of 32 analogy 2 ‘analytical’ philosophy 88 Anglo-Saxon humanities 87 Anglo-Saxon philosophy 87 Anglo-Saxon theory of knowledge 87 anxiety 195–8 Aristotle 44–5 Arno Penzias and Robert Wilson 109 artificial intelligence 208 art of governing 213, 215 Ashby’s law 112 asset-backed commercial papers (ABCP) 126 astigmatism of intuition 79–83 Atlan, Henri 10, 35, 75, 112 audible noise 158–9, 162 audiometers 145 AWGN. See additive white Gaussian noise (AWGN) axiomatic set theory 59 Bachelard, Gaston 8–9, 88, 117 background noise 109 Badiou, Alain 9 basis of noise 71–7

Bayesian data analysis 100 behaviour, defined 197 Bell, Eric Temple 58 Bernoulli’s law 1 beta-blockers 204 big data 2, 10 Bijsterveld, Karen 144 historical analysis 154 biosemiotics 11 bit 40 Black-Scholes option price model 123 Blanche, Robert 57–9 Boltzmann, Ludwig 4, 27, 31, 36, 200 entropy 32, 69, 93, 95 formula to physical reality 29 Brillouin, Leon 4, 17, 35, 65–9, 76, 182 negation of entropy 71 neologism 80 Canguilhem, George 10, 12, 87–90, 184–5, 191–2 concept of normativity 191–2 Cantor, Georg 58 Cartesian Method 10 catastrophe theory 89 catastrophic reaction 175, 180, 181, 187–9, 195–7, 204 definition of 215 Goldstein’s concept of 180, 191 to noise 199 psychiatric definition of 199 Cavaillès, Jean 88, 106, 116–17 CDOs. See collateralized debt obligations (CDOs) Central Limit Theorem 99 chemico-diffusional coupling 81 choice freedom of 57–60 logic of 57–60 clash of cultures 158 Clausius, Rudolf 4 clonidine 204

Index closed systems 81 closure 187–9 CMOs. See collateralized mortgage obligations (CMOs) cognition, processes of 207 cognition, process of 171 collateralized debt obligations (CDOs) 126–7 collateralized mortgage obligations (CMOs) 126 colonialization 172 combined sound energy 151 complex systems basis for 207 contemporary modelizations of 207 theory 211 confusion 169–77 contemporary humanism 44 contemporary philosophy 199 context of psychiatry 203–4 continental philosophy 86, 87 contingency 215 control 203–6 contingency vs. 203 cybernetic concept of 216 cybernetic notion of 203, 216 of noise pollution 155 controllability 216 limits of 210 control law 209 control mechanisms 207 mathematical analysis of 208 control theory 207, 209, 210 cybernetics’ origins in 207 emergence of 208 limits of 207 mathematics in 210 uncertainty in 211 Copernican revolution 46 cosmic background radiation 109–13 cosmic expansion 111 cosmic noise 109 cosmos 111–12 Crutzen, Jozef 44 crystallization 72 process of 72 cultural theory 144 culture 20 Cummings, David 6 cybernetician 112

229

cybernetics 1, 2, 5, 11, 16, 68, 170, 209, 211, 213 control of noise in 204–5 inclination 170–1 information model 69 ‘ironmongery’ 89 origins of 207 second-order 5, 205 theoretical influence of 206 theory of 205 Cybernetics (Wiener) 16 cybernetic theory 171 reliance on 171 declining life 187–9 deductive redundancy 57, 59, 82 Desrosieres, Alain 136 determinism 200 Diaz Nafria, Jose Maria 23, 24 discourse, ‘de-differentiation’ of 180 discursive ambiguity 17–18 discursive interpretation 17–18 dispersion of energy 3 docte ignorance 215 Duchamp, Marcel 104–5 dynamical state of system 209 eclecticism, impression of 171 ecology of code and vibration 161 economic development 149–50 ecosphere 2, 4 efferent noise 98, 127 Eigen, Manfred 81 Eisler, Zoltan 119, 141 electronic signal transmission 19, 31, 62 emotional agitation 204–5 empirical noise 167–8. See also noise empirical reality 8, 45, 49, 81, 94 energy in thermodynamic terms 4 entropy 4, 15–16, 39–40, 50, 62, 67, 73, 83 accidental 198 Boltzmann statistical concept of 95 definition of 32, 69, 136, 199 dissipation of energy 111–12 as ‘Freedom of Choice’ 23–6 of information (see information entropy) informational value of 79 of message 15, 98 negation of 66–7, 80, 112, 182

230 and noise 28, 202 perception of 80 physical 27–8, 132 statistical 3, 93, 95, 200 environmental noise 156 Enzensberger, Hans Magnus 115 epistemic humility 8, 140 epistemological noise 7–10, 33, 60, 69, 95 epistemological normativity 106, 207 epistemological obstacles 8–9, 64 epistemology 11 essential human trait 181–4 etymology of noise 213 Euclid’s geometry 58 exercise control 202 extrinsic noise factors 175–6 Floridi, Luciano 1, 11, 93, 105 Foucault, Michel 10, 214 revaluation of error 89 freedom of choice 16, 17, 28–30, 33, 39, 51, 53–4, 57–60, 69, 74–6, 82, 95, 182, 188, 197, 198, 202 French epistemology 87, 88 Freudian notion of regression 189 Galton, Francis 140 gambling behaviour 122 Gaussian noise 98 Gelman, Andrew 97, 100–1 Goldstein, Kurt 175, 195–8 catastrophic reaction 191, 192 clinical observations 175 Goodman, Steve 160 governance 213 political 214 Greene, Marjorie 59 Hainge, Greg 11 hearing impairment 155 consequences of 155 helmsman metaphor 207–11, 214, 216 Herbert, Martin 30 homeostasis 71, 170 physiological concept of 205 hormone noradrenaline 204 humanity 44 hylomorphism 44–5 hyperbolic geometry 58 hyper complex systems 76

Index ignorance, knowledge vs. 216 imitation 188 immorality of speculation 127–8 impairment, hearing 155 inaudible noise 159 inconceivable freedom of choice 198 inconceivable multiplicity of choice 197 individuation empirical aspects of 45 process of 43–50 theory of 48 industrialization 143 inforgs 2 information 8, 16, 41, 45, 117, 133, 134, 141 concept of 4, 5, 17, 69, 71 definition of 16, 17, 23, 134, 140 entropic conception of 15 entropic definition of 61, 76 epistemological equation of 134 ethics 11 iconoclastic definition of 20 vs. noise 61–2, 198 paradigm 2 philosophy of 11 physical concepts of 35–7 politics of 6 process of 39–41, 43–50, 63, 98 qualitative definition of 47 quality of 49 quantity of 67 ready-made 103–7 Shannon’s definition of 90 statistical conception of 16 systems 11 Wiener definition of 16, 82 informational potential 82 information culture 11 information entropy 3, 4, 15–19, 24, 26–9, 31, 33, 36, 48, 51, 53, 62–3, 64, 66–69, 71–7, 83, 90, 103, 105, 109, 182, 188, 189, 197–8 of complex systems 77 Shannon’s definition of 23, 68, 76 information process 54 information theory 1, 5, 11, 16, 143 noise 1–2 infosphere 2 internal chaos 167–77 internal disposition 175, 176

Index intrinsic noise factors 175–6 intuition 79–83 astigmatism of 79–83 irreversible/irreversibility 7, 201 Keats, John 9–10, 181 negative capability 182–4 Kertesz, Janos 141 Kindleberger, Charles 130 knowledge 133 about noise 145 academic organization of 12 Anglo-Saxon theory of 87 discipline of 141 domain of 94, 136 forms of 2, 106 vs. ignorance 216 lack of 215 theory of 6–7 krach 130–1 Krugman, Paul 130 Kuhn, Thomas 93–4 kybernetes 207–11 labour, division of 141 Lacanian psychoanalysis 172 lack of knowledge 215 law of errors 1, 137 law of possibilities 1 Leriche, Rene 171 Lillo, Fabrizio 141 linguistics, structuralist analysis of 172 Lobachevsky, Nikolai Ivanovich 58 Lockwood, Dean 161 Logical Positivism 59 logic of choice 57–60 Long Range Acoustic Device (LRAD) 157 LRAD. See Long Range Acoustic Device (LRAD) magisterial discourse 116 Manichean dualism 106 market variations, real causes of 121 Markoff process 39 mathematical formalization 17 mathematical noise 167–8 mathematical theory of communication (MTC) 3–4, 15, 24, 66, 67 theoretical relevance of 52 maximal entropy 28, 48, 71–2, 80, 82, 189

231

Maxwell, James Clerk 93, 208 MBS. See mortgaged-backed security (MBS) Mead, Margaret 133 Medical Protection survey 177 mental health, assessment of 177 mental state of noise 93, 169, 170, 173–4, 176, 179, 181, 183–4, 187–9, 195, 203–6, 215–19 definition of 188 message’s entropy 98 metaphor 7 critical use of 8 of helmsman 207–11, 214 of noise 8 role of 7–8 metastability 71–3 molecular entropy in thermodynamics 27 moral dichotomy 121, 129 moral philosophy 121, 125, 129 Morange, Michel 201 Morin, Edgar 109, 111, 113, 115 mortgaged-backed security (MBS) 126–7 motor governors 207–8 MTC. See Mathematical Theory of Communication (MTC); mathematical theory of communication (MTC) necessity 51–5, 57 conceptual presence of 52 negation of entropy 16–17 negative capability 9, 181–5, 189 definition of 181 negentropic attitude 197 negentropic function 188 negentropic process 112 negentropy 3–6, 33, 36, 65–9, 74, 76–7, 80, 82–3, 182, 184, 188, 189 new communication technologies (NTCs) 105 Newtonian physics 100 noise 8, 79, 117, 150 abatement 143–8 amplification of 205 avoidance of 173 basis of 71–7 catastrophic reaction to 191–3, 199 concept of 167–9, 170, 173, 199 control of 211

232 definition of 162, 169, 216 dimensions of 144 Dutch opinion on 146 entropy and 18–19, 28, 202 epistemological aspect of 197 etymology of 135, 213 in finance 119–32 in gap between narratives 115–17 impact on learning 156 information and 2, 198 information potential of 110 knowledge about 145 mental state of 173 metaphors of 7–8 military history of 11 multiple sources of 156 participation of 75 politics of 6 problem of 150 psychiatric notion of 172 quantification of 152 quantitative measure of 150, 152–3 scientific and cultural understanding of 15 scientific and technological mastery of 93–4 sources of 110 speculative dimension of 167 as spurious uncertainty 61–4 statistical 1 stochastic models of 10 in stock market 119 theoretical polymorphism of 144 transduction of 93–6 of ventilation system 151 well-worn metaphor of 161 Noise Matters (Hainge) 11 noise pollution 2, 3, 89, 149–54 concept of 152–3 control of 155 definition of 159 sources of 151 WHO report on 148 noise toxicity 155–7 definition of 155 noise traders 120–1, 123–5, 129 non-audible noise 159, 160, 162 non-classical mechanics 93–4 ‘non-Euclidian’ geometry 58 The Normal and the Pathological (Canguilhem) 10, 87, 191, 193

Index normativity 106, 191–2, 217 norms of living 217 no-touch torture 158 NTCs. See new communication technologies (NTCs) On the Mode of Existence of Technical Beings (Simondon) 20 openness 110, 177, 188, 189 ‘essential human trait’ of 181 open systems 81 ‘optimization’ for control 209 order 199–202 from noise 2 parallel postulate 58 parrhesia 214 Pascal, Blaise 1 Penzias and Robert Wilson 109–10 perception, cultural codes of 160 Petty, William 136 phenomenological astigmatism 83 philosophical discourse 7–8 philosophy, analytic objectives of 86 Phyllis Greenacre’s 1952 study 179 physical analogy 32 physical entropy 15, 27–8, 32, 33, 54–5 physical potential, nature of 81 physical spectatorship 159 physics information and informational concepts of 35–8 statistical method in 201 Piaget, Jean 188, 189 Picavet, Emmanuel 90 Planck, Max 28 Plato 45, 213–18 Platonic inspiration 87 Plato’s concept of form 44–5 pluralization of logics 59 political governance 214 pollution noise (See noise pollution) pollution, noise 2 post-modernism 86 post-structuralism 86 ‘post-truth’ elections 214 potential information 29–33 actualization of 33 potential, physical concept of 31 Pracontal, de Michel 129–30

Index pre-anxiety tensions 179, 187 Preda, Alex 122, 124, 125, 208 predictability, advantage of 99 predictable noise 97–101 probability 15, 200 process of individuation 43–50 protocols of observation 140 proto-mythological thinking 8 psychiatric illnesses 169 psychiatry 170 psychology 8 psychopathology, forms of 173 public health 149 qualities 139–41 quantity of information 23, 28, 36, 40, 51, 65, 68, 76 quantum physics 79, 88, 93, 94, 112, 200 Quetelet, Adolphe 139 statistical idealism 141 radio noise 109 radio silence 144 Ratey, John 171–4, 179–85, 187–9, 195, 203–6, 215 ready-made information 103–7 recuperated disorganizations 73 redundancy 51–5 deductive 57 Regnault, Jules 121–3, 125, 129, 139 moral dichotomy 121 rehabilitation 192 relative entropy 51, 189 relative uncertainty 106 remote controls 207 requisite variety 75 road and air traffic noise 144 role of teacher 214 Roubini, Nouriel 130 Ruyer, Raymond 2, 80 Samuelson, Paul 122–3 Sands, Steven 169, 171–4, 179–85, 187–9, 195, 203–6, 215 Schachter, S. 180 ‘schismatic morphogenetic’ process 111 Schrödinger, Erwin 67, 80 nascent theory 68 scientific discourse 21 Scott, Peter 99 second order cybernetic system 5, 124

233

SEL. See sound exposure level (SEL) self-control 215, 216 self-critical method 172 self-governance 215 self-organization 81 ‘self,’ presuppositions of 176 self-regulating systems, Winer’s cybernetic theory of 4 self-regulation 205 sense of governing 207 serendipity, instance of 97 servomechanisms 207 Shalizi, Cosma Rohilla 97, 100–1 Shannon, Claude 3–4, 11, 17, 23, 24, 46–7, 49–52, 62, 69, 103, 137 audacity 15 concept of information 4, 25, 32–33, 39, 46–7, 64, 68, 69 conceptual audacity 85, 105 conceptual debt 16 cultural significance of 90 definition of information 18, 20, 23, 25, 35, 36, 103, 136 designation of information 27 entropic ideas 18, 19, 25, 29, 35, 36, 66, 68, 85, 89, 90, 105 formula 40 information entropy 93 information theoretical concepts 112 MTC 24 quantitative measure of information 32 shift of emphasis 43 Shapiro, Meyer 181 Shostak, Stanley 201–2 signal-to-noise ratio 151, 155–6, 162 signal transmission 20, 109, 142 mathematical theory of 3 perturbation of 110–11 Simondon, Gilbert 20, 44, 45, 48, 49, 71, 80, 94, 95 epistemological appropriation 95 theory of individuation 47 simple tautologies 59 social life, technologies in 145 sociology 8 Socrates 217 Socratic maxim 85 sonic ecology 167 sonic weapons 157–9 sound. See also noise cannons 158

234 culture 147 ecology 161, 162 levels of 155 military deployment of 158 sound exposure level (SEL) 151 special purpose vehicles (SPV) 126 SPV. See special purpose vehicles (SPV) Stalinism 172 statistical analysis 15, 17 statistical entropy, formalization of 3 statistical noise 1, 2, 158 definition of 1 statistical physics 139 statistics 133–7 implications of 141 stereotypies 175 stimuli disposition 176 Stoermer, Eugene F. 44 subjective noise 167–8 subjectivism 137 subprime crisis 119, 125, 130, 131 symbolization 188 synthetic CDOs 126–8 systematic communication 63 systematic imitation 188 taxonomy, Conrig’s system of 136 teacher, role of 214 technical noise 167–8 technicity 20 technological innovation 143 teleonomy 201 terror 169–77 Tetlock, Paul C. 123–4 theory of individuation 48 thermal noise 143 thermodynamics molecular entropy in 27 principle of 111 Thompson, Christopher 126 Toscano, Alberto 9 toxicity of noise 159

Index traffic noise 151 transdisciplinary discourse 8 transformation, process of 45 true market value 122, 139 uncertainty degree of 23 philosophical cultures of 86 recalcitrant cultures of 86 urban white collar workers 150 ‘vicious whir’ of sensations 179–85 vision, cultural dominance of 159 visual spectatorship 159 Volcler, Juliette 157–9 von Foerster, Heinz 112 von Humboldt, Wilhelm 90 von Neumann, John 112 Walter, Christian 129–30 war on terror 158 Weaver, Warren 16, 39–40, 51, 52, 63–7, 97 definition of noise 103 idea of noise 64 white noise 79, 80 WHO. See World Health Organization (WHO) Wiener, Norbert 4, 16, 17, 69, 122, 205 concept of negentropy 47 cybernetic definition of information 69, 85 cybernetic theory 17 definition of information 16, 47, 82, 136 origins of cybernetics 208 understanding of information 135 Wilkins, Inigo 7 Wilson, Laura 159 World Health Organization (WHO) 148–51