The Unpredictability of Gameplay 9781501321627, 9781501321597, 9781501321603

Citation preview

The Unpredictability of Gameplay

ii

The Unpredictability of Gameplay Mark R. Johnson

BLOOMSBURY ACADEMIC Bloomsbury Publishing Inc 1385 Broadway, New York, NY 10018, USA 50 Bedford Square London, WC1B 3DP, UK BLOOMSBURY, BLOOMSBURY ACADEMIC and the Diana logo are trademarks of Bloomsbury Publishing Plc First published in the United States of America 2019 Copyright ©  Mark R. Johnson, 2019 For legal purposes the Acknowledgements on p. ix constitute an extension of this copyright page. Cover design: Holly Bell Cover image ©  iStock 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 Inc 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 catalog record for this book is available from the Library of Congress. ISBN: HB: 978-1-5013-2162-7 ePDF: 978-1-5013-2160-3 eBook: 978-1-5013-2161-0 Typeset by Deanta Global Publishing Services, Chennai, India

To find out more about our authors and books visit www.bloomsbury.com and sign up for our newsletters.

CONTENTS

Preface vii Acknowledgements ix

1 Introduction: Unpredictable Play 1 The era of outcomes and the era of systems 3 Unpredictability so far 9 Deleuze and game studies 13 Outline of the work 20

PART ONE 2 Games with Randomness 35 The initial conditions of gameplay 35 Lasting impact of initial conditions 38 The creation of significance 44 Granularity of interest 51 System and variable knowledge 57

3 Games of Chance 66 The ongoing system of gameplay 66 Codifying chance 68 The Random Number God 79

4 Games of Luck 88 The conclusion of gameplay 89 Levelling the playing field 95 Luck, skill and time 97 Mistaking randomness and chance for luck 106

vi

CONTENTS

5 Games with Instability 120 Unplanned unpredictability 120 Glitches and broken rules 125 Exploits and unforeseen scenarios 127 Tight coupling and weak coupling 131 Internal and external instability 133 Subverting and reconstituting the rules of play 136

PART TWO 6 Procedural Generation 149 Roguelikes 151 Depth, length and co-dependence 156 Identifying procedurality 162

7 Replay Value and Grinding 169 The value of returning 173 Ease and difficulty in grinding 176 Completionism and exhaustive play 181

8 The Negation of Unpredictability 199 Duplicate deals 200 Save scumming and permadeath 203 Gameplay seeds 211 The negation of unpredictability 213

PART THREE 9 Conclusion: A Framework for Non-deterministic Games 221 Deleuzean games 224 The ludic experience of unpredictability 227 Bibliography 231 Index 241

PREFACE

This book is the result of my academic thought of the last few years, my own ongoing development of a somewhat unusual procedurally generated game and the considerable period of time I previously spent playing a rather unpredictable card game as my primary source of income. It starts from what I first perceived as being merely a linguistic vexation, but which eventually grew into a somewhat broader one, a realization that the vocabulary to adequately and fully experientially describe games which deploy unpredictability in different ways did not yet exist. Or, rather, that the words appropriate to such a vocabulary did exist, but we were using them as synonyms rather than as clearly distinguishing terms. Are games which we say contain randomness, or chance, or luck, really all that similar to one another? And if – as I argue in this work – they are not, how do these factors vary across games and lead to sometimes quite profoundly different experiences? Studying the unpredictability of gameplay means understanding how players negotiate with, and understand, game systems which can be often opaque, and seemingly unfair. It means understanding why some players delight in the play of unpredictability, why other players attempt to constrain or curtail or circumscribe the play of unpredictability and why games ranging from slot machines to grand strategy board games, and from poker to role-playing games, are so compelling. It also means unpicking terms such as ‘games of chance’ into a larger sets of categories or terms, which allow us to discover conceptual and experiential distinctions between them. Most games – both contemporary and historical – contain some element of what I will define in this work as ‘unpredictability’, and understanding unpredictability is consequently crucial for understanding experiences of gameplay as a whole. This work can therefore be understood as the reconstruction of unpredictability from ‘first principles’ in order to identify some boundaries, distinctions, forms and margins of unpredictability that have perhaps previously been overlooked, and to generate an understanding of the subject of unpredictable games – which is to say, what is it like to play an unpredictable game? And what factors cause the answer to this question to differ? I answer these questions via the philosophy of Gilles Deleuze, whose

viii

PREFACE

metaphysics offers significant and previously unexploited insights into the experiences of playing games of all stripes, and especially into playing games of unpredictability. In turn, however, my intention is also to clarify Deleuze’s notoriously difficult thought and fill the work with practical findings, case studies and suggestions. The goal is therefore simultaneously to allow us to understand playing these kinds of games in a conceptual and theoretical sense, and also from a perspective of practical game design and player experience. As such, this volume should offer new insights into how the experience of playing unpredictable games differs according to the nature of their unpredictability, how the subject who plays these games constructs their understandings of game unpredictability and how these understandings can lead to a diverse range of personal reflections, cultural practices and even political contests. Unpredictability is a central element to so much of gameplay, but cannot be understood as just a single design practice or play experience; only by unpicking it in greater detail can we come to fully understand it, and the profound effect it has on player experience of games of almost every kind.

ACKNOWLEDGEMENTS

My thanks go to Robert Mejia, Rob Parker, Madelon Hoedt, Luke Ridley and Tom Brock for their extremely valuable commentary and critique on the work; to Mary Al-Sayed and Katie Gallof for their support and willingness to discuss structure, cover art and much else besides; and most importantly to my parents for giving me the freedom to spend every spare second of my time playing computer games.

x

1 Introduction: Unpredictable Play

Few games are fully deterministic – which is to say, games where a player performing a given action knows precisely what the outcome of that action will be upon the field of play. The player who moves a piece in chess knows where that piece will end up and whether it will take an opposing piece in the process; the player who places their stone at a particular vertex in Go knows whether it will make a group of stones alive, dead or unsettled, and whether or not it might result in a ko fight; the player considering their next move in draughts can always see how many captures are possible in a single sequence. The chess piece cannot be sometimes defeated by the piece it tries to capture; a Go stone cannot switch to the other colour upon a lucky roll of a die; draughts pieces cannot fail to capture the piece they jump over. The same action repeated across multiple instances1 of play is guaranteed to result in the same outcome each time. The encyclopaedia of such games is, however, modest compared to the great compendium of those which include at least one non-deterministic game mechanic – the rolling of a die, the flipping of a coin, the generation of a world map, the drawing of lots to determine which team plays first, randomized damage outputs, percentage-based drop rates, unpredictable artificial intelligence decisions, the shuffling of cards, virtual weapons without perfect aim, spawn point selection in multiplayer games, randomized placement of enemies and items, the drawing of unseen cards or tiles and many others besides. In all of these cases one similarity may be divined – a player performs a given action, but that action is no longer universally and inseparably shackled to a given outcome. Instead actions lead to a possibility space of potential outcomes, which may be small – a flipped coin effectively has two potential outcomes, discounting the obvious but unlikely third outcome – or large, as in the procedural generation of a world map which might have billions of possible forms. The input is given, and one possible outcome is ‘selected’ (a word which, as we shall see, rather masks the complexity of this process in most cases), and that particular outcome could not have been predicted before the act of selection is carried out. This

2

THE UNPREDICTABILITY OF GAMEPLAY

core experience, of being unable to predict the outcome for a given input, is one with profound importance for the study of game-playing experience, for it problematizes a wide constellation of related elements: the triviality or seriousness of play, player agency, notions of fairness, having something ‘at stake’ in play, the role (if any) of skill and many more besides. Let us consequently define ‘unpredictability’ as descriptive of the category of game action which can yield multiple outcomes for identical inputs, and that the ‘selected’ outcome from the possibility space of potential outcomes cannot be foreseen. Now take three gameplay scenarios in which this unpredictability unfolds: in the first, one loads up the latest iteration of the grand strategy nation-building Civilization (1991–present) franchise and starts a new game – the world map is procedurally generated by the game and the player’s nation is placed at an unpredictable location on this map, and this location then serves as the inescapable ‘starting point’ of the player’s civilization during all subsequent play. In the second, one is in the midst of playing a role-playing game (RPG), and upon slaying a powerful ‘boss’ enemy, the player wonders: will this foe drop the item I know they can drop x per cent of the time, or will I get nothing for my effort, again, as the last y times I’ve defeated them? In the third scenario, one walks into a casino and finds a slot machine to play, sits upon the stool, places a coin in the slot, pulls the lever and watches the wheels whirl towards their indeterminate denouement. To describe all of these unpredictable gameplay experiences we use the same unexamined appellations. We call them randomness, luck or chance – yet we surely cannot analytically justify using a given term – ‘luck’, for instance – to describe the unpredictable map creation for a twenty-hour game of complex strategy with thousands of individual steps and decisions, each of which is in some way dependent on the generation of the world map upon which they play out, and then apply that same term also to a heartbeat encounter with a one-armed bandit where player skill and thought play no role. This book is therefore concerned with a single task: to guide us in an escape from this obfuscating synonymy by identifying analytically distinct forms of unpredictability in game systems, and by developing a typology based upon these observations – one with both theoretical depth and practical application. This is the task of the first five chapters. Having done so, the book will then examine three particular case studies of unpredictability, their navigation by players and the player experience thereof: procedural content generation (PCG), replay value and grinding, and player practices designed towards the deliberate and directed negation of unpredictability. These will allow us to move from the theoretical to the empirical and carry out a broader analysis of where the elements of unpredictability arise in the vast milieu we currently term ‘games’, the perception (or rejection) of identity and difference between multiple acts of gameplay in unpredictable games, how players navigate and reflect upon games of unpredictability which can be opaque and even seemingly unfair and how we might develop

INTRODUCTION: UNPREDICTABLE PLAY

3

a comprehensive experiential understanding of ludic unpredictability from these enquiries. First, however, it is useful to address the history of games that feature unpredictable systems, consider some of the broad shifts in the design and social embedding of these games up to the present day and as a result begin to unpick some of the roles that unpredictability plays in modern game design.

The era of outcomes and the era of systems In the ancient Mesoamerican game of Patolli players were expected to bring with them six items and possessions they were willing to gamble, each item being represented by a piece on the board, and all items being taken by the eventual victor. Patolli was a ‘race game’ of both skill and chance – beans were thrown to determine how many moves a player’s piece may make around the cross-shaped board, while players had the choice over which of their pieces they advance, aiming to reach the end positions assigned to them. Playing the game was an event of immense importance, both because of the significant wagers required by both sides – especially if the players were not nobles and might have their material lives seriously impacted by a loss – and the game’s status as communion with the gods Macuilxochitl and Ometochtli. Popular belief held that the player favoured by the gods would emerge victorious – thereby transferring divine cachet into an improvement in their material status by the acquisition of items wagered by their opponent (one Mayan codex explicitly states that the winnings are a ‘gift’ from Macuilxochitl) – the victory having been achieved primarily through the gods’ largesse. One cannot speculate now as to the role ascribed by Patolli’s players to skill – were players seen to possess skill inherently, or was skill seen as something bestowed or loaned out by the gods (so if a good player wins through skill, not fortune, that is still seen as god-given fortune of a contrasting species) – but nevertheless this conflation of fortune, the ambiguous role of skill, and the notoriety and recognition (divine and worldly) accrued through victory in such games, is apparent. These intertwined themes repeat themselves visibly throughout the history of games, and many civilizations have believed that players of unpredictable games have received supernatural favour or disfavour through the act of play.2 The Royal Game of Ur (dating from before 2500 BCE) has been proposed as possessing a dual status as both play and divination;3 in the ancient Egyptian game of Senet, some anthropologists propose that the successful player was seen as victorious only under the benevolent aegis of several gods;4 ancient Greeks and Romans believed the  gods could determine success in games,5 while Greek religion recorded the distribution of the world between the gods via the outcome of a game;6 Zuñ i Native

4

THE UNPREDICTABILITY OF GAMEPLAY

Americans worshipped gods of war, associated each with a specific game and invoked them when gambling upon that game;7 the Sanskrit text Atharvaveda composed in perhaps 1100 BCE contains hymns to the gods for a range of purposes, including successful gambling.8 Equally, we can also appreciate moments of divination that could, outside of their cultural context, be readily mistaken for the play of games. The classical Western world saw the casting of small bones within a religious temple context to seek answers from the gods,9 although this divine will can be interpreted as arbitrary, rather than agential and deliberate;10 in the Tibetan Empire (600– 800 CE) the rolls of dice were commonly used to resolve legal disputes via, it seems, reference to divination texts;11 and numerous belief systems today continue to use the drawing and dealing of cards to prognosticate on the future,12 or invoke contests and games in their broader mythical narratives.13 The broader relationship between divination and games of chance remains contested, especially with regard to the causality or flow of ideas between the two,14 but they undeniably share ‘notions of the unknown, mystery, and fate, as well as imagery of suddenly receiving something of great value that changes life for the better’,15 exhibiting an ‘amazing parallelism’16 in their commonalities: being formal systems, entailing the modelling of reality, the parcelling of time into ‘sessions’, the spatial restriction of the game or holy place, and so forth. In early games of this unpredictable sort we therefore readily perceive a strange conflation: games of unpredictability have often been seen as having predetermined outcomes, and therefore what we now see as moments of mathematical indeterminacy were instead perceived as moments of theistic fatalism and certitude. Actions in these games were seen not as unpredictable per se, but simply unpredictable by humans – the outcome was always going to be the outcome, for the player blessed by the gods was destined to emerge victorious, even if the identity of the fortunate player could not be ascertained without channelling the will of the divine through the numinous moment of gameplay. In this regard it is not unreasonable to even ascribe a sacerdotal role to games – priests do not have a monopoly on interpreting the will of the gods when a game board can achieve the very same. Simultaneously, by ontologically positioning terrestrial non-determinism and non-knowledge as echoes of empyrean determinism and divine knowledge, the game’s outcomes rather than the game’s systems are emphasized: each instance of play gains character and noteworthiness beyond the mere abilities of its players through claiming to represent and illustrate the will of the heavens. The difference in outcome was what mattered, not the system (mathematical or spiritual) which underpinned it. However, this results-focused orientation to games of unpredictability, informed by a belief that understanding the outcomes was beyond human ken, was not to last. Many subsequent societal developments each shifted the discourse of unpredictable games further and further away from their

INTRODUCTION: UNPREDICTABLE PLAY

5

spiritual and results-oriented beginnings. Although a full exploration of this process would merit a book in its own right, here I wish to look at three. These are the Enlightenment and the beginnings of a deeper mathematical understanding of ludic unpredictability, the global emergence of casinos and the formalization of gambling games, and the development of game theory as a scholarly discipline. I certainly do not propose that these are a conclusive history of unpredictability in gameplay, or anything close to it, and there are many other events I elide from this brief summary; nor do I adhere to ‘grand narrative’ theories of history. However, I believe these three points in particular (three of many) do highlight some extremely important developments in how we think about game unpredictability, how players use (or deal with) unpredictability in their play experiences, and the relationship between our perception of game unpredictability and our understanding of the fundamental mechanics of a given game. The goal of relating these moments is therefore to start the process of thinking about how even the most basic non-deterministic systems such as coins and dice can lead to extremely complex cultural or ideological associations.17 These historical moments serve as valuable examples that highlight the themes and the kinds of meaning that can be assigned to playing such unpredictable games – themes and meanings, as I will show throughout the present work, which are in many cases still very much alive to the present day. The era we now call the Enlightenment saw the rise of a new approach to thinking about unpredictability, and unpredictability in a ludic context. In 1654 Blaise Pascal and Pierre de Fermat, following a discussion of an ageold gambling problem set to them by the French writer Antoine Gombaud, laid out the foundations of modern probability theory as the underpinnings of their reply. In the process they came to the fundamental realization that unpredictable events nevertheless occur in probabilistic distributions, even if those values may be extremely challenging to ascertain in complex systems, and therefore events which are yet to come may be predicted. This was a fundamental shift from the dominant ‘pre-modern’ orientation towards game unpredictability, which posited and enacted an epistemological divide between human reasoning and ludic outcomes. This conceptualization of unpredictability marked the beginning of the end of what we might call the era of outcomes (where little appeared predictable by humans), and the gradual dawn of the era of systems (where all seemed predictable by humans), a rationale that has continued in all detailed enquiry of unpredictable gameplay and game-like systems up to the present day. The expanding empirical rationalist episteme gave impetus to this mathematical research, which soon bifurcated – the questions explored by the new field of probability expanded outside the realms of the game, while also opening up new avenues of potential enquiry into specifically game-based systems whose depth was starting to be uncovered. In New Essays on Human Understanding, German polymath Gottfried Leibniz (an avid game player)

6

THE UNPREDICTABILITY OF GAMEPLAY

stressed the value of developing a comprehensive logic of probabilities in order to ‘pursue the investigation of games of chance’,18 and suggested the value of creating an ‘Academy of Games’ for the study of games of both unpredictability and skill, emphasizing that such games give the most realistic account of human life, and therefore merit the closest examination.19 Probability theory thereby ‘found its first and still paradigmatic elaboration in relation to games of chance’, for in games mathematicians saw a ‘form of human activity regulated by chance and not simply (as all human activity) subject to uncertainty’ (emphasis in original).20 In games, natural philosophers could readily see a duplicate copy of the newly scientific world now surrounding them; the appreciation of unpredictability thus began to shift away from the outcome, and towards the system that creates it. A second macro-level social change of interest to our enquiry was the growth and subsequent institutionalization of gambling games devoid of explicit religious content (although superstition remained) and the casinos in which they were played. Naturally humans have always gambled on games of chance, but it was in the seventeenth century in Europe that gambling games became a site of ‘scientific, rather than sacred, dramas’ (emphasis in original).21 Within the wider context of a growing mercantile society that boosted both affluence and the value of thinking about mathematics, statistics and finance,22 gambling games became a point of probabilistic study par excellence. In such games, the skilled player ‘weighs the stakes against the odds, or calculates the risks involved, at the same time predicting which play strategies are more likely, or unlikely, to pay off’23 – as in any form of investment or trade. As de Goede notes, both finance and gambling are strategies for dealing with and facing up against uncertainty, and lacked, in early-modern Europe, their now-clear divide.24 Subsequently, what we might call ‘true’ casinos, which is to say the structures and institutions we now recognize as such, emerged in the second half of the nineteenth century.25 These not only further reified the non-religious spaces such play would take place in, but also contributed to the now-clear divide between finance (as a legitimate pursuit) and gambling (as a deviant practice). These changes are noteworthy here not for their social or economic effects, but rather for perspective on unpredictable games adopted by a small, but over time increasingly significant volume of players. While ‘gamblers’ continued to lose, a new breed of player emerged – one that looked to the long term, understood in great detail both the mathematical permutations underpinning the games and the psychological decisions made by their opponents, and was able to conceive methods of overcoming the unpredictability of these games to produce statistically consistent victories.26 At this point systems thus became identified as not just something of mathematical interest, but also as the definitive site of cultural meaning and noteworthiness in such forms of play (even if the manipulation of the system was designed to induce the production of desirable outcomes), rather than that status being afforded

INTRODUCTION: UNPREDICTABLE PLAY

7

to the outcomes, which were almost secondary. The outcomes were only of interest insofar as being the manifestation and validation of the strengths, or weaknesses, of careful gameplay strategy. The highly public rise of gambling games marked a new visibility of unpredictable play, a wide expansion in their potentially profitable play for those skilled enough to navigate their unpredictability (whether through legitimate or illegitimate means), and arguably marked their formal emergence and clearest codification into secular, rather than religious, society, beyond the writings of intellectuals and scholars. These developments marked new occurrences in not just the ways of thinking about unpredictability, but also in terms of who engaged with unpredictability, how they engaged and why. This growing arithmetic perspective on unpredictable games achieved its zenith with the third historical moment (of many more than this in the grand historical progression of games and unpredictability) I highlight here – the advent of game theory as an intellectual discipline. Game theory, entailing the study of cooperation and competition between actors, was formalized as a field by John von Neumann, both through a paper in 192827 and more substantially in the 1944 ‘Theory of Games and Economic Behaviour’.28 Before too long, it became apparent that several of the ‘games’ considered by this new branch of mathematics had striking applications to real-world strategic situations, such as the growing nuclear stand-off between the Cold War’s two superpowers. Understanding the behaviours of unpredictable actors, and designing strategies to respond optimally to all possible unpredictable behaviours, became essential to the high-stakes ‘game’ of global brinksmanship. Forms of thinking and reasoning about the outcomes of unpredictable scenarios which began around a board scratched in the dirt now came to determine the nuclear annihilation, or otherwise, of the human race – with the worrying conclusion that the correct move was almost always, strategically speaking, to up the stakes. Within the intellectual domain of game theory, almost anything within the natural or social world might be modelled as a game-like system;29 the mathematical approach to understanding unpredictable game systems now grew beyond its initial limits, pushed by a fundamental metaphorical claim of the practical parity between ‘games’ (as a leisure activity) and ‘games’ (meaning any system of interactions between actors). Contemporary game theory maintains significant relevance and research in economics, computer science, political science and a number of other related sub-fields where strategic decision-making remains of paramount relevance; such disciplines have consequently become dominated by an understanding of systems and all their possible outcomes. In this episteme, unpredictability is seen as being indicative of a possibility space, and one through which supposedly rational choices can be made to secure the best outcomes, no matter the outcomes of unpredictable processes (whether mechanical unpredictability or through the uncertain actions of other actors).

8

THE UNPREDICTABILITY OF GAMEPLAY

We can therefore reasonably identify two discrete historical epochs in regards to unpredictable games, although the shift from one to the other was certainly slow and piecemeal, not rapid and binary, and nor has it been all-encompassing or complete. The players of the first epoch were concerned with the outcomes of such games; the players of the second with the systems which generated these outcomes and how those systems might be understood and mastered (even if such enquiry was seen as a way to control the outcomes). Players of the first epoch emphasized belief in the externality of unpredictable outcomes; players of the second epoch sought to examine the internality of these same outcomes. Players of the first regarded each instance of play as possessing its own distinction and identity, while players of the second looked at unpredictable games through the longer lens of hundreds, thousands or tens of thousands of iterations, seeking to identify commonalities, numerical underpinnings and optimal decisions. These eras are not mutually exclusive, absolutist distinctions – there remain many contemporary gamblers who place superstitious weight on the outcome of a die roll, and it seems deeply insulting to the intelligence of people from the past to suggest that Patolli players would not have consciously strategized over a game which counted so much towards their material conditions in life – but this broad change in the emphasis placed on the two central components of unpredictable games and the analytic focus of any who examined or observed such games is apparent. What, therefore, is missing from this picture? Our contemporary intellectual environment compels us to think of unpredictable games first and foremost as systems, and in doing so we have become less interested in the experience of unpredictability and the experience of the particular outcome that emerges through a single iteration. In the first era there were deep meanings assigned to individual gameplay outcomes; modernizing this emphasis, we might valuably look to consider the meanings players now append to individual outcomes, given the far larger range of unpredictable games we now enjoy. However, a return to the first era seems untenable, for such understandings were inextricably linked to supernatural conceptions of causality and complexity, which – although fascinating as the subject of study – have little place in building a critical scholarship. A new method needs to be found to address the experience of playing unpredictable games, without recourse to religious factors. Indeed, going further, such a new method would ideally integrate and shed light on a number of additional elements traditionally omitted from examinations of this first era, which tend to stress the meanings assigned to only the outcomes of games. What is it like to play an unpredictable game, knowing that your present situation is one of dozens, hundreds or millions or more? How do players treat one iteration of a game from the next, and might one iteration inform another? How will different kinds of unpredictability shape a game’s play experience? Outcomes are not the only site of meaning or significance in

INTRODUCTION: UNPREDICTABLE PLAY

9

such games, and a new critical engagement to the experience of playing unpredictable games could instead develop a framework for all elements of game unpredictability. Doing so would explore a topic that the analysis of game systems as mathematical models can never capture: what players actually experience when playing these games. Given that unpredictability ranges across almost the complete range of digital, board and card games, both contemporary and historical, this is an important building block in any analysis of what players actually do and feel when they play, and how these experiences shape their interactions with games themselves.30 I hope to answer these questions through the development of a fourpart typology of game unpredictability. I will term these ‘randomness’, for unpredictable initial starting conditions of a game; ‘chance’, for unpredictability during the play of a game; and ‘luck’, for unpredictability regarding the final outcome of a game. Although these terms have naturally been used elsewhere in game scholarship, and scholarship more generally, I seek to justify new definitions assigned to these words within the specific context of the experienced unpredictability from formal game systems. These particular words have rich sets of broader cultural associations assigned to them I will draw on in this process, and offer us a typology that is both immediately linguistically familiar and more detailed and specific than before. Then, addressing the fact that these three cover intended unpredictability in game designs without exploring unintended unpredictability, I will propose the existence of instability, through which games can gain unpredictability unplanned and unanticipated by their designers, which can quite fundamentally reshape the play experience. Each of these I understand as being a different ‘location’ or ‘place’ of unpredictability, moving beyond sweeping terms like ‘games of chance’ to see how different implementations of unpredictability lead to very different kinds of player experience. The location of unpredictability refers to the temporal stage of gameplay where unpredictability manifests – beginning, middle or end – and is central to unpicking the wide variety of game unpredictability, and its navigation by game players.

Unpredictability so far There have been several important attempts to date, coming from a range of disciplinary backgrounds and epistemological foundations, and with a range of objectives and justifications, at attempting to codify – depending on how we define them – unpredictability, uncertainty, indeterminacy, randomness, chance, luck and instability in games. These will be referred to when relevant throughout this work in order to build upon these previous engagements, identify commonalities and differences across these conceptualizations and typologies, and attempt to draw out in greater depth the range of forms that non-anticipated events in gameplay can take. However, before outlining the

10

THE UNPREDICTABILITY OF GAMEPLAY

foundation of this work in the metaphysics of Gilles Deleuze, it is valuable to give a brief overview of the field. Doing so will identify immediately the diversity in the major works in the area, note one prior engagement in particular that I see this work as building upon and developing into new analytic territory (video games as opposed to, or alongside, gambling games), and make clear the centrality that thinking through unpredictability has had for thinking through games of almost all kinds, both historically and within contemporary game studies and a number of cognate disciplines. From the earliest days of the then-nascent field of ‘game studies’, questions of unpredictability (going by many names) have been present. Johan Huizinga’s Homo Ludens (1938) has relatively little to say on games of chance, although he does argue that the tension felt by a player of chance is rarely communicated to onlookers (although any ethnographic engagement with a casino floor or betting shop would be enough to dismiss that claim), and that gambling games are fundamentally ‘unproductive’ in a cultural sense, ‘sterile [and] adding nothing to life or the mind’31 – as opposed to games of ‘skill’, which Huizinga defines and praises. By contrast, Roger Caillois in his Man, Play and Games (1958) allows ‘chance’, or ‘alea’, to occupy one part of his four-part typology. As he puts it, ‘alea’ encompasses ‘all games that are based on a decision independent of the player, an outcome over which he has no control, and in which winning is the result of fate rather than triumphing over an adversary’; the exercise of alea in gameplay thereby ‘negates work, patience, experience and qualifications’,32 but can be thrilling and compelling to the player who lets themselves go and gives themselves over to the externalizing of play outcomes.33 More recently, arguably the first game studies book to focus on this topic was Uncertainty in Games, by game designer Greg Costikyan (2013). He explores a wide range of forms of ‘uncertainty’ – uncertainty over what to do, knowledge of the game, the player’s performance, the accuracy of one’s puzzle solution, hidden information, analytic complexity and others – but situates these valuable design analyses within overly broad summaries of society, culture and history, causing the work to overlook some of the most interesting elements of the embedding of these elements in the life of a game-playing subject. From beyond the discipline of game studies, meanwhile, have come numerous analyses based on gambling studies, and cognate domains which take gambling to be their object of study (if not their subject of study). Thomas M Malaby’s Gambling Life (2003) examines how the unpredictability of gambling games is closely related to unpredictable non-gambling elements of players’ lives – death, business, romantic interest – and the flow of ideas and coping strategies between the two, arguing that unpredictability can be a beneficial and positive element of social life, rather than something to be feared and acted against. Natasha Schü ll’s Addiction by Design (2012) investigates the phenomenology of playing slot machines, focusing on how

INTRODUCTION: UNPREDICTABLE PLAY

11

the complex deployment of unpredictability through different temporal rhythms, aesthetics, and technologies shapes the play of these devices, and how this play can exploit human reactions to unpredictability and become addictive and damaging in the process. Fiona Nicoll’s Gambling In Everyday Life (forthcoming) and the wider emerging project of ‘critical gambling studies’ she is leading aim to move beyond the problematization of playing games for money to instead interrogate gambling and its politics across cultures, regions, jurisdictions, historical moments and the crucial yet oft-overlooked dimension of play, fun, leisure and enjoyment. Lastly, we cannot also exclude the range of work into these areas coming from domains outside of game studies and gambling studies, which are concerned with unpredictability and come from a tremendous range of disciplinary backgrounds, and writing styles that may be academic, journalistic or popular. Specifically, two other scholars should be considered. First, in After Virtue,34 moral philosopher Alasdair MacIntyre uses the term ‘systematic unpredictability’ to refer to four phenomena. First, the emergence of true innovations, such as ‘the wheel’, in contexts in which it did not exist, and could not have been predicted, for the moment it is predicted it has, in a sense, been invented; secondly, how the unpredictability of an individual’s choices creates unpredictability in the wider social world of others; thirdly, elements of social life that cannot be adequately modelled in a gametheoretic manner, such as vagueness and uncertainty over who and what is playing a certain ‘game’, the risk of infinite regression in trying to predict the actions of others, and the role of imperfect knowledge in the real world; and fourthly what he terms ‘contingency’, the complex and unpredictable, yet powerful and influential, ways that seemingly unrelated events can exhibit shared causation. In this regard I use the term ‘unpredictability’ in a manner closest to his first and fourth definitions; his work is obviously not primarily concerned with games, but brings up concepts surrounding originality and causation relevant to the discussion presented here. Equally, theorist of optimal experience ‘flow’ Mihaly Csikszentmihalyi has also on occasion examined similar topics. For example, he suggests that cultures and cultural practices are ‘defensive constructions against chaos’,35 designed to help us handle and contain the vicissitudes of unpredictability through developing and preserving norms and standards. In doing so they ‘rule out many alternatives’ and ‘limit possibilities’;36 in this we see something of an introduction to some of the later discussion in this work regarding gaming cultures that seek to contain, constrain or control the unpredictability in many games. Finally, arguably most important in the last domain of popular or journalistic writing is the work of Nassim Nicholas Taleb, risk analyst turned philosopher, with books such as Fooled by Randomness (2001) and The Black Swan (2007) committed to examining the psychological tricks played by randomness on the human mind, how individuals struggle to understand

12

THE UNPREDICTABILITY OF GAMEPLAY

a number of situations of unpredictability, and the effects and potential for anticipation of what he calls ‘Black Swan Events’ – events vastly outside of the norm of expected behaviour in a given system. All of these works offers valuable insights which will be returned to throughout this book, whether via conceptual developments, empirical examples or observations about player engagements with unpredictability in a range of personal and social contexts. We also note that game scholars agree – and simply playing a game twice confirms this observation – that games are indeterminate, which is to say their outcomes in a broad sense are not known when play begins, but this is a very different matter from what I aim to explore in this book. The work that comes closest to my project here, however, is Gerda Reith’s comprehensive monograph The Age of Chance (1999), the final two chapters of which are committed to examining, in her words, ‘what is it like to play at games of chance?’37 – note the close terminological similarity to what I propose in this work. Reith examines in detail the thrill of playing unpredictable games, the shifts of player identity in the ‘vertigo’ of playing for long periods with unpredictable outcomes, the perception of time for such players (which she argues is centred on the present, the immediate gamble, where antecedent plays are irrelevant and future plays unknown), and what she calls the ‘metaphysical significance’ of victory or loss in games of this sort. However, there are two elements of this work which, I believe, open the door to further examinations of this experiential question. First, The Age of Chance is focused on games that include wagering, betting or otherwise the deployment of money: as she states, ‘money must be present in games of chance’38 in order for the ‘generation of the affective tension – the excitement – in games of chance’;39 it is therefore suggested that the ‘anxious moment of wagering is altogether more essential than the outcome of the play’.40 In the present work I follow this analysis by moving beyond ‘games of chance’ on which people wager money in order to include, and indeed focus on, video games that include elements of unpredictability, and in doing so demonstrate that one need not have money at stake to generate this excitement: the risking of time, effort, potential stress and social status are very capable of generating that same compelling affective tension (and generate quite different cultural practices, player experiences, critical responses and so forth). Jesper Juul proposes the act of play as an ‘emotional gamble’ where we invest ‘time and self-esteem’ in our play activities,41 just as Thomas M Malaby notes that the ‘social status of the participants’42 is also embroiled in games; it is these observations I wish to develop further, alongside and without excluding the central role of money in many unpredictable games. Secondly, I want to unpick in greater detail the category described in The Age of Chance as ‘games of chance’. Reith rightly notes that ‘the heterogeneity of games is matched by the heterogeneity of players and motives’,43 acknowledging the tremendous variety of games, those who play them and the ways in

INTRODUCTION: UNPREDICTABLE PLAY

13

which they are played, but throughout ‘games of chance’ are treated as a single category – contingent, in part, on wagering – and are not unpicked in greater detail. This is a broader observation which applies to most work in this area: the category of ‘games of chance’ is never unpicked to identify different forms of games within that broader heading. This is not always an issue in scholarly works with other intellectual objectives, such as Reith’s, but does highlight the opportunity for a new and deeper exploration of what precisely the games in this category look like, and how they differ. With these prior examinations acknowledged, how should we begin to consider unpredictability in game systems, the locations of that unpredictability and the creation of an unpredictable-game-playing subject?

Deleuze and game studies To do so, the scholar who affords us the most complete ontology of difference (and the attendant experience of difference) is Gilles Deleuze (1925–95). There have been a number of prior explorations of Deleuze in relation to games,44 but Deleuze cannot be considered (as of yet) an integral part of the game studies theoretical canon. Nevertheless, as Colin Cremin has argued, ‘Deleuze’s work particularly lends itself to video game analysis with his focus on nonrepresentation, affect, and movement.’45 Cremin’s recent monograph on the topic, for example, takes a wide look over a range of games and political questions surrounding games, such as the games industry, and the relationship between players and the games they play.46 In Nick Dyer-Witheford and Greig de Peuter’s Games of Empire (2009), Deleuze and Guattari (1930–92) are drawn upon to theorize the actions of ‘minority’ participants in gaming spaces, the relationships between global capital and the video games industry and how computer simulations can offer the opportunity for alternative, possible worlds. In a similar vein, Tauel Harper examines the emergence of political communities around play, utilizing the work of Deleuze and Guattari to consider the emancipatory potential of games which is not inherent to their play, but can emerge from playing particular games in particular manners.47 The doctoral thesis of game designer Tom Betts, meanwhile, examines processes and phenomena of emergence in generative systems. Deleuze, he notes, ‘frequently discusses themes of repetition, permutation, emergence and the virtual’,48 making him an ideal theorist for exploring such ideas in a video game context. Coming at Deleuze from a different angle, Chiel Kattenbelt and Joost Raessens draw upon Deleuze’s work on cinema and different kinds of cinematic image to explore the variety of the phenomenological experiences that players encounter in computer games.49 Across all of these studies and many others, Deleuze’s individual work and that written alongside Felix Guattari have been put to numerous tasks in the study of games and play: their aesthetic

14

THE UNPREDICTABILITY OF GAMEPLAY

dimension, their political dimensions and their social dimensions. However, I believe the present work will be the first game studies book to put Deleuze to work exploring unpredictability, in doing so hopefully bring to light more fully the importance of this thinker to understanding the ontological and experiential components of gameplay. It will be helpful to set out here the Deleuzean terminology which will be used in our examination of gameplay unpredictability. Deleuze has been universally noted as a thinker whose work is highly challenging to read;50 for being almost ‘wilfully obstructive’51 to the reader’s progress; for producing texts of distinct difficulty and a lack of clarity;52 for having a ‘love of provocation’;53 for displaying a noted ‘terminological exuberance’54 in his ever-changing use of language; and much else besides.55 I consequently present here my own readings of key terms, and brief summaries of their intended relevance to this work. Most of these are drawn from Difference and Repetition (2014 [1968]), which is the primary Deleuzean text relevant to the present enquiry, although The Logic of Sense (2004 [1969]) and small segments of his collaboration with Felix Guattari A Thousand Plateaus (2004 [1974]) will also be utilized when appropriate (insofar as my reading of them does not offer a conflict with Deleuze’s original ontology of difference as outlined in Difference and Repetition). Although I have endeavoured to outline this vocabulary in a logical sequence, some definitions inevitably point to other definitions later in the list, but Deleuze’s densely interwoven thought rendered this unavoidable. Nevertheless, a subjective consideration of such games and such forms of play is of great interest irrespective of any ‘rightness’ or ‘wrongness’ vis-à -vis an underlying generative system;56 thinking of human action only in terms of modelled rationality or selfinterest can only ‘lead to a limited picture of human life’ when tackling what I call unpredictability.57 By contrast, Deleuze’s focus on subjectivity opens a door to a more thorough understanding of the play experiences which unpredictability creates, which have been all but ubiquitous throughout the history of play. Given the conceptual density of Deleuze’s work I have sought to make this summary as succinct as possible, but it is important to establish many of his ideas up-front, as I will be drawing heavily upon them in the remainder of the work. Repetition and Generality: Deleuze opens Difference and Repetition defining what he calls generality and repetition. These two concepts will be integral to understanding gameplay unpredictability. He states that ‘generality presents two major orders: the qualitative order of resemblances and the quantitative order of equivalences’.58 To identify a generality, Deleuze states that ‘resemblances are unpacked in order to discover an equality’59 – this equality allows for the exchange of terms or things (such as bodies of water) that are defined as being subject to the same laws. What Deleuze calls repetition, by contrast, is not repetition of the same, of an ‘original self-identical thing’,60 but rather the repetition of difference.61 Far

INTRODUCTION: UNPREDICTABLE PLAY

15

from repetition being the observation of when matters are the same, as in traditional philosophy, Deleuzean repetition takes place when matters are not the same,62 but something new is created from a shared source or origin. As Deleuze puts it, ‘To repeat is to behave in a certain manner, but in relation to something unique or singular which has no equal or equivalent.’63 James Williams gives a clear example of this – the boundaries of the territory that an animal prowls are only defined by the repeated prowling of that territory, although each act of patrol is different; the territory does not exist prior to the repeating,64 but rather the repeater and the repetitions emerge from what Tom Betts calls a ‘self-reflexive feedback loop which underlies the construction of meaning’. The animal behaves in a manner in relation to something singular – its territory – which is created through these repetitions. Things cannot repeat without something that causes this repetition,65 such as – at the risk of anticipating future analysis – someone shuffling a deck of cards, running a procedural generation system or flipping a coin (or patrolling a territory). In this way repetitions repeat similar things (e.g. a deal of cards), but do not reproduce the same thing (a particular deal), since there must always be difference in order for a repetition to take place. Whereas traditional philosophical repetition presupposes the concept that is being repeated, for Deleuze, the concept being repeated is produced through the repetitions.66 As I will argue in this work, this is crucial for our understanding of how game-playing subjects construct their understandings of the game being played, the nature and pattern of its unpredictability, and how that unpredictability is consequently experienced. Identity and Difference: Deleuze seeks to challenge existing Hegelian67 and Aristotelian68 emphases on the pre-existing transcendent identities of things. Deleuze instead proposes that ‘all identities are only simulated, produced as an optical “effect” by the more profound game of difference and repetition’.69 By this he means that identity is never inherent, and such a model of thought detracts from the unique specificity of lived experiences by attempting to force all phenomena into pre-existing models. He argues that we must ‘take seriously the nature of the world as it is perceived’70 because there is nothing ‘behind’ the world (as in a Platonic perspective founded upon pre-formed identities). Rather than supposing that difference is only found when perceiving that two objects have identities and that these two are not the same identity, Deleuze inverts this causality and asserts that we only perceive identity through an endless sequence of differences between all sensible objects: ‘Resemblance and identity are only functional effects of that difference which alone is originary within the system.’71 Difference is consequently Deleuze’s term for the ‘emergence of form, which cannot be captured within the structure of the already formed’.72 If we are therefore to seriously study the gameplay of unpredictable games in a Deleuzean sense, ‘every object, every thing, must see its own identity swallowed up in difference, each being no more than a difference between

16

THE UNPREDICTABILITY OF GAMEPLAY

differences’,73 and such a position allows us to examine unpredictability through the perceptions and assessments and subjectivity of the player, not through the underlying mathematics (except insofar as the player considers those). The repetition of difference is, I will argue, the process through which players experience unpredictable games, and a crucial theoretical turning point for understanding how players become subjects that play unpredictable games, and understand and experience those games in particular ways. Possible, Real, Virtual and Actual: Deleuze draws two important distinctions between what he calls the possible and the real, and the virtual and the actual.74 The possible is everything that might be – I might play poker today or tomorrow – and the real is what actually takes place and exists (e.g. I played poker today).75 The real and possible are therefore of the same sort, even if the number of things in the real is miniscule compared with the number of things in the possible. The virtual and the actual, however, are different. The virtual is an aspect of reality which is not physicality instantiated, but is nevertheless real – a meaning of a word, for example. Virtuality ‘contains only the rules for the production of objects’76 which might be actualized; whereas the possible resembles the real, the actual does not resemble the virtual and the two are instead quite different. Deleuze proposes that we should not define something according to its already-actualized forms but rather by also understanding the virtual – we cannot understand what it is for a deck of cards to be dealt by only looking at previously dealt permutations. The virtual content of a deck of cards would include the system for the production of every possible deal, for the full set of possible deals defines what kind of deck it is, even if most have never physically occurred (been actualized), but it does not contain possible deals from decks of cards which contain different kinds of cards. As James Williams puts it, the properties of an actualized thing are virtual, but when you touch or otherwise engage with it, there is ‘something of all the other things’77 with that same virtual content; exploring one dungeon has something of all the other dungeons that the player has experienced, and the other dungeons that might have been, and might still be in the future. Deleuze’s understandings of virtual and actual will further contribute to a rich toolkit for exploring game unpredictability, specifically vis-à -vis the relationship between multiple different instances of one unpredictable game, how players think about the possibilities of those unpredictable games that were and were not actualized in a particular playthrough, and how we think about the formative systems that underlie unpredictable games. Ideas/Multiplicities and Singularities: Deleuze uses the terms ‘Ideas’ and ‘multiplicities’ to refer to the same things.78 I find multiplicity to be both the most descriptive and least abstract of these two terms – and one that seems highly appropriate to structures like decks of cards or procedural generation systems – and will therefore use multiplicity throughout this work rather than Idea. A multiplicity is ‘pure virtuality’.79 A multiplicity can only be

INTRODUCTION: UNPREDICTABLE PLAY

17

perceived by an individual subject through regarding the actualized forms that emerge from it and in the differences between it and other multiplicities. A multiplicity consists80 of differential elements (such as dungeon rooms or enemies), differential relations (how those rooms might be arranged or how enemies might be placed) and the potential for resultant ‘singularities’ (noteworthy combinations of actualized elements and relations, such as a very challenging enemy placed in a challenging room which results in a gameplay situation far more challenging than either of those on their own). These distinctive permutations that emerge from a multiplicity – the singularities – have particular importance for a Deleuzean study of games. Singularities ‘emanate from that aleatory point which every time condenses the whole of chance into one time’81 – by this Deleuze means that an aleatory (‘contingent’) point is the intersection of elements and relations in a way that produces a noteworthy result (e.g. a challenging enemy in a challenging room), while the singularity is the result itself that emerges from that point (e.g. a very high level of difficulty). Anthropologist of religion Katherine Swancutt defines Deleuze’s aleatory points in a game context as being ‘configurations [that] allow players to make a winning move’,82 but I believe a slightly broader definition is in order. These points could also include particular combinations of layouts and enemies which produce strikingly challenging, original, compelling or even amusing instances of gameplay, not just those with the potential for competitive ascendancy – they are gameplay permutations constructed from differential elements and relations that yield a unique and distinctive singularity. They are ‘turning points and points of inflection; bottlenecks, knots, foyers, and centers; points of fusion, condensation, and boiling’,83 fundamentally antithetical to any notion of ordinariness or routine. Multiplicities therefore contain within them elements and relations and from these the potential for singularities, which are actualized in particular deals or particular generations when the elements and relations intersect in a noteworthy way. It must also be stressed again here that Deleuze’s project is one of subjectivity, and therefore any multiplicity in question is the one perceived by the subject playing the game: for one person the multiplicity of the ‘same’ actual deck of cards will differ from what someone else sees in that deck. I will therefore use the term ‘generative system’ to refer to the material fabric of a game itself and the systems which yield and shape its unpredictability, and ‘multiplicity’ to mean the subjective perception of that system, slowly and gradually constructed, in the mind of a player.84 Multiplicities represent the sum total of player understanding about a system of unpredictability – the very nature of what they think that system is and what they think it can produce – and are therefore crucial for the subsequent analysis in this work. Differentiation and Differenciation: These two terms, although differing by only a single letter in their English translation, are crucial to Deleuze’s

18

THE UNPREDICTABILITY OF GAMEPLAY

thought and to much of the discussion of this book (especially the latter – differenciation). Differentiation is the process by which the virtual content of one multiplicity is distinguished (‘determined’) from another – for example, every use of the term ‘first-person shooter’ (FPS) in relation to a particular context serves to differentiate the multiplicity of ‘the FPS’ from everything else in the world. In turn, we come to understand how FPS, RPG and real-time strategy (RTS) games are differentiated as distinct genres. Similarly, differentiation marks the boundary between ‘dungeon’ generators and ‘world generators’, or ‘standard 52-card deck’ against ‘Tarot deck’, or between board games and computer games. Differenciation, meanwhile, is the actualization of the virtual, resulting in (for example) a series of games all defined as ‘FPS’, the definition of which has been virtually differentiated from other genres. This results in a heterogeneous (but similar) series of things which take the form of repetition. This is not generality, because in each differenciation something new is created. Differenciation is therefore central to the following analysis of unpredictability, being predicated on the creation of varied outcomes from a generative system, while differentiation is of less value, but will nevertheless be useful in our consideration of the player’s mental differentiating of concepts and aspects of unpredictable systems. Rhizomes and Arborescence: These final two concepts are the only terms in this list drawn from Deleuze’s later collaboration with Felix Guattari, the two-volume Anti-Oedipus (2004 [1972]) and A Thousand Plateaus (2004 [1974]). Systems, they propose, can be structured in a rhizomatic manner, or an arborescent one, drawing metaphorical relevance from cellular biology: arborescent systems are akin to trees, with a linear central trunk and all else being dependent on that trunk, while rhizomatic systems are more akin to colonies of fungi, or hive-building insects. ‘Arborescences are hierarchical, stratified totalities which impose limited and regulated connections between their components’, while rhizomes are ‘non-hierarchical, horizontal multiplicities which cannot be subsumed within a unified structure, whose components form random, unregulated networks in which any element may be connected with any other element’.85 The word ‘random’ is used here to refer to what I have termed ‘unpredictability’, as an overarching catch-all phrase for all phenomena of non-determinism or indeterminacy or stochasticity, rather than what I propose as randomness, but these definitions clearly highlight the distinction between these concepts. The concept of the rhizome ‘allows for an autonomy of the “reading material” without having to organize the user in a restrictive way’86 – which is to say, whereas games that lack unpredictable elements in their systems can only organize the user’s behaviour in one way, the ‘reading material’ – the game – gains a certain quasi-autonomy when unpredictable elements are introduced, shaping itself in new and different ways and thereby enabling a greater range of possible interactions on the part of the user. Two rhizomes represent alternatives,

INTRODUCTION: UNPREDICTABLE PLAY

19

not opposition,87 and can be switched out and replaced more easily than elements in a hierarchic, arborescent structure. Rhizomatic elements may therefore be alternated, exchanged; arborescent elements bifurcate and open up a range of options, through which (in most cases) rhizomatic elements are in turn distributed. The system that procedurally generates a world map is arborescent, while the elements it distributes into a final map are rhizomatic. These terms and the structuring and ordering of variables they imply will be of value particularly in the exploration of luck, as well as to a lesser extent in the exploration of the generation of what I will call ‘significance’ (important elements) in randomness and chance, for they help us conceptualize how unpredictable games distribute their elements, and how players move through those elements. With this terminology and its relevance to games established, it is interesting to note that Deleuze does himself make references to games on several occasions including giving a prominent place to questions of what I term unpredictability. In The Logic of Sense, Deleuze proposes a definition of games, which importantly argues that ‘rules determine hypotheses which divide and apportion chance, that is, hypotheses of loss or gain’, and that such ‘hypotheses organise the playing of the game according to a plurality of throws’, each of which ‘brings about a fixed distribution corresponding to one case or another’.88 In these passages he understands games as being first and foremost what we might think of as vehicles for the distribution of outcomes, according to sets of rules. He, however, emphasizes the role of unpredictability in those rules, whereby players lose and gain through unpredictable processes that in the Deleuzean sense repeat, yet due to their core unpredictability, repeat difference with each throw, rather than repeating (in the traditional sense of the word) any form of generality. In Difference and Repetition, meanwhile, he suggests that when players are given a ‘situation of chance or multiplicity’, one seeks to ‘impose limits upon it’ and ‘bring about the return of the same’,89 which is to say particular outcomes mentally noted as being desirable. In this Deleuze notes that we find ourselves immediately trying to reset or counter unpredictability whenever we find it – even if it is that same unpredictability which first compelled us to consider the game in question. This is a theme which runs throughout the present work: overcome by the opacity, confusion, apparent unfairness and capriciousness of unpredictable systems, players do much to change their gameplay practices, the discourses they understand their gameplay through or the very accepted design of the games they play. However, these are the limits of Deleuze’s prior engagements with games and play, which more often take the form of metaphor or ideal system than an examination of real games played by real individuals; this is a gap which will be bridged by the present work. Finally, before proceeding to the outline of the work, a crucial critique in my theoretical choice must be anticipated and addressed. Deleuze

20

THE UNPREDICTABILITY OF GAMEPLAY

acknowledges that it is impossible to ever fully list or conceptualize or hold in one’s mind all the differences between two repetitions. In our consideration of games, this is certainly the case, for the player will be a little older each time, will have different experiences to draw on, the surrounding temperature might have changed, they might be more or less hungry, an ambulance might be driving by with its siren on, the gravitational pull of the earth might have shifted one iota – and so on. There is always an infinite set of other repetitions related to anything or any matter. Consequently, we cannot ever grasp the infinite set of differences between one thing and its relation to an infinite number of others, and another thing and its relation to that same infinite set of others: ‘The difference is resistant to actual identification.’90 In this regard, if we consider ‘factors which change with each playthrough and affect play, even in the smallest way’, what game scholars call the ‘magic circle’ – a separate space, ‘either materially or ideally’,91 marked off for the play of games – becomes unfathomably huge,92 and riddled with uncountable differences of every sort that divide each repetition from the next (just as all the external factors mutually shape the territory of the prowling animal). Why, therefore, focus solely on differences and repetition written into the game’s code (or formal rules and the actions of play, as in a card game)? The answer is that unlike the examples listed above which might affect gameplay in some sense, these repetitions are central to the play of these games: they are not incidental, they are fully ‘within’ the technical or formal fabric of the game, and do not exist outside it. They take place either from the game’s fabric functioning as anticipated or expected, or in the case of what I later term ‘instability’, from that fabric functioning ‘incorrectly’93 in some manner; but they are nevertheless part of the game as an item, as an artefact, as a set of practices, as a way of mobilizing circuitry within a computer, as a way of giving meaning to fifty-two pieces of cardboard with pictures and symbols drawn on them. As a result, these forms of repetition and difference are actively regarded and considered and thought about by players; they are not incidental, outside things, which have an effect on the game but are not, except within a broad definition of the magic circle, part of the game. They enable players to take an in-game action, and be unsure – even with perfect knowledge – of the outcome. These elements make their contribution to gameplay experience distinct, and an appropriate analytic focus in their own right.

Outline of the work As noted earlier, the purpose of this work is to create a (primarily experiential and ontological) framework for considering gameplay unpredictability, and to develop an understanding of the game-playing subject when confronted

INTRODUCTION: UNPREDICTABLE PLAY

21

with unpredictable games. It is concerned with clarifying the many forms unpredictability can take and subsequently proposing models to understand the player experience of engaging with unpredictable games. However, it is also interested in the more ‘material’ objective of categorizing overlooked aspects of game design and offering a framework for critically minded game designers seeking to understand the varied forms that unpredictability may take, identify which forms they wish to include in their games and develop systems (or avoid accidentally developing systems) of the sort explored in this book. The second chapter identifies the first of the work’s central three-part typology – that of ‘randomness’, defined as the unpredictability of initial starting conditions. It first analyses how games create differenciated starting conditions from a virtual multiplicity of possible elements and methods of element distribution. This leads to a discussion of the importance of the initial conditions of gameplay and the extent to which any one instance of these reverberate throughout a given instance of play. The chapter then explores the creation of significant elements through the distributions of unpredictable elements via unpredictable processes, all of which draw their importance from the other elements they are distributed alongside, with a particular focus on how the sets of possible interactions in unpredictable games are, themselves, unpredictable. It then introduces the concept of the ‘granularity of interest’ – the volume of unpredictable components that are interesting and the chance that they will form themselves into interesting constellations – as a method of assessing the extent to which the differential elements and differential relations in a particular system create (or fail to create) singularities as perceived by each individual player. The chapter concludes by examining how players perceive these differenciating processes via the assessment of perceptions, and how these assessments yield system knowledge (how elements are distributed) and variable knowledge (what elements are distributed). Several discussions of the second chapter could be potentially placed in the third chapter (on chance), but studying the creation of significance, the granularity of interest and the acquisition of system and variable knowledge seem more usefully considered at this earlier point. Studying the creation of significance through the distribution of unpredictable variables is best positioned in a discussion of randomness, not chance, for in the former in most cases a vastly greater number of variables are distributed, and yield significance, at any one time (the start of the game) than in chance (ordinarily slowly, and bit by bit). In the case of the granularity of interest, randomness is generally of greater and longer-term consequence to the player, and creates sweeping structures of significance and importance rather than piecemeal interventions throughout play, making randomness an ideal point to examine the concept. In the third case, the notion a player builds up of randomness is ordinarily more complex than that of chance,

22

THE UNPREDICTABILITY OF GAMEPLAY

owing to the far greater potential for emergent phenomena and singularities in the former, thereby allowing us to explore how players gain knowledge of unpredictable systems in the greatest depth through considering it via randomness. The third chapter identifies the second part of the central typology – ‘chance’, defined as unpredictability during ongoing gameplay. The chapter first assesses how moments of uncertainty are distributed throughout a game, and examines some of the dominant kinds of chance in games, utilizing illustrative case studies for each, before summarizing in a broader sense the overall purposes of introducing chance into the ongoing play of a game. These are broken down into four potentially overlapping categories: the distribution of outcomes which are qualitatively similar but quantitatively distinct; the distribution of outcomes which are qualitatively distinct; the distribution of outcomes which might include a null outcome that effects no change in the game world; and the potential for all outcomes to be positive for the player, all negative or a mixture. Focusing back on how players respond to unpredictability, the chapter concludes with an analysis of the concept of the ‘Random Number God’, a personification of the ‘will’ of systems that contain only mathematical differenciation that is invoked whenever moments of unpredictability, specifically chance, are perceived as having been unfairly cruel and ending or negatively affecting a player’s experience. This sheds further light on how players therefore experience, and come to create narratives and internal psychological models of, in-game chance. The fourth chapter identifies the third part of the work’s typology – ‘luck’, defined as the unpredictability of the eventual outcomes of an instance of play, and therefore the extent to which player actively during gameplay can or cannot influence its resolution. I define luck as the existence of lines of arborescence that stem from unpredictability in a game and endure until the end of that instance of play. The chapter explores the use of luck in games as a means to levelling a playing field between game players of different skill levels, and unpicks both the motivation of this game design goal and its effects on player experience of unpredictability. This is closely related to questions of the long term and short term in gameplay, and so the chapter explores how games with tremendous luck in the short term might, over hundreds or as many as millions of moments, become dominated instead by skill. In addition, I posit that despite the strong presence of skill in many games of luck, randomness and chance are often mistaken for luck, and the chapter concludes by addressing this issue, exploring in detail a valuable case study – modern ‘roguelike’ game FTL (2012) – of this common player confusion. Unpicking the concept of luck thereby allows us to look more closely at the extent to which unpredictability does or does not shape gameplay, and the importance of considering luck as a distinct location of unpredictability from two others mentioned here.

INTRODUCTION: UNPREDICTABLE PLAY

23

The fifth chapter – on instability – distinguishes between intended and unintended unpredictability (identifying the existence of exploits and glitches in the latter), and then seeks to develop a critical separation between ‘glitches’ – a breakdown in a game’s systems – and ‘exploits’ – the correct functioning of a game’s systems, but in an aleatory point unanticipated by a game’s designers. It also covers debates over the definitions of an ‘exploit’ and a ‘strong strategy’, noting that this line often blurs and that conclusions of such debates often come down to just how strong a strategy represented by a particular singularity is. The chapter then develops the concepts of ‘tight coupling’ and ‘weak coupling’ to describe games (tightly coupled) with a small number of simplistic rules and elements, which tend to be highly stable (such as draughts), and games with a large number of complex rules and elements (weakly coupled), which tend to be highly unstable (such as digital RPGs). I then note the existence of both ‘internal’ and ‘external’ instability – for example, the difference between a glitch in a video game, and someone stealing a chess piece from the board, when the non-game world intrudes into the game – and what these mean for player experience, and the restructuring of a player’s multiplicity for the game they are playing. The chapter then concludes by examining how instability can be embraced by players, redefining and shifting the accepted rules of a game, and the enjoyment to be found exploring game instability. The sixth chapter is the first of three chapters to apply this theoretical analysis to three practical cases, beginning with the consideration of ‘procedural content generation’. Specifically, it focuses on some of the experiences that might be gained playing a PCG game which rarely, if ever, can be found in other games with randomness and chance, primarily due to the complexity of procedural generation systems and the tremendous range of gameplay purposes to which they have been put. To do so, the chapter focuses on a case study of ‘roguelikes’, the game genre most closely associated with procedural generation, and which deploy such processes in the greatest range of scenarios. Having outlined this genre, the discussion then proceeds to discussion of the relationship between three aspects I term depth, length and co-dependency in PCG, aiming to develop a toolkit through which we can examine some of the experiences that are almost, or entirely, exclusive to games with PCG. ‘Depth’ refers to the number of steps in a generative system that leads to its outcome; ‘length’ refers to the number of steps in a chain of connections and influences between multiple procedurally generated elements; ‘co-dependency’ refers to the extent to which factors in that chain are deeply, or loosely, causally interwoven. These aspects, in turn, allow for new experiences to be had by players: confusion and puzzlement, a sense of scope and scale, and a sense of grounding and realism, respectively. The chapter then concludes by exploring player perception of procedural generation by examining how players identify procedural from non-procedural elements in games, developing a notion of

24

THE UNPREDICTABILITY OF GAMEPLAY

a ‘procedural aesthetic’ in such games and considering how these judgements (whether accurate or inaccurate) affect player experiences across multiple playthroughs of a game. The seventh chapter examines the notion of ‘replay value’ and the practice of ‘grinding’. Both are contingent on the deployment of randomness and chance in a game design, and are brought into being through the game design use of these unpredictable elements in very particular ways. The chapter first defines each of these – ‘replay value’ is the notion that some of the ‘value’ in a game is contingent on being able to play it more than once, while ‘grinding’ entails performing repetitive in-game actions waiting for a fortunate roll of chance (or, less often, randomness). The chapter explores the values assigned to, and the impetuses towards, both of these practices, and how they intersect with the possibilities within specifically unpredictable game systems. I then consider the tension between ease and difficulty in grinding, specifically the existence in many games of systems to make grinding more palatable, without actually removing the practice altogether, and what this shows us about both player and developer expectations of grinding as a practice. It then explores the notion of ‘completionism’ – exhausting all the possibilities within a particular game – and what pursing the goal of completionism via replay within an unpredictable game means for player experience, and particularly a player’s experience of chance. The eighth chapter deals with what I term the ‘negation of unpredictability’, gameplay practices designed to force an unpredictable game into offering identical experiences to multiple players (or to the same player multiple times). The chapter explores three of the most well-used and well-tested methods for players wishing to transform unpredictable games into semi-deterministic or fully deterministic games: ‘duplicate deals’ used in some competitive card game competitions to mitigate the randomness of traditional shuffling and dealing, and thereby establish a shared baseline for play between teams; the practice of ‘save scumming’, which entails creating a backup of one’s save in a game featuring ‘permadeath’ (the permanent loss of one’s character upon death), and then restoring that save if one’s character perishes or one is faced with a displeasing selection of unpredictably selected elements; and ‘gameplay seeds’, strings of numbers that allow more than one player to play a given unpredictable game in the exact same way in two different physical locations. The chapter concludes by summarizing the social and game design contexts surrounding such forms of luck-heavy play, and considers how the desire to negate unpredictability by altering the design of the games themselves shows the importance of wider setting, as well as individual experience, to player responses to unpredictability. The ninth and final chapter summarizes the work’s theoretical foundations, typological development and the case studies through which this four-part framework has been elucidated, and its value hopefully demonstrated. The chapter begins with an overview of the work before returning to the

INTRODUCTION: UNPREDICTABLE PLAY

25

Deleuzean foundations of the study, and recapping the definitions proposed of randomness, chance, luck and instability. It also notes why such a separation into four forms is necessary to explore in greater detail the wide range of gameplay experiences they create and underpin. The chapter concludes by highlighting the importance of studying the experience of unpredictability, some of the core findings of the work, and the overall range of present and future insights that might be gathered through the deployment of such a typology. Before proceeding to these analyses, a couple of acknowledgements should be made about both the methodological and epistemological foundation of this work, and also the specific kinds of games the work primarily uses as its case studies. In the first case, it is usual in the discussion of many subjects for scholars for talk about ‘you’, ‘one’, ‘we’ or the ‘imagined reader’. Addressing what exactly these terms mean in a work such as this is essential. It is of course correct to have a specific subject population in mind when discussing psychological or affective states: however, in this work my focus is the conditions of possibility for such states. For example, in talking about the rules of games as setting out the conditions for fear or excitement in players, a subsequent discussion would not look to attribute those states to particular players, but rather to possible players as a whole. I am not talking about a single subject, but rather about possible game-playing subjects, in many cases subjects whose particular perspectives are shared by large communities (see the discussions, topics, cultures and norms I cite later in this book). The adoption of a case studies approach throughout the work is thus especially valuable for helping show the reality, accuracy and commonality of many of the possible states subsequently explored. These principles and sources of knowledge therefore do not discount the obvious tremendous variety in player experiences, but rather offer a cumulative look at both common experiences and the wider range of experiences that players have in games. Second, given the nature of my topic, much of my discussion of unpredictable computer games focuses on a reasonably small number of genres: roguelikes, RPGs more generally and strategy games. In terms of analogue games, my focus is inevitably on card games, to a lesser extent where relevant, casino games and occasionally tabletop games. These games are the most common to use unpredictability (although rarer examples can be found in almost every game genre), they use unpredictability in the greatest range of ways and to the greater number of purposes, and they offer the most compelling and unusual case studies, a large number of which are developed throughout this work to illustrate and ground the broader theoretical and conceptual discussion. This is of course not an exhaustive summation of all games of all sorts, and not even all games which exhibit systemic unpredictability, but these selections offer us the greatest variety in the implementation and gameplay effects of unpredictability, the most

26

THE UNPREDICTABILITY OF GAMEPLAY

potential for detailed study, and in all of these games unpredictability is central instead of tangential. Much of what I aim to do here is unify our examination of these kinds of games from a Deleuzean perspective, highlighting their commonalities and differences, and the different gameplay experiences that emerge from the different ‘locations’ in which unpredictability might be found. My focus is on the lived experience of playing such games, the many different forms of unpredictability manifested within these games, and how we might come to understand and typify these forms and move towards a comprehensive typology of ludic unpredictability. This is the task I have set myself in this work.

Notes 1 Concepts of the ‘beginning’ and ‘end’ of play, distinctions we might draw between a ‘session’ and an ‘instance’ of play, and how players conceive of and understand their play activities through such structures, are ripe for further theorization. In the case of card games, we might understand an instance of play as a single hand of poker, the time spent playing without a break, the time spent playing including a break so long as the player considers themselves to be ‘taking a break before returning to poker’ rather than ‘doing something else’, or the time spent within a tournament or within a certain cash game regardless of external actions or occurrences. In video games, an instance of play might mean one attempt at navigating a section of that game, the period during which the controller is in one’s hands, the period during which the games console is turned on, a single match in a multiplayer game and so forth. For the duration of this work, however, I proceed on the basis that an instance of play can be understood through a systems sense, identified by when the player finds themselves at the same part of a game twice at different times – two hands of poker, two procedural generations of a game world, clicking the ‘New Game’ button twice, reloading a game to attempt the same segment again, and so forth. This position seems especially valuable for considering unpredictable games, as these are games fundamentally informed by, as I argue in this work, the difference across instances of play. 2 John M. Roberts, Malcolm J. Arth and Robert R. Bush, ‘Games in culture’. American Anthropologist 61, no. 4 (1959), 601. 3 Irving L. Finkel, ‘On the rules for the Royal Game of Ur’. In Ancient Board Games in Perspective: Papers from the 1990 British Museum Colloquium, with Additional Contributions, edited by Irving L. Finkel (London: British Museum Publications Limited, 2007). 4 For discussion, see Peter A. Piccione, In Search of the Meaning of Senet (Boston and New York: Archaeological Institute of America, 1980), and W. John Tait, ‘Were there gamesters in pharaonic Egypt?’ In Ancient Board Games in Perspective: Papers from the 1990 British Museum Colloquium,

INTRODUCTION: UNPREDICTABLE PLAY

27

with Additional Contributions, edited by Irving L. Finkel (London: British Museum Publications Limited, 2007). 5 Roberts, Arth and Bush, ‘Games in culture’, 601–2. 6 D. S. Robertson, ‘The Delphian succession in the opening of the Eumenides’. The Classical Review 55, no. 2 (1941), 69–70. 7 See Matilda Coxe Stevenson, ‘Zuñ i games’. American Anthropologist 5, no. 3 (1903), and Stewart Culin, Games of the North American Indians (Courier Corporation 1975 [1907]), 468–97. 8 Kenneth G. Zysk, Asceticism and Healing in Ancient India: Medicine in the Buddhist Monastery, vol. 2. (New Delhi: Motilal Banarsidass, 1998). 9 Florence Nightingale David, Games, Gods and Gambling: The Origins and History of Probability and Statistical Ideas from the Earliest Times to the Newtonian Era (New York: Hafner Publishing Company, 1962), 15. 10 Mihai Spariosu, Dionysus Reborn: Play and the Aesthetic Dimension in Modern Philosophical and Scientific Discourse (New York: Cornell University Press, 1989), 15. 11 Brandon Dotson, ‘Divination and law in the Tibetan Empire: The role of dice in the legislation of loans, interest, marital law and troop conscription’. In Contributions to the Cultural History of Early Tibet, edited by Matthew Tom Kapstein and Brandon Dotson (Leiden, The Netherlands: Brill, 2007), 1–78. 12 Maya Burger, ‘Drawing cards, playing destiny: Karma and play in new divinatory practices’. In Religions in Play: Games, Rituals, and Virtual Worlds, edited by Philippe Bornet and Maya Burger (Zü rich: Theologischer Verlag Zü rich, 2012). 13 René  Girard, Violence and the Sacred (London and New York: Continuum, 2005), 99. 14 The most common proposition is that games of chance emerge out of practices which originally entail divination. However, this is not a universal perspective. For the argument that divination is the antecedent element of the two, see Mihaly Csikszentmihalyi and Stith Bennett, ‘An Exploratory Model of Play’. American Anthropologist 73, no. 1 (1971), 45–58. For an argument that the two are more deeply interwoven, that board games have been converted into divination practices as well as divination practices yielding games, and that any kind of chronological organization is both highly challenging based on existing anthropological and archaeological evidence and might inevitably overlook deep geographical-cultural specificities, see Win M. J. Van Binsbergen, ‘Time, space and history in African divination and board-games’. In Time and Temporality in Intercultural Perspective, vol. 4, edited by Douwe Tiemersma and Henk Oosterling (Amsterdam: Rodopi, 1996). 15 Per Binde, ‘Gambling and religion: Histories of concord and conflict’. Journal of Gambling Issues 20 (2007), 145. 16 Van Binsbergen, ‘African divination and board-games’, 107. 17 Deborah J. Bennett, Randomness (Cambridge, MA: Harvard University Press, 1998), 11.

28

THE UNPREDICTABILITY OF GAMEPLAY

18 Gottfried Wilhelm Leibniz, New Essays on Human Understanding, translated and edited by Peter Remnant and Jonathan Bennett (Cambridge: Cambridge University Press, 1981 [1765]). 19 In Leibniz’s essay ‘An Odd Thought Concerning a New Sort of Exhibition’ he describes a range of games and game-like, or at least playful, enterprises and contraptions and systems that would merit display to the public, through which public understanding of technologies and systems might be developed. Although war games are mentioned as an integral part, he notes that an ‘Academy of Pleasures’ might be more akin to what he had originally imagined, but would be less enticing to the public; from this we note an immediate contrast between the playful nature of games and the real-world purposes a game-like understanding of decision-making and complex systems can be put to, and his era, was already being put to. For more on this, see Philip P. Wiener, ‘Leibniz’s project of a public exhibition of scientific inventions’. Journal of the History of Ideas 1, no. 2 (1940), 237. In turn, Deleuze briefly addresses the notion of the ‘academy of games’, although he calls it a ‘disturbing institution’; see Gilles Deleuze and Charles J. Stivale, ‘Vincennes session of April 15, 1980, Leibniz seminar.’ Discourse 20, no. 3 (1998), 77–97. This is because such an institution would inevitably serve a multiplicity of purposes, at least as Leibniz imagined it within the historical episteme he operated within (and co-constructed). As well as being akin to a ‘zoological or botanical garden’ where one might regard and consider ludic endeavours as one would a rare plant or exotic animal, such a place would also, Deleuze argues, resemble a ‘police enterprise’. He declines to expand on this claim, but we see an echo in Lazardzig’s claim that an academy of games would be a place which would portray ‘strategic deception and illusions of perspective’; see Jan Lazardzig, ‘The machine as spectacle: Function and admiration in seventeenth-century perspectives on machines’. In Instruments in Art and Science: On the Architectonics of Cultural Boundaries in the 17th Century, vol. 2, edited by Helmar Schramm, Ludger Schwarte and Jan Lazardzig (Berlin and Gö ttingen: Walter de Gruyter, 2008). An academy of games would inevitably teach its attendees how to control populations, make game-theoretic choices, regard society with a powerful mathematical and systemic eye and so forth; they might ‘establish order’, as noted in Wiener, ‘Leibniz’s project’, 238. In all of these cases we see something of this tension between games as play and games as an understanding of complex worldly systems, and the purposes to which the latter might be put. For a broader and excellent investigation of Leibniz’s greater role in the history of games, their ideological and mathematical foundations, and the proto-game-theoretic ways of viewing the world that early war games inculcated, see Philipp Von Hilgers, Philipp and Ross Benjamin, War Games: A History of War on Paper (Cambridge, MA: MIT Press, 2012). 20 Gerda Reith, The Age of Chance (London, USA and Canada: Routledge, 1999), xiv. 21 Reith, Age of Chance, 29. 22 Marieke De Goede, Virtue, Fortune, and Faith: A Genealogy of Finance (Minneapolis, MN: University of Minnesota Press), 2005.

INTRODUCTION: UNPREDICTABLE PLAY

29

23 Spariosu, Dionysus Reborn, 211. 24 De Goede, Virtue, Fortune and Faith, 50. 25 Reith, Age of Chance, 74. 26 For examinations of skilled gamblers, see Jeremiah Weinstock, Carrie E. Massura and Nancy M. Petry, ‘Professional and Pathological Gamblers: Similarities and Differences’. Journal of Gambling Studies 29, no. 2 (2013) 205–16, and David M. Hayano, ‘The professional gambler: Fame, fortune, and failure’. The Annals of the American Academy of Political and Social Science 474, no. 1 (1984), 157–67. 27 John Von Neumann, ‘Zur theorie der gesellschaftsspiele’. Mathematische Annalen 100, no. 1 (1928), 295–320. 28 John Von Neumann and Oskar Morgenstern, Theory of Games and Economic Behavior (Princeton, NJ: Princeton University Press, 2007 [1944]). 29 Spariosu, Dionysus Reborn, 211. 30 Of course, many other things affect gameplay experience, and the individual player does not exist as an ‘unencumbered self’: they are not always alone (e.g. play with friends, with family, over live-streaming software), they play within institutions (e.g. competitive teams or guilds), and they sometimes play for a diverse range of reasons entangled with political economy (such as gold farming). Naturally these complications of context exist – and indeed, most of my scholarly work is actually about these issues – but these are not the focus of the present work. I am concerned instead with considering the experiences and thought processes of players when facing unpredictable games, and in the latter half of the book, the cultures, practices and norms that have arisen as a result of these interactions. 31 Johan Huizinga, Homo Ludens: A Study of the Play-Element in Culture (Kettering, OH: Angelico Press, 2016 [1938]), 48. 32 Roger Caillois, Man, Play and Games (USA: University of Illinois Press, 2001 [1958]), 17. 33 Caillois, Man, Play and Games, 78. 34 Alasdair MacIntyre, After Virtue: A Study in Moral Philosophy (Indiana: University of Notre Dame Press, 2007). 35 Mihaly Csikszentmihalyi and Isabella Selega Csikszentmihalyi, Optimal Experience: Psychological Studies of Flow in Consciousness (Cambridge, UK: Cambridge University Press, 1992), 185. 36 Csikszentmihalyi and Csikszentmihalyi, Optimal Experience, 185. 37 Reith, Age of Chance, 127. 38 Ibid., 147. 39 Ibid., 146. 40 Harvey Ferguson in Reith, Age of Chance, xvii. 41 Jesper Juul, The Art of Failure: An Essay on the Pain of Playing Video Games (Cambridge, MA: MIT Press, 2013), 14.

30

THE UNPREDICTABILITY OF GAMEPLAY

42 Thomas Malaby, Gambling Life: Dealing in Contingency in a Greek City (Champaign, IL: University of Illinois Press, 2010), 18. 43 Reith, Age of Chance, 126. 44 See Tauel Harper, ‘The smooth spaces of play: Deleuze and the emancipative potential of games’. symploke 17, no. 1 (2009), 129–42; Colin Cremin, Exploring Videogames with Deleuze and Guattari: Towards an Affective Theory of Form (Oxford and New York: Routledge, 2015); Tom Betts, ‘An investigation of the digital sublime in video game production’ (PhD diss. University of Huddersfield, 2014); Chiel Kattenbelt and Joost Raessens, ‘Computer games and the complexity of experience’. In Level Up: Digital Games Research Conference, edited by Marinka Copier and Joost Raessens (Utrecht: Faculty of Arts, Utrecht University, 2003); Nick Dyer-Witheford and Greig de Peuter, Games of Empire: Global Capitalism and Video Games (Minneapolis, MN: University of Minnesota, 2009); Ian G. R. Shaw and Barney Warf, ‘Worlds of affect: Virtual geographies of video games’. Environment and Planning A 41, no. 6 (2009), 1332–43. 45 Colin Cremin, ‘The formal qualities of the video game: An exploration of Super Mario Galaxy with Gilles Deleuze’. Games and Culture 7, no. 1 (2012), 73. 46 Cremin, Exploring Videogames. 47 Harper, ‘The smooth spaces of play’. 48 Betts, ‘Digital sublime’, 32. 49 Kattenbelt and Raessens, ‘Complexity of experience’. 50 Joe Hughes, Deleuze’s ‘Difference and Repetition’: A Reader’s Guide (London and New York: Bloomsbury Publishing, 2009), 14–15. 51 James Williams, Gilles Deleuze’s Difference and Repetition: A Critical Introduction and Guide (Edinburgh: Edinburgh University Press, 2013), 2. 52 Henry Somers-Hall, Deleuze’s Difference and Repetition: An Edinburgh Philosophical Guide (Edinburgh: Edinburgh University Press, 2013), 2–3. 53 John Protevi, ‘Deleuze and life’. In The Cambridge Companion to Deleuze, edited by Daniel W. Smith and Henry Somers-Hall (Cambridge: Cambridge University Press, 2012), 239. 54 Manuel DeLanda, Intensive Science and Virtual Philosophy (London and New York: Bloomsbury Academic, 2013), 196. 55 I certainly do not claim a stronger handle on Deleuze than any of these scholars, but I believe the reading of his work I put forward – situated between a ‘traditional’ philosophical understanding and the primarily mathematical understanding proposed (or perhaps, developed) by Manuel DeLanda – gives as true a reading as any, and certainly the reading of most value to the project of this work. 56 John Cohen, Chance, Skill, and Luck (Wokin and London: Pelican, 1960), 7. 57 Malaby, Gambling Life, 10. 58 Gilles Deleuze, Difference and Repetition (London and New York: Bloomsbury Academic, 2014 [1968]), 1.

INTRODUCTION: UNPREDICTABLE PLAY

31

59 Deleuze, Difference and Repetition, 4. 60 IEP, ‘Gilles Deleuze (1925–1995)’. Internet Encyclopaedia of Philosophy. Available at http://www.iep.utm.edu/deleuze/. 61 Clayton Crockett, Deleuze Beyond Baidou: Ontology, Multiplicity, and Event (New York: Columbia University Press, 2013), 53. 62 Daniel Smith and John Protevi, ‘Gilles Deleuze’. In The Stanford Encyclopedia of Philosophy (2015). Available at http://plato.stanford.edu/archives/win2015/ entries/deleuze/. 63 Deleuze, Difference and Repetition, 1, emphasis mine. 64 Williams, Critical Introduction, 12. 65 Deleuze, Difference and Repetition, 28. 66 Hughes, Deleuze’s Difference and Repetition, 35. 67 Deleuze, Difference and Repetition, xix. 68 Ibid., 32. 69 Deleuze, Difference and Repetition, xv. 70 Cliff Stagoll, ‘Difference’. In Deleuze Dictionary Revised Edition, edited by Adrian Parr (Edinburgh: Edinburgh University Press, 2010), 75. 71 Deleuze, Difference and Repetition, 159. 72 Somers-Hall, Deleuze’s Difference and Repetition, 23. 73 Deleuze, Difference and Repetition, 71. 74 Ibid., 275. 75 Todd May, Reconsidering Difference: Nancy, Derrida, Levinas, and Deleuze (University Park, PA: Penn State Press, 1997), 186. 76 Hughes, Deleuze’s Difference and Repetition, 142. 87 Williams, Critical Introduction, 7–8. 78 Deleuze, Difference and Repetition, 321; 364. 79 Ibid., 365. 80 Ibid., 364. 81 Ibid., 259. 82 Katherine Swancutt, ‘The ontological spiral: Virtuosity and transparency in Mongolian games’. Inner Asia 9, no. 2 (2007), 249. 83 Gilles Deueze, The Logic of Sense (London and New York: Continuum, 2004 [1969]), 52. 84 In more recent Deleuzean work, such as that of Manuel DeLanda, multiplicities exist outside of perception if need be; multiplicities ‘specify the structure of spaces of possibilities’ (Intensive Science, 3), but can take the form of entirely physical processes without the necessary presence of a subject. As Joe Hughes puts it, DeLanda ‘‘‘reconstructs” Deleuze entirely within the language of science and mathematics, and tends to ignore the fact that the whole of Difference and Repetition is a theory of subjectivity’ (Deleuze’s Difference and Repetition, 183). As such, in this work, although I find

32

THE UNPREDICTABILITY OF GAMEPLAY

DeLanda’s reading highly valuable, I will adhere to the ‘standard’ Deleuzean reading: a multiplicity is the player’s perception of the repeating generative system they are currently engaging with, through a ‘progressive determination’ based on new sensations and new uses of a game’s system, and not the mathematical specifics of that system (although, of course, the perception of the former will hinge substantially) on the latter. 85 Ronald Bogue, Deleuze and Guattari (London: Routledge, 1989), 17. 86 Alice Van der Klei, ‘Repeating the rhizome’. SubStance 31, no. 1 (2002), 51. 87 Damian Sutton and David Martin-Jones, Deleuze Reframed: Interpreting Key Thinkers for the Arts (London and New York: IB Tauris, 2008), 4. 88 Deleuze, Logic of Sense, 69. 89 Deleuze, Difference and Repetition, 147–8. 90 Williams, Critical Introduction, 12. 91 Huizinga, Homo Ludens, 10. 92 A note on the form of magic circle adopted in this work is useful here. As previously stated, I am specifically looking at games where, within the formal, material fabric of the game – the pieces, the code, the ruleset – unpredictability (the ability to put in one input and get a range of outputs as a result of mechanics or permutations within the game system) is present. This ability to get diverging outputs from identical inputs is what marks ‘unpredictability’, as I define it, out from other comparable concepts; for example, Costikyan’s ‘uncertainty’ is focused on a player not knowing what will happen next, which might emerge from input–output variability, but also simply from not knowing what other players are thinking, or not understanding the meaning of some important in-game element. This is therefore different, and focuses us in on the interaction between the non-human elements of the game, in which we feed input and get outputs, and the human element that experiences this. This might appear to be an unduly narrow scope when game unpredictability – if defined differently – is a topic of tremendous scale. However, other scholars have covered most of the elements under a hypothetical wider umbrella already, while what I term unpredictability specifically in game systems, and its effects on players, has not. Thinking about this input–output unpredictability therefore gives us a new way to think about a major segment of this broader field of indeterminacy not yet fully examined. Games are of course not ‘apart’ from the wider world of their play, but those elements are not my focus, except where parts of the wider world – such as gaming cultures – are clearly influenced by unpredictability, which is the focus of the second part of the present work. 93 The notion of ‘correct’ or ‘incorrect’ functioning of a game is not fixed, but contested; this will be explored primarily in Chapter 5, which examines what I term ‘instability’, which is to say the unpredictable flexibility of acceptable game rules and the presence of unanticipated possibilities inherent within that game’s differential relations and differential elements.

PART ONE

34

2 Games with Randomness

The initial conditions of gameplay There are eighty unvigintillion – which is to say, an eight with sixty-seven zeroes following it – possible shuffles of a standard fifty-two-card deck of playing cards. This means that despite all the times since the invention of the modern playing card that a hand of blackjack, bridge, rummy, whist, euchre, spades, baccarat or poker has been dealt, either live or across the uncountable hundreds of billions of hands played online, it is almost impossibly unlikely that the same hand has ever been dealt twice. A simple system – fifty-two cards, placed in an unpredictable order – can achieve incredible variation. There are many ways of shuffling a deck, and although the importance of understanding these systems may seem trivial to the casual player, they have a significant impact on what kind of unpredictability is created in the eventual sequencing of the cards. If cards are riffle shuffled ‘perfectly’ – the deck is split into two halves which are then perfectly interlaced in an ABAB pattern – the outcome is very predictable, even if performed multiple times (indeed, after fifty-two such shuffles, the deck will return to its original configuration). Alternatively, if the riffle shuffle is not perfect, there remain rising sequences: taking a single suit of cards 23456789TJQKA, riffling it once, might result in 293TJ4Q56KA78, which we note contains from left to right two rising sequences (2–8 and 9–A), preserving much of the original card order. Multiple riffle shuffles of a non-perfect sort, however, will eventually cause these sequences to disappear. Other shuffling techniques exist: the pile shuffle involves dealing cards into separate piles and then recombining them, wherein the level of unpredictability in the eventual deck is contingent on the pattern (or rather, the lack of pattern) with which the cards are dealt and eventually reconstituted into a full deck; an ‘Indian shuffle’ involves stripping out packets of cards from the deck which are then assembled bottom-up, resulting in a reordering of these packets but the preservation of rising sequences within them until performed multiple times; while online

36

THE UNPREDICTABILITY OF GAMEPLAY

poker sites rely on random number generators which use unpredictable data – the millisecond at which an unpredictably selected player last clicked their mouse, for example – to generate their number patterns, which then serve as the basis for the shuffling of each virtual deck of cards (and there are tens of thousands of ‘decks’ in play at any given time). Casinos instead tend to ‘wash’ the deck, which means the dealer places every card face down on the felt and then rapidly mixes them around with their hands before forming them back into a deck. Although less aesthetically elegant than the riffle and less mathematically elegant than the online ‘perfect shuffle’, this is considered as effective as the latter, although it leaves the possibility for a skilled player to attempt to visually track certain cards. In this introductory examination of playing cards we can therefore appreciate several crucial points: first, that even a system with only fifty-two components can readily generate an effectively unlimited volume of starting conditions for play; secondly, that there are many possible techniques for unpredictably reordering a set of constituent variables, which appear broadly indistinguishable and equally adequate to the untrained eye but contain some which are ‘more random’ than others and therefore more challenging for a skilled player to exploit; thirdly, that certain methods of generating unpredictable initial conditions are at least partly predictable, and that these methods result in recurring patterns being perceived; and fourthly that an understanding of all the components which go into the unpredictable system (i.e. the standard fifty-two cards) is of tremendous value to players’ strategizing, irrespective of how those components later come out of the unpredictable system. However, shuffling playing cards is not the only way by which the initial conditions of gameplay are varied. Numerous games, explored in greater depth throughout this work, utilize a technique known as ‘procedural content generation’ to create (most commonly) spaces of play in video games. This might mean distributing potential rooms in a building, the corridors in that building, and the items and creatures occupying those rooms and corridors, all of which are placed unpredictably. As significant as the potential volume of shuffled card permutations is, the potential volume of video game spaces produced by even the simplest form of PCG is orders of magnitude higher; if one considers the distribution of an unknown number of rooms, an unknown number of corridors, all of different sizes and connected differently, alongside items and creatures and any other features – which might themselves be unpredictably actualized from a multiplicity of virtual rules – along with connections to other areas of the game world, it is easy to see how a gameplay space of this sort has a far larger possibility space than shuffling cards. Therefore, given that a single dungeon level has vastly more permutations than a single deck of cards, and that a complete dungeon may consist of dozens, hundreds or a potentially infinite number of such levels, we readily perceive that the

GAMES WITH RANDOMNESS

37

randomness of a procedurally generated  game immeasurably exceeds, by incalculable orders of magnitude, the randomness that may be actualized through a deck of cards. Nevertheless, two orders of ‘near infinity’ of which the shorter contains sixty-seven zeroes are reasonably comparable for our purposes, so these can be conflated into a simpler statement: games with any level of significant unpredictability in their starting conditions – through a wide range of unpredictably selected components, a complex system which selects and connects those components, or both – can, very easily, offer a near-infinite space of possible starting situations for the unfolding of subsequent play. This kind of game-start unpredictability I term ‘randomness’. Distinct from ‘chance’ (ongoing unpredictability throughout a game) and ‘luck’ (unpredictability of the game’s outcomes), which I will explain the linguistic justifications for later, I believe randomness is the ideal term to describe this variation in initial gameplay conditions. The word ‘randomness’, or the idea that something ‘was random’, implies a degree of surprise, or of shock, which I feel is appropriate to the entire initial conditions of gameplay – rather than, as in chance, something passing, something likely to be a smaller element of gameplay – being presented to the player, who cannot anticipate what form it might take. Equally, randomness, more perhaps than chance or luck, speaks to the tremendous perceived breadth of possibilities that initial conditions can take. A game’s underlying systems and components can seem to create very random playing conditions, each of which might seem quite unrelated to others a player has previously seen (although this sense declines with time). As such, we will proceed using the term ‘randomness’ for the initial conditions of gameplay; as noted earlier this requires something of a redefinition of an existing term with use in both game studies and beyond, but given the lack of precision with which ‘randomness’ and its synonyms are presently used, and their potential to be separated and lead to greater analytic clarity, this will hopefully be a valuable redefinition. Randomness therefore occurs when two players might begin playing two copies of the same game, and what the game gives them to begin their gameplay experience – whether a differently shuffled deck, a differently built dungeon or anything else – is not identical. A single input (beginning play) leads to divergent outputs (the starting scenario of play). Having established the definition of randomness, the chapter now proceeds to considering the lasting impacts of initial conditions: which is to say, to what extent does randomness affect and shape gameplay? This allows us to begin to think about randomness as a distinct location of unpredictability, with distinct effects. This then develops into a deeper engagement with the creation of significant elements in games that include randomness, examining how the distribution of elements in different constellations can lead to profoundly different levels of importance for those elements, and consequently for the ebb and flow of player experience

38

THE UNPREDICTABILITY OF GAMEPLAY

in such games. The chapter then develops the concept of the ‘granularity of interest’, which is to say what proportion (or how many ‘bits’) of random conditions are interesting to players and what proportion are not, and how both the granularity of interest and the nature of that interest are mutually intertwined. The chapter concludes by returning to the earlier observation in this section about the development of knowledge over systems in variables in games, and unpicks how developing understandings of these influences how players deal with games with randomness, especially over multiple plays of the same game. It is therefore designed to identify randomness as a distinct location, identify trends in player experience that are solely or primarily present in the perception of randomness (rather than chance or luck), and begin to develop a more general appraisal of the importance of separating out unpredictability according to its location in a game’s design.

Lasting impact of initial conditions Few games offer the players what we might call absolute temporal symmetry – the ability to perform an action which is the exact opposite of the previous action and thereby reset the gameplay to the precise scenario which existed prior to that action (by which I mean rewinding within the act of play, not the non-diegetic act of accessing a menu and reloading a previous save). It is not often possible to remove items one has acquired, to strip one’s character of a power-up, to return board game pieces to their previous locations and so forth. Instead the game continues, and even performing a perfect reversal of a previous move in terms of the dispositions and locations of one’s own pieces may not return to the game to its previous state, as other actors within the game space, whether human or artificial intelligence or simulation or incidental, will have moved on in the meantime. This unshakeably linear temporal movement through the game means that the initial state of a game with randomness will continue to influence every moment of play until the game is concluded. As Alessandro Romano puts it, ‘In chaotic systems, small differences in initial conditions cascade through various iterations into drastically different outcomes.’1 The earliest decisions in such a game will be influenced only by the conditions of play and orientation of gameplay components generated by the game’s randomization processes – a world map, or a particular deal or cards and so on – while later decisions will be based upon both the conditions of play and all decisions made by players since the start of the game (each influenced by those conditions and all earlier decisions). As play continues the direct influence of initial conditions therefore gradually fades and the importance of player choices gradually increases (as play becomes further and further removed from the earliest actions which had only the initial gameplay conditions to base their decisions upon, and the number of player-made impacts on the game world

GAMES WITH RANDOMNESS

39

increases). However, each of these choices, if taken back to their origin, will nevertheless return to the first player who made the first move after the generation of the conditions of gameplay, and whose first move will in turn have been inevitably influenced by those conditions. This gives rise to several related questions this section will examine: How much influence do initial conditions have on an entire sequence of gameplay? How long does this influence last? Does this vary from game to game, and if so, why and in what ways? In doing so this section will continue developing the factors that make randomness highly distinct from chance in gameplay purpose, theoretically position randomness with regard to the actualization of possible gameplay experiences (both immediately and throughout a session of play), and attempt to trace how unpredictable starting conditions lead to unpredictable gameplay outcomes – or, sometimes, have far less effect on gameplay. First, by considering the randomness systems in different games, and how and how often players ‘return to’ or ‘use’ the initial conditions of gameplay, we can see that across different games and genres the length of time that initial conditions matter, and the ‘rhythm’ or ‘resolution’ in which they matter, varies. For example, in modern ‘roguelike’ games (which will be explored in greater detail later) such as The Binding of Isaac (2011) or FTL (2012), for example, the very earliest conditions of play – the layout of the first level, the items distributed across that level, the enemies distributed across the level and the challenges they pose – have only a small influence on the overall continuing play of the game. The items acquired in the early floors or systems of Isaac or FTL will be often replaced or superseded, and the map layout and the strategies it implied (and challenges provided) will soon be left behind. Although they establish the basis of a playthrough, it is not required to continue using what was acquired in the first area. There are ample opportunities to change one’s character/ship and to regain any lost health, and these are games structured around adapting and refining one’s strategy according to the unpredictable developments across an entire playthrough. Equally, the inability to return to that first level means that much of what might have been enacted as a result of that particular level will actually never take place, and upon departing, we might reasonably state that everything about that level except what the player takes with them – any items, quests begun, reduced health – ceases entirely to ‘matter’ to the continuation of that player character’s adventure,2 and their chances of victory or defeat. By contrast, in a game in the Civilization series (1991–present) there is a single world map created on which the entire game plays out: nations are distributed, cities are founded, battles are fought, resources contested and randomly created but strategically important elements of the world’s geography – mountain passes, rivers and ocean trade routes – will be identified by players and will shape strategy until the conclusion of the game.

40

THE UNPREDICTABILITY OF GAMEPLAY

In some versions of the franchise natural resources remain undiscovered until a prerequisite technology is ‘acquired’ by a faction – such as in Civilization IV (2005) where each faction requires knowledge of ‘Bronze Working’ to reveal the copper resource on the map, knowledge of ‘Scientific Method’ to reveal the oil resource, and knowledge of ‘Physics’ to uncover uranium – meaning that new elements of the initial conditions will continue to come to light as the game progresses, rather than those elements being generated later in gameplay. The initial conditions in such a game therefore strongly shape gameplay until its conclusion. They will also shape gameplay in different ways as a playthrough of Civilization continues: what appeared at first to be an unimportant strip of land will later attain importance when oil deposits are discovered, for example, giving the initial conditions of play multiple kinds of significance over time, and making players return to and readdress those consequences, as play continues. When proceeding to a new solar system in FTL or a new floor in Isaac, the player only carries with them a comparatively minor reflection – their health, the items they possess – of the initial conditions of gameplay, whereas when proceeding to a new technological level or proceeding into a new conflict in Civilization, by contrast, everything which has gone before will have shaped, and will continue to shape, gameplay. Nothing, such as a past level, is left behind; and any part of the initial conditions might be returned to in the future.3 Secondly, any consideration of the lasting impact of the initial conditions of play upon play as a whole cannot ignore the question of whether the initial conditions affect the final or eventual outcomes of play – which is to say, does randomness create luck? Consider once more generating a world map for a game of Civilization, but in this case, instead of the generation of a realistic world with a diverse range of climates, terrain features, and civilizations spread out in an ‘organic’ manner across the virtual world, this world instead appears to have been handcrafted by some technologically superior civilization. The landmasses of this world are perfectly symmetrical; the climate zones are absolutely identical in both the north and the south of the planet; civilizations arose in the same exact places on the different land masses of the world; and natural resources, rather than being distributed according to billions of years of geologic processes (or rather, their simulation), are equally parcelled out to the world’s inhabitants. We could safely say, now that this game world is sufficiently symmetric to be akin to a game of chess or another board game where players start with precisely equal forces of precisely equal distribution, placement and ability, that there will be no lasting impact of the initial conditions on the outcome of gameplay in favour of one player or another. Such conditions will still shape gameplay – consider the possible experience of playing repeated games of chess where each iteration has a differently shaped though symmetrical board, but identical forces for both players are distributed in the precise same way on the two sides of the board – but they do not directly affect the outcome of

GAMES WITH RANDOMNESS

41

gameplay for or against any actor through a different distribution of ingame advantages and disadvantages (although it is still possible one player might be better at playing with particular kinds of initial conditions than another player, but the game has nevertheless offered all players the same resources and affordances). On the other hand, consider a game world where one player begins next to every resource on the world map, and another player begins next to none, or a pseudo-chess game where one player’s army of queens faces off against the lone king of their opponent. It is immediately clear that these initial conditions will influence gameplay, and do so significantly, perhaps to the point in its most extreme iteration of reducing a game to complete luck, which is to say there is either no room for gameplay, or gameplay cannot alter the state of play sufficiently to alter the outcome of the game. Symmetrical starting conditions will affect the progression of gameplay, but not its conclusion; asymmetrical starting conditions will affect progression and outcome of gameplay, up to the point of completely determining gameplay. As a result, trivially rather than severely asymmetric starting conditions – one player has one piece more, one player starts a little closer to an important strategic point, one player starts with a better hand of cards – will have some influence on gameplay, but the precise amount is extremely difficult to qualify or quantify. On one level, this is obvious (of course giving one player an advantage will influence gameplay), but our concern here is specifically with the relationship between the initial conditions of gameplay and the remainder of gameplay. An asymmetric condition which is favouring one player, but is not so extreme that one player is guaranteed to win, will shape gameplay significantly for both players, with different in-game abilities and opportunities, the need to develop different strategies and so forth. The initial conditions in games of this sort, when the initial conditions can vary between players in one playthrough rather than for the one player who takes part in each playthrough, will shape player experiences in one game in multiple different ways, as well as shaping their interactions with one another. In all cases, therefore, initial conditions in games with randomness-created unpredictability can have an effect on gameplay for all players, can have different effects on different players as well as different effects in different playthroughs. Having considered how different gameplay rhythms and structures change the extent to which initial conditions continue to matter, and how asymmetric starting conditions for different players will lead to different gameplay outcomes, we might next ask: What if an initial condition of gameplay is ignored? How should this be conceptually understood and how does this affect continuing gameplay, if at all? Consider again a strategy game with a randomly generated world map on which all of gameplay plays out, such as Civilization – as noted above, such games are the obvious archetype for video games where randomness has the largest long-term impact on the

42

THE UNPREDICTABILITY OF GAMEPLAY

unfolding of gameplay. This generated world might contain, for example, an island far away from where any of the players begin their empires, lacking in resources and not cutting across any major route of naval warfare or trade that develops. It is not hard to imagine a playthrough of such a game where that island is entirely ignored – no player settles it, no conflicts are fought over it, no ships have to move around it (for there are better trade routes) and nobody factors that island into their decision-making. For all intents and purposes, the presence of that island is indistinguishable, in the resulting gameplay its presence influences, from an open patch of ocean. Similarly, consider a hand of poker where the player is dealt three strong cards, and one weak card, and the weak card is never used in gameplay, and simply sits in the player’s hand until the conclusion of the hand. This is very possible in subtypes of poker such as Omaha hold ’em, in which a player is dealt four cards but can only actually use two of them to make the strongest possible hand. Did an irrelevant card dealt to a player affect gameplay, did it not affect gameplay or did it only affect gameplay by the absence of another possible card which would probably have positively (rather than negatively) affected gameplay? Of course, in both of these cases the irrelevant element was certainly not guaranteed to be irrelevant when it was generated or dealt; as we will explore in greater depth shortly in this chapter, the significance assigned to each randomly placed element, in these examples virtual islands and playing cards, is contingent upon the constellation of other elements they exist within. That same irrelevant island might have been the absolute centre of gameplay had the rest of that world map been generated differently; that same irrelevant playing card might have been the crux upon which that player’s strategy rested had the rest of the deck been shuffled and distributed differently. Nevertheless, in the particular permutations of differential relations and differential elements we are discussing here, these were indeed irrelevant components; but how should we understand them? To answer where we should position irrelevant components, we should begin by noting that the absence of a significant randomly placed element is ontologically no different from the presence of a significant randomly placed element; it still affects gameplay, but does so through players not using it or taking any advantage of whatever opportunities it might offer, rather than doing so through players actively using it or taking some advantage of it. Whether a player strongly engages with a random element and uses it, draws on it, builds a strategy upon it for the remainder of play or a player entirely ignores a random element (which they might not have, had that element been different), these are theoretically the same in that they shape gameplay; whether through presence, or through absence. Let us call those which create an effect ‘active’ random elements, and the other sort ‘null’ random elements. At first glance it might be proposed that there is no actual difference here, and that non-use equates to use of a different sort. There

GAMES WITH RANDOMNESS

43

is, however, an important distinction to be made between the relationships between active elements and other active elements, and null elements and other null elements, and where these two kinds of random elements should be conceptually positioned, which demonstrates that they are in fact of two different orders. All elements with an active impact on conditions will usually have a different active effect, although in some unusual cases it would be possible for them to have the same positive effect, so long as they share a characteristic or trait which is the basis for the effect – an obvious example would be the ability for a ‘Jack’ of any suit in a game of poker to form a straight, therefore allowing four different cards to exhibit the precise same effect. However, all elements with a negative effect on conditions will have the same negative effect, which is to say they are simply not used. Succinctly: active elements that could be actualized in a given context vary in their effects, but all null elements have the same effect. They could have been something which was used, but they were not, and now united in their irrelevance, all are not used in the same way, and are thus indistinguishable from one another in their lack of effect on gameplay. Given the Deleuzean underpinning of this work, we can therefore further suggest that games of randomness possess points of arborescence which flow from each generated initial condition of play, all situated within the broader rhizomatic nature of the possibility space denoted by randomness. These are what I have termed the ‘active elements’, actualized variables which, in the context of other actualized variables in the initial conditions of play, cause gameplay to be changed; as a result, future gameplay is built upon them. One possible tree of gameplay decisions is then played out in one instance of gameplay, but the roots from which that tree originates are only one in a vast forest of potential roots for the same game – all the other active elements that might have appeared in that context instead. It is also a tree which (in all but the simplest of randomization systems) will only be explored once, so vast is this forest and so varied the roots that can be found within it. As Deleuze and Guattari put it, ‘trees have rhizome lines, and the rhizome points of arborescence’4 – the construction of the initial conditions of a game is rife with the distribution of active and null elements, and then the active portion of those elements give rise to new sequences of gameplay. Randomness is rhizomatic (in its selection and orientation of different elements), but each rhizome selected only becomes the foundation of future gameplay if it matters. If a rhizome matters and has a direct impact on gameplay, it is consequently of the order of repetition; if it does not matter and gameplay would be no different had another rhizome been selected (an ocean instead of an uninteresting island, one unhelpful card instead of another unhelpful card), then it is of the order of generality. In failing to become repetition, the null random element cannot be the point from which new arborescence, in the form of a sequence of gameplay influenced or determined by that gameplay element, begins. An overlooked island or

44

THE UNPREDICTABILITY OF GAMEPLAY

an ignored card in one’s hand cannot actively change the gameplay of that instance of play if none of the players ever settle on that island, for example, or none of the players elect to use the uninteresting card. Null and active elements are therefore quite clearly differentiated (not alike in form), for all active elements have distinctive effects and all null elements have the same effect; in turn, null elements are undifferentiated in effect although certainly still differenciated in their actual manifestation (an irrelevant island looks unlike an irrelevant mountain, but if neither factor into any gameplay decisions, they serve the same lack of a purpose), while active elements are always differentiated in their effect and differenciated in their manifestation. However, the gameplay arborescence produced from the active elements can also be cut short and only influence some, rather than all, of the remaining gameplay; this is a topic we will return to later exploring luck and the unpredictability of game outcomes, but is important to first raise here for it is a central element of the lasting impacts of initial game conditions. Not all rhizomes, even those initial conditions which do affect gameplay, affect gameplay for the entire span of a game. For example, one randomly placed item acquired early in a game might readily be replaced with a stronger randomly placed item acquired later in the game; at that point its impact upon continuing gameplay is cut short. If all other possible items would have also been replaced by a player, and it would always have been replaced whatever it was, then the nature of the early-game item ceases to matter. If the later item found was of the order of randomness and was determined at the start of the game, then the initial condition the player saw – the first item – was always going to influence gameplay up to a certain point, and then cease to matter. If the second item was of the order of chance (explored fully in the following chapter), and decided unpredictably later in gameplay, then the tree born from that initial condition might, or might not, have been cut short. We therefore see that the initial conditions of gameplay, even from active elements, can range from influencing the entire play of a game to influencing only a portion, and that this variety is contingent, as with much else, on the interactions between multiple orders (randomness, chance) and multiple contexts of unpredictable elements. The initial conditions of play generated by randomness will therefore always have an effect on gameplay, but this effect is contingent on a range of factors, and the causal origins of those effects on gameplay can be located in many active elements, few active elements and – in a way – in the null elements as well.

The creation of significance We have now shown that some random elements can have consequence in a game of randomness, and others can lack direct consequence but continue to possess consequence by representing the absence of other possibilities,

GAMES WITH RANDOMNESS

45

while at the same time those same permutations which do not offer possibilities might have offered possibilities in a different context of other, related permutations. It is therefore now time to explore this final point, and consider why an element – the same element – in one permutation is entirely irrelevant and becomes reduced to generality which has no ‘positive’ impact on the flow of play, and why that same element in another permutation would be deeply significant, could influence gameplay or even become the core of the most important aleatory point in the entire game. We need to therefore consider more fully how significance or importance emerges from the distribution of variables by systems, and through player knowledge of those variables and systems, as well as the particular orientation of variables they are facing during a particular playthrough of an unpredictable game. What is or is not ‘significant’ or ‘important’ in this case refers to the player acting upon some singularities and not others as a result of the differential elements and differential relations actualized in one instance of a game, and how the different orientations of these factors can lead to the same element (or the same relation) taking on great or minimal significance, and thereby being important or irrelevant in the subjective experience of the player. In other words, a significant unpredictable element is one which impacts upon a hypothetical player’s experience of that game: this might be something a player notices, or something which affects what they later do, or what they think, remember, or is considered of value when making a later choice. Of course, a card (for example) that is unused in gameplay is not without significance in one sense, for a player will still engage, consider and discard (mentally) that card. It possesses affective content and will change their view of the hand, even if in terms of gameplay mechanics it lacks consequence to the strategizing player, who elects not to use it. We are therefore talking about significance in a systemic sense, which is to say which parts matter to gameplay in terms of the interactions between differential elements and relations offered to the player, and their orientations vis-à -vis each other. Considering a world map in a game where the world map is fixed, not generated, allows us to see that, following Deleuze and Guattari, any fixed initial condition exists within a ‘milieu of interiority’5 which lends inherent structural certainty to its presentation and operation across any number of gameplay instances. What makes parts of a map noteworthy is inherent to those parts, as selected and developed by the level designer, and how they are distributed. By contrast, a procedurally generated world map instead exists within a ‘milieu of exteriority’, where each tile or component on the map takes new significance in each generation according to the functions and identities of its neighbours within close proximity, and the wider shape of the map as a whole, and the actions of all actors (whether human or artificial) within the game world. One cannot speak before the game begins of the strategic importance of controlling a particular part of the map; one can only speak of combinations of tiles which would lend themselves to the

46

THE UNPREDICTABILITY OF GAMEPLAY

generation of parts of the map (aleatory points) which may be perceived as being important and have an effect upon the player (singularities). No tile will inherently be of consequence, nor be irrelevant; all tiles adopt their properties from the combination of all surrounding tiles, not any fixed or predetermined orientation. In A Thousand Plateaus (2004), Deleuze and Guattari develop these notions by, in fact, reference to game pieces. They note that a chess piece contains interiority, for its moves – the ability to move indefinitely in any orthogonal direction as a rook, the diagonal equivalent as a bishop and so forth – are inherent to those pieces. They might be deployed in different contexts, of course, such as a minor tactical skirmish in the corner of the board, or a checkmate, but ultimately ‘chess pieces are coded’ as bishops, rooks, knights and so forth; ‘they have an internal nature and intrinsic properties from which their movements, situations, and confrontations derive.’6 This is quite distinct from pieces in a game such as Go, which ‘have only an anonymous, collective, or third-person function’; they are ‘elements of a nonsubjectified machine assemblage with no intrinsic properties, only situational ones’. Deleuze notes that Go is more akin to ‘guerrilla warfare’ than a ‘war of extermination’; one does not catch one’s adversary (the King) as one does in chess, but rather one must encircle the presence of one’s opponent, who lacks any obvious location or manifestation on the game board, but whose presence is rather distributed wider across the entire field of play.7 As they put it: Within their milieu of interiority, chess pieces entertain biunivocal relations with one another, and with the adversary’s pieces: their functioning is structural. On the other hand, a Go piece has only a milieu of exteriority, or extrinsic relations with nebulas or constellations, according to which it fulfils functions of insertion or situation, such as bordering, encircling, shattering.8 This discussion by Deleuze and Guattari takes place at the level of the game pieces, or in a broader sense, the game elements (elements of video games and card games are rarely referred to as ‘pieces’, but ‘elements’ adequately captures what we might call the ‘moving parts’ of all kinds of games). However, when considering games of randomness and other kinds of unpredictability, we need to move outward from this focus on the relationships between game elements, and whether those emerge from interiority or exteriority, and shift to considering the set of all possible relationships between game elements, and whether those emerge from interiority or exteriority. Although the forms of relations present in chess and Go are quite distinct, as Deleuze and Guattari convincingly argue, the set of possible relations, whether interior or exterior to the individual game elements, is the same in these and their other mechanically deterministic cousins – it is a matter of interiority. When

GAMES WITH RANDOMNESS

47

one acquires a new game board, or acquires a new set of pieces, or plays physically or digitally, the set of possible relations – whether interior or exterior to the pieces whose interactions constitute that set – that might be realized is the same. The ontological source of those relations is different, but the full set of possible relations is unchanged. However, in a game with randomness and/or other forms of unpredictability, the set of possible relations changes based on the elements actualized in that particular playthrough. As I wish to argue here, the distinction between interiority and exteriority in the relations between game elements is a distinction which can also be fruitfully applied to the full set of elements and their relations (and the virtual content those relations represent). A valuable way to visualize this argument about interiority and exteriority, and one which will shortly allow us to understand the differences between deterministic and unpredictable games in this context, is through considering actions, outcomes and significance. In a deterministic game, we can know that one action will lead to one outcome rather than an unpredictable outcome, and that the ‘significance’ of that action the player perceives, which is to say the virtual content it refers to, will be of a certain sort. This action might be setting up the board and starting the game – which will be set up in only one way, without any randomness – or taking an action during the game – which will play out in only one way in one context, without any chance. As such, for a deterministic game, for each context, the relationships depicted in Figure 2.1 are the case. Figure 2.1 therefore shows the traditional relationship between actions, outcomes and their significance or importance to the player in fully deterministic games. There is a distribution of outcomes and significances into categories that will not change across multiple playthroughs. Action A (a particular move within a particular game state) leads via x1 to Outcome A, and there will only be one outcome that x1 can possibly lead to; in turn, that leads via x2, which there will only be one version of, to Significance A. Significance A is the virtual content that is actualized when Action A takes place (and therefore Outcome A takes place). This means that when one action is performed, prior knowledge of a game means that a player would be able to predict the significance of that move before the move is completed, or before the outcome is played out; this is both because the same significance is tethered to the same action, and because that meaning itself is unchanged. Figure 2.1 thus, as the caption notes, portrays a ‘sedentary’ distribution. Deleuze distinguishes between what he calls ‘sedentary’ and ‘nomadic’ distributions: a sedentary distribution is one where judgement ‘operates by allocating things to different pre-established categories’, while a nomadic distribution defines placement according to where things end up; it is ‘resistant to external or generic hierarchies’.9 The first is therefore a question of ‘where does this belong?’, while the second is an observation of where elements seem to accumulate, which might then be labelled. Figure

48

THE UNPREDICTABILITY OF GAMEPLAY

FIGURE 2.1 Actions, outcomes and significant parts of games without unpredictability.

2.1 therefore portrays a sedentary distribution, with outcomes and meanings being slotted into categories which can be known before play, and which do not shift. By contrast, in an unpredictable game, the relationship between these three tiers of actions, outcomes and meanings becomes dislodged. This can be seen in Figure 2.2. In this model we can therefore see that not just is Action A severed from a predetermined Outcome A, and Outcome A may be one of any number of things, but the multiplicity of hypothetical outcomes means that the significance of each outcome is also severed, and significances come to circulate just as freely as outcomes within the possibility space of the algorithm. A given outcome for a given action in one game may be deeply consequential, and in another entirely irrelevant – which is to say, a particular pairing of Action A and Outcome A will still differ in significance between two instances of gameplay because they draw their importance from the other Action– Outcome syntheses, which themselves also differ. Action A and Outcome A may form a temporary island of stability across multiple instances of play, but when the remainder of the ocean continues to shift, this stability is protean and deceptive, for any significance of this ‘stability’ remains contingent on the remainder of the ocean – the externality – rather than on any given A–O pairing – the internality – even if that pairing happens, by non-deterministic fluke, to persist across multiple repetitions. One kind of significance no longer comes with one outcome, but rather all significances are dependent on all other significances. Significance is of course tethered to context in a deterministic

GAMES WITH RANDOMNESS

49

FIGURE 2.2  Actions, outcomes and significant parts of games with unpredictability.

game as well – the significance of moving one piece is determined by the coordinates of all other pieces – but when context is deterministic, as noted above, we can always predict it and perfectly reproduce it, and thus Action– Outcome–Significance will always be identical after a given record of all ingame Actions performed, whereas in an unpredictable game, reproducing the same inputs neither reproduces the same outputs, nor the same importance of those outputs. In such a game the context cannot be predicted, as the context is not deterministic, but rather co-constructed from the interplay between outcomes, their significances, and the location of those outcomes and significances in the context of all other outcomes and significances. Each action, outcome and significance in the set of all possible relationships in each playthrough might not be possible in the next, even if the same sequence of moves was faithfully reproduced. As James Williams puts it, a nomadic distribution is ‘resistant to external or general hierarchies’ because that which is sorted cannot be comparable to one another, or to categories into which they should fit, for one does not have ‘access to the principle guiding the distribution’ of these elements.10 Consequently, ‘any fixed representation of a nomadic distribution’ is inevitably illusory: in unpredictable games these elements described above consequently adopt a nomadic distribution, evading categorizing and reforming themselves with each permutation. The distributions of outcomes and significances cannot be categorized into existing categories, but can only be grouped once the playing subject perceives the emergence of significance in a particular given playthrough. As such, we can now move from addressing the interiority and exteriority of game elements and their relations towards considering the interiority and

50

THE UNPREDICTABILITY OF GAMEPLAY

exteriority of the sets of game elements and their relations. In a deterministic game, as we have seen, the complete set of relations emerges from interiority, being inherent to the game and fixed. There is one possible set of relations, a set of relations which is an integral part of the game, a set of relations which contains all the possible chains of activity that might be actualized within a play of such a game. However, in an unpredictable game there might be any number of sets of relations, millions or billions or an almost uncountable number, each of which would contain a full volume of possible chains of activity that might be actualized in the play of that game. Whereas the set of relations in a deterministic game emerge from interiority, these sets of relations are a matter of exteriority, for the particular set of relations present and the particular virtual content present are contingent on the distribution of game elements, actions and the outcomes that take place as a result of each action. Through these distributions of actions, outcomes and significances, each instance of play of an unpredictable game becomes distinct not just because of what takes place – which elements of unpredictability are actualized – but also due to the set of potential significances they adopt. In this section I have sought to demonstrate how what I term ‘significance’ emerges through the distribution of elements in unpredictable games, using Deleuze and Guattari’s notions of interiority and exteriority. In terms of pieces, relations are derived from the interior nature of the piece, or from the exterior constellation of elements they find themselves in. Extending this distinction to the set of possible relations, in deterministic games these come from the interiority of action–outcome–significance relations that can arise in a game, where the set of action–outcome–significance relations are fixed because there is no randomness or chance present in the game. By contrast, in unpredictable games the sets of possible relations vary, because each action–outcome–significance relation is unpredictable and reciprocally determined through its interactions with other, equally unpredictable action–outcome–significance chains. In this case I have explored significance through the example of randomness, although it just as readily applies to the concept of chance – ongoing durational unpredictability, instead of initial pre-play unpredictability – which we consider in the following chapter. This notion of ‘significance’, as we shall see throughout the work, is central to player experiences of unpredictability, whether something as foundational to unpredictability as a world map in a computer strategy game, or the more complex social practices we explore later such as completionism and the collective cultural response surrounding games of ‘permadeath’, where the meaning of a single playthrough comes not just from the constellation of elements and their relations and singularities, but from the hardwired uniqueness of that orientation, and the impossibility of visiting it again. By understanding both how significance is created in unpredictable games and how and why it differs in both potential and actual experience, we begin to get some idea of the ever-changing experiences of playing such games.

GAMES WITH RANDOMNESS

51

Granularity of interest What remains unclear, however, is how these significant and important portions of randomness systems, which is to say the elements that have an active rather than negative effect on gameplay, are experienced by players, and how this might vary across players, and why. Let us call this the granularity of interest – how many of the potential components are actually interesting to players (or viewers)? How many arrangements of those components from which gameplay proceeds are interesting? If there is a high granularity of interest, then we might suggest that the majority of randomized play conditions will be interesting and compelling, have a high number of active components and a player will always find interesting things that can emerge from the possible or real gameplay in any permutation; with a low granularity only a small number of permutations (or a small part of each permutation) are of interest. In this second point, we see that we can further subdivide granularity of interest: are we dealing with a randomization system in which some conditions are interesting, and some are not – which is to say that the interest (or not) on one part of the outcome is readily generalizable to the whole, or that the whole is sufficiently ‘small’ that it cannot readily be broken down into constituent parts – or are we dealing with a system in which all starting conditions are sufficiently varied and expansive that all permutations will contain some interesting parts and some non-interesting parts, and the question is therefore of the relative weightings and percentages of these parts in each permutation? Understanding the granularity of interest is therefore closely related to our previous two enquiries in this chapter, but is distinct, for it focuses more fully on permutations across playthroughs, not just within a playthrough. Equally, it can be considered for a spectator as well as for a player, as the consideration of gameplay interest is not inherently predicated on acting within that game, simply on observing gameplay interactions taking place and permutations being created. To begin examining the granularity of interest we can take an illustrative example from the game of poker. When a player has moved all of their ‘chips’ (tokens denoting currency or other value with which the game is played) into the pot and is awaiting the outcome of the hand, they are ‘allin’. When one is all-in, that player is unable to make any further actions until the conclusion of that hand of poker is reached. If all other players fold to the all-in bet, the all-in player wins the pot as they normally would if all other players folded; otherwise, they can no longer affect the play of the hand and must await the dealing of any remaining cards and the arbitration of winning and losing their emergence heralds. If they are victorious, they will remain at the table with (normally) at least twice as many chips as they started the hand with. If they are not victorious after all the cards have

52

THE UNPREDICTABILITY OF GAMEPLAY

been dealt, they will be eliminated from play if in a tournament, or will have to leave the table or ‘reload’ (pay more money to gain more chips) in order to continue. This dramatic tension – of a player seeing a striking improvement in their fortune, or seeing their fortunes dashed – gives the ‘allin’ moment a deep excitement for many, both as players and as spectators. Indeed, for several years this was most obvious during the television broadcasts of the annual ‘World Series of Poker’ tournament series, where all-in moments were marked out through the sponsorship of the deodorant brand Degree, resulting in the reification of these moments into the ‘Degree All-In Moment’, and marked out as being those elements most worthy of attention, consideration and broadcast. Jay P. Childers’s analysis of this moment presents it as a ‘celebration of risk’ – which is to say, taking actions within a context of unpredictability – which, even when acknowledged as certainly one of the more ‘appealing dramatic aspects’ of the game of poker, is highlighted to an unusual degree in these broadcasts.11 The granularity of interest in the overall game of poker these broadcasts reflected and reproduced was, therefore, understood to be quite low, at least for the players these broadcasts were designed to appeal to. There are few exciting moments, it suggests, and therefore we must focus on those. Interestingly, in later broadcasts the emphasis on all-in moments actually reduced, although they remained dominant, but the sponsored moment shifted towards one based on the exercise of poker skill through deception or skilled reading of one’s opponent. This, in turn, became known as the ‘Planters Good Instinct Moment’, sponsored by the eponymous peanut company (for reasons which remain obscure). Certain moments were therefore still reified and marked out as being special, but the moments marked out as interesting then had a greater degree of skill and strategic decision-making, rather than relying on the outcomes of the cards; nevertheless, all-in moments still dominated a disproportionate volume of the broadcasts. Therefore, for many players and spectators of poker – or these players as television executives perceive(d) them – the granularity of interest of poker is quite low. Many hands are actualized, but few, seemingly, as especially compelling. The average ebb and flow of chips around the table are deemed to make far less captivating television than the absolute, all-ornothing, everything on the line, moment of an all-in. It is easy to find poker broadcasts – this is especially true of the World Series of Poker’s television presence – where almost every hand broadcast (perhaps two dozen in an episode) is an all-in, representing an edited version of a sequence of play that might have included many hundreds of hands, only a few of which will have included all-in situations. In watching these broadcasts time can seem to jump, as the relative chip positions of the players shift and change off-screen but without narrative explanation, being deemed insufficiently interesting for mass consumption. This ‘skipping’ broadcast, although it might seem strange to one kind of poker player watching for the highly skilled plays,

GAMES WITH RANDOMNESS

53

is interesting because it represents a broadcast devoid of uninteresting moments; this is, perhaps, what it would be like to play poker and only experience really interesting hands, rather than the endless hours at the table which can get rapidly dull. Nevertheless, we must acknowledge that this misses a tremendous volume of the game, when we define ‘interesting’ according to this metric of drama and the maximum volume of chips taking place; the moments missed are both integral to the flow of chips around the table, as evidenced by the strange nature of the flow rhythm of these broadcasts, and the story cannot be fully understood with only these more interesting elements. Any model of the granularity of interest privileges and emphasizes some elements of the gameplay over others, either within one instance or play or across multiple instances of play, and shapes the perception of a game’s overall flow. A second valuable example of granularity of interest, and how deeply it can affect direct gameplay experience rather than the spectating of such, can be found within slot machines. The operation of such machines and their ‘gameplay’ is extremely simple, and can be described extremely briefly. One inserts a coin, or a token or some other form of currency, and then pulls a lever, pushes a button or performs some other action which ‘starts’ the game. The wheels on the slot machine spin, they reach a conclusion (either after time or when the player presses a button), and the player is either paid or not paid. The only interaction the player has with the machine is the insertion of a coin or equivalent – which does not actually have a gameplay effect, beyond allowing the game to take place, which is therefore the equivalent of downloading a video game or unwrapping a pack of cards. As such, in a slot machine – or the flipping of a coin, or the drawing of lots – randomness and luck are brought together, squeezing out any possible role for chance. Nothing takes place between the initial generation of play conditions and the outcome of play; one simply looks at those initial conditions, and the winner or loser is thus determined based on the configuration of such elements. Relevant to our discussion here is the fact that slot machines have garnered particular controversy as a result of the presence of the ‘near miss’ and the ‘false win’. A near miss is a spin where four out of five of the ‘jackpot’ wheels line up, for example, but all five would actually be required in order to receive the jackpot. A false win, meanwhile, entails the machine paying out, but paying out less than the player put into the machine. These practices shed light on how the granularity of interest in unpredictable gameplay can be manipulated or deployed, and how the subjective significance of game events can be separated out from the ingame impact those same unpredictable events have. In the case of the near miss, Natasha Schü ll in her work Addiction by Design (2012) argues that these are designed with ‘reconfiguring loss’12 in mind. They transform a spin that loses the player money into one which almost won the player money, and change loss from an absolute failure into something which was both

54

THE UNPREDICTABILITY OF GAMEPLAY

almost a victory, and which hints towards an inaccurate perception of a high chance of future victory. In the case of the false win, the ‘continuous flow of small gains’13 cause the surrounding losses to be overlooked, setting up a tight reinforcing loop of apparent victories which both create the feeling of having ‘won’, inculcating what we might call a post-win feeling in the player; this is what it feels like winning a small amount, so imagine what it would feel like to win a lot. However, these practices shed light not just on the problematics of slot machine game design from critical, ethical and ideological standpoints – on which Schü ll’s analysis and that of other scholars have generally been focused  – but also on the granularity of interest as a critical tool for understanding game unpredictability. In near misses, there is no gamemechanical difference in outcome between these four jackpot wheels which yield no jackpot, and a totally random selection of icons which also yields no jackpot. Both give no money back to the player, and both are the conclusion of a sequence of events where the player commits their money (and their time) to a moment of unpredictability, and their money, at least, is not rewarded at its conclusion; they have ‘lost’. There is, however, a systemic difference, which is to say a different system of randomness underpinning it which generates slot machines’ particular – and deliberately skewed – granularity of interest. The deliberate generation of seeming nearmiss outcomes in slot machines is designed to inculcate a feeling of having almost won in the machine’s players, and thereby create a false (incorrectly high) sense of the chance of victory and consequently encourage further spinning. Technical systems designed to distribute randomness in certain ways – to more often than would be expected align elements in a near-win configuration – come to boost the granularity of interest of slot machines. Instead of one in one hundred being of particular interest – a near miss, the fleeting glimpse of a possible jackpot – this number becomes perhaps one in twenty. This boosts the granularity of interest in the game-playing subject beyond the granularity of interest in a purely mechanical sense (changes in the state of the game). False wins, in a similar way, are designed to create the feeling of having won, even though one has actually lost money on that particular spin. This is a game-mechanical difference, for the machine behaves differently during a false win (paying out) compared to a loss (not paying out) – which is to say, giving the player some money. In the case of the slot machine the casino ‘intervenes heavily in how players experience their losses and gains’14 – and consequently, inevitably, the Ideal synthesis through which events and reflections on those events come to constitute the player’s understanding of the multiplicity producing them, which is to say that the generative system of the slot machine has a greater chance of producing a ‘true win’ than it actually has. Whether or not there is a difference in the material player–machine relationship (coins are rewarded, coins are not rewarded, the player gains coins, the player does not), the presentation of

GAMES WITH RANDOMNESS

55

wins and losses affect both the experienced granularity of interest in the moment and the expectation of future interest. From these two examples, we can see three important points. First, granularity of interest is not a fixed and universal value to be assigned to a system of game randomness, but is rather individually subjective and dependent, alongside other factors,15 on the experience and preferences of the specific viewer or player. The higher a player’s personal sense of the granularity of interest of a given game, predicated upon knowledge of that game and interest in that game, the higher the portion of that game’s random permutations appear compelling; the lower those elements are, the fewer permutations yield subjectively interesting gameplay. This is one of many ways in which the broader category of ‘games of chance’, used in many prior scholarly engagements with these topics, needs to be unpicked and more tightly focused on how and where unpredictability is deployed. The granularity of interest, then, is an element of the multiplicity that each player imagines represents the generative system. The granularity of interest will change the virtual content a player holds in their mind; it might reduce the chance of unpredictable events considered interesting (as, I would argue, a deeper appreciation of slot machinery would do for most players), but might also increase the number of unpredictable events considered to be interesting, if the player develops a mental model of why previously uninteresting moments are now, actually, of consequence to the play of the game. Secondly, granularity of interest emerges from the interplay between systems and perceptions (which is, across all forms of unpredictability, a central theme of this book); it is dependent on how systems display randomness, and how players consequently construct mental schemata for understanding that randomness. The construction of a subjective multiplicity of a game’s generative system does not move the player in an unproblematic manner towards a ‘more accurate’ notion of the granularity of interest; this path can be upset and disrupted through the presentation of the unpredictable moments that constitute the game. Mechanically uninteresting moments can be reconstituted into subjectively interesting moments if and when shown to the player in a particular manner. What ‘should’ be perceived as generality becomes perceived as significant repetition, by the generative system actualizing outcomes that speak to virtual content in the slot machine which is, in a strictly formal generative sense, not present. Players learn one set of virtual content – that spins suggesting a near-miss, or spins where one wins a small amount – come up unusually often (which is ‘true’), but this virtual content implies the presence of other virtual content, which is that genuine wins will be actualized more often than is the case. In this way we note that players will not necessarily perceive the virtual content of a generative system, but rather perceive singularities and draw conclusions about the originating multiplicity that are not neutral and unbiased, but shaped by the

56

THE UNPREDICTABILITY OF GAMEPLAY

presentation of supposed virtual content, which is to say the rules for the production of content.16 Thirdly, the interest in the granularity of interest is itself contingent upon its granularity: as Deleuze would put it, these are subject to reciprocal determination. By this, I mean that periods of lower interest appear necessary for periods of high interest to be interesting; a game where each permutation is of roughly equal interest is unlikely to sustain interest for long, because monotony comes to dominate, and the promise of difference is not fully realized. By contrast, a game that sometimes procedurally generates a flat, drab and monotonous world map, but sometimes creates an intricate and fascinating topographical mosaic, will generate more interest in a player when its most interesting permutations appear. This is a particularly crucial concept for our later understanding of ‘completionist’ players: those who wish to experience or accomplish everything interesting a game has to offer. Nevertheless, more broadly, the kind of ‘interest’ which is split into a granular form depends on the potential extremes of a randomness system (the noteworthiness of its aleatory points), and how often and rarely those aleatory points are sighted. There is therefore a two-way relationship between granularity and interest: the granularity of interesting permutations influences how interesting those permutations are, and the interest of those permutations affects the perceived granularity of interest, which is to say how many of a games permutations are actually interesting. It is therefore difficult to say whether a higher or lower granularity is ‘better’ – more interesting permutations might be each less interesting (but more common), while fewer interesting permutations might each seem more interesting (but rarer). Randomness is fundamentally implicated in these player perceptions, being the arbiter and the distribution mechanism by which initial permutations are actualized; we therefore see again how the placement and distribution of unpredictable elements will have an immense effect on player experience. To summarize, I propose that the granularity of interest is an important element for our understanding of randomness, and how randomness shapes our experience of play. Naturally the granularity of interest can also be applied to chance and indeed to luck, but in randomness the granularity of interest sets up the foundations for a complete instance of play, and  – as we have shown – elements of randomness are likely to reverberate throughout a player’s experiences. By contrast, as we see in the next chapter, moments of chance tend to be greater in number and smaller in effect, whereas randomness is normally highly consequential to the player, regularly mechanically or psychologically returned to as play continues, and able to create broad and complex structures of significance, rather than staging occasional interventions through the course of play. Granularity of interest in games of randomness therefore offers us the most comprehensive and most play-impactful way to address and develop the concept; it is more important, and more routinely and starkly perceived. Perhaps most

GAMES WITH RANDOMNESS

57

importantly, and it is here we move into the next section’s discussion, it also points our attention to the role of game knowledge in shaping our perception of game unpredictability. Players of different skill levels in poker perceive a different granularity of interest, and locate that interest in different game moments. Having examined significance (the singularities players perceive) and the granularity of interest (which singularities are interesting, which are not and why), it is the acquisition of knowledge about unpredictable gameplay systems, the learning about these systems by players, to which we now turn in greater detail.

System and variable knowledge In this chapter we have identified three key conceptual understandings of randomness thus far. First, we saw that randomly selected elements which do and do not affect gameplay (when looked at against other possible elements which could have occupied those same positions and roles) are conceptually different, and their effects on gameplay varied and diverse. However, our analysis of lasting impact has thus far focused on lasting impact in one instance of play and how initial conditions of various sorts can shape one playthrough of a game with randomness, although this has left unexplored the question of what impact random permutations have on multiple playthroughs. Secondly, we have seen that the significance of elements and experiences in games that contain randomness is a complex question, constructed through the emergent relationships between actions, outcomes and the significance assigned to those actions, their outcomes and their action–outcome sequence. However, how players note these differences across multiple instances of play, and the variation of significance created across those instances, is less clear. Thirdly, the interest players have in games of randomness is contingent on player knowledge of the game being played or observed, the interplay between systems and the perceptions of those systems, and that the level of interest in random initial permutations and the granularity of that same interest appear to be closely interwoven. But how is that knowledge obtained by players, whether or not mathematically ‘correct’? Which is to say, how does experiencing a generative system and its variables lead to a player’s multiplicity? Manuel DeLanda notes that multiplicities are ‘by design, obscure and distinct’, for the singularities that define one emerge gradually and unpredictably in a manner that will ‘progressively specify the nature of a multiplicity as they unfold during recurrent sequences’:17 each singularity that a player experiences contributes to their notion of the multiplicity from which they emerged, but does so slowly, gradually, unevenly and itself in an unpredictable manner. We will now therefore bring the above three findings together to address these questions. In this section we will consider two interrelated final

58

THE UNPREDICTABILITY OF GAMEPLAY

questions that are crucial to our understanding of gameplay randomness: how does knowledge of the randomized components, and knowledge of the systems by which those components are distributed, affect player experiences over multiple playthroughs of an unpredictable game? The first half asks how players understand what components remain and can therefore be anticipated and can come to factor into player decision-making in games of randomness, while the second question moves into a greater understanding of the randomness systems underlying gameplay, knowledge of which allows players to estimate which components are the more or less likely, and when. First, how does player knowledge of variables (differential elements) build up? Which is to say, differential elements: by what process does a player discover what these are? As Deleuze show us, ‘learning’ takes place through the synthesis of memory with the present through recognition. In this case that recognition might take several forms. First, recognition of an element already seen which is in the same place, in which case it might not be an unpredictable element; secondly, recognition of an element already seen in a new context, implying either unpredictability or duplication as a variable; thirdly, recognition of an earlier context with a new element where once another element stood; fourthly, a new element within a new context. Across these four possibilities players are able to build up both knowledge of variables in their individual natures, the range of possible variables and also what is and is not a variable, for this is not always obvious.18 However, we must note that such knowledge must inevitably remain contingent, and incomplete, without a specific examination of the code (or equivalent) underlying the game: de-compiling the executable program, looking through the deck and so forth. It remains contingent because variable knowledge is only built on the variables that have appeared so far. As more variables appear, the percentage of possible elements that have been actualized rises, although one can never know what that percentage is, only its status (higher or lower) relate to previous assessments. Equally, one cannot know whether entire categories of variables have never appeared, as well as not knowing how many variables remain unseen. In turn, it remains incomplete because there are always permutations one has not seen which might contain other variables; equally, not just does the list of variables remain incomplete, but knowledge of that list, which is to say the point at which it is complete, cannot be acquired. As generality (the same element appearing over and over) grows, so too does one’s confidence in knowledge of the underlying system or phenomenon, in this case both the nature of variables, and how many possible variables are present that might be slotted into a particular ‘place’ in a particular game. In addition, there are interesting subtleties of this practice: consider a game one plays one thousand times, and nine-hundred and ninety-eight times, Variable X appears at a certain point. In the other two plays, Variable

GAMES WITH RANDOMNESS

59

Y and Variable Z appear. This leads to a particular texture of unpredictability the player is shown when it comes to the distribution, and the number, of variables. A fair assumption would be that X is almost always actualized, but roughly one in five-hundred times, some other variable might be; Y and Z might be the only two, but there might also be a tremendous library of variables for this one-in-five-hundred variation. One has therefore learnt, specifically, that one cannot make an informed judgement about the set of differential elements due to the unusual statistical distribution in this example, but given how rarely those unusual elements appear, and more than one has appeared, one can make an informed judgement that there might be other extremely rare elements in the virtual content of the game. However, until that point the player does not even know that this is a moment of unpredictably in the first place. In a card game one has a model of the generative system in one’s head which is quite complete (shuffling a deck), even if a precise understanding of the statistics might be elusive. By contrast, in games with obfuscated generative systems one might not just be surprised by specific moments of randomness or chance within an understood set, but also by the set of things thought to be vulnerable to unpredictability to begin with. On the opposite end of the spectrum, it is easy to imagine a scenario in which judgements over the size of the set are obvious, but the nature of the set is not: many games contain ‘Guidebooks’ or ‘Bestiaries’ or ‘Encyclopaedias’ or equivalent which list, according to some ordering principle, what the player has thus far encountered in the play of a game. If the categories in these structures are broad enough, players can readily know that elements have not been encountered – there are empty pages in the digital guidebook (whether literally or metaphorically) – without being able to hypothesize about the nature of that variable. It is also readily possible for one to make a judgement about both content and size of the set: if one were in a context where one saw twenty variables over a long sequence of play, each represented by a different letter of the alphabet with no apparent link between the letter and the generative system placing the variables, one could reasonably make judgements about both the size of the set and the remaining content of the set. However, to return to the above insight about the contingency and incompleteness of all variable knowledge of this sort, one will always have an uncertainty which only reduces, but never disappears, as more experiences are synthesized by the game-playing subject. Subsequently, one can make judgements about that which is unseen, but not always: sometimes unseen variables can be anticipated, sometimes unseen size of the set of differential elements, and sometimes both and sometimes neither. One can therefore learn much about the databases of variables in unpredictable games, but acting upon that knowledge must be tempered with appreciation of the process by which that knowledge was gained, and knowledge of some variables can in fact speak to others unseen. In this way unpredictability can be synthesized and learnt about by the

60

THE UNPREDICTABILITY OF GAMEPLAY

player, but the possibility of what Nassim Nicholas Taleb calls a black swan – something unforeseen and outside of prior models – remains omnipresent. Having considered components, how does player knowledge of systems (which repeat) build up? Four elements of gaining system knowledge stand out, which is to say four things that players can look for across playthroughs, or within one playthrough, to boost their understanding of the underlying generative systems, and thereby fill in the gaps in their multiplicity of that system. First, the distribution of variables, which is to say how the system chooses to distribute variables; which ones appear where and why? Secondly, the extent of the system, which is to say what does change and what does not change, so how far do the game’s unpredictable systems go and how much of the game is determined by those systems? Thirdly, by observing when numerous elements are changed, or only one of those elements is changed, the player can begin to perceive the relationships between systems, and some of the contingencies between systems whereby some variables are altered, and some are not, depending on other outcomes in that system or outcomes in other systems. Fourthly, systemic biases and preferences: do certain variables appear more or less often, do certain systems tend to produce broadly similar kinds of outcome when viewed at a particular scale and so forth. From all of these elements the player develops a notion of the likely future components, how they will be spread and the contingencies on which they rest. The more complex the systems, the harder these elements are to divine (this is a point which will be returned to later). Three other factors in this discussion are now important: the renewable or non-renewable nature of differential elements in their distribution by generative systems, the kinds of patterns that will develop in a player’s Idea of a game’s generative system, and also the level of confidence (touched upon above) players can have in their assessments of both variables and systems. In the case of renewal, we must note that some randomness systems distribute variables by taking them out of that list when selected for actualization, while other randomness systems distribute variables by leaving them in the list when selected, thereby allowing for the same component or variable to appear multiple times, and possibly in multiple forms, in the same permutation. A simple comparison here is, once more, between a deck of cards and dungeon generation in a video game; in the first case, once a card is shuffled into a certain place that same card cannot also appear elsewhere in the deck, assuming the deck is not faulty. By contrast, a dungeon generation system might be specifically designed to place multiple versions of a simple default room, for example, and only more rarely add more unusual kinds of rooms to the eventual level. A randomness system can therefore be finite or replenishing, depending on whether alreadyused elements can be used again, or not. Acquiring an understanding of this dimension is also important for players seeking to understand a game’s systems, and therefore its distribution of variables.

GAMES WITH RANDOMNESS

61

This has significant consequences for our understanding of a player’s experience and their attempts to understand and synthesize the data they have been given into a comprehensive whole. If one perceives what appears to be a system that distributes non-renewable variables, one can only make an estimate about whether those variables are indeed of that class. The larger the number of seemingly non-renewable variables one sees, the more likely it becomes that a potential renewal has simply not been seen yet, because the overall possibility space is larger; if only a few variables are actualized and seem not to repeat, the chance that they never repeat becomes higher. Equally, as soon as a renewable variable is sighted, previous synthesized observations lose some of their generality: the texture of variables is not what it first seemed, with the as-of-yet unknown potential for duplicates hovering beneath the surface. As such, a player can know that renewable variables are possible, but can never know that renewable variables are impossible, simply from play and observation; these both have consequences for the construction of a game’s multiplicity that the playing subject develops, and not just their synthesis of past experiences, but their anticipation of the texture of an unpredictable future. The unfolding of gameplay is thus affected by the constitutive initial components, the system which orders and distributes those initial components, player knowledge of the components and player knowledge of the system. The latter two take the form of patterns, for that which the player perceives and commits to memory are turned into generality, which is consequently used as the basis for a pattern of regularity and predictability on the part of future gameplay, and therefore used for prediction. These kinds of patterns, as the above discussion is intended to show, are complicated and contingent on a number of factors, and overlap and intersect in several ways. Given the complexity of these elements and the relationship between how they play out in the formal generative system of the game and the internal multiplicity of the player, some terminology is valuable at this point. Let us say that ‘systemic’ patterns are patterns in the distribution of variables, while ‘commutable’ patterns are patterns in the lexicon of variables. Systemic patterns represent the player’s appraisal of how, where and when variables are distributed, and the development of an idea of how those variables might be anticipated in the future. Commutable patterns are commonalities identified in the variables themselves; perhaps variables that appear in one context are always of a certain form, or variables in another area always appear in a certain sequence or other ‘discoveries’ of this sort. In both cases knowledge builds up through perception of generated phenomena, not through direct access to the underlying elements or relations themselves: ‘Habit does not operate in terms of that which generates impressions, but rather in terms of signs.’19 There are also other kinds of patterns which players perceive as well as these. Variables will ‘gain’ patterns outside themselves if they are distributed in a distinctive

62

THE UNPREDICTABILITY OF GAMEPLAY

way by their systems (the player will perceive some of the pattern being in the variables, rather than in the system), and systems will ‘gain’ patterns outside themselves if the components distributed happen to form a pattern in their content which is separate from the regularity of their distribution. In the first case variables which do not have their own patterns (such as a sequence) when viewed dispassionately appear to take on the properties that would suggest a pattern when they are distributed in a certain manner. In the second case, variables of a certain sort distributed in a certain way – even if a far greater range might have been – will seem to lead to new patterns in the system, even if those patterns are the result of unpredictable distributions, rather than a specific intention of the generative system. In summary, we might propose the existence of four kinds of patterns which players can detect: interior system patterns (the distribution of elements), exterior system patterns (connections between the elements they distribute), interior variable patterns (the relationships between variables in a given dictionary) and exterior variable patterns (the distinctive way a set of variables is distributed). Unfortunately, the question of confidence upsets the value of these patterns. How sure can one be in one’s predictions, and what are the patterns that one is predicting? When a player knows the set of variables, they can attempt to predict the next variable without a perfect knowledge of the system; when they understand the system they can predict the next appearance of a variable, without knowing anything about those remaining variables. Although these may seem similar, it is important to distinguish between variable hypotheses and relational hypotheses: one relates to the ability to guess a variable but not its specific instantiation or consequence, while the other relates to the ability to estimate the importance or location of the next variable without knowing what precise form that variable will take. Variable hypotheses allow players to synthesize their past experiences of variables with the present gameplay scenario in order to guess, or potentially even know, what variables remain in a system or what variables will happen the next time an event takes place, even without necessarily knowing what that event might be of when it might occur. Relational hypotheses allow players to draw conclusions from past experience – each level of this game has a variable from a certain library in the middle of it, and this has happened every time so far – to predict when another variable might occur, but not necessarily the nature of that variable. A player can therefore come to conclude four different values: their estimation of the remaining future components (based on their understandings of the variables), their estimation of the sort of component likely to be selected next (understanding of the system), their estimation of how accurate their understanding of the components is and their estimation of how accurate their understanding of the systems is. The first two in this case relate directly to what will occur in the game, while the second two constitute a meta-level reflection on the player’s part back onto

GAMES WITH RANDOMNESS

63

their own understanding of a game’s variables and systems, and how well those can be assessed and anticipated. In conclusion: all processes of learning and synthesizing unpredictable experiences into a model (multiplicity) of randomness systems (or other unpredictable systems) are contingent on coming to understand the generative system lying behind an unpredictable game, which repeats with each new playthrough and distributes differential elements according to (normally) profoundly opaque rationales. Deleuze argues that once perceived, phenomena become a reflective past which is subsequently deployed to estimate the future.20 A player’s information on a system forms the multiplicity of that system in the mind, and that consequently affects several things. It affects gameplay, by shaping a player’s decisions when encountering something known (something reproduced and compared to that previously apprehended) and when encountering something unknown, by attempting to fit it into their growing multiplicity for the game in a logical place. However, basing judgements on induction is risky, for a counterexample is always waiting to happen21 for which previous experiences cannot have prepared one. The player’s information also shapes the anticipation of what comes next, which might be a known element – the greater the perception of generality, the greater the confidence in the predictability of the possible set of upcoming unpredictability – or might be an unknown element, but is likely to be within constraints, as discussed above, that a player now has experience of. In this chapter I have sought to unpick what I term ‘randomness’, which is to say unpredictability that establishes the initial conditions of a game. I first considered the initial conditions of play in unpredictable games, the tremendous range of forms these can take and how the process by which those conditions are created shapes the conditions themselves. The chapter then considered the lasting impact of initial conditions, looking to explore how initial conditions can shape not just the start of gameplay, but the entirety of gameplay by creating and shaping the ‘field’ of play, and when and why lasting conditions sometimes have strong or weaker effects, or shorter-lasting or longer-lasting effects. Next, the chapter unpicked the creation of significance through randomness, and why different elements of initial conditions can have a profound or a minimal effect when situated in the context of other random elements; this led onto a discussion of the granularity of interest, the volume of potential experiences in an unpredictable game which are of interest to the player and an exploration of the complex two-way relationship between granularity and interest as independent but interwoven elements. The chapter lastly considered how the game-playing subject constructs concepts of variables and system knowledge, and a number of other kinds of patterns which can emerge during this process, helping us to understand the experience of playing unpredictable games and how players might construct many layers of many different kinds of mental

64

THE UNPREDICTABILITY OF GAMEPLAY

models from available gameplay data. In the following chapter we will next come to consider the question of chance, which is to say unpredictability during the course of gameplay, rather than unpredictability which takes place at the start of gameplay or before gameplay. This will further develop the relationship I argue for between unpredictability location and the experiences created by that unpredictability, and in this regard, much of chance is highly distinct from the randomness explored here.

Notes 1 Alessandro Romano, ‘God’s dice: The law in a probabilistic world’. University of Dayton Law Review 41 (2016), 66. 2 This is the case unless those elements are drawn from a finite and nonreplenishing library, in which case previously actualized elements affect future to-be-actualized elements: which is to say, the elements which have already appeared will not appear again. 3 This is even reflected in the presence, in Civilization V (2010) and Civilization VI (2016), of an in-game unit known as the ‘archaeologist’. In both games players direct their archaeologists to ‘antiquity sites’ and excavate them, which subsequently yields cultural value, artefacts and other potential rewards. The locations of these sites sometimes depend on earlier events within that playthrough, such as major battles or the razing of cities, and thus reference the earlier parts of the game when one reaches a later ‘era’ that makes the archaeologist an available unit. Although it is unlikely that players will strategize based on this element – one is generally surprised at where antiquity sites are located, their original importance forgotten – it nevertheless causes the initial randomness of the game, and how the earliest parts of the game played out, to visibly resonate once more towards the end of a playthrough. 4 Deleuze and Guattari, A Thousand Plateaus, 38. 5 Ibid., 389. 6 Ibid. 7 Gilles Deleuze, The Fold (London and New York: Continuum, 2006), 77. 8 Deleuze and Guattari, A Thousand Plateaus, 389. 9 Williams, Critical Introduction, 71. 10 Ibid., 71. 11 J. P. Childers, ‘Going all in on the global market: The rhetorical performance of neoliberal capitalism on ESPN’s World Series of Poker’. In Sporting Rhetoric: Performance, Games, and Politics, edited by Barry Brummett (New York: Peter Lang, 2009), 150. 12 Natasha Dow Schü ll, Addiction by Design: Machine Gambling in Las Vegas (Princeton, NJ: Princeton University Press, 2012), 92.

GAMES WITH RANDOMNESS

65

13 Michael Callon, ‘Devices of attachment’. In Markets and the Arts of Attachment, edited by Franck Cochoy, Joe Deville and Liz McFall (Oxford and New York: Routledge, 2017), 190. 14 Callon, ‘Devices of attachment’, 190. 15 As noted previously, these include cultures (gaming or otherwise) players belong to, what their objectives are when playing a game (such as the pursuit of profit or other concerns of political economy) and so forth. Even factoring in such elements, the granularity of interest will still vary from individual to individual, and remain deeply influenced by the nature of unpredictable games themselves; these broader cultural, economic or political elements and their influence on the perceptions of unpredictability are topics for future publications. 16 These singularities might, of course, also be experienced elsewhere, such as from watching the play of others, through conversations, viewing film and television and other media dealing with slot machines and so forth. Especially interesting here is the micro-genre of ‘slot machine reviews’ on video-sharing websites such as YouTube, in which reviewers – whose relationships to casinos remain opaque – record themselves celebrating victories (yet never or rarely defeats, and even recorded losing sessions tend to be minimal in their losses, and bracketed by wins). Nevertheless, players are still given a particular set of singularities, through all of these methods as well as their own play, with which to decipher the workings of a slot machine. 17 DeLanda, Intensive Science, 8. 18 For more discussion of this element, see Identifying Procedurality in Chapter VI. 19 Somers-Hall, Deleuze’s Difference and Repetition, 65. 20 Deleuze, Difference and Repetition, 94–5. 21 Michael Kaplan and Ellen Kaplan, Chances Are: Adventures in Probability (London, England: Penguin, 2007), 7.

3 Games of Chance

The ongoing system of gameplay We have now established randomness as referring to the unpredictability of initial starting conditions of a game. Although we saw that these initial conditions continue to reverberate throughout an instance of play, randomness does not account for new moments of unpredictability beyond the unpredictability that set up a particular instance of play. Consequently, unpredictability within ongoing conditions of play is distinct; whether or not randomness is present, unpredictability can take place during gameplay, and will shape player experience and player decision-making as play proceeds and continues. This form of unpredictability I shall term ‘chance’. Whereas randomness was chosen for its connotations of surprise and shock, the potential breadth implied by the term, and the often profound and sweeping unpredictability of starting conditions in games with randomness, I have chosen the term ‘chance’ due to its sense of briefness, fleetingness and – almost – irrelevance. Whereas ‘luck’ (which we explore in the next chapter) brings a sense of a trait someone possesses, an overall assessment of a life or something deeply out of one’s control, ‘chance’ is more passing, more transient, as in the phrase taking a chance. As unpredictable elements of ongoing gameplay often take place repeatedly and consistently in games which feature any such elements, I believe chance is an appropriate label for these events. ‘Chance’ acknowledges that chance tends to affect gameplay in any one moment less than randomness and less than any underlying luck within the game (although the two are naturally interwoven, and luck might manifest through chance); it also acknowledges that, because many games that include chance include a tremendous amount of chance which the player is confronted with regularly and almost constantly – rolling a die for every attack a character performs, for example – chance is often seen as a brief, throwaway thing, and one chance may be forgotten as soon as the player then moves onto the next. Nevertheless, as we will see in this

GAMES OF CHANCE

67

chapter chance is an integral part of unpredictability in games, is present in a range of games that often do not possess randomness or luck, and yields and enables very different kinds of gameplay. By examining chance, the purpose of this chapter is to continue to address the two major concerns of this work beyond the development of the typology itself: how unpredictability is used in games (its location) and how player experiences are consequently shaped by these different uses of unpredictability. Although randomness appears at first glance to be more dramatic than chance, concerned as it can be by the creation of entire worlds and the setting-down of gameplay elements which will continue to reverberate through one instance of gameplay, and in the longer term across multiple instances, chance is actually by far the more common manifestation of gameplay unpredictability. It therefore merits a separate examination, albeit one building upon a common theoretical and conceptual background, for the placement of chance into games takes on a very different texture to that of randomness. The structure of this is as follows. I begin by proposing the most visible, well-used and consequential examples of chance systems within games. This is in terms of the gameplay purpose to which chance is put, not a typology of the underlying mathematical systems; this is because relating all possible chance-creating systems in games would be beyond the scope of even the largest volume, and partly because my interest in this first section lies in understanding where and how chance is deployed in games as a design decision, not understanding the mathematics behind these choices, except where that mathematics significantly alters experiences. I break the use of chance in games down into three categories: the ‘Degrees of Outcome’ model, through which the outcome for a given action is known but the degree of that outcome varies unpredictably and quantitatively; the ‘Different Outcomes’ model of chance, in which one action may yield an unpredictable range of qualitatively different outcomes; and the ‘Outcome or Failure’ model of chance, in which a given action unpredictably yields a choice between an outcome which affects the game state (which might be of either of the previous forms) and no outcome whatsoever. In doing so I explore why each of these models is used, and the different gameplay experiences these different models of in-play unpredictability can create. It concludes by relating the personification of chance (and much more rarely randomness) in the popular notion of the ‘Random Number God’, a capricious deity who is seen to offer up bad (and more rarely good) fortune through the chance, and consequently luck, present in unpredictable games. Although intended by players to be a humorous tongue-in-cheek concept, the final part of this chapter will demonstrate the relationship between this concept and more traditional notions of fortune and fate, what it shows us about the agential assigning of game success and failure, the role of striking moments of unpredictability in determining this and how even the

68

THE UNPREDICTABILITY OF GAMEPLAY

desire to assign moments of unpredictability to something ‘beyond’ one’s actions or the actions of the system remain prevalent even in our supposedly rationalistic contemporary era. The chapter thereby considers the subjective player experience of encountering moments of chance in unpredictable games, the forms these take and how players (both individually and collectively) respond to these distinctive moments.

Codifying chance Chance – the ongoing unpredictability of gameplay after the game’s initial conditions are created (whether through randomness or handmade design) – is present in games in a wide range of contexts, and is deployed in pursuit of a wide range of gameplay purposes. Indeed, this range of contexts and range of gameplay purposes is so wide it is difficult to narrow these down to a distinct and manageable set of forms or purposes. However, I believe there are three ways of distributing inputs and outcomes within the rubric of chance which are theoretically distinct, in terms of the relationships between actions and possible outcomes, and the relationships and differences between those possible outcomes. It is through this initial foothold that we can then begin to examine chance as a whole, how these three forms of chance differ in the player experiences they create and the design goals to which they are put. It must be noted that there is blurring and overlap here, for the same gameplay goal can sometimes be met, to a greater or lesser extent, by more than one of these mechanics, but the distinct distribution of elements and the actualization of unpredictable outcomes nevertheless mark out three distinct forms of gameplay chance, to which we now turn. First, consider a scenario in which an outcome – a change – is guaranteed from a moment of chance, but the degree of that outcome will vary according to a process of unpredictability. For example, perhaps a player character is wielding a weapon whose damage output is unpredictable, and ranges from between two values, without dipping down to zero; the use of that weapon in any one instance of conflict is therefore guaranteed to deal some damage to one’s opponent, but the precise amount of damage might vary, and be unpredictable on any one particular attack. This might have a small impact – during the middle of a moment of conflict, for example – or a far larger impact; if the player’s character is themselves near death, and the player requires only a final blow on their enemy to defeat them and survive, a low or high damage output on one swing of this hypothetical weapon might be the difference between death and survival. The quality of the outcome is the same, which is to say the dealing of damage to an opponent within the game world, but the precise quantity of the outcome takes the form of chance, and ranges between two possible values. Let us call this the Degrees of Outcome model of chance; in this variation an outcome is guaranteed, but the degree

GAMES OF CHANCE

69

and severity of that change – both within its own terms, and in terms of its impact on the game world and a player’s experience – will vary. Then, consider a scenario in which an outcome is guaranteed from a moment of chance, as in the previous model, but there is more than one such outcome which the moment of chance might produce, and these outcomes are qualitatively rather than quantitatively different. For example, one might be guaranteed to find an item of some sort at a point in a game, but the precise sort is not set until a process of chance has been completed. The player will always leave that encounter or that gameplay moment in possession of one item more than they previously were, but the extra item they leave with will vary. A change is guaranteed, but the qualitative nature of that change will differ each time that moment of unpredictability is played out, and each possible change is divided from the other changes within the same virtuality by their quality, not their quantity. Let us call this the Different Outcomes model of chance. Each potential outcome will be qualitatively different from the other, and the difference is in kind rather than in degree; these rhizomes will give rise to arborescence lines that are unlike one another and travel in different directions, instead of lines which travel in the same direction but with a different strength, as in the previous model. Lastly, consider a scenario where an in-game action can have two possible (sets of) resolutions, which are unpredictably selected. In the first the player is awarded with a discrete and distinct outcome which will enact a change in the state of the game, or a change in the current condition of their player character, their faction or the like. In the second, nothing happens; the game world is unchanged, and the player or the player’s character are also unchanged, save for having expended time, energy and/ or in-game resources, on completing the in-game action which offered the potential of these two solutions. In this scenario of chance, during the play of a game an unpredictable decision is taken within the game’s system and there are two outcomes from that unpredictable moment – a change and no change – which are clearly different. In the first case a potential change is actualized, and in the second case it is not. Let us call this the Outcome or Failure model of chance; either an element of the game’s virtual potential is actualized or it is not. If it is not, then no sense of change in the material conditions of the player’s state is created. This does not mean the null outcome cannot create a change in the player’s experience or decision-making, thereby creating identities for the pre-null and postnull states where different gameplay situations existed, especially if this moment of chance cannot be repeated or ‘tried again’; but it does mean that the technical, material, formal game state is not altered by this moment of chance. Equally, there might be many possible change-creating outcomes, but the possibility of no outcome is the crucial element of this variant. This model might be close to the previous model if both had, for example, 100 possible outcomes – in almost all iterations, there would be no difference

70

THE UNPREDICTABILITY OF GAMEPLAY

between experiencing one or the other. However, the presence of a null outcome in this third and final nevertheless marks it out as different, for it maintains the possibility to actualize nothing, and to fail to change the game state in any way. The differences between the structuring of unpredictability in these three models of chance can be shown in the following diagram. In the Degrees of Outcome model, players are presented with one possible kind of outcome which varies in numerical degree (what Deleuze would call extensity); in the Different Outcomes model, players are presented with a selection of qualitatively distinct outcomes (what Deleuze would call intensity); in the Outcome or Failure model, there is a possibility that no outcome will take place, and therefore in absolute terms the situation of the game will be unchanged upon the conclusion of that moment of unpredictability. All of these models can of course blur and be combined – a version of the Different Outcomes model could readily contain several outcomes which differ by degree rather than by form – in which case elements of both models are present, and might be actualized. These differences are shown in Figure 3.1. With this typology of chance established, we can now consider the purpose(s) of each of these kinds of chance, and to what gameplay purposes they are put. This will both help develop our understanding of chance and lay the foundation for the later discussion of several gameplay practices that make up the case study of this work’s typology in the latter half of the book. Inevitably, such an accounting of gameplay design purposes and player experiences will always be incomplete, for no matter how extensive one’s game knowledge and game research, it is likely that some purpose to which one of these three models has been put will have escaped observation or consideration. Nevertheless, I believe most of the main purposes of chance are outlined below, with appropriate acknowledgements where one gameplay design purpose can be implemented across multiple forms of chance. This analysis is the foundation of subsequent considerations of

FIGURE 3.1  Three forms of chance, separated by the different genera or species of outcomes they can actualize, and the possibilities of those outcomes.

GAMES OF CHANCE

71

chance in the present work, and further develops one of the central claims of this work – that where unpredictability is implemented into a game, its location, will have significant effects on how players react to it and the experience of playing with, or against, that unpredictability.

Degrees of outcome We first consider the Degrees of Outcome model, where qualitatively there is only one possible outcome, but that outcome might be of a more or less ‘severe’ form. Perhaps the most common use of this system is in the dealing of damage from one in-game unit, character or creature to another. There are more examples of this than one can possibly recount in a paragraph, but these are all united by the idea that the same in-game weapon, or the same in-game attack, or other form of damage affecting one’s opponent, should vary in effectiveness across multiple ‘hits’, instead of always being at the precise same level of effectiveness. Rather than something always dealing five damage, for example, it might deal between four and six damage, with an equal possibility of all outcomes, resulting in the average being the same as the unchanging-damage model, but adding in a small amount of chancebased variation into each individual iteration of the attack. Another simple implementation of this model of chance is in the collection of numerous kinds of items or item-like pickups by the player, which vary in number. This commonly means picking up an amount of in-game currency where the precise amount varies, or perhaps an amount of ammunition for a weapon, a number of tokens or points or tickets, a collection of one-use items or anything else. Over time, picking up a range of these will average out towards a median point, assuming that all the piles of currency or ammunition are generated by the same system. However, varying them gives a different texture to the player’s experience of collection, a texture which might be very ‘coarse’ (if the parameters of currency generation are extremely wide) or very ‘fine’ (if the pickup is always almost exactly the same, but just varies by a fraction). What are the gameplay purposes of this first implementation of this chance? At first glance, the gameplay purposes of the Degrees of Outcome model are the hardest to divine: what is the value in offering what are normally quite minute variations in the same kind of outcome, especially if they average out in the mid-range over a long enough period of time? When these kinds of chance intersect with other outcomes that require or draw upon the degrees produced, it becomes apparent why such systems can gain an importance within a game that outweighs the initial appraisal of these variations. First, when implemented alongside certain other systems, even a small variation in these kinds of outcomes can seriously matter. Consider a game where one has a choice between two weapons, one of which deals

72

THE UNPREDICTABILITY OF GAMEPLAY

four to six damage and another deals a flat five damage, and one is faced with an enemy whose armour always reduces one’s damage by five. The second weapon is useless, while the first brings with it a chance of dealing damage; against an enemy whose defensive capabilities function differently, of course, this will play out differently, but in this context what seems like an arbitrary distribution of damage around a mean average shifts from adding a small amount of incidental variety to perhaps providing the player with the opportunity to challenge an enemy they otherwise would not face. Secondly, these kinds of variations can become strategic decisions in a context where one needs to achieve a certain degree of outcome, and gameplay options with wide degrees of outcomes offer the opportunity to do so faster, or with less danger, than gameplay options with unchanging but lower outcomes. An example can be seen in the acquisition of in-game financial reward from a game action: if one is performing sequences of gameplay which take an hour to complete, and distributes an unpredictable amount of virtual coinage at its conclusion, and the player requires x coins to meet their objective, different sequences of gameplay with different unpredictable and predictable currency outcomes – such as different quests – might become actual strategic decisions. Does one pursue five sets of quests guaranteed to yield a certain amount each, or pursue only one quest which might, if one hits the right chance, reward the player enough after a single sequence? In this way unpredictable chance events within the Degrees of Outcome model can be seen as important gameplay choices, giving players options between numerous ways to construct a set of quantitative outcomes needed for their gameplay objectives. Thirdly, and related to this second point, we might also propose that this model of chance affects the feel of different weapons and the like: a weapon with a high range of damage feels more unpredictable, potentially more exciting, more nerve-wracking, more like a gamble, than one with a smaller range of possible damage outcomes. Similarly, there need not always be an unbroken chain of possible outcomes – one’s weapon can deal between ten and twenty damage with an equal distribution, for example, but one can also readily have disconnected outcomes: there is a 10 per cent chance of dealing 100 damage, and a 90 per cent chance of dealing almost 10. A weapon of that sort would create a very different feeling in the player, one with rare but extremely noteworthy outcomes as opposed to a closer meshing of those outcomes. Fourthly and lastly, this model also seems to reflect a tacit intuition of the real world more readily than other forms of chance perhaps might. We are aware that not all sword swings are created equal in the real world, no matter how skilled the wielder; the Degrees of Outcome model reflects this and creates a mechanical analogue of the inevitable variation in repeating real-world events. Doing this adds what we might call a certain texture of realism to such unpredictable outcomes, with outcomes that acknowledge the inevitable variation in many kinds of repetition; the gameplay purpose here

GAMES OF CHANCE

73

is therefore a subtle, tacit one, not one created through its interaction with other gameplay systems or strategic choices as suggested in the other three explanations here.

Different outcomes We now consider the second model of chance, in which there are a range of qualitatively different outcomes (and there might also be quantitative variation within those, but the qualitative difference is the important element of this discussion). In this model the player is still guaranteed a change in the game state of some sort, as in the first model, but whereas all outcomes of the first model were different severities or degrees of the same sort of outcome, in this case we see the player having the possibility of a range of different, mutually exclusive and non-exchangeable, outcomes. Whereas the Degrees of Outcome model represents a difference in species (ten, eleven, twelve) within one genus (coins, damage), this model represents a difference across multiple genera which might or might not be actualized in the moment of chance–unpredictability. This genus might be a weapon, a piece of armour, a one-use item, a piece of information, a key item; these genera are mutually exclusive, do not serve the same purpose and are not mutually interchangeable in the gameplay function they serve. Game artificial intelligence (AI) is a central site for implementation of the Different Outcomes model of chance. In many cases AIs in video games will have a range of actions they can perform in a given context which are qualitatively distinct from one another. Take, for example, the AI granted to a humanoid enemy in an average first-person shooter game. Enemies might elect to attack the player, and if they do so, they might decide to shoot at the player, throw a grenade, call for backup or move into close combat to attempt a melee attack. They might alternatively choose to jump across a gap or jump or climb up to a higher vantage point; they might retreat; they might utilize an ability, such as healing themselves or reloading their weapon; they might encourage some of their allies to move forward while they remain back. Each of these is entirely different, for throwing a grenade and shooting a weapon cannot be interchanged in the effect they might have on the game world, and therefore represent qualitative differences in action that give rise to genera; by contrast, firing once, twice or thrice, or throwing a grenade nearby, far, or extremely far, are not. The Different Outcomes model is consequently deployed to make the decisions of AI actors unpredictable. With players perceiving generic rather than specific difference across the decisions, such decisions will have a considerable impact on the player’s experience with these actors. In this regard, the Different Outcomes model can be seen as a way of mirroring

74

THE UNPREDICTABILITY OF GAMEPLAY

human behaviour, which is to say, making choices within a wide space of possible distinct choices, rather than doing the same thing to a greater or lesser extent. The Different Outcomes model is also common in games where a particular moment of unpredictability plays out – often using dice in physical board games and tabletop RPGs – most often during an action by the player or another character, where the outcomes are qualitatively distinct from one another. In the cyberpunk Shadowrun board game series (1989–present) it is possible to produce a ‘glitch’1 by at least half of the dice rolled at one moment coming up as ‘1’s, which leads in turn to a set of positive or negative, but quite distinctive, effects for the player. In 2004 board game Betrayal at House on the Hill, the unpredictably selected ‘haunts’ determine much of the structure of the game (the goals of players) and the thematic setting (zombies, cannibals, werewolves) behind that particular playthrough.2 The year 2014’s Dead of Winter, a zombie-apocalypse board game, features a twelve-sided dice of six blank sides with no effect, five sides which ‘wound’ the character and one side which results in the immediate death of a player’s character.3 There is consequently a wide qualitative difference between an absent effect, a gradual cumulative effect which might lead to a future death, and an immediate death which, on a poor chance roll, cannot be evaded. A number of cards in trading card games Magic: The Gathering (1993– present) break with the determinism of the numerical gameplay of most of the game (the drawing of cards from a shuffled deck notwithstanding) by encouraging the player using the card to flip a coin to determine whether a particular card attacks or does not attack, or can defend, or whether you are dealt damage or get to use a special power and so forth.4 The ‘Wand of Wonder’ item from Dungeons and Dragons (1974–present) requires a player to generate a number between 1 and 100 (a ‘d100’), with wildly divergent effects depending on the outcome: one’s character is stunned, heavy rain begins to fall, a neutral animal appears nearby, the area darkens, hundreds of butterflies are summoned, one turns invisible, one’s body shrinks down in size.5 In the Reaper (2008) expansion for the board game Talisman (1983– present), the ‘Grim Reaper’ enemy, when encountered, requires players to roll a single six-sided die, with outcomes ranging from a boost to the player character’s statistics (the Reaper ‘ha[s] plans’ for that character) to instant death.6 This second model of chance therefore produces a wide divergence in the range of actualized outcomes compared to the Degrees of Outcomes model. As the player synthesizes these into an understanding of the unpredictability of the game, these outcomes become seen as entirely different genera, as outcomes which are not of the same sort but of a qualitatively different kind. Whereas the Degrees of Outcome differenciates a single element of a game’s virtuality – a monetary system, a damage system, something else numerical and able to be separated into clear variations within one concept

GAMES OF CHANCE

75

of generality which is easily understood – the Different Outcomes model represents instead the differentiation of the game’s virtual content, and in turn, the expanding of the player’s experience to understand the wider set of possibilities a game contains, not just the degrees to which it contains a given one of those experiences or aspects.

Outcome or failure We now come to consider the Outcome or Failure model, where there are two possible resolutions: either an outcome which creates a change (there may be one such outcome, or many) or a null outcome which creates no change; the moment of chance is triggered, but the unpredictable system elements provide no noticeable outcome. Within the change-producing outcome, there might be only one option – in which case the player receives X, or receives nothing – or there might be many options, which might vary qualitatively or quantitatively, but the spectre of the possibility of receiving nothing at all remains ever-present. Perhaps the most common implementation of the Outcome or Failure model is in ‘drop rates’. In many games, especially RPGs of almost all kinds, defeated or slain enemies will have a chance to ‘drop’ something upon their demise. This is an item or selection of items which can be collected as a prize for one’s victory, either from the corpse of the vanquished foe or via the reward(s) automatically appearing in one’s inventory. In some cases certain drops are guaranteed from certain foes,7 but in many cases these drops are a form of unpredictability, specifically chance and specifically of the Outcome or Failure sort. Many enemies in such games will sometimes drop a particular item, and other times yield nothing. Deploying chance in this spot is a way to distribute consequential items in an unpredictable manner, changing the rhythm and flow of gameplay in a particular playthrough when certain items do (or do not) drop. For example, the game Dark Souls (2011) features a number of foes in the earliest stages of the game – such as the ‘Black Knights’ and the ‘Taurus Demon’ boss enemy – with small chances to drop unusually powerful weapons for the player to wield. Although these weapons require the player’s character to have a particular set of stats to use effectively, this is not impossible, and by potentially offering the player such powerful items at the start of the game it is quite reasonable to expect that players, having become aware of these items, will then structure their continuing gameplay in order to reach a point where these powerful items can be used. Deploying chance in this way also serves to limit the player’s acquisition of items to a slower pace and to extend the playtime required to access all items, access a particular item and so forth. Although we examine this in greater detail in Chapter 7’s exploration of grinding and the notion of replay value, it is important to address here as a decision motivation for the

76

THE UNPREDICTABILITY OF GAMEPLAY

Outcome or Failure model of chance implementation. Such an approach is often criticized for both the ‘arbitrary gating’ of game content – which is to say, preventing the player from accessing something until an unpredictable decision decides they can access it, instead of gating according to, for example, skill – and for making gameplay last longer without actually giving new experiences to the player. However, this approach can extend the period of time players play a particular game, which is naturally appealing to many game designers, especially in games with a monthly subscription cost or other ways in which a more ‘committed’ player base, at least in terms of time spent, is understood as being highly beneficial. The application of the Outcome or Failure model to drop rates and conceptually comparable gameplay systems therefore stretches out the play experience, and likely slows a player’s synthesizing of game data into a multiplicity that closely resembles the game’s generative system. Did I not receive an item because there was no item to receive, or was it because I did not pass the game’s internal and unseen die roll? Another common implementation of the model involves the player being given the opportunity to make some choice which has a chance of yielding an outcome, or a chance of yielding no outcome. For example, some games have ‘speech check’ mechanics through which the player character can, for example, attempt to ‘persuade’ another character of something, using their skill of ‘charm’. In a video game, for instance, one might be offered the prompt ‘Explain that you are here on behalf of the Queen and should be let in [34%]’. In this case the player character attempts to make their case with approximately a one-in-three chance of success, but if unsuccessful and the other character is not convinced – 66 per cent of the time – then another strategy must be considered. Failure in these cases can also mean an actively negative consequence (as we explore in the next section), in which case it is a form of repetition rather than generality, but this is not required. Another form of the Outcome or Failure model comes from strategy games and closely related genres, where this is used in multiple ways. For example, in Europa Universalis IV (2013) one might be given the opportunity to take a charitable action, accompanied by the chance of one’s ruler gaining the ‘charitable’ trait. In many cases this also means committing to some payment or some other loss in order to gamble on the potential gain, but if the virtual dice roll poorly, that payment is lost for good. Alternatively, a unit or creature might have a ‘to-hit’ chance which represents how likely they are to land a blow (or accurately shoot a foe, etc.) a foe; this is different from the damage calculations explored in our earlier consideration of Degrees of Outcome, and in games with both, this first calculation takes place before the second is even considered. In this case the potential outcomes that a player will face become sharply polarized, and far from the ‘smooth’ variation in damage outcomes one will find in most calculations of the first sort, one will either deal some damage, or none, allowing in turn for the emergence of

GAMES OF CHANCE

77

long and extremely beneficial chains of hits, or long and potentially deadly streams where repeated misses build up in a short sequence of time. This model thereby enables a greater degree of unpredictability in the rhythm of outcomes, with high peaks and low troughs of a sort unlike either of the previous two models.

Different valence outcomes One final element to consider, which cuts across the models already outlined here, is the question of whether negative outcomes are included in the mix. In the Degrees of Outcome model, some of the outcomes of the same qualitative nature might be negative, such as the loss of an amount of money instead of a guaranteed but uncertain in quantity gain (although it is highly rare for negative outcomes and positive outcomes to mix in such a model – how could one lose money from opening a treasure chest?). In the case of the Different Outcomes model, some of these different outcomes might be negative; an example can be found in the Europa Universalis (2000–13) series of grand strategy games where one is often faced with decisions that might increase one’s nation’s ‘stability’ or decrease one’s ‘administrative power’, and weighing up the severity of those pros and cons, and the likelihood of those pros and cons, is central to the game’s strategy. In the Outcome or Failure model, the outcomes might similarly include negative outcomes, or in a scenario with only two possibilities – a change, and a null outcome – the change might be so undesirable that the null outcome, which is to say nothing happening, might be actively desirable. Of course, in many cases simply receiving nothing in the Outcome or Failure model can be seen as a negative outcome, or receiving quantitatively unusually low outcomes in the Degrees of Outcome model or the weakest option in the Different Outcomes model can similarly be seen as a negative outcome, when seen in contrast to what might have been. As such, here we are talking about a negative change in ones ‘absolute’, rather than ‘relative’, position in the game: one’s relative position in the game might be worse after a null outcome because one has, for example, exhausted one of the only moments in the game where one has a chance of acquiring an unusually powerful item, and few if any such chances remain; but it stayed unchanged. Here, however, an outcome has happened which has effected a material change in the game world, something has been actualized, and the player’s position is less strong than it originally was in some way. This is therefore different to negative actualizations being present in a randomization process: in that case one sees only the outcome and not the process, whereas here the player perceives a before and an after, and consequently the location of these unpredictabilities in chance yield a quite different player experience.

78

THE UNPREDICTABILITY OF GAMEPLAY

Repeating and individual chances There is now an important difference to be made between moments of chance which can be repeated indefinitely and moments of chance which can be ‘attempted’ only once, or a finite number of times, before they cease to be available to the player. A common example of the former might entail repeatedly slaying an enemy that regularly respawns in the hope it will drop a particular item, while a common example of the latter might entail progressing to a new floor of a procedurally generated dungeon, which will be created and filled with items and enemies only once. In the first case, when one is able to trigger and perform the same moment of chance many times, we see a hint towards the book’s later discussion of ‘grinding’ – performing the same chance-based in-game activity repeatedly in the desire for a particular non-guaranteed outcome which that particular moment of chance might produce. Without the ability to repeat (in the traditional, non-Deleuzean sense) the same chance, one must get the desired result on the first try or go without; with the ability to repeat, in a sense such repeating chances cease to become a matter of true repetition, of the repetition of difference, and instead become generality until the desired outcome is achieved. If one can try again, one can trigger a Degrees of Outcome chance until the highest or lowest (depending on what is desired) outcome occurs; one can perform the Different Outcomes model until the best outcome takes place; one can perform the Outcome or Failure model until any outcome at all takes place, and something is actually affected by the player’s actions. This ability to redo moments of chance also changes the ability of the player to learn about both the variables and the systems at play in a given unpredictable game; one can get access to all possible outcomes in close succession, without other moments or types of gameplay in the middle. Repeating the same ‘experiment’ gives us ‘knowledge of longrun behaviour’, while other ‘experiments can only be tried once’,8 limiting our ability to perform such mental calculations. However, this slows down the player’s advancement through the overall content of the game, when one piece of content is iterated upon until a desired conclusion is achieved. As such, this means that in the same block of time one comes to see more singularities that can be produced out of fewer aleatory points, rather than fewer singularities from a greater range of aleatory points in total. These lead to very different rhythms of gameplay, and shake up the speed by which the player learns about the game’s chance. Nevertheless, in many cases unpredictable moments of this sort only come up once in a game, or a finite number of times within the game. In Chapter 2’s exploration of the lasting impact of initial conditions, we considered how not getting the outcome one desires in a moment of unpredictability, if the number of potential moments is finite (as in randomness), can have a significant effect on the continuation of gameplay through denying the

GAMES OF CHANCE

79

player a potential benefit. However, as well as being present in games with randomness, this can also readily be present in games of chance, especially if the game entails some form of ‘permadeath’ system which actively prevents the player from reloading and trying that moment again. This form of chance does ensure difference, however, for a null outcome is not just a null outcome in and of itself, but also a chance lost to have another, active outcome; one’s condition has therefore changed, for prior to that moment of chance there was the possibility of repetition and change, and after it that possibility is gone, and any of the beneficial outcomes it might have given can no longer be acquired. In formal mechanical terms this is no different to no moment of chance at all, but it will affect the player’s thinking about strategy, future moments of chance and so forth. Therefore, if one has only one try, active or null elements can be created; if one can have many tries, a null element can be avoided if one simply attempts or triggers the same action a sufficiently high number of times; if one can have many tries but declines or exhausts the number of tries one is willing to endure of the same gameplay activity, null elements are once again possible. Experiences produced by these kinds of non-repeatable moments of chance are therefore quite different to their repeating cousins, forcing the player to accept their judgements, forcing strategic consideration to be made in the light of a failed role and foreclosing on pursuing any alternatives.

The Random Number God Throughout much of the work we have explored, and will continue to explore, how players deal with being confronted by game systems which can be opaque, difficult to unpick, and can often appear to be ‘unfair’. In turn, failing a game makes a player feel inadequate, for we are implicitly (or overtly) taught that ‘good’ games, or well-designed games, will always give the player a fair chance.9 What structures and concepts do players consequently create in order to navigate unpredictability and to become more comfortable with the problems such games present? In this vein, we conclude this chapter by examining a personification of chance (and, to a far smaller extent, randomness) regularly cited within communities who play unpredictable games, and a personification which is used for several of these purposes. This is the ‘Random Number God’, a deification or perhaps divine anthropomorphizing of the perceived fickleness, capriciousness or perversity of the game’s systems of chance. The Random Number God is a play on words that riffs off the more traditional ‘random number generator’ or ‘RNG’: a crucial component of any system through which elements of randomness and chance are shuffled and dealt in games – almost exclusively video games – that contain random elements. RNGs, as the name suggests, are designed to produce numerical values that cannot be predicted, and

80

THE UNPREDICTABILITY OF GAMEPLAY

which can then be put to a range of purposes in the pursuit, in this case, of creating unpredictable play experiences. The notion of the Random Number God is designed to take these random numbers and present a sense of agency and meaning to their emergence and the impact(s) they produce within the game world. This assigning of agency does not take place in any unpredictable context; it takes place in the context of any kind of randomness or chance that produces luck, which is to say unpredictability which significantly changes the outcome of the game. However, the Random Number God is rarely if ever mentioned when it comes to the randomness in unpredictable games,10 but is instead almost always mentioned in reference to a sudden and unexpected moment of chance which – it is thought – was a major part in a player’s downfall, or the sole determinant of their downfall, or in rarer cases caused the player character to become unusually powerful unusually quickly. Players will invoke the deity when it comes to unexpectedly encountering a particularly powerful enemy, for example, the battle with which then leads to the player’s death; it will also be invoked in a context where the player has, for example, a coin-flip chance to perish or survive, and the fall of the coin seals their fate instead of granting them another chance (which was, or was seen to be, equally likely). In rarer cases a string of unfortunate chance events might be ascribed to this divine ludic will, such as a chain of moments in the game where the player might have received valuable items, but instead of a ‘normal’ distribution of strong and weak items (perhaps three strong and three weak), all six items in a row were weak items, leaving the player’s character unusually under-powered moving into a later section of the game, which was later perceived as heading led to their death, or significant extra difficulties, and the like. I propose the invoking of this capricious game-playing spirit for chance rather than randomness reflects an implicit claim about the function of chance in these kinds of games, and how players see the role of chance in their gameplay successes or failures: by only invoking the Random Number God when an element of chance occurs and the player is subsequently defeated, there is a perception that ‘everything was going fine until the Random Number God decided to ruin my play’. The initial conditions of play are not blamed and are not assigned to the Random Number God; the same can be said of previous moments of chance leading up to the critical negative moment of chance; but only the negative moments of chance are the fault of the Random Number God. It is as if the deity is seen as almost selecting that particular instance of gameplay precisely because it was going so well, and deciding to maliciously intervene. Surveying the online commentaries that mention the deity demonstrates clearly this particular element to the phenomenon, and this sense that a victory, or at least continued progression, was denied. This anthropomorphizing or personalization is simultaneously very tongue-in-cheek, and yet also utilized extensively as a framing device for

GAMES OF CHANCE

81

an implicit or explicit claim that the outcome of an instance of play was beyond one’s power to influence; the power of the Random Number God was such that the human player was unable to overcome its decision to bestow victory or sow defeat. The invocation of the Random Number God positions agency outside the player, and gives the player something else – even if jokingly – to blame for their defeat. The implication is that the player’s skill dominated until the Random Number God intervened, at which point the ability to act within the game world was taken away, and replaced with a negative, anti-agential, kind of unpredictability. As the above demonstrates, there is an element of malice present in these conceptions of this god, or, perhaps, of dark humour, of toying with playthings, or leading the player on to believe they might emerge victorious through their own actions, before subsequently pulling the rug out from beneath their feet. I believe there is a depth here in the phenomenon of the Random Number God which allows us to understand in greater detail player perception of randomness, chance and their relationship with luck, and one which merits our consideration as part of understanding the player experience of unpredictability at different locations in a game’s design. To begin with, it is not a new observation to note that global mythologies are replete with ‘trickster gods’. Major trickster gods in the Western tradition include Loki, Hermes and Prometheus,11 the first two of which are known in particular for their mischief – sometimes benevolent, sometimes malevolent or anarchic, wicked, or spiteful – while the latter defies the existing order of the universe, upsetting the rule of the gods. Anansi, the spider trickster god of the West African Akan people, is a deity who routinely tests the boundaries of reasonable behaviour, ‘wreaks havoc in the human world’,12 and deliberately and provocatively casts doubt on established notions of reality. When Legba – a god of ‘crossroads’ and one who serves as an intermediary – moves, the accompanying ‘transformation that takes place creates a world for man’;13 he shakes up and disrupts the existing order. In these examples we see that trickster gods are understood as those who upset expectations and assumptions, and do in a way that is, often, malign and unpleasant, even if they understand or present their actions as a form of humour. As Helen Lock14 puts it most crucially for our discussion, ‘The true trickster’s trickery calls into question fundamental assumptions about the way the world is organized’ (emphasis mine). Where players previously believed their skill and the unpredictable offerings of the game thus far were sufficient for victory, the Random Number God is mentioned in contexts where this is no longer the case, and an unpredictable event either catastrophically undesirable and often causing the player’s defeat, or strikingly positive, has taken place. It is difficult to trace the origin of the personification, but one of the earliest references to be found – and I select this one due both to its antiquity and the description of the god it offers – comes from roguelike developer Chris Kern, who states that the RNG will ‘laugh in your face’, ‘taunt you’, that it

82

THE UNPREDICTABILITY OF GAMEPLAY

‘knows your deepest desires’ and will do its utmost never to offer them to the player.15 As such, the Random Number God is a modern trickster god, especially given the sense one gets from reading accounts of its nefarious actions that it seemingly delights in torpedoing otherwise promising play experiences. In turn, I think there is much we can learn from this orientation for understanding players’ attitudes towards these important moments of chance, their experiences of these moments of chance and how they are perceived irrespective of their systemic probability. Mihai Spariosu’s notion of pre-rational values, and specifically prerational play, is incisive in unpicking any assigning of game causality to a divine figure. Spariosu describes what he calls ‘pre-rational values’ as ‘an arbitrary and violent play of forces, the melting of the self into the ecstatic play of the world, necessity as an affirmation of pure chance’.16 Although the term pre-rational carries with it a certain veneer of colonial ontological arrogance and an implicit acceptance of a positivistic model of history we might find somewhat distasteful, terminology aside the distinction he draws is incisive and convincing. For example, the god Zeus, Spariosu suggests, is seen as a divine power who ‘scatters good and evil fates or lots among men not according to any principle of fair play, let alone divine justice, but according to his own whims’ – precisely as the Random Number God supposedly does. To unpick this statement a little, good (victorious) or evil (losing) outcomes are distributed not according to ‘fair play’ (according to a ‘the most skilled player wins’ model of fairness, although, as I explore in detail in Chapter 4, this is not the only possible form of ‘fairness’ for gameplay), and certainly not ‘divine justice’ (the most ‘deserving’ player wins?), but simply according to ‘whims’: the arbitrary selection of who, this time, will come out on top. As Lorraine Daston puts it in her analysis of the personification of ‘Fortuna’ as an ongoing theme of divine attribution in Western society, Fortuna was seen as having no vested interests, yet also unfair; everyone was equal, but no special rewards were given in exchange for extra effort, work or merit.17 This is echoed by de Goede, who notes the ‘capricious and fickle’ nature of this same quasi-deity.18 Where ‘rational’ values emphasize skill, ability and expertise, pre-rational value systems emphasize a complete acceptance of what Spariosu terms ‘chance-necessity’.19 This is the term he uses to refer to the concept that, in a mentality of ‘pre-rational’ play, the actions of the gods or the cosmos are little more than a ‘divine lottery’20 through which victory and failure, and happiness and tragedy, are handed out to human beings. In such a model, ‘randomisation, the blind goddess, fate, fortune, call it what you will, is an accepted part of life’.21 Game players are therefore understanding the strong effects of the RNG as a kind of pre-rational play, one that comes with no notions of fairness, distributes according to whims (mechanistic if not humanistic), arbitrary, fickle and capricious, and with no concern for the experience of the person playing it (which is, implicitly, supposed to be one where the player triumphs over adversity).

GAMES OF CHANCE

83

In turn, ascribing this agency and this blame specifically and particularly to a god that hands out unpredictable outcomes, rather than some other form of force beyond the domain and agential scope of the human (a scientific discourse that could hinge upon chaos theory or quantum mechanics, for example), should not be overlooked. In the introduction to this work we examined the close connections between games of unpredictability (and unpredictability more generally) and concepts of the divine. As we saw, a substantial number of religions from all parts of the world exhibit two commonalities in this regard. First, a belief that unpredictable events are not unpredictable per se, but only unpredictable to human perception, and would be highly predictable if one were able to perceive the will and (arbitrary) whims of divine powers. The Random Number God displays or enacts what Spariosu would call the ‘unconstrained and arbitrary play of the will’22 when it decides a player’s playthrough has ended, or is about to receive an incredible benefit. Secondly, that unpredictable games and communication with the divine can be closely connected, with a number of beliefs positing that game outcomes are contingent on the wills of the gods, while others perform rituals whose formal structural characteristics and use of unpredictable systems appear almost indistinguishable from the play of unpredictable board or card games. As such, placing this blame in a god is not an arbitrary choice or fluke, but rather deeply in keeping with existing trends in both religion and games. This shows us that, as noted in the introductory chapter, ideas about unpredictability have not entirely shifted to a supposedly ‘rational’ way of thinking. The apparent victory of a universal modernization resulted in the promise that ‘superstitions’ would decline, resulting in ways of approaching unpredictability in life, such as ‘fate’, being replaced more universally with the notion of ‘risk’, and an implicitly mathematical and ‘rationalistic’ appraisal of unpredictable events and futures. According to Anthony Giddens, this reflects a shift in the socioeconomic and political circumstances of people around the globe, and an attendant newfound orientation towards decisionmaking, and thinking about decisions, which would herald such a change. He proposes that ‘the more social activities are structured by what has been done in the past, the more people tend to think in terms of fate. The more people take active decisions about future events, the more, whether they are aware of it or not, people think in terms of risk.’23 The failure of this apparent new rationalism in casinos and card games is well documented with extensive literatures on the superstitions and dubious probabilistic beliefs of these players, but in the specifically divine nature of the Random Number God we find another continuing recourse to action and fate outside of the player’s control, which is seen as being deliberately and specifically directed at causing difficulty in one’s playing experiences. As interesting as I hope this analysis is to the reader: why does the Random Number God matter to our understanding of unpredictable games?

84

THE UNPREDICTABILITY OF GAMEPLAY

It matters because one of the core objectives of this work is to explore the nature of the game-playing subject, specifically the game-playing subject who spends their time playing unpredictable games, and in doing so must inevitably find themselves gathering data – if only subconsciously – about that game, and generating a mental model of the multiplicity that actualizes the gameplay experiences they undergo. The subject imagines a certain relationship with the random number generator through the deployment of the RNG concept: a relationship that is not mutually beneficial, but rather one which is personal, an omniscient and antagonistic god who sees the attempts of the player and works against them. It matters because it shows where players assign agency to the unpredictability of a game: when play continues in its customary way, players remain confident that their abilities have helped them navigate the game world, but a subsequent loss is reframed as something beyond the subject’s control. It shows what some players think of the repetitions that game offers, for some are seen as unusually capricious and cruel, while other repetitions (although all are part of the same multiplicity) are normalized and considered to be part of the ordinary play of the game. In turn, it shows how agency can be assigned to the differences between these repetitions, and helps us unpick why some actualized elements might be supposedly fair, and others unfair. Players having recourse to the Random Number God expected and anticipated generality, but got a true repetition, an emergence of something new – and in this case, something which defeated them. It also shows the wider cultural concepts players draw on, even if subconsciously and humorously, in the play of unpredictable games, and how specifically chance, as one location of unpredictability, draws on these concepts in particular ways. To conclude this section, I must once more stress that assigning blame or responsibility to, or in any way invoking, the Random Number God must be understood as an entirely humorous recourse. No player – at least, I assume no player – truly believes in the existence of a deity whose sole divine calling is the meting out of misery (and, more rarely, benevolence) to players who spend their hours playing games that contain unpredictable systems, and who is truly fulfilled by this rather minimal impact on the comings-and-goings of the human race. Nevertheless, as I hope to have demonstrated here, I believe the existence of this tongue-in-cheek deity is not merely an incidental cultural aside for players of unpredictable (video) games who wish to reflect with some humour on their gameplay experiences; instead, I believe an analysis of this concept of the sort I have presented here can shed significant light on the thought processes of those who play these games. Specifically, I believe it shows us how players address the relationship between unpredictability and its singularities, between chance and luck, and between the different forms of causality ascribed to success or failure within unpredictable games, as well as the extent to which individual moments, rather than longer-term gameplay decisions, are inevitably more

GAMES OF CHANCE

85

visible and more easily ascribed significant impact. Rather than seeing them as certainly extreme occurrences, but within the normal play of an unpredictable game and therefore something to be tackled by a player, these events are considered to be outside the normal play of the game, unfair and unjust (if negative), or unusually kind and generous (if positive). It also reflects the human tendency to shift agency for some unpredictable events to powers outside of human control, conflating the event being out of human control with the response to the event being out of human control – which is to say, that the success or failure or a playthrough came down in large part to the decision of the Random Number God, rather than the player’s response to that decision. Some of these issues highlighted by the existence of the Random Number God concept will be returned to in greater detail later – primarily at the end of the next chapter, which examines specifically how randomness and chance can be mistaken for luck – but chance is the almost-exclusive domain of this deity, and it therefore should be included in our analysis of chance. In this chapter we have examined the second location of unpredictability I propose in this work, which I call chance, taking place during the progression of gameplay. The chapter focused on identifying different kinds of chance, focusing on a four-part typology: a Different Outcomes model where each possibility is qualitatively distinct, a Degrees of Outcome model where the player is offered a spectrum of potential outcomes which share a given quality and commonality, an Outcome or Failure model where in some cases no change to the game’s state will be enacted upon the end of the moment of chance, and scenarios where outcomes are not only positive for the player (although these are the most common) but might be actively negative, changing the game state, rather than being simply negative through omission or absence. Each of these is used in different ways in game design, and each can generate different experiences for the player. To examine how particularly intense moments of chance, of any sort, are experienced by the player, the chapter then addressed itself to studying the notion of the ‘Random Number God’, which I argued was central to players’ negotiation of moments of chance perceived as being deeply consequential. The personification allows for the shifting of agency and harks back to numerous more ancient concepts of fate and fortune, reproducing a prerational notion of play where seemingly harsh and capricious events can be rationalized and accepted, preserving the sense of agency and skill that the player wishes to maintain. As a whole, chance is therefore a location of unpredictability quite unlike randomness. Where randomness creates tremendous variety in a single moment, chance creates gradually divergent lines of arborescence over time; while the first is generally regarded as a fair foundation for gameplay, the second is far more contested and runs a greater risk of upsetting a player’s notions of their agency and the emotional associations of the game’s decisions; and whereas randomness can only lay

86

THE UNPREDICTABILITY OF GAMEPLAY

the foundations for play, using unpredictability at the location of chance can serve a greater range of gameplay purposes, and generate a wide variety of experiences in the player according to the nature of chance, the distribution of chance elements and the impacts of those elements.

Notes 1 Shadowrun Wiki, ‘Shadowrun system’. Shadowrun Wiki. 1 May 2015. Available at http://shadowrun.wikia.com/wiki/Shadowrun_system. 2 Glassco, Bruce, ‘Betrayal at house on the hill’. Avalon Hill, 2004. 3 Plaidhatgames, ‘Dead of winter: Rulebook’. Plaid Hat Games, 2014. Available at https://www.plaidhatgames.com/images/games/dead-of-winter/rules.pdf. 4 MTGWiki, ‘Flipping a coin’. MTGWiki, 14 May 2016. Available at http://mtg. gamepedia.com/index.php?title=Flipping_a_coin. 5 Roll20, ‘Wand of wonder’. Roll20.net, 2017. Available at http://roll20.net/ compendium/dnd5e/Wand%20of%20Wonder. 6 Talisman Wiki, ‘Grim reaper’. Talismanwiki, 10 February 2014. Available at www.talismanwiki.com/wiki/index.php?title=Grim_Reaper. 7 All kinds of drop rates can encourage farming and grinding, one of the practices explored in much greater depth in Chapter 6. 8 John Haigh, Taking Chances: Winning with Probability (New York: Oxford University Press, 1999), 9. 9 Juul, Art of Failure, 7. 10 Alongside the reasons proposed for why the Random Number God is associated primarily with chance, there are two hypotheses as to why it is not equally associated with randomness. First, although the concept is a modern iteration of numerous historical gods of fortune and trickster gods, the terminology is, as noted, a riff on the ‘Random Number Generator’ – an element specifically of computing which is designed to mirror the shuffling of cards, the rolling of dice and the like. Video games almost never conflate randomness and luck, for doing so removes any actual gameplay; the only exception there is video game copies of existing games, such as video game slots, but (a) those are copies of randomness-luck conflation games, and (b) often have extra accoutrements in their video game form anyway. As such, the perception of one’s luck being created from randomness is common in analogue and pre-digital games, but exceedingly rare in video games, reducing the need to invoke the Random Number God in relation to randomness. Secondly, and to a lesser extent, we might reasonably propose that there is little to no investment in a game which has just started, whereas the capriciousness, fickleness and cruelty of the Random Number God are best seen during gameplay, when a player has already become attached to their current playthrough. For this reason too, the Random Number God is a god of what I term chance, not a god of randomness.

GAMES OF CHANCE

87

11 Joseph Russo, ‘A Jungian analysis of Homer’s Odysseus’. In The Cambridge Companion to Jung, edited by Polly Young-Eisendrath and Terence Dawson (Cambridge: Cambridge University Press, 1997), 240–55. 12 Emily Zoebel Marshall, ‘Anansi, Eshu, and Legba: Slave resistance and the West African trickster’. Human Bondage in the Cultural Contact Zone: Transdisciplinary Perspectives on Slavery and Its Discourses, edited by Raphael Hö rmann and Gesa Mackenthun (Mü nster: Waxmann Verlag, 2010), 178. 13 Robert D. Pelton, The Trickster in West Africa: A Study of Mythic Irony and Sacred Delight, No. 8 (Berkeley, USA: University of California Press, 1989), 127. 14 Helen Lock, ‘Transformations of the trickster’. Southern Cross Review 18 (2002), 1–7. 15 Chris Kern, ‘Re: Angband Rule Number One!’. games.roguelike.angband, 25 July 1998. Available at http://groups.google.com/forum/#!msg/rec.games.rog uelike.angband/MtMK_E__A4M/IOW_FdRdOfUJ. 16 Spariosu, Dionysus Reborn, 11. 17 Lorraine Daston, Classical Probability in the Enlightenment (Princeton, NJ: Princeton University Press, 1988), 151–2. 18 De Goede, Virtue, Fortune and Faith, 29. 19 Spariosu, Dionysus Reborn, 12. 20 Ibid., 15. 21 David, Games, Gods and Gambling, 12. 22 Spariosu, Dionysus Reborn, 22. 23 Anthony Giddens and Christopher Pierson, Conversations with Anthony Giddens: Making Sense of Modernity (Stanford: Stanford University Press, 1998), 102.

4 Games of Luck

Having covered the start and what we might call the middle of gameplay – what should be said of the conclusion of gameplay, vis-à -vis unpredictability? For this question, I have selected the term ‘luck’. A particular set of associations make this label especially appropriate. First and most obviously, there is a significant number of cultural associations with luck and its close partners, fate, fortune and destiny, which I believe make it an appropriate term for considering outcomes. These all refer to things which affect the outcomes of actions and the creation of events that happen to the individual, or to the player, which deeply shape their life; this emphasis on the resolutions and impacts of events on the individual, and on their lasting effects, makes it an appropriate term to use here. Secondly, we speak of people being ‘lucky’ or ‘unlucky’ in life, as if luck is a trait one is born with which continues to impact one’s life from cradle to grave. This can be seen as something outside of oneself, such as historically the interest or lack of interest a deity might take in one’s life (or as in the fickle decisions of the Random Number God), or as something one is born with, part of one’s allotment by the universe of good and ill experiences and therefore something which cannot be avoided. In all cases these notions reflect the idea of a ‘set or immutable pattern to the world’,1 a prior determination of the path one will take, and therefore something affected not by the vicissitudes of various unpredictable or chaotic systems, but rather by an external actor or element affecting one’s luck. Thirdly, following on from the discussion of the Random Number God in the previous chapter and the notion of outcomes being determined thus, ‘luck’ is an appropriate term because we use it to refer to the outcome of some contestation or challenge, to refer to how things played out, in the process legitimizing or more often delegitimizing the perceived ‘validity’ or ‘fairness’ of the outcome. This is therefore quite distinct from chance and randomness; randomness is a sorting of elements, a chance is an event, whereas luck is a trait, or a state

GAMES OF LUCK

89

or a summation. As such, I believe luck aptly covers the unpredictability of game outcomes, and the extent to which player actions do, or do not, affect the nature of those outcomes. All of this is of course not to say other terms could not have been considered, but none seemed ideal for this definition. ‘Fate’ has a sense of the religious and extra-worldly I would like to avoid, although it is essential for our understanding of how we think about luck in games; ‘fortune’ can evoke gambling or monetary outcomes, which are almost never present when we consider the entire range of games that might be played, and is thus perhaps not entirely appropriate; ‘destiny’ has an unnecessary grandeur, to put it mildly, and is a term never used within games discourse to date, and would mark a rather strange terminological departure. As such, luck is the obvious answer. In this chapter I first outline the importance of the conclusion of gameplay to the experience of playing and completing unpredictable games. I then proceed to consider one of the primary motivations for introducing luck into gameplay design, which is to say the levelling of the playing field – a game with a high volume of luck will allow anyone to play, which is seen as a method of reducing differences between players and thereby enabling all players to experience a winning gameplay moment. However, I will demonstrate that there are two forms of ‘fairness’ – making games contingent on luck and making them contingent on skill – which reflect two distinctive political philosophies, attitudes towards play, and beliefs about where power and agency in these kinds of games should be located. The chapter then unpicks the relationship between luck, time and skill, focusing particularly on games of short-term luck but long-term skill (such as poker and the emergent video game practice of ‘speedrunning’, entailing the fastest possible completion of a game) and the discourses and cultural notions developed by players of these games to come to terms with the role of luck in their play, and to reframe the navigation of luck as something which is not anathema to skill, but rather something which can be overcome or enhanced through skill, and something whose very presence actually demonstrates the skill required to play the games in question. The chapter then concludes by exploring how and why players often mistake randomness and chance for luck via studying the global player responses to 2012’s highly popular roguelike release FTL. This examination allows us to explore in greater detail the relationships – both mathematical and more importantly subjective and perceptual – between the three elements of the unpredictability typology thus far developed and the subjectivity in all perceptions of such.

The conclusion of gameplay Before considering unpredictability in the conclusion of gameplay, we should first consider fully the more general nature of the conclusion

90

THE UNPREDICTABILITY OF GAMEPLAY

of gameplay: What happens when games conclude and how does unpredictability affect it? What importance – personal, social, financial, structural – is assigned to these events and how does this differ for games with non-trivial amounts of luck? Understanding the first question will help us to see why specifically luck and its associations are quite distinct from those of randomness and chance, and consequently merit separate attention, and also how luck is mobilized alongside skill, the implicit element in this work that runs throughout our exploration of all kinds of unpredictability. Understanding the second question, meanwhile, will make clear why considering what I term ‘luck’ is important for any exploration of game unpredictability from the perspective of player experience. To do so we will consider games of complete luck, then games devoid of luck (the extent to which the game’s systems directly influence which player emerges victorious) and finally games with some luck, the cultural associations around the outcomes of which can draw on either of the previous two domains. The outcome of a game of unpredictability, although only the final moment in what might have been an otherwise lengthy period of play, takes on a striking importance in the cultural associations attached to the conclusion of gameplay, and therefore takes on a profound importance in both game design and our understanding of player experiences enabled by different kinds of game unpredictability. It is therefore an important ‘location’ of unpredictability to explore – whereas we have seen certain entanglements of randomness and chance can be deployed equally as randomness or chance, the entanglements of luck are quite distinct, on account of concluding and thereby drawing a line underneath gameplay, consequently preventing further impacts or further intersections with other elements of play.

Deleuze and luck I propose that luck occurs when prior moments of unpredictability, whether randomness, chance or both give rise to arborescence lines that are guaranteed to progress and endure all the way to the end of the game, irrespective of the actions of the player. In turn, regardless of when in gameplay these lines begin, they or their effects might be felt continually and gradually until the end of gameplay, or only at the exact moment of gameplay when their true consequence becomes apparent. This is a very particular definition, and one which requires further detail. As with chance and randomness, ‘luck’ has been used by other scholars to mean unpredictability as a whole; however, I am using it to refer to a particular kind of unpredictability, that of the end of the game, and one which is predicated on actions earlier in a game continuing to shape matters all the way until the end, instead of being overturned by later unpredictable events, or made irrelevant by later choices.

GAMES OF LUCK

91

By this, I do not mean a player’s choice which continues to reverberate until the end of the game, such as the choice of a set of mutually exclusive options, each of which is preserved in the player’s character (or civilization, or faction, or storyline, or whatever else) until the game’s end. These do produce comparable arborescence lines, but are selected by the player within the confines offered by a game; luck is produced by randomness and chance, by particularly consequential manifestations or moments of randomness and chance. The greater the number of these moments that affect a game outcome, or the greater the extent the outcome is shaped by these moments, the greater the luck involved. However, beyond that luck is difficult to state with any degree of certainty, for games are deeply complex things, and tracking back causality to its earliest actualization or emergence within the game system is a deeply challenging task. Figure 4.1 shows this understanding of luck as being consistent arborescent lines drawn from game unpredictability: some lines emerged (circles) from the game’s formal unpredictability and ‘died’, ceasing to have an effect on the outcome of the game, although they might continue to influence the game’s play, its story and so forth. Of course, each line was one line of many possibilities, but we are talking here about continued impact of the lines that were actualized, not the possible lines which might have been actualized. ‘Luck’ therefore entails the outcome or impact of an unpredictable choice within the game’s fabric continuing to have an effect until the end of the game takes place, and having some effect on that outcome: such as who wins and loses, the distribution of ‘score’ or ‘points’ or other rewards, the resolution of a narrative, the ‘ending’ the player is given in a game of multiple endings, whether the entire game is ‘completed’ or the game ends earlier (as with permadeath) and so forth. It is from such a basis that this chapter and other chapters explore luck, how the end of moments of gameplay can be shaped by the unpredictability within

FIGURE 4.1  The emergence and continued impacts of unpredictable moments in gameplay.

92

THE UNPREDICTABILITY OF GAMEPLAY

those games, how players deal with that, try to negate or evade that, and why luck – as with randomness and chance – is a very different location of unpredictability, and one with entirely different impacts and associations.

Luck and the meaning of ending gameplay We consequently turn to considering the cultural associations of games where these lines do and do not exist – in other words, games which do and do not contain luck. This will highlight the importance of considering unpredictability at this particular location in games, which is to say unpredictability best understood as playing out throughout games but being most fully realized at the end of games, and why unpredictability of this sort differs from the others explored in this work. In games of complete luck, the conclusion of gameplay allows for the arbitration of decisions through a system that appears fair and unbiased towards the contestants. This is readily appreciated when considering ‘games’ such as flipping a coin or high card draw. A central appeal of such games is that they require few if any accoutrements to perform; the overwhelming majority of people will have a coin to hand which can be readily flipped, while most of the contexts entailing high card draw involve card play anyway, and even if not, playing cards remain common analogue game-playing and household items. The outcome of a game is also incredibly rapid, with only a few seconds required to carry out either of these examples. This makes such games very appealing for making often trivial decisions, and might arguably make such games, despite their lack of actual gameplay, some of the most universally known and played in the world. In turn, players do not – superstitions aside – believe that their actions had any impact upon the outcome, and there is no suggestion of skill here in ‘decision-making’ games of this kind. However, we must also note that when monetary or other wagers are made on games of pure luck, they shift from being valued or at least accepted as legitimate equal-opportunity forms of decision-making to being ideologically maligned. For centuries the practice of gambling has been associated with a range of negative associations, most obviously those associated with what is nowadays termed ‘problem’ gambling and the need for ‘responsible gambling’ among those who might be affected, although a full review of this extensive literature is far outside the scope of the present work. By contrast, in games devoid of luck, the conclusion of gameplay is seen as the exercise of skill. This is the case for board games such as Go or chess, is somewhat the case for card games where systems have been implemented to reduce or come close to eradicating luck, such as duplicate deal bridge, and in a number of video games played at a competitive or professional level which have been designed to avoid the presence of all but the smallest

GAMES OF LUCK

93

amount of unpredictability. With no unpredictability to speak of, the responsibility for the conclusion of play must be placed purely within the player, rather than within any external system in the game. This is a positioning which inevitably comes with consequences from the extremely positive ideological values attached to these games, in contrast to the almost overwhelmingly negative ideological values attached to games of pure luck (where money is involved), or the negative-to-neutral ideological values where such games are used merely as decision-making aide (such as flipping a coin). Skill in games such as chess and Go, for example, has long been a site of substantial cultural and intellectual value, status and distinction.2 Comparisons have been often made throughout history between games of strategy and strategic decision-making in a range of contexts, including war, business and politics, all generally understood – in these comparisons – as highly desirable pursuits and abilities, or at least domains of work and activity that will offer success to people with significant game skill. Leibniz drew similar comparisons during his many and lengthy reflections on games, noting that both in games of ‘pure reason’ and games ‘partly of chance’, one can ‘see clever players determining more or less what is better to do, as is done in military matters or in medicine, using reasoning that is more broad than deep, which is also part of the art’;3 in doing so he also explicitly mentions clever players, building up a comparison of intellect and skill across games and game-like domains, and particularly those with high social and cultural status, such as military endeavours and medicine. Similarly, Brian Sutton-Smith and John Roberts argue that games of skill are seen as rehearsals for ‘related adult cultural forms’ – games of physical skill give one practice in the ‘exercise of speed and power’, while those of intellectual decision-making ‘rehearse hope for success by wise or clever decision making’,4 and offering us models of ‘real competitive styles’ we can practice against.5 In all of these cases, games without luck are seen to reward the most worthy, and in a broader sense, are seen to represent and make manifest the broader worth of the playing, and to prepare that player for success in more ‘productive’, if apparently structurally similar, domains. Lastly, in games with some luck, the conclusion of gameplay can be seen in several ways. A player might ordinarily expect the outcomes of such games to be seen as the interaction between skill and luck; skill played a part in the outcome, luck played a part and the particular balance between the two will vary from game to game, and will vary from one instance of play of a given game to another instance of play of that same game. Instead, in many games with luck and skill, the conclusion of a game is most often understood as being one or the other, rather than the far more complex and far more difficult to pin-down interaction between the two. Poker is the ideal example of this duality and cross-applicability. On the one hand, the player of poker has become strongly associated with North American cultural–historical tropes centred on the capacity for anyone to ‘make it’ through their skill

94

THE UNPREDICTABILITY OF GAMEPLAY

and ability,6 the quintessential American dream and the appeal of becoming a ‘self-made man’7 (poker is also highly gendered and male dominated8 and almost completely devoid of advertising aimed at women), and the idea that fortunes can be made in a domain that doesn’t require elite schooling, elite social status or elite wealth. Yet at the same time, the ability for anyone to win a particular instance of poker play is one of the fundamental appeals of the game (as with many forms of gambling), and it is often noted in both public discourse and in scholarly examinations that one needs only to ‘get lucky’ in order to defeat professional players9 in a single game of poker, suggesting the aspirations of the quotidian ‘everyman’10 to get rich quick without, necessarily, the deployment of any particular kind of skill or ability. Especially in an American context, scholars have suggested that the dominant ‘cultural code’ of gambling per se in the United States is that of the winner, especially if from the perspective of an ‘underdog or everyday hero’.11 Games which exist at these intersections between luck and skill are thereby able to perform an interesting act of ontological ambiguity, or perhaps of an ontological double-life, where depending on the context, the audience, and the point being made, tropes more traditionally associated with pure skill and pure luck games can be readily, and perhaps more importantly convincingly, deployed; but rarely both in the same context, for these two sets of ideological associations are inevitably opposed to one another. One last important point about the conclusion of gameplay is that it is not sufficient simply for luck to make unpredictable the outcome of play; in a game context this must take place, and must be signalled to the player, specifically at the end of play. If one knew the conclusion of an unpredictable game, especially a long game, at its start, the interest in playing the game would be lost. Consequently, luck is a form of unpredictability that must take place at the end of the game (as opposed to randomness at the start, and chance throughout) in order to have its true effect. What is compelling about games of luck is both that one plays out the full game without knowing what the conclusion is, that one can take actions that might to a greater or lesser extent affect the outcome of the game, and that the apparent chances of victory or defeat for all actors will likely vary as the game progresses before reaching their final conclusion. Play therefore retains a veneer of indeterminacy until its conclusion, or extremely close to its conclusion, at which point the actual role of luck re-appears and it becomes apparent what outcome will become actualized in this particular instance. As such, we see the importance of the specificity of the location of luck: even if players want the sense of eventual indeterminacy over who will emerge victorious and who will not, this runs up against the desire for action and decision-making throughout,12 meaning that luck must inevitably be manifested, or revealed to the player, at the end of gameplay. This makes it a location entirely unlike randomness and chance, and as such befitting of its own independent consideration.

GAMES OF LUCK

95

Levelling the playing field As touched upon above, luck is seen as a method for levelling the playing field of a game – by giving all players, no matter their skill level, the opportunity to have the experience of emerging victorious in a game contest. The greater the luck in a game, which is to say the greater the extent to which player actions do not affect the outcome, the greater the possibility of a less skilled player claiming victory against either another human player or a computer opponent. As Reith puts it, ‘Part of the “unique appeal” of games of chance is their absolute democracy. Just as it abolishes the efficacy of skill, so chance abolishes inherited or acquired difference, as well as those based on merit, patience, hard work or education.’13 Deborah J Bennett takes a perhaps less positive appraisal of this dimension, arguing that elements such as dice have always been ‘mechanisms for the purpose of removing human will, skill, and intelligence from their play and serious decision-making’.14 In a similar vein, Caillois argues that in games containing what I call luck, the player is ‘passive’ and ‘does not deploy his resources, skill, muscles or intelligence’; the play of such games consequently negates ‘work, patience, experience and qualifications’.15 It does not matter what one did prior to the play of the game, who one was or is, nor one’s successes or failures in other walks of life that might be considered game-like (contrast here with the cultural associations of skill). In such a game, ‘individual skill, intelligence or habitual difference have no influence on the opportunities for winning’,16 and therefore anyone has an equal chance of winning the game, of presumably enjoying the game, and deriving some pleasure from that game, or seeing areas of the game that are rare or unusual. This latter example is interesting because it points us towards not just thinking about skill and luck as questions of winning and losing, but also in terms of what the game-playing subject gets to experience in the play of the game in question. Luck, as Caillois tells us above, is an ‘insult to merit’17 – so what of skill? We tend to see skill rather than luck as the primary source of our successes in life18 and the source of the successes of others we seek to emulate. We associate skill with concepts of ‘knowledge and training’ as cultivated and performed by the meticulous adherent to one’s craft,19 we are keen to compare achievements made in games of skill,20 and skill can lead to dominance and visibility within one’s immediate interpersonal setting: for example, ‘through the display of skill and knowledges, gamblers …  can become experts within their social group and so win the respect and admiration of their peers’.21 In this regard skill is certainly not a level playing field: the play of the game is affected by the inputs a player performs or can bring to the table, and the consequent rewards are distributed unequally among its players. As Deleuze puts it, whatever parts of games are not resolved by unpredictability must be resolved by skill, which he calls the ‘mechanical development of consequences’, understood as ‘the art of causality’,22 which is to say actions

96

THE UNPREDICTABILITY OF GAMEPLAY

leading to guaranteed outcomes, rather than unpredictable ones. The equality of true luck is thereby lost because those who take winning actions will win, and those who take losing actions will lose; and indeed in a broader structural sense, the game simply possesses actions which are inherently and inescapably winning and losing actions, rather than only possessing actions that might or might not be winning or losing actions. However, although it appears that luck and skill are profoundly different with regard to equality, they are both deeply interwoven with each other and this concept than it might first appear. This similarity is most clearly expressed by Caillois, who states that the result of skill ‘is necessarily uncertain and paradoxically must approximate the effect of pure chance, assuming that the chances of the competitors are as equal as possible’.23 Developing further this notion that absolute skill resembles absolute unpredictability against equal competitors, it has been noted that the more skilled competition becomes, the greater the role of luck actually becomes. As Michael J Mauboussin argues, as one’s abilities improve, ‘performance becomes more consistent, and therefore luck becomes more important’24 – if two evenly matched highly skilled individuals compete, it is likely to be luck, which is to say extraneous unpredictable factors, that determine the outcome, for neither has a greater ability than the other. This he calls the ‘paradox of skill’; a skilled player facing an unskilled player allows skill to dominate, but between equally skilled opponents something must arbitrate on the results of any contest, and that something will be luck. As such, skill levels the playing field in the individualist sense of ‘only one’s ability, only elements within one’s control matter’, but also levels the playing field by, perversely, actually boosting the role of luck in competition when levels of skill are equal. However, we must note that this second conception of skill’s balancing role is rarely talked about compared with the first, where skill is lauded most often for its clear ideological associations with the desirability and inherent justice of competition, not just in games, but in societal games more generally. As such, both luck and skill propose an equality of players, either through unpredictability, through all players entering a contest with the same (which is to say, no) external assistance, or at high levels through the commensurability of the abilities of two actors and the challenge of finding a distinction between them. The divide in political philosophy between the concepts of ‘equality of opportunity’ and ‘equality of outcome’ is a valuable tool to bring into this discussion at this point. Equality of opportunity refers to the notion that all positions should be open to any individual who is most equipped for that post, based on some metric of merit. Under most definitions the unpredictability of origins – gender, race, sexuality, social class, education and so forth – should not be allowed to reduce opportunities to aspire or achieve. Equality of opportunity is instead concerned only with symmetric ‘starting conditions’ and makes no guarantee of symmetric outcomes. By contrast, equality of outcome seeks fair and balanced outcomes for all, irrespective of the asymmetry of

GAMES OF LUCK

97

starting conditions. We can therefore see that both skill and luck represent an equality of opportunity – all are able to have the same opportunities to win, albeit by different implicit metrics. On the one hand, that opportunity is understood as having nothing to do with the formal systems of the game, and is therefore entirely contingent on the individual, who brings nothing but their skill to bear and anyone who has cultivated sufficient talent can emerge victorious; on the other hand, that same opportunity is understood as being contingent on the notion that no matter what advantages and disadvantages one might be born with and consequently bring to the field of play, anyone might still emerge victorious through luck. However, equality of outcome – or something which, depending on definitions, is very close to it – is only possible with games of luck, rather than games of skill. Anyone can win when a game is played where everyone faces the same distribution of possible outcomes every time they play. In turn, in a zero-sum game where one is playing human opponents without a ‘house’ actor involved, perfect equality of outcome (returning to where one started, wins and losses evening out) becomes increasingly likely the more one plays, with the probability trending towards, although never actually reaching, an absolute certainty as time goes by. Even in a game against a house actor where one will lose in the long term, all players are statistically channelled into losing the same amount over the long term. The combination of these two factors means that games of luck provide a compelling equality of outcome, so long as one plays sufficiently long for the unpredictability, which will vary from game to game, to balance out. Equally, when considering the central role of chances in life more generally and their relationship to equality of outcome, it is interesting to note that game-like terms, such as ‘luck of the draw’, are sometimes used in such discussions. This illustrates – not for the first time in this work – the bleeding of game terms and game concepts into domains of social, political or religious life not ordinarily understood as ‘games’, but which can readily be presented to an audience metaphorically through an apparently fitting ludic comparison. Opportunities will always be a question of both luck and skill, for some of those opportunities are distributed by the ‘luck of the draw’ of birth, but equality of outcomes, in a game context, can only be achieved through luck that mitigates this initial luck and replaces it with concluding luck. Equality and luck are thus deeply interwoven, and luck is seen often as a central way to make games fair, and allow all players the same chance to experience the joys of victory – but also, of course, the lows of defeat.

Luck, skill and time Having established the conceptual underpinnings of luck, the ideological and cultural associations with games of luck and the role of luck in levelling the playing field of games, we now consider more fully the relationship

98

THE UNPREDICTABILITY OF GAMEPLAY

between luck and skill (understood here in a broad sense to mean agency, the ability to affect the game despite its unpredictable elements) in greater detail. Specifically, this will be achieved by focusing on the category of games we explored above where both luck and skill play at least some role in the outcome of play. How do players act to navigate games of both skill and luck, where one’s actions have some effect on the distribution of rewards but not a complete effect on their distribution, and what are these games like to play for the subject? Games such as poker, Backgammon or many tabletop and board games are games of this sort. They entangle both skill and luck, offering spaces of play where players certainly maintain some agency, but where some arborescence lines produced by unpredictable elements will continue to exert an influence or some importance until the conclusion of one instance of play. Such games occupy uncertain territory between the two extremes of luck and skill, where discourses and play experiences proliferate and brush up against one another, confusing any easy assignment of agency and blurring the values that become assigned to both the players and the unpredictable game itself. This section is therefore concerned with how the game-playing subject experiences luck and deals with it, and tells themselves luck is acceptable, and redefines traditional parameters of ‘winning’ and ‘losing’ – as well as traditional parameters of what entails ‘skill’ – in order to establish a model of game outcomes that is deemed psychologically acceptable. We can begin by distinguishing between games where there is luck, but only in the short term – which is to say that the strongest players will win in the longer term over dozens, hundreds or thousands of games, but any player may emerge victorious in a single session of play (the majority of card games belong to this category) – and games where a skilled player can expect to emerge victorious every time they play, and therefore the level of unpredictability inscribed into the game’s randomness and chance is minimal enough (and/or the level of potential mastery high enough) that no amount of ‘bad luck’ will affect the chance of success for commensurately skilled players. In the first category of games, players often take the possibility for any player to win an individual game as evidence for the game’s overall luck. In other words, the observation that ‘an unskilled player will sometimes win over a skilled player’ is incorrectly transmuted into the statement ‘unskilled players win as often as skilled players’, which is demonstrably spurious, but pervasive. This immediately shows us the tension felt in allowing for even the smallest amount of luck in games which are otherwise designed to be contingent on skill, and marks out an examination of these games as especially valuable. We will examine them primarily via two case studies. In the first case, I will examine poker, and how the intersection of player agency and game-system agency is navigated by its players, both in terms of immediate experience and the deployment of cultural and discursive norms and practices. In the second case, I will examine a particular speedrunning

GAMES OF LUCK

99

community, known as The Elite, whose behaviours epitomize the paradox of skill: that at especially high levels of skill, luck actually comes to dominate over skill, meaning that high luck can actually be discursively positioned as indicative of high skill, and therefore the pursuit of luck is in fact an element of skilled practice. Both are consequently discourses and practices of ‘reconciling’ or ‘acceptance’ of the challenges that luck poses to a player who wishes to have and maintain a sense of their own agency, and their abilities, when faced with game unpredictability. Poker is perhaps the game of skill and luck par excellence. On the one side there are many thousands of professional players whose skill is sufficient to ensure reasonably stable long-term incomes, poker training websites charging hundreds of dollars per hour for expertise, thousands of poker training videos, and at least hundreds of books on poker strategy written by many of the world’s most accomplished, decorated and ultimately profitable players. On the other side, there are tens of millions of players who rely on the turn of the card rather than the decision of strategy to determine all of their wins and losses, abrogating the potential agency afforded by the game. What is crucial to the interplay between luck and skill in poker is the differentiation – the splitting into two distinct concepts – of these two virtual notions over time. As Peter Bernstein accurately puts it, ‘Time is the dominant factor in gambling.’25 If one is a losing player, time put into gambling will influence the rate at which one loses money; if one is a winning player, the time put into gambling positively correlates with the likelihood of actually emerging from one’s gambling activities with a profit. John Haigh concurs, ‘If you play a game once only, the average outcome is useful to know, but it will be well short of the full story. Over one game, or just a few games, the variability of the outcomes can be much more important.’26 Consequently, the notion of expected value in game theory is integral to the understanding of skill and luck in unpredictable games and how players have come to tackle the relationship between luck, skill and time in poker. Expected value is the average value in the long run that will occur by playing out the same scenario an infinite or near-infinite number of times. In a single case, playing out a single hand of poker for ten dollars will only ever result in one player taking the ten dollars and the other player going home with nothing. If one player has a nine-to-one favourite hand, that player is more likely to leave with the money than the other, but the outcome is still absolute; ten dollars to one player, nothing to the other.27 Expected value has been central to scholarly examinations of poker-playing strategy, where it has long been used as the strongest metric for assessing a player’s ability in poker and their long-term profitability,28 even if in the short term the cards may fall against them. However, in poker, even a decision that appears to have positive expected value, being ‘+EV’ in pure game-theoretic terms is not necessarily always understood as the best decision. For example,

100

THE UNPREDICTABILITY OF GAMEPLAY

in poker discussion around 2005 and the years immediately following, a concept now known as the ‘Gigabet Dilemma’ was subjected to extensive debate by some of the world’s best card players. Succinctly, a noted poker player known online as ‘Gigabet’ proposed that a skilled player should actually sometimes deliberately take −EV decisions in poker tournaments, because on the times – less than one-in-two – when they emerge victorious from those −EV decisions, their skill will make up for the long-term loss of chips, and thus make losing chips in the long run an actually profitable decision due to their greater ability to profit with the extra chips on the times when the game’s unpredictability rules in their favour. In other words, a −EV chip decision might be a +EV money decision. This was met with strong challenges, arguing that taking −EV decisions is, by definition, never the correct play, while other players rallied to the original claim and supported the notion that although even the world’s best players should never take a ninety-percent-to-lose situation for all their chips, perhaps a fifty-five-percent-to-lose situation where the alternative outcome doubles their chips will, in fact, be profitable in the long term for a player skilled enough to put those extra chips, when acquired, to good use. In this way it proved difficult to define precisely how much more or less skilled one player is than another, the extra expected value (if any) that gains them, and how best to quantify that value in such a complex mathematical system. Expected value is therefore not an unproblematic objective matter in a game such as poker, for identifying game theory optima requires the modelling of systems of tremendous complexity and inherent unpredictability, as well as the unpredictability of actors whose behaviours are difficult to quantify. However, it is also integral to the practices and individual experiences of poker players as game-playing subjects. It is common to hear players discuss a hand where they lost, then to subsequently remark: ‘but it was +EV’, or some equivalent. By this they mean that although they lost in an immediate, physical, practical, financial sense, in another sense – a more rarefied, abstract, theoretical, skilled one – they actually won. Decisions, rather than outcomes, become what matter in the application of skill in a game of significant luck. Whereas several scholars have argued for the importance of score keeping as a metric for comparing abilities,29 poker players in this way have almost entirely rejected the role of money, which it to say how often one actually wins at poker,30 as a score-keeping tool, except in the extremely long term. These players are discussing what Deleuze calls the possible rather than the real, and are making decisions as if the real does not truly matter. This means such players have developed an understanding that one might make the ‘correct’ play but still lose money (or in games more generally, fail to see the outcome one desires). Data mining is a player practice that has become an essential part of navigating this sort of unpredictability for players, and one which reinforces the role of expected value as a dominant discourse within poker play. In my own time as a professional online player,

GAMES OF LUCK

101

I recorded and studied a little over two million hands of poker, examining which hands and which positions I played with the least skill, common situations where I lost more money than I should, scenarios where I failed to extract maximum value, traps I regularly fell into, and the like. By the standards of players who play a greater number of tables each day and have played for longer, two million hands is relatively small, but these numbers themselves are a crucial part of the unpredictability-navigating concept of expected value. The greater the sample size, the greater the accuracy with which can judge one’s ‘true’ ability at poker, and therefore crucially, the more comfortable one can feel in one’s play and the stronger the sense of one’s skill (and therefore agency) determining outcomes becomes. It is perhaps impossible to truly quantify how much of one single hand of poker is ‘luck’ and how much is ‘skill’, for these terms are contingent on the situation of that hand within previous hands and the anticipation of future hands, inevitably entangling any hand of poker with the synthesis of others. However, a modicum of mathematical knowledge is sufficient for starting to think about how much of the long-term play of poker is skill, and how much is luck; or, to put it another way, how many times one must play poker until good luck and ill luck will have somewhat, mostly, or almost entirely, balanced out. It is generally accepted in the poker community that one needs to play at least one hundred thousand hands, at the minimum, until luck will normally have ‘evened out’ and one can accurately judge whether one’s successes are due to high skill or good fortune, or whether one’s losses are due to low skill or poor fortune. As a result, when taking stock of their abilities, professional players don’t talk in terms of hands31 – except when bragging about a particularly impressive play – but they talk in terms of hundreds of thousands, or millions, of hands.32 EV is therefore a luck-navigating discourse, one which captures luck as being a question of the short term, and redefines winning away from actually winning money and towards winning EV instead. There is even a tongue-in-cheek name in poker for this: as opposed to real dollars, one wins ‘Sklansky Bucks’33 – named after poker theorist Mike Sklansky – when one makes a gametheoretic correct decision but loses money in the particular instance. The notion of +EV is therefore mobilized both as a genuinely valuable strategic tool for understanding the values of decisions in an unpredictable game with an undeniable element of luck (created through the nature of the game’s randomness and the structure of its subsequent play) and as a discourse which enables players to make their peace with the nature of the game. This foregrounds skill over luck and foregrounds skill as a technique to overcome luck, even if it might take a significant length of time for that overcoming to actually be manifested in the game (and in the case of poker, in the bank account of the player). It tells players that although they came into poker to win actual money, their objective should instead be the acquisition of theoretical money, for that is always completely within their

102

THE UNPREDICTABILITY OF GAMEPLAY

control no matter how unpredictably the cards might choose to fall in a given instance; correct decisions are always +EV, and correct decisions will inevitably lead to financial profit if made with regularity over a sufficiently long frame of time. Equally, this is a discourse that makes sense only to handle luck, not to handle randomness and chance, for it is the placement of this unpredictability at the conclusion of the game – how much control do I have over how I be rewarded for my in-game actions? – which makes the +EV concept so valuable. This highlights the value in separating moments of unpredictability in these games into their three temporal forms (randomness, chance, luck) for each one necessitates and encourages different responses and reactions from players. This absolute foregrounding of skill even when one actually loses is quite distinct to how skill manifests in scenarios of randomness and chance. Skill in randomness means the ability to navigate, in that playthrough, the unpredictable starting conditions with which the player is presented. Skill in chance can mean taking the best gambles a game offers you, making-do with what the game provides and moving on, or sometimes repeating a moment of chance to reach the desired outcome. By contrast, skill in luck, in the example of poker, entails accepting the role luck plays, looking to the long term, and understanding skill as being a question of process rather than outcome, for one remains master of one’s choices while equally being at the mercy of uncontrollable unpredictability. For my second example, I turn to video games. Speedrunning is ‘the practice of completing a video game as quickly as possible without the use of cheats or cheat devices’.34 One of the oldest speedrunning communities in the world to still remain active in the present day is The Elite. Almost exactly two decades old, The Elite are a community – now of dozens, in their heyday of at least hundreds – dedicated to speedrunning GoldenEye (1997) and Perfect Dark (2000), both ground-breaking first-person shooter games for the Nintendo 64 noted for their complex level design, wide variety of mission settings and objectives, and for speedrunning being already ‘built into’ their technical fabric. For most games players must do two things to enable speedrunning: deploy external timers which the player starts and stops at the appropriate moments, and reach community agreement over what moments count as the ‘start’ and ‘end’ of one playthrough. However, GoldenEye and Perfect Dark both feature in-game timers: one can see while playing how any seconds have elapsed (in Perfect Dark) and after finishing a level, the game reports one’s time to within one second (in both). These facilities have made these games easier to speedrun than others, and have also resulted in a community that counts to the second rather than the millisecond, opening the possibility (very common in The Elite) of tied world records. However, most important for our discussion here is the interplay between unpredictability and skill in The Elite’s speedrunning competition, and how members of The Elite both conduct their speedrunning practice and ‘rationalize’ the role of luck in their play, reconciling themselves to it

GAMES OF LUCK

103

by – as with poker players – foregrounding particular counter-discourses of skill and the navigation of luck. Both GoldenEye and Perfect Dark are relatively simple by modern video game standards, including when it comes to the complexity and sophistication of their unpredictable elements, but there are several elements of unpredictability which substantially complicate their speedrunning. First, in GoldenEye and Perfect Dark being shot by an enemy causes knock-back, much like in any first-person shooter games. However, the effect here is surprisingly pronounced, and more significant than many games; here, it feels almost as strong as explosives in many other games. This knock-back – or rather, knock-forward – has become central to speedrunning these games. The players of The Elite call the effect this has on their speed ‘boosting’. Being shot in the back and being propelled forward is a ‘boost’; being shot in the front and pushed back is a ‘back boost’, even one of which causes a large enough part of a second to be lost that it will normally end a worldrecord speedrun attempt. In many levels it is essential that one receives boosts in order to compete for records, sometimes as many as half a dozen or more, but this is not trivial to achieve. One must line oneself up between an enemy and the direction one wishes to be boosted; one must maintain that alignment, since enemies in these games do not ‘track’ the player and adjust their aim after they begin firing; and one must get lucky, for the guards in these games (especially GoldenEye, it seems) are extraordinarily poor marksmen. Competing therefore entails hoping for beneficial chances where boosts land successfully and are plentiful, and back-boosts either miss or are very few in number. Since this is almost completely a matter of chance creating luck, many tries – especially in levels where many boosts are needed – are the norm. This unpredictability present in every level is further complicated by a range of game design elements, which – although clearly intended to be minor, perhaps even all-but-irrelevant details – have caused The Elite’s players to address these incidental elements with the utmost attention. This is particularly apparent in GoldenEye. In the level ‘Streets’, there are several sets of enemy soldiers who spawn ordinarily with assault rifles, but one in eight times they instead appear with grenade launchers, weapons which allow the clever player to boost themselves via the grenade explosions. This means the best possible speedruns are only possible on the one-insixty-four times both sets of guards have grenade launchers, which are then acquired and used by the player. Alternatively, in the level ‘Frigate’ one is tasked with saving hostages; when saved, they select a point in the map to flee to, and when they get there, they are then ‘rescued’. However, selecting these points is managed by an unpredictable system, and not all end points are equidistant from where the hostages begin. It is estimated that only between one in every few hundred and one in several thousand tries35 will the hostages escape in the fastest possible way. Perhaps the most interesting

104

THE UNPREDICTABILITY OF GAMEPLAY

is the fact that how an enemy soldier dies affects speedrunning. When killed, enemies select one ‘death animation’ of several, and then act that death animation out. In the level ‘Control’ an objective cannot be accomplished until all enemies are dead – the speedrunner must therefore hope that all enemies offer a quick death animation rather than a slow one, for the latter might cost valuable seconds. In summary, therefore, clearly minor elements of the game’s design – the occasional unpredictable weapon, the occasional unpredictable movement path, the occasional unpredictable exhumation – add up along with boosting to create sufficient randomness and chance with inescapable arborescence, thereby turning into luck, that acquiring worldrecord times needs hundreds or sometimes thousands of tries until these elements align correctly to enable the ‘perfect run’. Therefore, here, as in poker, skill comes not just from playing well within the confines of the game, but doing so every time until the game’s unpredictable elements are actualized in such a way as to create a sufficiently rare aleatory point that, with the addition of skill, a singularity can be achieved – victory in a tournament, or a new world record. The Elite speak in terms of playing well every time until luck aligns; in one sense this is precisely the same as poker, but the rhythm of the interplay of skill and luck across the two is very different. Although poker takes hundreds of thousands of hands to even out, the forms of randomness that it produces tend to be a smooth spectrum of days of extreme profit, days of extreme loss, days of middling profit or loss, and days of break-even financial status. By contrast, in The Elite’s speedrunning practice a run is only of true value, which is to say able to potentially yield a world record for the skilled player, one-in-a-hundred or one-in-a-thousand times, while all other runs do not produce less profit or less loss in a regression of mean, but are all losses. This is therefore quite different from what the long term in poker looks like, or perhaps feels like, to the playing subject. The Elite also suggest that one of the skills that marks out the greatest players from weaker players is simply perseverance, and the time that one commits to one’s play, and not becoming disheartened after long sequences of speedruns where none of those playthroughs had a chance to become a world record, even with the greatest possible skill deployed by the player. Again, there is a similarity here with poker, but again also a divergence. In poker, players attempting to convince others of the skill of poker tend to downplay the elements of luck, stressing the number of opportunities that players have to make choices, the variety of forms those choices can take, the ability to bluff, the psychological and physical elements of reading one’s opponent and so forth. By contrast, players of The Elite often emphasize elements of luck, using it as proof of the competitive nature of GoldenEye and Perfect Dark speedrunning: the level of competition is immensely high, and this is shown by the fact that world records have been optimized to the point of needing a one-ina-thousand alignment of unpredictable elements to have even a possibility

GAMES OF LUCK

105

of success. The immense luck of a world record is reformed into a signifier of the skill with which these players play. As we saw earlier, in contexts between two equally skilled opponents it is often a question of luck as to who wins; in deterministic games this might mean the more well-rested player or the player more fully comfortable with a given climate, and in unpredictable games such as these it might mean who happens to be playing at the moment the required unpredictable elements of the game happen to align. It requires skill to compete at this level, but once one is at a high skill level occupied by numerous players, luck once more becomes a determining factor. There are three things we can learn about these discourses, their functions, and what they demonstrate about the gameplay experiences of players of games with luck. First, these discourses show that players perceive a need to ‘know’ they are doing well, that they are responding well to the unanticipated repetitions they face, that their mastery of the game’s multiplicity is real and significant, rather than feeling they are being taken for a ride by a game that is fundamentally out of their control. This shows that although unpredictability is central to the appeal of these games, it is also something that can make players doubt themselves, causing the agency behind one’s successes and losses to become somewhat blurred. Players consequently look for a way to reframe their actions, and the outcomes of those actions, in an unambiguously positive light. Secondly, these discourses encourage players not to become disheartened, and encourage not an acquiescence to the harshness and vicissitudes of fate, but rather a re-doubling of one’s efforts. In poker, one must continue to play longer so that one’s profitable play (if one is indeed profitable) will come to actually show the profit it should. In GoldenEye and Perfect Dark speedrunning, one must continue to play longer so that when the elements of luck align, one is there to take advantage; if the game is not played, that rare constellation of unpredictable elements can never be taken advantage of. Thirdly, they suggest that players need not just understand whether or not they are making the ‘right’ decisions, but to also understand how much they have to play in order to profit (make money, or secure a world record). In both case studies players are ‘reconciling’ themselves to luck, producing discourses practices to enable them to continue their play practices (maintaining a high level of poker play even if one is losing; playing a level hundreds or thousands of times without reward), and in turn to convince themselves of their skilled play even when the results are not what they would otherwise desire. Having observed that luck can ruin the most carefully laid plans or the most intricately executed move in games like these, players have developed these kinds of discourses to promote their own agency (even if that agency takes the rhetorical form of ‘I choose to play this 10,000 times’) and promote skill, or at least lengthy hard work (as Jesper Juul notes36), in these kinds of games. As Deleuze scholar James Williams notes, ‘Deleuze believes that

106

THE UNPREDICTABILITY OF GAMEPLAY

we have to find the right ways of acting in terms of underlying processes of repetition that we cannot fully control or understand.’37 The interactions between these discourses and the practices they ‘protect’ are designed to do exactly this: unable to fully control, and potentially fully understand the luck within these games, players have found ways to act that enable them to continue playing these games and continue a feeling of positivity about their own gameplay activities and their skill, even when results are mediocre, loss or defeat.

Mistaking randomness and chance for luck At this point randomness, chance and luck should be clear. ‘Randomness’ is the term proposed here for unpredictability in the initial conditions of play; ‘chance’ is the term proposed for unpredictability during play; and ‘luck’ is the term for unpredictability of the outcome(s) of play, which is to say who wins, who loses and so forth. However, there is one last phenomenon we must consider in the explication of luck: cases of players mistaking randomness and chance for luck. Or, to put it another way: players believing that the initial unpredictable conditions of play, and the ongoing unpredictability of play, have been having a greater impact on the outcome of play than is actually the case.38 This means players have the perception of arborescence lines which emerge somewhere and progress until the end of the game, while in the formal generative system rather than the player’s multiplicity which represents that system, none actually exist. Although at first glance perhaps something to be written off as a minor question of judgement, this question is of surprising importance in any consideration of unpredictable games; the level of luck a player believes to be present can come to shape their reception of the game as a whole, their level of engagement and enjoyment, the extent to which they believe they have agency within that game world, and is therefore a fundamental question if we seek to understand the nature of unpredictable games, and how players perceive that nature. Conceptualizing luck is therefore not only important to the development of the typology here, but allows us to address another issue: when, why and how might players see one form of unpredictability and read it as another? Perhaps the most extreme example in recent years of this confusion between multiple forms of unpredictability can be found in the 2012 independent game FTL: Faster Than Light. FTL is, depending on definition,39 a ‘roguelike’, ‘roguelite’ or ‘roguelike-like’ game, which is to say it features permadeath (the player cannot reload after a failure), procedural generation (much of the game is distributed through complex systems of randomness, explored in greater detail in Chapter 6), and a high level of complexity (there are a large number of moving parts, many sophisticated systems, many paths to victory, and these moving parts do not exist in a vacuum but

GAMES OF LUCK

107

intersect regularly with each other). The player controls a spacecraft and the crew of that spacecraft – both human and extra-terrestrial – through a series of solar systems and encounters with hostile vessels, friendly vessels, unusual events, stations, shops, natural interstellar phenomena, and more, until a final climactic duel with the ‘Rebel Mothership’, victory or defeat in which determines victory or defeat in the playthrough. Throughout the player’s journey they acquire and lose crew members, upgrade their ship, acquire or sell weapons and various abilities, and make a range of other choices. One playthrough, depending on player skill, style and the difficulty level being played (‘Easy’, ‘Normal’ or ‘Hard’), can take on average between thirty minutes and two hours. The game quickly became known for the fact that many players considered ‘Easy’ mode to be extremely challenging, and even to this day, players speak of struggling to complete the game once on that difficulty. Difficult games are nothing new, of course, but what makes FTL noteworthy is the discourses deployed by these struggling players, the responses deployed by players who did not experience these same hurdles, and the interplay between the two: all of which are predicated on the misattribution of randomness for luck. A brief and entirely informal examination in the process of writing this book identified at least thousands of forum threads, tweets and articles claiming that even Easy mode is ‘all luck’, or some differently worded variation on that central claim – which is to say, the player’s actions have minimal effect on the outcome of the game, and whether a player wins or loses a particular playthrough is dependent on whether the game’s randomness systems are ‘generous’ or not, and whether or not the game provides the player with the tools required to win. In the hundreds of these threads I opened, attempts by skilled players to point out that even Hard mode can be mastered to the point of an almost 100 per cent win-rate tended to be overwhelmingly ignored, overlooked, argued against or resented. This was the case most strongly in threads written near the game’s release; over time, the resistance of players to the idea that FTL is not entirely based on luck does seem to have softened somewhat, although not completely. There has been since the game’s release, and there still remains, a certainty for many that victory or defeat in FTL is a question of luck – through the game’s randomness and chance coming to fundamentally affect whether the player can win. Why, therefore, will players in games where the most skilled players are demonstrably able to win every game persist in stating that such games are luck? Or, rather, how are these players experiencing unpredictability and consequently drawing one conclusion about the virtual content of that game’s generative system – its degree of luck – which other players have demonstrated is not the case? There is a comparison here with the portrayal of one body of repeating virtual content in slot machines masking another body of repeating virtual content, or rather, with players reading a game’s

108

THE UNPREDICTABILITY OF GAMEPLAY

virtual content in a particular manner as a result of the presentation of that content. In this way the multiplicity, understood as a mental element of the game-playing subject, differs from the generative system itself which produces a set of gameplay experiences that suggest to players (albeit unintentionally) a different body of virtual content. In the case of FTL, players with one model of virtuality are experiencing randomness as luck, as a negative factor, something that prevents their skill; as such, victory is contingent upon the roll of the dice, not upon the player’s agency. Deleuze himself states, ‘The bad player counts on several throws of the dice, on a great number of throws. In this way he makes use of causality and probability to produce a combination that he sees as desirable. He posits this combination itself as an end to be obtained, hidden behind causality.’40 As Deleuze scholar Todd May interprets this comment, bad players (which in this context we use to mean ‘less skilled’ players, a term hopefully with a less hostile valorization) ‘seek a particular combination, and hope that the throw of the dice will offer it to them’,41 and are only able to progress in the game they wish to play, in the manner they wish to progress, if that particular roll of the dice comes to pass. For less skilled players, it appears that this is the only method of progression in the game; in other words, it appears that a game such as FTL is a game such as poker or GoldenEye and Perfect Dark, where one’s success in any specific instance of gameplay is contingent on the unpredictability of the game, not solely on the ability of the player. As such, if reconciling skill and luck are difficult for many players, might there be a superior way to integrate these two concepts, develop an understanding of how skill nevertheless functions in a context of luck and overcome such divides between a generative system and the multiplicity it inculcates in its players? In an influential article from 2012, game designer Daniel Cook proposed the concept of ‘mastery’ as a way of overcoming the classical dualism between games of luck and games of skill.42 Cook argues, correctly, that randomness and the perception of that randomness are never one and the same thing, for player perception of any game is a complex and ever-changing construct. This construct in the case of a non-deterministic game, as we see throughout this work, is composed of a player’s collective experiences in direct playing of the game, the unpredictable elements which have or have not been experienced to date, the hypotheses players consequently make about both the lexicons of variables and the systemic distribution of those variables, the significances assigned to particular orientations of elements and moments in gameplay, any insights drawn from conversation with others about that game or reading third-party information about that game, and much else besides; these are mechanical or psychological aspects, but aesthetic and thematic elements also shape how a player thinks about the game’s systems.43 Crucial to what Cook calls the ‘player modelling of randomness’ (in my terminology, unpredictability) is the distribution of elements, the

GAMES OF LUCK

109

successes and failures players experience with particular elements in play distributed in certain ways, and the conclusions players draw, therefore, about what led to their successes and failures. This is therefore a question of discerning ‘cause and effect’ – moments in gameplay which shaped the player’s experiences, whether a player’s decisions or the outcomes of a game’s unpredictable systems – and ‘noise’, which is to say what Cook calls ‘uncertainty, extraneous elements or unmastered complexity’ that confuse a player’s perception of cause and effect by introducing additional elements between the mechanically significant elements of the game and the player’s perception of those mechanics. What Cook defines as ‘mastery’ means coming to truly understand cause and effect in the game one is playing, and consequently being able to respond to whatever challenges the game throws out. This is not the ‘mastering’ of physical reflex in a deterministic video game where one knows precisely what challenge will come up on each playthrough, but rather mastery over the game’s systems such that no matter what unpredictable elements are deployed, the skilled player is able to succeed, and this success is based on their detailed understanding of the unpredictable systems underpinning the game as a whole. Developing his concept of mastery into a Deleuzean conceptual orientation, we can define mastery as being a twofold transformation in the unpredictable-game-playing subject: first, aligning the virtual content in their imagined multiplicity with the formal, mechanical specifics of the generative system; secondly, learning how to navigate whatever might be actualized from that system, which in large part is contingent on understanding the virtual content (‘How likely was this?’, ‘What might happen next?’). This is the foundation of the stark divide between those who believe the game to be primarily luck and those who do not. As such, in exploring what has caused some FTL players to develop mastery while others did not, I argue that two elements are responsible for this. First, this is down to what we might term the rhythm of gameplay, which is to say the pace at which a player experiences important moments in the game, the pace at which moments of gameplay occur which push the player towards failure. Secondly, I will argue that this results from the contrast between the cultural expectations of traditional roguelike players and the cultural expectations of those – often inexperienced in roguelikes – who were coming to play FTL for the first time. These combined to make FTL an exemplary study of how and why players mistake randomness and chance for luck, and therefore the importance of conceptualizing not just these placements or moments of unpredictability as different, but also the relationship between a generative system and the multiplicity it creates in the mind of the player. To understand the placement of game unpredictability we do not only need to understand where it is in a game’s design, but also where players subjectively perceive that same unpredictability, and when these differ, why they differ should be examined.

110

THE UNPREDICTABILITY OF GAMEPLAY

First, let us consider the rhythm of the game’s play: the pace and cadence with which differential elements are actualized and shown to the player. FTL follows a gameplay rhythm that is distinct from most ‘classic’ roguelike games (explored in greater depth in Chapter 7). In most classic roguelike games, gameplay intensity is at a low level for the majority of one playthrough; the player moves through the dungeon (or equivalent), encounters primarily or exclusively challenges that are easy to overcome given the present strength and abilities of the player character, and slowly but surely the game is completed. Then, very rarely, the player will encounter a sudden crisis. This might be from a singularity created through the distribution of randomized elements, such as a particularly challenging enemy which happens to have appeared next to a range of other enemies whose abilities happen to synergize extremely well with the strong enemy, or it might be a fixed, non-randomized element of the game which is designed to push the player’s abilities to their limits. If this crisis is survived, the regular ‘calm’ gameplay resumes for a period until the next handmade challenge or aleatory point where a fusion of randomized elements propose a potentially fatal threat; if it is not, the player’s character is slain, and the player must return to the start to play again. The crucial point here is that the player is given a highly visible event to assign the causality of their failure to, if they are indeed defeated by this challenging moment: it was this single encounter that slew them. Players can then reflect on how they might have otherwise survived this particular encounter – perhaps an item should have been used they didn’t use, perhaps they could have acquired help from another character or creature, perhaps the use of the spatial environment could have been better, or perhaps they should have simply fled, used an ‘escape item’,44 or taken longer to debate what they should do before making a move (most roguelike games are turnbased). One comes to focus on the singularity that ended one’s playthrough, and to understand either how one could have escaped that singularity (through different tactical choices) or how that singularity could simply have been avoided altogether (through different strategic choices). However, it is not accurate to say that the player’s reactions in the challenging singularities produced by such aleatory constellations of game elements are the sole determinant of success or failure. Significant importance can be found in the longer-term, slow-burning, strategic decisions that take place gradually and regularly throughout the play of a game, to the eventual success of a given playthrough. Nevertheless, roguelikes are generally understood as being games of tactics over games of strategy, and this is indeed the dominant form of play; indeed, in most classic roguelikes one’s character automatically regains health as one progresses through the dungeon, which means that one normally enters each new conflict at full strength. By contrast, rather than entering a battle at full strength and being defeated, in FTL one slowly loses one’s health over a sequence of slightly sub-optimal decisions,

GAMES OF LUCK

111

and although the final moment might appear unavoidable, it was not the final moment to which responsibility can be most accurately assigned. Think back to our previous understanding of luck as enduring lines of arborescence throughout the temporal movement of a player through a game: If we look at a death in game such as FTL, we will see lines going all the way back to many earlier events, often even choices made right at the start of a playthrough which have each added up to a player character which is behind the intended ‘power curve’ of the game, and slowly bleeds out, leading to an eventual defeat. In classic roguelikes the final moment carries much less responsibility, any lines of luck are likely to be shorter and more curtailed than in a modern roguelike such as FTL, where instead is a longer, slower death. It is not a death that takes place fully or primarily in the heat of a battle, but rather a death by inches. This is the difference between responsibility being found (primarily) in a single moment of unpredictable challenge, or distributed across many; the rhythm of causality in the latter is far harder to perceive, and therefore can appear to simply not be present at all. There is also a cultural difference alongside this variation in gameplay cadence. To both commemorate on deaths and to reflect back on the strategic or tactical decisions that might have led to their avoidance, roguelike players have taken to posting stories known as ‘YASDs’ – which stands for ‘Yet Another Stupid Death’. In these stories, players recount a narrative of their character’s adventure, often written in a detached thirdperson perspective, more comparable to a history book than a recounting of experiences the player was actually a part of. They describe what their character encountered, how they fought and overcame each challenge (until they didn’t) and what battle or encounter led to their demise. In many cases these are accompanied by a description of alternatives that might have been pursued, solutions the player only realized after their character’s death, and therefore why the death was an avoidable and ‘stupid’ one. The existence of the YASD denotes what we might call an acknowledgement of blame – by defining one’s own death as stupid, and placing the agency for that death in the player rather than the capriciousness of a game gifting the player unpredictable random elements, the YASD makes it clear that players who write them understand their death was an error in skill, not a matter of luck. It is through such concepts and rituals that players, as Thomas M Malaby would put it, deploy ‘tropes of accountability’ to ‘both construct and situate themselves in momentarily structured patterns of outcomes’.45 In turn, Taleb argues that ‘just as we tend to underestimate the role of luck in life in general, we tend to overestimate it in games of chance’,46 and the practices surrounding YASDs demonstrate the easy applicability of this observation in video games, when such large communities of players are regularly and routinely able to reduce the luck involved in the play of roguelikes for their fellow players, and demonstrate that what appeared to be luck is actually, merely, chance.

112

THE UNPREDICTABILITY OF GAMEPLAY

In turn, therefore, one can improve on a YASD by avoiding that same ‘stupid’ death the next time a similar situation arises. Such players believe they will be victorious when they become better at the game, not when the game’s systems of randomness or chance give them a better roll of the die. This has given rise to a striking culture of self-improvement within the classic roguelike community, with many individuals giving assistance and suggesting new strategies on roguelike forums, as well as the creation of in-depth ‘wikis’ for all the most popular games which give extensive information on enemies, items, strategies and so forth. The presence of a supportive community and these large knowledge-bases of information further demonstrate that ‘failures’ are seen as the responsibility of the player, not the system; with more information, and making better decisions, a player will be more successful. The language of ‘luck, fate and fortune’ is arguably less concerned with the challenge of being faced with an unknown future, and more concerned with ‘the focus it enables on the question of responsibility’.47 As such, when we say that ‘someone has fallen on bad luck’ or someone had a ‘run of good luck’ or some other equivalent construction, we ‘relieve that person of any responsibility for what has happened’.48 By contrast, in roguelikes the fatal moment makes for a clear, concise, and contained ‘problem’, to which a ‘solution’ which is the responsibility of the player or the spectator can be offered; equally, in rare cases where there is no tactical solution, community members will often instead suggest that longerterm strategic choices, such as the in-game ‘skill trees’ or equivalent that the player committed to, might have been in error. Learning therefore takes place as the player makes connections across the virtual elements of the game – the significance and import of those actualized, those which might have been, the different actions they might have taken. In classic roguelikes, this learning is collaborative and supportive, takes place in a context where points of defeat can be readily identified, and within a community used to dealing with such emotions, for all roguelike players have had them. However, FTL – and its contemporaries, such as The Binding of Isaac (2010) and Spelunky (2008–14) – was so noticeable because it transcended the boundaries of the traditional roguelike community, gaining recognition, interest, accolades and players from far outside the roguelike’s traditional epistemic confines. The game sold at least many hundreds of thousands of copies, far outstripping the ‘sales’ of even the most successful non-freeware classic roguelikes in recent years. These games have been recognized as bringing the gameplay of roguelikes to a wider audience,49 perhaps otherwise put off by the impenetrability of traditional roguelike graphics or the extremely long playthrough times50 that still carried the risk of permadeath. This meant that a lot of those playing FTL had little if any experience with the roguelike formula prior to experiencing the game for the first time, making FTL the ‘introductory’ roguelike for many players and creating two different demographics of players who both explored the same game. Indeed,

GAMES OF LUCK

113

the divide between FTL’s players and more traditional roguelike players was made stronger because many classic roguelike players actively rejected the rise of these modern roguelikes, believing that new graphical styles, their shorter playthrough lengths and the ability to unlock new elements across playthroughs marked too significant a divergence from orthodoxy. All of these factors meant that many of FTL’s players were not used to thinking about challenging, highly unpredictable games in the way that a traditional roguelike player might; equally, relatively few players with those bodies of experience found themselves playing FTL, with many preferring to focus on the pursuit of victory in older, more well-established, roguelike games. As such, two aspects of classic roguelike thinking would have been valuable. First, thinking about defeat in the manner of a YASD, which is to say as a failure on the part of the player, and one to be used as data from which to improve; secondly, the concept of mastery51 (over unpredictability) rather than pure ‘skill’ (over predictability) as a means to understanding one’s play practice, the navigation of these challenging spaces and the role that luck plays in one’s experiences. Therefore, I propose two causal factors that have led to, in the case study of FTL, the misattribution of luck for randomness and chance. Considering this misattribution is important for understanding how the three locations of unpredictability proposed in this work do not serve as absolute norms, but are rather present in the generative system and emerge in a player’s multiplicity, but the two need not always be the same; a player’s experience will fundamentally be of unpredictability as they experience it, rather than necessarily as the generative system ‘truly’ produces. One is that those who insist FTL is a question of luck tend not to be classic roguelike players, while those insisting that the element of luck in FTL is minimal, if even present at all, tend to be those with existing experience in classic roguelikes. These players consequently have less experience of dealing with what is, still, a quite distinctive and unusual game design pattern when viewed in the wider context of the video games industry as a whole. The other is that FTL is a game where one dies slowly but surely across a long period; even if the final blow might be ‘unavoidable’, once a certain level of strategic errors has been reached, one’s defeat is guaranteed, if implicit. A long sequence of battles with minor errors will slowly deplete one’s ship and crew; there is rarely a single climactic battle during which the player’s chances go from ‘excellent’ to ‘zero’, as can (and often does) happen in a classic roguelike. Considering in great detail the moment of death might serve one’s purpose to a greater extent in a classic roguelike, where the rhythm of gameplay means that individual moments of near-death do matter profoundly for the player’s overall success or failure, but this is far less appropriate in a game structured in the way that FTL is. The player focus of cause and effect upon any ‘luck’ in the moment of death, instead of the many strategic decisions that brought them to that point, has led to this stark rift in the FTL community between

114

THE UNPREDICTABILITY OF GAMEPLAY

those who pursued an understanding of mastery when they addressed themselves to understanding the game’s unpredictability, and those who pursued a more traditional understanding of skill as a dichotomized opposition to luck, leading to the (from the perspective of the generative system) misattribution of unpredictability in one location for another. In summary, ‘good players do not know what [the universe] will give, but they give themselves over to the game, in each throw’.52 Skilled players of games with randomness do not assign their successes or failures to luck, but understand that the core gameplay of such games entails the navigation of randomness, which is itself a skill – or mastery. As Katherine Isbister puts it, ‘Great games can take hours to play, and years to master; sometimes you can’t appreciate them without a high level of skill.’53 This points towards thinking about ‘skill’ in a broader sense as not just to do with the skilled navigation of in-game elements and challenges, although this is obviously and inevitably central to any such definition, but also where there is skill to be cultivated in how one addresses oneself to the game. To put it another way, making the transition from skill to mastery is itself a skill, and one which requires a distinctive understanding of the relationship between one’s actions and the outcomes they create; with insufficient mastery, randomness and chance readily begin to look like luck. An unpredictability in the initial or ongoing play of the game is taken to also create unpredictability in the final state of the game; a continued branch of arborescence selected through randomness or chance by the game, which despite the player’s actions will continue affecting the play until the play’s conclusion. All of this highlights once more the central importance of considering the player’s experience of unpredictable gameplay. This can not just diverge from a mathematical generative system in the eyes of the individual, but can even lead to significant cultural gaps and unresolved tensions in gaming communities, made more difficult to resolve amicably by the intertwining of contentious concepts of agency, ability and skill, into the mix. In this chapter I have sought to demonstrate that luck – the unpredictability of the outcome of a game irrespective of player action – in games can be seen in many places, has a range of effects on the game-playing subject and is responded to both individually and culturally in a diverse range of ways. Luck is a way to level the playing field according to one political ontology of fairness (the others being a game of pure skill, and indeed a game of pure work), and in doing so enable all players to potentially experience the joy of victory without one’s previous demographics or experiences impacting upon the play experience. In turn, although these forms of level playing field posit luck and skill as dichotomous opposites, luck and skill actually possess a far more complex relationship; games can be predominantly luck in the short term, but predominantly skill in the long term, highlighting the importance of considering the temporal dimension to luck (and, as explored throughout this work, to ludic unpredictability more broadly). We then explored how unpredictability

GAMES OF LUCK

115

in one part of a generative system, but not another, can be transformed in a player’s subjective multiplicity of that game into unpredictability in another location in the game, which can fundamentally transform a play experience. We proposed the concept of mastery as opposed to skill when considering the player’s actions in unpredictable games as a way to understand the player’s ability to take actions in contexts of profound unpredictability, and to identify the differences between games whose generative systems show ‘true’ luck and games whose generative systems generate a sense of luck in the multiplicities created by their players. Up to this point in the analysis, everything we have explored has been either played deliberately in a game’s generative system or perceived by players as being part of that generative system; but what about unpredictability which wasn’t intended to be present in a game, and certainly wasn’t meant to be seen by players?

Notes 1 Lisa Ann Raphals, ‘Fate, fortune, chance, and luck in Chinese and Greek: A comparative semantic history’. Philosophy East and West 53, no. 4 (2003), 537. 2 For an examination of the aesthetic dimension to chess play and an argument that chess exemplifies a particular form of ‘intellectual beauty’, see Harold Osborne, ‘Notes on the aesthetics of chess and the concept of intellectual beauty’. The British Journal of Aesthetics 4, no. 2 (1964), 160–3; for an examination of the premodern values assigned to Go in the Japanese context, with a particular focus on the cultivation of play as an almost collaborative, beautiful experience where the moves of both players and the performances of those moves are valued, see Andrew Feenberg, ‘Alternative modernity? Playing the Japanese game of culture’. Cultural Critique 29 (1994), 107–38; for an exploration of chess as a site of increasing cultural capital, and the entanglement of chess with notions of intelligence, ‘genius’, ‘prodigy’ and its site as a training ground for young people being encouraged towards adult eminence, see Gary Alan Fine, Players and Pawns: How Chess Builds Community and Culture (Chicago: University of Chicago Press, 2015). There are many other works in this area besides these, and many which explore cultural and intellectual capital in games with skill and unpredictability, such as David M. Hayano, ‘The professional poker player: Career identification and the problem of respectability’. Social Problems 24, no. 5 (1977), 556–64; and Breigh Radburn and Rachel Horsley, ‘Gamblers, grinders, and mavericks: The use of membership categorisation to manage identity by professional poker players’. Journal of Gambling Issues 26 (2011), 30–50. 3 Quoted in Edith Dudley Sylla, ‘The emergence of mathematical probability from the perspective of the Leibniz-Jacob Bernoulli correspondence’. Perspectives on Science 6, no. 1 (1998), 41–76. 4 Brian Sutton-Smith and John Roberts, ‘Rubrics of competitive behavior’, The Journal of Genetic Psychology 105, no. 1 (1964), 14. 5 Sutton-Smith and Roberts, ‘Competitive behavior’, 15.

116

THE UNPREDICTABILITY OF GAMEPLAY

6 Hayano, ‘Career identification’. 7 Rex L. Jones, ‘Poker and the American dream’. The American Dimension (Sherman Oaks, CA: Alfred Publishing Company, 1976); Aaron Duncan, ‘Reimagining the self-made man: Myth, risk, and the pokerization of America’. Western Journal of Communication 78, no. 1 (2014), 39–57. 8 Brett L. Abarbanel and Bo J. Bernhard, ‘Chicks with decks: The female lived experience in poker’. International Gambling Studies 12, no. 3 (2012), 367–85. 9 Childers, ‘Going all in’, 139. 10 Kathryn A. LaTour, Franck Sarrazit, Rom Hendler and Michael S. LaTour, ‘Cracking the cultural code of gambling’. Cornell Hospitality Quarterly 50, no. 4 (2009), 475–97. 11 LaTour et al, ‘Cultural code of gambling’, 486. 12 Even in slots, one still often clicks a button to stop the wheels spinning, even if this is simply an action which brings the game to a close without actually affecting its outcome in any real mechanical sense. 13 Reith, Age of Chance, 94. 14 Bennett, Randomness, 27. 15 Caillois, Man, Play and Games, 17. 16 Jens Beckert and Mark Lutter, ‘All you need is a dollar and a dream: Explaining the inverse relationship between socio-economic position and lottery play’. In Random Riches: Gambling Past & Present, edited by Manfred Zollinger (Oxford and New York: Routledge, 2016), 147. 17 Caillois, Man, Play and Games, 17. 18 Nassim Nicholas Taleb, The Black Swan: The Impact of the Highly Improbable (England: Penguin, 2008), 152. 19 Reith, Age of Chance, 93. 20 Kaplan and Kaplan, Chances Are, 32. 21 Reith, Age of Chance, 134. 22 Deleuze, Logic of Sense, 70. 23 Callois, Man, Play and Games, 18. 24 Michael J. Mauboussin, The Success Equation: Untangling Skill and Luck in Business, Sports, and Investing (Boston, MA: Harvard Business Press, 2012). 25 Peter L. Bernstein, Against the Gods: The Remarkable Story of Risk (USA and Canada: John Wiley and Sons, 1998), 15. 26 Haigh, Taking Chances, 12. 27 The poker-playing reader will note that there are other scenarios here – the pot might be split between two equal hands, in which case both leave with $5; or the kind of poker being played might be a ‘high-low’ or ‘Hi-Lo’ variety, in which case potentially one player could leave with $7.50 (for winning one half of the pot and splitting the other hand) and the other with $2.50. However, for the sake of the illustration, the absolute nature of the single outcome is

GAMES OF LUCK

117

what required emphasis, even if that outcome might in fact entail distributing winnings to more than one player. 28 For example, see Michael Bowling, Neil Burch, Michael Johanson and Oskari Tammelin. ‘Heads-up limit hold’em poker is solved’. Science 347, no. 6218 (2015), 145–9, and Darse Billings, Neil Burch, Aaron Davidson, Robert Holte, Jonathan Schaeffer, Terence Schauenberg and Duane Szafron, ‘Approximating game-theoretic optimal strategies for full-scale poker’. In International Joint Conference on Artificial Intelligence (2003), 661. 29 Kaplan and Kaplan, Chances Are, 32; Gary Smith, What the Luck?: The Surprising Role of Chance in our Everyday Lives (New York, NY: The Overlook Press, 2016). 30 Money is, however, used as a means of ‘keeping score’ between gamblers; see Mark R. Johnson, ‘Comparing the professionalization of pro gamblers and pro video game players’. Occasional Paper Series, 40 (Las Vegas: Center for Gaming Research, UNLV University Libraries, 2017), 1–10. 31 Individual hands of card games, as with aleatory points and singularities in all unpredictable games, form the centre of compelling stories that players tell each other and share more widely through online forums, video-sharing websites and the like. However, an interesting moment in gameplay where the multiplicity of a game reveals something uniquely interesting, and a player’s subsequent skilled navigation of that moment, do not imply anything in the longer term about the fundamental abilities of that player. Such individual hands or equivalent moments make for a compelling story, but do not necessarily make a longer-term winner. 32 These are the volumes of hands required for luck to balance out over the long term, especially in tournament play where prize money is disproportionately heavily weighted towards the smallest number of the highest finishes. 33 Rachel McKinnon, The Norms of Assertion (Basingstoke, UK: Palgrave Macmillan UK, 2015). 34 Rainforest Scully-Blaker, ‘Re-curating the accident: Speedrunning as community and practice’ (Diss. Concordia University, 2016). 35 RWhiteGoose, ‘Frigate 00 Agent 1:06 and Lower Hostage Completions REPOSITORY’. The Elite, 8 February 2017. Available at https://forums.theelite.net/index.php?topic=21774.0. 36 Juul, Art of Failure, 74. 37 Williams, Critical Introduction, 92. 38 The other side of this, of course, is believing that randomness and chance have less impact on luck than they actually do, and thereby failing to attribute luck where there is luck. This is most obvious in gambler superstitions or ‘systems’ designed to ensure victory in games which are actually purely or predominantly luck, and either way give a long-term mathematical house to the casino hosting the game. One need only mention a visit to Las Vegas or Macau to a small number of people and you can guarantee that someone will tell a story of this sort, namely a world-changing system to win roulette that only they have worked out, a lucky toy that once allowed them to win the

118

THE UNPREDICTABILITY OF GAMEPLAY

jackpot at the Bellagio, and so on. The attribution of randomness, chance and luck, and how players factor these elements into their reflections on gameplay, can therefore err on both sides of the mathematical distribution of play experiences present in any given game system; however, whereas the reduced attribution of luck has been studied extensively, the increased attribution of luck has been explored much less often to date, and is a question centred primarily on player perceptions of video games with unpredictable elements. 39 The definition of the term ‘roguelike’ is notoriously difficult to pin down, and hotly contested – see Mark R. Johnson, ‘The use of ASCII graphics in roguelikes: Aesthetic nostalgia and semiotic difference’. Games and Culture 12, no. 2 (2015), 115–35. 40 Gilles Deleuze, Nietzsche and Philosophy (New York: Columbia University Press, 2006 [1962]), 27. 41 Todd May, Gilles Deleuze: An Introduction (Cambridge, UK: Cambridge University Press, 2005), 64. 42 Daniel Cook, ‘Understanding randomness in terms of mastery’. Lost Garden, 31 December 2012. Available at http://www.lostgarden.com/2012/12/ understanding-randomness-in-terms-of.html. 43 See Chris Solarski, ‘The Aesthetics of Game Art and Game Design’. Gamasutra, 30 January 2013. Available at https://www.gamasutra.com/view/ feature/185676/the_aesthetics_of_game_art_and_.php. 44 ‘Escape items’ are items in roguelikes designed specifically to allow players to avoid and escape moments of unpredictability which might throw a challenge at the player which is extremely difficult, or impossible, to ‘solve’ (at least upon a first encounter with that challenge). In doing so, escape items – which often let players teleport, or become briefly invincible, or invisible, or other effects of this sort – are meant to reduce the luck in a game with a high level of randomness and chance, and the presence of escape items represents a tacit acknowledgement that not all possible outcomes can be predicted in the game design process. Interestingly, games with high randomness but an emphasis on longer-term decisions – such as FTL – tend to have fewer escape items, while games whose rhythms entail long periods of regularity and short moments of immense importance, such as ‘classic’ roguelikes, tend to have more; these item distributions reflect the different cadence of gameplay, and where the risks of permadeath are found. 45 Malaby, Gambling Life, 14. 46 Taleb, The Black Swan, 129. 47 Esther Eidinow, Luck, Fate and Fortune: Antiquity and Its Legacy (London and New York: IB Tauris, 2011), 153. 48 Bernstein, Against the Gods, 197. 49 Jack Forde, ‘The evolution of the roguelike’. IGN, 21 January 2016. Available at http://za.ign.com/the-binding-of-isaac-rebirth/97192/feature/theevolution-of-the-roguelike; Extra Credits, ‘Roguelike returns – how to revive a genre’. Extra Credits, 11 June 2014. Available at https://www.youtube.com/ watch?v=gk8ZknjYmek.

GAMES OF LUCK

119

50 Shorter classic roguelikes can easily take tens of hours for inexperienced players, and up to ten hours to complete for all but the most elite experienced players. Some with extensive story content, such as Ancient Domains of Mystery (1994–present), can take multiple days of play in order to complete some of their more extensive and challenging sub-quests, all done while in danger of permadeath. Some players of modern roguelikes sometimes suggest that the length of classic roguelikes is a primary reason behind their disinclination to play them; it is one thing to have the risk of permadeath for perhaps two hours at the most, and quite another to play with the same character over many days and still endure the psychological risk of their potential, permanent, demise. 51 Cook, ‘Understanding randomness’. 52 May, Gilles Deueze: An Introduction, 65. 53 Katherine Isbister, How Games Move Us (Cambridge, MA: MIT Press, 2016), xvi.

5 Games with Instability

Unplanned unpredictability We have thus far examined games whose unpredictability is an essential constituent of their gameplay design. Any card game would become trivial after a few plays were the same deal repeated every time; many computer games would be simple to complete if the AI’s behaviour could be easily and perfectly codified by the player; gambling games would never be gambled upon were there not some recourse to luck in their design. These are all forms of intended unpredictability – the game designers in all cases were aware of both the inclusion of unpredictability and were (they believed) cognizant of all effects from the unpredictability they had admitted into the game design. If they were not aware of the full scope of this unpredictability during the design process, it is reasonable to believe that either it was accepted by the designer once its true extent became apparent during testing, or accepted by the community of players who found the game compelling or interesting with its forms of unpredictability. However, such a definition immediately sheds light by omission upon a component of unpredictability that has thus far been neglected – that of unintended unpredictability. This can apply to games that are intended to be entirely deterministic within the strictures of an exacting ruleset but fail, in actual play, to measure up to this goal, and to games containing forms of unpredictability that go beyond that which was originally intended in the game’s construction. Any game can therefore become unstable in the right circumstances. These circumstances can be broken down into two categories: (1) situations where something occurs in gameplay that is technically within the designed ruleset but clearly unintended by the designer (an ‘exploit’) and consequently against what we might call the ‘spirit’ of the game, and (2) situations where a designed element does not function as intended (a ‘glitch’). First, to ground this discussion, we need to develop a concept of the ‘normal’ functioning of a game. James Williams highlights Deleuze’s

GAMES WITH INSTABILITY

121

sometimes-implicit argument that interacting with anything is ‘accompanied by the expectation that the thing will maintain some degree of consistency’1 – this could mean interacting with some element of a system, but in this case, the same observation is valuable for considering the game as a whole, and the identification of instabilities. Any kind of consistent or normal functioning is, however, an inevitably elusive notion. Deleuze himself notes the potential arbitrariness between acceptable and unacceptable points of chance: were one to lean over a roulette wheel and blow upon the ball, the player ‘would be stopped, thrown out, and the move would be annulled’. And yet, ‘What would have been accomplished, however, other than to breathe a little more chance into the game?’2 As we will see throughout this chapter, there are many valid reasons to propose that there is actually no such thing as the normal functioning of a game. However, in attempting to problematize this notion, I believe a valuable starting point is the ludic philosophy of Bernard Suits, particularly in his work The Grasshopper (1978). Suits argues that the act of playing a game means doing so only by doing so through and within the rules laid out by that game, even while different players might be more or less able at manipulating or acting within those rules and thereby more or less likely to take those rules and emerge victorious in a contest.3 He consequently defines a game as ‘an activity in which observance of rules is part of the end of the activity’,4 and thus even if one is pursuing the most formal goals of a game – get a ball into a net, for example – one must pursue it within the constraints of the game in order to be playing the game, rather than pursuing a non-game activity with the same strict end goal. Using the example of high jumping, Suits demonstrates that for high jumpers, ‘their goal is not simply to get to the other side [of the pole], but to do so only by using means permitted by rules’; and, in turn, ‘their reason for accepting such rules is just because they want to act within the limitations the rules impose’, and ‘they accept these rules so that they can play this game’.5 He concludes that ‘we may therefore say that games require obedience to rules which limit the permissible means to a sought end, and where such rules are obeyed just so that such activity can occur’.6 In this regard we can see that when one addresses oneself to the play of a game, one is presented with a set of rules, which are designed to construct a particular experience. The player gets the impression of what this experience is meant to be by learning and considering these rules, from external sources such as discussions with friends or readings of game manuals, and most obviously simply by playing the game and testing the boundaries of the constraints and restrictions the rules place on what one can do. One’s broader knowledge of other semantic domains, both in games and elsewhere, also contribute to this understanding: for example, all games within differentiated genre labels such as first-person shooter or real-time strategy have a similar spirit to them. One assumes that the experiences one has through play represent, broadly speaking, the ‘normality’ of the

122

THE UNPREDICTABILITY OF GAMEPLAY

game being played, and the more one plays and the more that subsequent experiences confirm to form, the stronger this perspective becomes.7 As such, one quickly gets an idea of both the letter of the game (the rules) and the spirit of the game (the game they are meant to produce, and which could not be played without those rules). When one plays a game, the letter and the spirit generally produce the same thing, for they reciprocally determine one another. However, with the spirit of a game established, it can be the case that a rule can be broken (a glitch), in which case both letter and spirit are broken, or something can emerge which keeps to the letter, but breaks with the spirit (an exploit). In this regard, the spirit of the game entails two elements of virtuality: both ‘discovered virtuality’, which is to say what a player has so far uncovered of the virtual content of the game, the rules by which content is produced, and what we might term the ‘expected virtuality’, which is to say the circle of gameplay experiences and the rules that generate them which are thought to be reasonable based on the virtuality thus far seen. These are connected in a closed loop; the discovered informs the expected, and when the expected conforms to type, it becomes accepted as part of the discovered virtuality, and a part that syntheses well with all the other parts. As Deleuze puts it (emphasis mine), a singularity ‘is by nature transgression or exception, always revealing a singularity opposed to the particulars subsumed under laws’;8 so both exploits and glitches represent such a transgression of the spirit of the game, the exception to the kinds of experiences that the game’s rules are supposed to produce. With our ‘baseline’ established, we can now return to the matter of instability. Given our focus here on unpredictability, I propose that distinguishing between the unpredictable events that give rise to these kinds of instability, which is to say the two kinds of singularities which allow for the emergence of instabilities, is the most valuable definitional difference. That is, glitches arise from unpredictability within game systems, and exploits arise from unpredictability at the intersections between game systems. In Deleuzean terms, a glitch is an unexpected and unintended aleatory point within a game system, while an exploit is an unexpected and unintended singularity constructed at the aleatory point between several game systems, all of which are individually functioning as intended, but collectively yielding something new which also, crucially, disrupts gameplay. These are both forms of unpredictability, but new kinds of unpredictability within games not previously considered in this work up to this point. Nevertheless, within my definition of unpredictability as being something within the fabric of the game that causes actions and outcomes not to be perfectly connected and anticipated, this form of unpredictability must be included. The precise meanings of these terms are debated in game studies and gaming communities, but I believe my use of these terms here summarizes this two-part typology well. By my definitions, exploits break the implicit rules of the game (‘given the rules written, it should not be possible for the

GAMES WITH INSTABILITY

123

player to hold Item A and Item B at the same time, although there is no rule that specifically and only states that Item B cannot be held if Item A is also held, and vice versa’) while glitches break the explicit rules of the game (‘we have coded a specific rule that Items A and B can never be held at the same time’). There are specific guards in place against glitches (which fail when a glitch occurs), while there are no specific guards in place against exploits, for these arise in the gaps between the established rules. Two concrete examples may illustrate this distinction better. In the RTS game StarCraft 2: Wings of Liberty (2010), the ‘Protoss’ faction possesses two particular units: one is the ‘Mothership’ which possesses a ‘Vortex’ ability that sucks in a group of enemy units into a small area, and the other is the ‘Archon’, which is a comparatively weak unit that deals splash damage (damage over an area). Players soon discovered a strategy which acquired the tongue-incheek moniker of the ‘Archon Toilet’ due to the swirling animation of the Vortex ability: by having one’s Mothership pull in a group of units, then having one’s Archons attack the units while they are extremely close (almost overlapping), damage is dealt to all the closely packed units simultaneously that far exceeds the intended damage-dealing abilities of the Archon, even though no explicit rules of the game have been broken. All units functioned correctly, but something emerged in their relations which actually worked against the kind of gameplay that Starcraft II’s rules were supposed to construct, which is to say, a balanced competitive strategy game. Exploits are thus unexpected (and therefore unpredictable) discoveries that often lead to an alteration of game balance once discovered (the Archon Toilet became impossible in later versions of the game due to developer intervention). Glitches, meanwhile, are of a qualitatively different nature. In The Legend of Zelda: The Wind Waker (2002), upon the death of ‘Link’ (the player character), by rapidly pressing (‘mashing’) the A and B buttons on the GameCube controller at high speed (ideally close to a dozen times a second) the apparently ‘dead’ Link will start to rise up in the air. By utilizing this curious strategy of post-mortem ascent the player is able to access many areas of the game’s dungeons (which are often designed with vertical movement in mind) far sooner than intended. This glitch is known as the ‘zombie hover’. Flying when dead should, naturally, be impossible, and the death and gravity systems in the game were both designed in such a way that anything resembling a zombie hover should be impossible; nevertheless, this glitch exists. Exploits therefore occur when the full possibility space of a given ruleset has not been anticipated by the game’s designers, no formal rules are broken in performing them, and there are possibilities in the virtual content of the game that enable actions that were not intended to be present. If any rules are broken they are unspoken rules, such as ‘the player should not be able to kill enemies so quickly’, which are implicit but never codified. Glitches, meanwhile, occur when a rule is broken due to that rule or system being inadequately coded, and the potential for a rule or a system to be

124

THE UNPREDICTABILITY OF GAMEPLAY

actualized in an unusual way. From these definitions, ‘bugs’, ‘errors’ and other such terms are all taken to be additional synonyms for ‘glitches’, while an exploit is of a qualitatively different form. Games with such elements I define as games with instability. We shall see that this definition encompasses a large range of games – almost all games, in fact, and it is very possible that instabilities (in Deleuzean terms: unanticipated singularities) that have not yet been discovered exist in games currently considered entirely stable – as too do randomness, chance and luck. As discussed earlier, few games are truly fully deterministic, and once we acknowledge the role of instability in the gameplay experience, this list shrinks even further. Instability should not be thought of as a fourth categorization alongside randomness, chance and luck, but rather as something of a second axis. All the unpredictability we have examined thus far has been intended, and although the scope of instability may be rendered sufficiently concise to merit just one chapter, it is nevertheless a form of unpredictability that is both analytically intriguing and extremely common across the full spectrum of games. How do players respond when instabilities arise? How do they study these forms of unpredictability and learn about what triggers them and how they might be used? In considering what the game-playing subject does when confronted with an instability, these instabilities should not necessarily be thought of as errors or mistakes. As Deleuze puts it, ‘Error …  is a very artificial notion, an abstract philosophical concept, because it affects only the truth of propositions which are assumed to be ready-made and isolated.’9 Deleuze’s philosophical project encourages us to abandon pre-formed identities and instead to simply assess the experiences players have in games, without constraining ourselves with prior notions of correct and incorrect functioning of a generative system – which, in some cases with instability, wasn’t even meant to be a generative system in the first place. By looking at instabilities as just another kind of unpredictability, we can bring them into our understanding of different locations of unpredictability, examine how these affect both player experiences and cultural practices, and expand our definitions of generative systems in a game-playing context. The chapter now proceeds to examine glitches and exploits in turn. Exploring glitches, I particularly focus on both the unpredictability of the glitch and the impossibility of anticipating one beforehand, and also on the instability that emerges in the game and cultural or community assumptions of how a game should be played that a glitch can create. The chapter then considers their closely related cousin in the form of exploits, where all game systems continue to function as intended individually, but yield a singularity which was both unpredictable and renders the game unstable in a similar manner. The chapter then proposes the concepts of ‘tight coupling’ and ‘weak coupling’ to describe the systems and the relationships between systems in games, wherein games of the first sort have a smaller number of simpler rules, and games of the second sort have a larger number of more complex

GAMES WITH INSTABILITY

125

rules (and are more prone to instability). I then briefly consider the question of internal and external instability, where the former entails the glitches and exploits of the sorts discussed, while external instability refers to an intrusion into the game system through an element not ordinarily understood as being part of that game system, whether something which supports the game but does so behind-the-scenes, such as computer hardware, or something entirely beyond the traditional magic circle of the game, such as a passer-by knocking over one’s chessboard. Lastly, the chapter explores how glitches and exploits can lead to the reconstitution of gameplay practices into something new, focusing once more on how players perceive and deal with unpredictability, and the discourses deployed to change what might be game-breaking mistakes into another kind of unpredictability embraced by players. Across the chapter I seek to demonstrate that instability is an important aspect of unpredictability, one which makes unpredictable games previously perceived as being deterministic, and reconstituting these games as being unpredictable; equally, in games with known unpredictability (randomness, chance or luck) instability adds new kinds of unpredictability potentially at any or all of those three ‘locations’, and in doing so changes the virtual multiplicity players perceive as underlying the game’s differential elements and differential relations.

Glitches and broken rules Let us first consider the glitch in more depth. The most common understanding of the phenomenon is that ‘glitches are software errors, either programming bugs or design flaws’.10 Eben G. Holmes offers a similar and compelling definition of the term, suggesting that glitches represent ‘the spectral appearance of failure in the virtual worlds of computers, cyberspace and videogames in which paradoxes and extremities of distance, geometry, velocity and shadow fold in on a single object/ surface/function in the [game] world’.11 As Thomas Apperley notes, the glitch ‘makes it explicit that complex processes are underway beneath the assumed user-friendliness of the contemporary interface’.12 These processes need not be processes of traditional unpredictability, whether processes that create randomness, chance or luck, but can be extremely complex seemingly deterministic processes, whose actual unpredictability is highlighted through the possible emergence of glitch instability. Equally, I would suggest that the use of the terms spectral and failure above both give some hint as to the role that unpredictability plays in understanding the glitch: it is passing and fleeting and might appear at any moment, and it represents a break in expected functioning that might occur suddenly when matters are otherwise proceeding as they ‘should’. As philosopher of science Ian Hacking states, the notion of normality does not denote how things

126

THE UNPREDICTABILITY OF GAMEPLAY

are, but rather how things should be: ‘The normal stands indifferently for what is typical, the unenthusiastic objective average, but it also stands for …  what shall be, our chosen destiny.’13 The use of the term ‘failure’ thereby immediately valorizes the emergence of the glitch from the perspective of the designer (and therefore undesirable), rather than from the perspective of the game-playing subject. By contrast, the definition of glitch which is most closely in line with my purpose here is that of Rosa Menkman, a leading glitch artist and art theorist, who proposes that a glitch is ‘an unexpected occurrence, unintended result, or break or disruption in a system’.14 Rather than an error or a mistake or something which should be resolved in order to return to a notion of correct functioning, this is instead taken to be a positive thing which opens the door to potential moments of unexpected creativity. This highlights their unpredictability and the attendant surprise on the part of both players and developers which is elicited when one is encountered, and the instabilities they can create in how a game is meant to be played, semantically read, thought about, engaged with or navigated. As Rainforest Scully-Blaker provocatively and accurately puts it, when a glitch is discovered ‘a wall that is implied to be impassable is shown to be a suggested boundary’;15 intended game rules cease to function as absolutes and become mere ‘suggestions’ whose implied constraints can continue to be used by the player, or can be surpassed and stepped around by players with the knowledge and the ability to execute a glitch. This knowledge and ability are not, however, easily gained; glitches by their nature are rare and ephemeral things, and rarely as easy to reproduce as ‘intended’ unpredictability in the forms of randomness, chance and luck deliberately written into a game’s fabric. As such, numerous individuals and even large communities of ‘glitch hunters’ exist, who understand the act of playing a game to be, at least in part, a question of discovering these singularities which nobody beforehand knew about. Some communities pursue glitches in the ‘sense of liking to find bugs, being motivated by the potential that a game bug might exist, and hence so motivated as to spend hours trying various quirky button and movement combinations in hopes of revealing a game bug’.16 In other but similar cases, where a glitch is discovered accidentally rather than deliberately but the replication of that glitch proves difficult, ‘the nature of glitch discoveries is such that runners will be unsure what exactly caused the new trick to occur and so the community of people who speedrun the game in question will work to figure out exactly why the glitch works or at least how to execute it reliably’.17 Finding a glitch is therefore a discovery of something truly new, that a game’s developers had no idea existed, and – if one is the first – that no other player knew existed, let alone had previously encountered or experienced. It then consequently becomes something that, once sighted, must be studied and understood and more fully comprehended, in a process akin to the learning processes that take place in any other form of

GAMES WITH INSTABILITY

127

unpredictability as explored in the previous chapters. Doing so brings back a sense of order, and can be understood as deliberately transforming a genuinely creative repetition into a generality which can be synthesized and stored and understood, and no longer brings surprise to the player. For some players a glitch is rare enough that it can be ignored, doesn’t disrupt their perception of their play and may well be fixed by a developer; for others, a new source of unpredictability is something to be embraced, for all unpredictability can lead to the generation of new experiences.

Exploits and unforeseen scenarios We now come to considering exploits, which I would understand as a particular singularity which was still unanticipated (so a game’s developers did not know the exploit existed), unintended (so developers did not want the exploit to exist), but one in which all game systems are working, in a systemic or mechanical sense and when assessed individually, correctly. Glitches, as we have seen, emerge when a game system does not function correctly, in the sense of something taking place which is a clear opposite to desired functioning: a creature supposed to only walk on the ground is now flying, a particular graphical element is inverted or invisible or miscoloured and so forth. In an exploit, by contrast, nothing of that sort is happening: the ground-dwelling creatures remain tethered to the earth, and everything displays as its designers intended, and yet something deeply unexpected has emerged which is counter to the intended play of the game. This is a case of emergence: where something unanticipated and unforeseen emerges from anticipated and designed rules and elements, but rules and elements whose possible singularities were not all foreseen. As with glitches, this causes a reconstitution of the virtual multiplicity of a game possessed by a player to include specific new singularities, and to include the very possibility of emergence and other new singularities being found later on. How, therefore, does the game-playing subject decide whether a new exploit is a singularity in keeping with their image of expected virtuality, of what they believe fits into a differentiated generative system with a particular set of rules, or rather an interloper that, although constituted from the letter of the game’s rules, is quite clearly contrary to their spirit? In somewhat more game-theoretic terms: is this discovery an exploit, or simply an unusually strong strategy? Understanding what generates these decisions allows us to understand how players perceive the unpredictability of exploits, and the social processes through which these perceptions are negotiated. In debates between exploits and strong strategies, there appear to be two kinds of social and political processes which lead to this resolution. In the first case, an instability which significantly reduces the variety of gameplay is more likely to be understood as being an exploit; by contrast,

128

THE UNPREDICTABILITY OF GAMEPLAY

an instability which makes little difference to the variety of gameplay, or which – although unintended and unanticipated – makes little difference to gameplay per se tends to be accepted as a strategy. To put this another way, does this unpredictable instability actually reduce the unknowability, whether from systemic unpredictability or the uncertainty of player actions, of what takes place in the game as played? After the release of Command & Conquer 3 (2007) it took little time for the player community to discover a strongly dominant strategy that only belonged to one faction (the ‘Scrin’). This involved the use of the ‘mastermind’ unit (able to capture a building from a distance), the ‘wormhole’ ability (creates a wormhole between two points on the map), the ‘signal transmitter’ building (allows the player to build a mothership directly above said signal transmitter), the ‘mothership’ unit (moves extremely slowly but fires a devastating weapon that causes a chain reaction across targets in close proximity) and the game mechanic that allows the player to construct buildings next to other buildings they own, regardless of their location on the map. A strategy was therefore found involving opening a wormhole between one’s base and that of one’s foe, sending a mastermind through it, having it instantly capture an enemy building, then constructing a signal transmitter in the enemy base next to one’s newly captured building that now no longer belongs to the enemy, immediately using the signal transmitter to summon a mothership, and then having it immediately target one’s own signal transmitter, which would explode – and being in the middle of the enemy base, the chain reaction would rapidly spread and decimate all the enemy buildings in sight. This strategy therefore negated the intended game balance attributed to the mothership unit – it is immensely powerful but immensely slow – by finding a way to instantly construct it within the base of one’s foe. Due to this strategy there was a period in which majority of players would select the Scrin and compete to be the first to deploy this strategy, and although there were a few tactics that would reduce the devastating impact of such a move – spacing one’s buildings further apart, for example, or defending one’s base with a range of anti-infantry defences in the hope of catching the mastermind before it could capture a building – this strategy remained the one that many players expected to win them the game. Fewer and fewer people played as the opposing two factions (‘GDI’ and ‘Nod’) until the developers stepped in and removed this dominant strategy. In Patch 1.07 (released in August 2007) this move was severely weakened by lengthening the period of time required for a mothership to successfully deploy; this gave the defending player ample time to sell their buildings near the enemy mothership, preventing this chain reaction. This encouraged the player base back towards the other two sides and significantly enlarged the variety of gameplay seen in the game, meaning the difference across matches, now that the Scrin lacked a strategy as dominant as this one. This strategy had therefore been strong enough to seriously discourage players from pursuing

GAMES WITH INSTABILITY

129

any other tactic and was thus defined as an ‘exploit’ – it was clear that the developers had not considered the existence of this singularity (although no explicit rule existed to prevent it and all units performed their intended roles correctly) and swiftly moved to remove it. By contrast, there are also examples of unintended singularities which were deemed so unimportant to gameplay – which is to say, so readily countered – that they were not categorized as exploits by their player communities. Perhaps the classic example of an exploit which can be easily countered, and has consequently not been acted against (although there are many other reasons for this as well18), is the Scholar’s Mate. In chess, the Scholar’s Mate is a simple four-move checkmate, which is to say a checkmate that takes place far faster than one might think intended, given the size of the game’s board, the number of pieces, the number of potential gameplay situations and so forth. Instabilities are therefore a kind of unpredictability which can arise even in entirely deterministic games, which is to say games with no systemic unpredictability; chess has no randomness, chance or luck, but instabilities, which is to say unintended and unanticipated singularities, can still arise. However, despite the potential power of the Scholar’s Mate, it is a simple attack which can be countered without much effort.19 This checkmate therefore does not reduce play variety in any way; it is not such a dominant exploit that competition is changed at all, and in the most amateur level of gameplay, we might reasonably suggest that competition is actually increased through the presence of this potentially simple checkmate – it is an easy-to-learn tactic that even new players can deploy, encourages players who are surprised by it to develop counters and might function as a primer for deeper study of the game. In the case of the Scholar’s Mate an unintended and unexpected singularity was discovered, but it was of little to no strategic value, and thus was not defined as an exploit that needed to be resolved. It is therefore still a strategy (albeit an almost useless one), not an exploit. A third valuable case study would be the potential collections of items one can acquire in a playthrough of The Binding of Isaac (2011). Although the game has gone through many versions, all versions have at least several hundred items that players are able to acquire to support their endeavours at navigating the game world. These range from the extremely weak – which offer a mediocre special ability, or simply boost the player’s speed or range a little bit – to the extremely strong – which transform the player’s weapon into something that damage opponents across the screen, grants significant boosts to all the player’s statistics and abilities, offers temporary invincibility and so forth. Given that there are hundreds of these items in even the most basic version of the game, it is extremely rare for any particular combination of items to be acquired in the same playthrough; this same tremendous variety, however, means that a number of combinations were clearly unintended. For example, one upgrade recharges a player’s item

130

THE UNPREDICTABILITY OF GAMEPLAY

whenever they are wounded; another item can be used to refill a player’s health. As rare a combination as any other, when acquired it gives the player an effectively infinite volume of health in the first half of the game, and then still effectively doubles a player’s health in the more challenging latter half of the game when the damage dealt by enemies increases. In many games these would be considered unacceptable exploits for trivializing the challenge intended by the game, either through tremendous defensive capabilities or tremendous offensive capabilities, or – in some cases – both. However, the variety of gameplay increased as a result of these instabilities, and with the game’s designer adhering to a particular understanding of what an aesthetic of unpredictability20 looks like – unanticipated singularities should be encouraged, not constrained – these remain unusually strong strategies, which might ordinarily have been exploits, but are not deemed to have had a negative effect on the game’s variety of gameplay, but rather have had a positive effect worthy of encouragement. We therefore have four kinds of elements in games with potential exploits. First, players have the differenciated (actualized in varying forms) elements they have previously witnessed. Secondly, players have the unseen differenciated elements which are anticipated, and which if they saw, they would consider to be differenciated as part of the same virtual content. Thirdly, there are actualized elements uncovered from a generative system, which don’t seem to relate to the same virtual content players had in their minds before, but might nevertheless be considered acceptable; and fourthly, there are actualized elements which are the same as the third version, but are considered unacceptable, for the gap between the virtual content they speak to and the space of ‘acceptable’ virtual content is too far. When this happens, the letter of the game is maintained, but the spirit of the game appears violated. In the case of the Scrin, this fourth case occurred because players expect variety in games, and do not expect differenciations to all be of a general rather than repeating nature. In the case of scholar’s mate, that one differenciated element stands out only for its lack of effectiveness and nobody minds a trivial strategy; we might not expect such a simple and potentially game-winning strategy to exist, but doesn’t do much to change the value or impact of other actualizations (the third variant listed above). In the case of The Binding of Isaac, players came to accept a model of virtuality which could actualize what would normally be considered ‘extreme’ outcomes as being within the scope of reasonable differenciation from a shared virtual source. Exploits are therefore quite different from glitches, although both take the form of singularities. In a formal sense, as I argued above, glitches emerge from a flaw in the implementation of an aspect of a generative system, whereas exploits emerge between generative systems which might be combined in a way that affects the game to a far greater extent than previously anticipated. In both cases, therefore, unpredictability can emerge from these sources, but whereas the contestations over randomness, chance

GAMES WITH INSTABILITY

131

and luck we have thus far seen (and almost all the others we will see later in this work) involve understanding or navigating or accepting or generating a particular discourse about unpredictability, here the question is over the acceptance of these kinds of unpredictability being in the game at all, and therefore whether they represent positive or negative unpredictabilities in player experience.

Tight coupling and weak coupling All the examples thus far used of glitches and exploits that were anything other than minor (as opposed to, say, the Scholar’s Mate) have been in video games. At this point I would like to propose the concepts of tight coupling and weak coupling for addressing why some game systems are more prone to instability than others – a trend that can clearly be seen across the history of games as a whole, and most precisely across the history of video games – and in turn how this coupling of gameplay systems and gameplay elements (in this case Deleuze’s differential elements and differential relations) influences player experience. Taking a broad view of the history of games of all forms, we can readily perceive an increase in the complexity of games as time has progressed. This can be broken down into the complexity (or number) of relations and the complexity (or number) of elements. In the first case, the earliest games known are believed or known to have had relatively simple rules. Many are race games, taking the form of a progression around or across a board which might be interrupted, challenged or delayed by other rules. Such games were played in the ancient Middle East, in Buddhist monasteries, in Mesoamerica and beyond, illustrating the appeal of quite simple relations between the player’s actions, generally the rolling of dice or knucklebones or equivalents, and movement on the board. Equally, most traditional board games display a small number of relations, most clearly in games like draughts, and while a game such as chess has a greater number of relations (moving, capturing, promoting) and some special unusual relations between pieces (castling and en passant), the overall number of potential relations, which is to say rules, remains small and generally singular in nature. In chess, pieces can only move, and capture, although they do so in different ways. Over the last few centuries, war games of tremendous complexity have gradually evolved,21 leading the charge towards both a greater volume of rules and a greater complexity of those rules (which is to say more special contexts, more variation between pieces, more exceptions, more possible interactions). At their most extreme such games required dozens or even hundreds of pages of rules to be played, and as many pages that denoted the specific interactions, such as the damage dealt by a certain weapon wielded by a certain soldier firing in a certain way, that game pieces could engage in. Most recently, video games have been the source of

132

THE UNPREDICTABILITY OF GAMEPLAY

a tremendous expansion in the number of rule categories, many of which would be hard, impossible or unnecessarily time-consuming to model or implement without digital technology, and consequently also an expansion in the numbers and kinds of interactions between game systems. From the interactions of simulated physical laws to the choices of artificial intelligence actors, and from the random spray of bullets to the modelling of weather and other natural processes, relations take many different forms, can be between asymmetrical kinds of elements and emerge unexpectedly from the interactions between those forms. In terms of game elements, meanwhile, many of the earliest games contained a handful of elements, such as game pieces, game boards and dice or their equivalent, with some requiring only a die to play. In turn, these were generally symmetric – all game pieces were of the same form, all players controlled the same kinds of pieces, there were no pieces that affected some parts of the game world but not others and so on. As games have developed into the complexity of war games and other board games, into card games, and subsequently into video games, asymmetrical game elements of many kinds have been differentiated from another. These include game elements or ‘pieces’ that can heal, shoot, move, capture, expand, upgrade, be sold, combine, fragment, climb, change form, change shape, exchange, fly, walk, swim, drive, be opened, be closed and so on, or enable or deny those abilities or any combination of them to other ingame actors. The earliest of video games released on home computer were rarely so weakly coupled that glitches and exploits would readily emerge once interacted with by players; without the ability to download patches or bug fixes, these games had to function correctly first time, and in almost all cases, they did. By contrast, it is rare to encounter a modern computer game (with its far greater complexity) of all but the smallest size or simplest nature that does not soon have glitches and/or exploits found by its players, and in some cases, patches or bug fixes are deemed necessary to resolve these. With a greater number of differential elements undergoing a greater number of differential relations, these more modern games are more weakly coupled than their antecedent cousins, and consequently possess a greater possibility for instability. As such: tight coupling refers to games with few rules, simple rules, few elements and simple elements; weak coupling is the opposite. Tightly coupled games can produce a smaller number of singularities than weakly coupled games, but a greater percentage of singularities produced, potentially up to 100 per cent, will have been anticipated by a game’s designers and will not surprise a player who is developing a concept of the game’s multiplicity which is partly built, but still incomplete. By contrast, weakly coupled games allow for unexpected singularities in the form of both glitches and exploits; the more complex each element becomes and the more other elements or a game it interacts with, the higher chance of something malfunctioning within that element (causing a glitch) and the greater the number and variety

GAMES WITH INSTABILITY

133

of differential relations, the higher the chance of something emerging from those rules which seems to go against the game the rules are meant to be constructing (causing an exploit). As Levi Bryant puts it, ‘You can’t discern singularities simply by looking at things, …  you have to interact with things to get a sense of what singularities are,’22 and these interactions can, in weakly coupled games, allow us to discern singularities that specifically could not be seen only by looking at the game. Understanding strong and weak coupling lets us see in greater detail the contingencies and necessities for the generation of instability, and therefore what kinds of games are more or less likely to have them, and what games will be more likely to give players the ambiguity and contested experiences explored in this chapter.

Internal and external instability A further conceptual divide is also necessary between internal and external instability. Internal instability is what has been examined in all the previous sections of this chapter, and will be examined in the rest of this chapter after this section, and all the subsequent chapters after this one. Internal instability takes place when an unpredictable gameplay mechanic or gameplay ability arises within a game, through the emergence of glitches or exploits. Both of these arise from the unintended functioning of intended gameplay systems, and this form of unpredictability is therefore contained within the technical structure of the game artefact. However, we must also acknowledge the existence of a second class of instability which can disrupt expected forms of gameplay in unpredictable ways, which I term external instability. External instability takes place when something outside the game’s system causes a change within the game system, and causes the intended gameplay systems to break down or become altered in some way. For example, a chess piece might be missing from a chess set, or might be knocked from the board and players are subsequently unable to recall its precise location before this disruption; a game controller might be broken or cease functioning correctly, or a problem with the underlying computer hardware might cause lag (the unpredictable and irregular slowing of in-game actions) to affect gameplay; cards might become marked intentionally or unintentionally, reducing the integrity of the incomplete-information component of the game. There is a key difference in the range of external instabilities that can occur in digital and analogue games. In analogue games, as noted in this chapter, it is very simple for a piece to be removed or for the play of the game to be clearly disrupted. The overwhelming majority of such games contain elements that are materially disparate, which is to say one element can be removed or damaged or changed without the other elements undergoing removal or damage or change. Instead of being digital elements within one material infrastructure, as in a video game running off a particular computer

134

THE UNPREDICTABILITY OF GAMEPLAY

or games console, analogue games consist of material elements lacking any broader overarching infrastructure. Nothing tethers together chess pieces, Go pieces, playing cards and the like; one aspect can be removed while leaving the other aspects intact, even if the structure of the game which is contingent upon all aspects being present might be damaged by the removal of just one. By contrast, all of a video game runs on the same physical hardware substrate, of circuitry, programs, electricity, drives and cards, controllers and so forth. Any of these could malfunction, ranging from one of the buttons on one’s mouse breaking to crucial system files that construct the game going missing. In this regard, video games are likely to display binary external instability – the game functions on the platform it is being played on, or it doesn’t, or something which cuts across the entire game malfunctions (e.g. the ability to press ‘X’ to interact with things) as opposed to one single part of the game malfunctioning. By contrast, in physical games where the elements of gameplay are not tethered to one another through being simulated within the same system, it is easy for one card, or one piece to go missing; or for a disgruntled player to disrupt the placement of pieces deliberately, unhappy with their fortunes in the game. In cases of external instability, we therefore see the non-game world beginning to intrude on the game world in various ways, causing unpredictable effects on how play continues, and whether or not play can continue at all. This means the expanding of a player’s appreciation of game’s multiplicity, which is to say many more relations and elements are conceptually part of the same multiplicity. Recall the earlier discussion in Chapter 1 with regard to the infinite series of differences between all repetitions: of course, even an enlarged multiplicity does not come close to encompassing all elements that might actually affect the play of a game, but comes closer, and can be triggered by the player’s experience of external instability. The impacts of these additional virtual rules and contingencies (in this now-enlarged multiplicity) are very rarely actualized, as the number of board or card games which proceed without an external hitch far exceeds the number that are disrupted through the intrusion of some external factor, or through a problem with the ‘set’ of game pieces, but they are nevertheless sometimes actualized. These moments show the multiplicity of the game to be much broader and more varied than initially and traditionally understood. Interference from outside the game as a form of instability and therefore unpredictability – somebody knocks over the chessboard, one’s computer crashes, the football is accidentally kicked over the fence, a card is missing from the deck – are not issues in a game’s design,23 but rather issues in the immediate sociotechnical environment within which a given instance of play is embedded. Once these external instabilities occur and the game is consequently rendered unpredictable, we see the emergence of game-playing practices which entail ‘work-arounds’ for dealing with this external instability, and

GAMES WITH INSTABILITY

135

either accepting it and adjusting the game or attempting to return to the game to its previous functioning. Such fixes might entail goal posts out of shirts in a game of football, recruiting another object to serve as a piece in a board game, creating one’s own playing cards when official playing cards are missing, finding another way to trigger an input on a game controller with a missing button and so forth. Having grown up on home computers rather than the generation of consoles that would be appropriate to my age, I have a clear recollection of developing methods to overcome external instability. If a game came without some of the security apparatus (such as code wheels24) required to put in the right code to play the game, which therefore reduced the potential gameplay only to moments where the right code was correctly guessed, I would photocopy a copy from a friend or create my own version from paper, cardboard and tape. If the tape player for the home computer lacked a button, resulting in the cassette sometimes failing to play correctly, a pen would be jammed into the slot to duplicate the action of depressing the button. All of these were ways of attempting to reconstitute the intended gameplay of the game in a situation where external elements were impacting upon the games. In many cases, therefore, these fixes are designed to return the game to almost exactly its intended state in pure gameplay terms – the primary purpose of a goal-post is to denote the boundaries of the space in which goals can be scored, and in this regard piles of shirts function almost as well as metal rods (for anything outside of a formal, competitive context, where precision is crucial). On the other hand, creating one’s own playing cards will be more problematic if the created cards differ in appearance from ‘official’ cards and thereby make some cards identifiable from their backs. Making physical alterations to a controller might restore intended functioning, but might also result in the inputs transferred from controller to computer being changed to a sufficient extent that the game becomes easier or harder than intended, or the player is able to otherwise achieve inputs that are supposed to be impossible. This attempted reconstitution of the previous gameplay, to keep the multiplicity as the accumulation of differential elements and relations it originally seemed to be, is therefore another distinct way that players navigate the unpredictability of gameplay, yet also simultaneously a way in which gameplay is made more unpredictable than it originally was. Ergo, external instability represents another element of unpredictability in games, as a subset of instability. Although at first glance this instability appears not to be a systemic form of unpredictability, which is how I have defined the term ‘unpredictability’ in this work (as opposed to the uncertainty of an opponent’s decisions), it is. However, understanding it as such requires us to expand what we think of as being the game’s system; the value of such an expansion is made visible through disruptions or breaks in the fabric of the game, which take a different form to either glitches or exploits, and might therefore be appropriately termed ruptures in the game. This work

136

THE UNPREDICTABILITY OF GAMEPLAY

is concerned with unpredictability in the systems of play and how they are perceived by those who engage with them, and in external instability we see these systems being expanded to include the integrity of the physical artefacts of play themselves. Generative systems are larger, multiplicities must therefore be expanded, and singularities which affect the game are not just present within the ‘game’ as traditionally understood, but in broader networks which simultaneously implicate the ‘game’ and its ‘context’ in one, breaking down traditional divides based on the magic circle and enlarging the range of actors and contingencies that might affect gameplay. There appears to be differentiation between that which is within and without the magic circle, but these moments show that differentiation as being (of course) a subjective notion that can be easily upset; both elements are part of a wider game multiplicity that can have an unpredictable effect on the progression of play and significantly affect player experience. External instability, therefore, is a part of the systemic and experiential unpredictability of gameplay, and can shape player experiences in ways quite unlike any other kind of unpredictability explored in this work.

Subverting and reconstituting the rules of play To conclude this chapter, we now examine how an internal instability in a game can come to be regarded not as a flaw, an error or a problem in gameplay or game design, but rather as laying the foundation for an entirely new form of gameplay. We have thus far been pointed towards two models of virtuality, which I term ‘discovered’ virtuality – where actualized elements in a playthrough thus far experienced point a player towards some part of the virtual content of that game – and ‘expected’ virtuality, where one has not explicitly seen something which demonstrates the presence of particular virtual content, but actualized elements imply elements of virtuality. However, we have also seen that glitches and exploits are not simply seen as mistakes, as divergences from expected virtuality too extreme to be acceptable; communities of glitch hunters and political contests over exploits both show strongly the willingness of players to adjust their mental models and ‘capture’ instabilities as acceptable kinds of unpredictability, in keeping with the spirit of a new, adjusted, game multiplicity. How precisely does this experience of changing one’s expectations of play manifest? Ordinarily in software engineering, the response to the discovery of a glitch is to eliminate that glitch;25 this might happen during the initial development of a piece of software, or if a glitch is not discovered until after the software has been released, then a sufficiently severe glitch will ordinarily be fixed using a ‘patch’ – a small, generally downloaded change to the program’s code which removes, resolves or works around the glitch in question. However, sometimes these factors outside expected virtuality

GAMES WITH INSTABILITY

137

do not remain outside that expectation, but are instead embraced and the virtual content of the ‘game’ proper, the spirit of the game, is enlarged. There are numerous potential case studies we might draw upon to illustrate this reconstitution of gameplay rules through unpredicted instability, but here I will consider two of the well-known examples. In the first case we will consider a community of players who have used glitches to change the acceptable rules of gameplay – that of ‘speedrunners’, who we previously visited in Chapter 4 – for which I focus on consideration of The Legend of Zelda: The Wind Waker and its remastered version, Wind Waker HD (2013). In the second case we will consider a scenario where an exploit has changed the acceptable rules of gameplay, both in the games where it originally arose, and consequently actually came to be written into games – the practice of ‘rocket jumping’. These two examples will also serve a second goal: they will show us how, even after players accept glitches and exploits as either universal stabilities or what I will term metastabilities of embraced unpredictability appropriate to particular communities, the story does not always end there; the agendas of game developers can come into play, and affect how these kinds of unpredictability remain available, or not, for players to experience. Until this chapter we have always examined unpredictability which is within the ‘spirit’ of the game – either explicitly known to the player or which appears when encountered to clearly be within that spirit. These case studies help us to understand how players respond when they encounter unpredictability beyond the spirit of the game; in some cases it can be fought against or rejected or fixed, but in other cases it is the spirit, not the technical fabric of the game, which shifts.

Speedrunning Glitches are essential to the practice of speedrunning. We have previously seen that speedrunning involves the most rapid possible completion of a game or a portion of a game, but while it is sometimes played according to the intended rules of that game, it can sometimes also become a space of play in which ‘“normal” play is virtually absent’.26 It first emerged as an ‘extra-official form of competition’27 through which the gameplay ability of multiple players could be compared, both in games which only have singleplayer modes – thereby lacking any intended multiplayer component – and in the singleplayer modes of games which did have multiplayer components, but which seemingly did not meet all the requirements of competitively minded players. As Thomas Smith, Marianna Obrist and Peter Wright note, speedrunning began ‘simply as acquiring sufficient skill to play a particular game as quickly as possible but now encompasses exploiting bugs in a game to skip as much game content as possible’28 – speedrunning therefore emerged out of skilled play of games in their stable forms, but later came to embrace the glitch-based

138

THE UNPREDICTABILITY OF GAMEPLAY

instabilities in those same games. Scully-Blaker draws a similar contrast between ‘finesse’ runs (in a stable game) and ‘deconstructive’ runs, where the game’s instability is deemed to be beneficial to the act of speedrunning, and the game’s instability is deconstructed until it is subsequently reconstructed as a new, speedrunning-specific, oasis of metastability.29 Metastabilities emerge from instabilities and represent new kinds of normal gameplay; if they are accepted and come to resemble ordinary gameplay for many or all of a game’s players, they form a new stability, for some players, or for all players; if they are rejected as cheating or illegitimate, they become a mere instability which might be fixed by a game’s designers in the future, or are otherwise rejected and considered unacceptable by a game’s players. Those which remain contested – such as an exploit in a multiplayer game which some players use and accept as a new strategy, while others reject it as cheating and complain to the game’s developers – remains a metastability, in flux and consistently politically contested, until it is fixed (and becomes a past, passing, forgotten instability) or accepted (becomes a new, innovative, accepted stability). When a game becomes stable once more, that stability is not the stability intended, imagined or envisaged by the game’s designer, but is nevertheless a new stability. For players who accept it, this represents the spirit of a new, better game. A metastability is therefore a new way of playing that seeks to establish itself as a stability, but remains contested. In this way using glitches has ‘become written into the standard lexicon of gameplay’ in speedrunning30 for some players, and glitches are seen by those who accept them as little different to any other kind of randomness, chance or luck, depending on the placement of the glitch in question. However, the perspective on unpredictability from the game-playing subject is not the only perspective which affects player experience in situations of this sort. As such, although the focus of this work is on the subjectivity of the player and examining the politics of game development might appear to be something of a tangent, the roles of game developers are central to player experiences of glitches and exploits when those glitches are anything other than simply rejected. What is perhaps the ideal case study of developer responses to glitches (we will address exploits shortly) is offered by Rainforest Scully-Blaker. In The Legend of Zelda: The Wind Waker, the player explores a vast world consisting of islands dotted across a large ocean, many of which contain buildings, underground areas and are much larger in terms of the space which can be explored than their small size on the world map might suggest. For much of the game, moving between these islands involves the player’s character, Link, sailing from one to another in his boat. Sailing is a relatively time-consuming process given that it is presented as the transition from one central ‘node’ of gameplay to another, with it normally taking around five minutes to shift from one island to another. The game’s developers originally attempted to instil interest in this process by dotting the open ocean with a range of enemies of various sorts,

GAMES WITH INSTABILITY

139

short challenges such as collecting gems floating on the ocean’s surface, along with a small number of rare but very striking and compelling semiunpredictable ‘encounters’, such as the player’s ability to meet a ‘ghost ship’ and a number of hostile giant kraken-esque sea monsters. Nevertheless, many ‘traditional’ players who played within the intended stability of the game’s designed rules found sailing to be quickly tedious, and those who speedrun games and thrive on the speed and rapidity of play found it even worse.31 Seeking a way around this apparent problem, speedrunners and glitch hunters soon discovered the ability to ‘super swim’ – this was a glitch which entailed setting Link up in a particular state, dropping into the ocean and then being ‘catapulted’ across the ocean at tremendous speed. When performed correctly, moving between islands no longer takes minutes, but only seconds – perhaps fifteen seconds for the setup, and another ten seconds for Link’s rapid transit across the open sea. It is unclear whether or not Nintendo were aware of this glitch, but in the later release of Wind Waker HD the errors in the game’s code which enabled super swimming had been removed – and a new item, the ‘Super Sail’, added, which sped up the transit across the game’s oceans (although it was still slower than the use of the glitch). As Scully-Blaker notes, ‘Many in the [speedrunning] community were unhappy with the decision to remove what was perceived as a glitch so obscure it could only matter to speedrunners.’32 In essence, this claim does two things with regard to questions of stability: first, it acknowledges that super swimming currently represents, and will always represent, a metastability which is accepted by some of a game’s players but not all, and is certainly not universal enough to represent a new stability in Wind Waker as a whole. Indeed, it is not even a stability for speedrunners, since glitch-free runs continue to appeal to some players, although these players are rare. This is especially the case in The Wind Waker, primarily because of the slowness of the central sailing mechanic. Secondly, that metastabilities can exist without disrupting the play of anyone else who doesn’t seek out those metastabilities; they are minor enough, limited to a small enough community, and hard enough to accidentally stumble across, that they pose no ‘threat’ to the game’s intended stability. A glitch of this sort, speedrunners claim, is only beneficial to them; it is not detrimental to others. Something beyond the spirit of the original game was embraced, but a game’s developers returned the game to their original notion of its spirit. This has little effect on the experiences of those who reject glitches and either ignored them when they occurred, or simply waited for a fix to be deployed, but has a significant effect on the ability of glitch-embracing players to actually continue to experience the unpredictability which they consider to be part of the game. The available unpredictability in a game can thus shift and change in the presence of instability, and the players are not the only actors with interest in the outcomes.

140

THE UNPREDICTABILITY OF GAMEPLAY

Rocket jumping In terms of exploits, the case study of the practice of rocket jumping is illustrative. Rocket jumping, as with all exploits, occurs at the intersection between several gameplay mechanics. The first of these was the physical push-back effect associated with explosions in early first-person shooter games. Explosive weapons, such as ‘rocket launchers’, would both deal damage to the player and physically move the player away from the direction of the explosion; this was an attempt to design a sense of the force as well as the damage produced by an explosive detonation. Under normal gameplay circumstances, therefore, the push-back effect of explosions would serve to perhaps knock players off walkways they were previously on, send them away from their opponents, and generally function as a second-level type of ‘damage’ – strategic, not directly in terms of health – dealt to the player at the receiving end. The second important mechanic was the presence of the ability to jump in first-person shooter games; jumping would produce an effort similar to the pushback, except directed and determined by the player performing the jump, allowing them to cross short gaps, move upwards towards areas that couldn’t simply be walked to, and generally navigate the in-game space more rapidly, smoothly and skilfully than merely running would allow. Rocket jumping varies just slightly from game to game in the precise application of these mechanics, but broadly speaking it entails jumping and launching a rocket at one’s feet at the same time; one’s character will take damage from the explosion of the rocket, but the force of the jump, combined with the force of the rocket’s explosion, can propel the player to heights quite unintended in these earliest FPS games, allow the player to cross wide gaps, and in a broader sense can transform player navigation of the virtual space. Although again unintended, the application of damage at the same time as the push-back effect meant that rocket jumping became not just immediately strategically essential when discovered, due to the new abilities it offered players, but it was also something with benefits that had to be weighed and considered against its downsides: which is to say, hurting one’s own character (and expending one shot of a powerful weapon which could have been turned against the enemy, although this is generally understood to be a lesser concern). Having originally emerged as an exploit, rocket jumping (and later grenade-jumping) was subsequently deliberately implemented into a tremendous range of future FPS games; the distinctive, compelling and highly skilled gameplay they offered were seen by developers as a positive, rather than negative, development. In this case players were able to continue experiencing the unpredictability they had come to regard as a standard part of the game, although it had started out life as a clear challenge to the intended spirit of these first-person shooters.

GAMES WITH INSTABILITY

141

The sociology of design helps us to conceptualize these communityaccepted changes in ‘correct’ gameplay practice that can emerge in games with instability. The sociology of design has demonstrated that the act of designing anything must also entail both making predictions about its intended use and inscribing those predictions into the design of the final product.33 This means that the users’ behaviours are anticipated34 by a product’s designers, envisioning future use and how users will address and respond to the technical affordances of the eventual material artefact. However, there is a ‘persistent narrowness’35 in the conceptualizations of user abilities and user desires when imagined by designers, who rarely anticipate either users wanting to use a technology in a way unlike they imagined, nor anticipate users having the ability to subvert their intended forms of use. Such subversion of the intended uses of technologies can take many forms, and the notion of the ‘technological frame’ from science and technology studies helps us further in thinking through this reconstitution of gameplay rules as a result of exploit-created instability and unpredictability. A technological frame consists of the sum total of all ‘meanings of technical artefacts’36 as well as all elements which influence behaviour in social groups who use that technology, or have some broader social stake in that technology, such as being inventors, maintainers and the like. Such a frame is the manner in which a social group interprets a particular material artefact,37 and can differ significantly across multiple groups when it comes to the same artefact. Obvious examples can be seen in highly public contestations over the value of certain technologies – such as nuclear power, drone warfare, genetic modification and other explicitly ‘controversial’ technologies – but is applicable to all technologies and technological artefacts, however minor, subtle or limited in their impact and social embedding. Now consider two individuals who watch a speedrun video with numerous glitches, or a competitive match of a game with rocket jumping. Where one individual sees original and clever uses of game mechanics, the deep unpicking of how the game functions and fascinating redefinitions of how the game is played, another player will only see someone ‘breaking’ these games, failing to play them ‘properly’, and therefore a speedrun or competition which is not truly completing or contesting the ‘game’, as they understand it. In this way two groups with different frames for the same artefact can consider it to be functioning ‘correctly’ and functioning ‘incorrectly’, while the actions or affordances of the artefact in a purely technical sense remain unchanged. There is therefore no inherent or internal blueprint present in how artefacts are used and the sets of meanings assigned to artefacts. In Deleuzean terms, technologies do not come with their own identities, but these identities are instead constructed through the differences in their use; a functioning and non-functioning technology is not an identity that already exists and players simple consume, but one created through the different ways people use them, and the different ways one

142

THE UNPREDICTABILITY OF GAMEPLAY

social group might regard how another social group elects to use the same technology. As argued earlier in this chapter, the ordinary spirit of the game is not pre-formed but is constructed through play and the synthesis of past and present events experienced by the player; in most cases the spirit of the game perceived by players matches up pretty well with the spirit of the game imagined by its designers. However, in the case of glitches and exploits being a source of gameplay redefinition, we see a form of what Johan Redströ m terms ‘design-after-design’,38 responding to instability by embracing, rather than rejecting, the new unpredictability it brings to the experiences of the game-playing subject. These practices take what was originally a form of instability and therefore unpredictability, and reconstitute it as not just something which is not an instability, but in many cases something which is not even unpredictable – rocket jumping, in particular, tends to actually be deterministic, despite its origins as a source of unpredictability. This chapter has been concerned with exploring glitches and exploits as forms of unpredictability, identifying games whose rules are ‘tightly coupled’ (extremely stable) and ‘weakly coupled’ (unstable), the differences between internal and external instability and therefore what formal characteristics give games these properties, and exploring a body of player experiences that actively embrace moments of ludic instability as a route to creating and defining new gameplay rules and expanding the spirit of a game to involve new forms of play which are predicated on the unpredictable emergence of extra gameplay possibilities. Whereas the core three-part typology of this work entails forms of unpredictability which are within the spirit of the game in question, instability represents unpredictable moments, mechanics and possibilities that are outside the spirit of the game as initially understood, which can lead to a range of different kinds of political contests over whether play should be continued, halted, stopped, reformed or returned to its original course. Such responses to instability represent another kind of negotiation over game unpredictability, and instability is consequently an important element in our broader consideration of player experience of unpredictable games. Having now identified the four different locations of unpredictability, at the start, middle, end of gameplay and inside or outside the spirit of gameplay, we now turn to three case studies of unpredictability. These will allow us to apply these more theoretical constructs to real-world case studies, demonstrating both the applicability of the frameworks and the tremendous range of methods players use to navigate unpredictable games, each predicated on the deep differences between different locations of unpredictability and their effects upon players.

Notes 1 Williams, Critical Introduction, 96. 2 Deleuze, Logic of Sense, 61.

GAMES WITH INSTABILITY

143

3 Bernard Suits, The Grasshopper: Games, Life and Utopia (Toronto, London: University of Toronto Press, 1978), 22–3. 4 Suits, The Grasshopper, 27. 5 Ibid., 31. 6 Ibid., 32. 7 A player’s overall experience with games as an activity, and particular genres of games, will also shape the player’s idea of the spirit of a game. The adherence to the rules and the notion of the game those rules are meant to construct are not just limited to an individual game, but exist in a dialogue with all games in broader differentiated genres (‘first-person shooter’, ‘ball game’, ‘racing game’, ‘tabletop RPG’) which share some common notions of rules, and of ends. 8 Deleuze, Difference and Repetition, 6. 9 Deleuze, Logic of Sense, 63. 10 Wilma Alice Bainbridge and William Sims Bainbridge, ‘Creative uses of software errors: Glitches and cheats’. Social Science Computer Review 25, no. 1 (2007), 62. 11 Eben G. Holmes, ‘Strange reality: On glitches and uncanny play’. Eludamos: Journal for Computer Game Culture 4, no. 2 (2010), 256. 12 Thomas H. Apperley, ‘The body of the gamer: Game art and gestural excess’. Digital Creativity 24, no. 2 (2013), 147. 13 Ian Hacking, The Taming of Chance (Cambridge: Cambridge University Press, 1990), 169. 14 Rosa Menkman, The Glitch Moment(um) (Amsterdam: Institute of Network Cultures, 2011), 18. 15 Scully-Blaker, ‘Re-curating the accident’, 58. 16 Chris Lewis, Jim Whitehead and Noah Wardrip-Fruin, ‘What went wrong: A taxonomy of video game bugs’. In Proceedings of the Fifth International Conference on the Foundations of Digital Games (ACM, 2010), 108. 17 Scully-Blaker, ‘Re-curating the accident’, 26. 18 Most obviously, chess does not have any explicit game developers or game designers who could arbitrate, from a central source, on changes to the game’s design. Nevertheless, chess has evolved significantly from its original forms, and it is not unreasonable to suppose that if an exploit of this sort was somehow only now discovered – a simple checkmate which trivializes the game, or an equivalent – chess’ global community would likely adapt to find a way to adjust the game and preserve the possibility for deep and meaningful competition. 19 David Hooper and Kenneth Whyld, The Oxford Companion to Chess (Oxford: Oxford University Press, 1992). 20 For more discussion of this element, see ‘Identifying Procedurality’ in Chapter 6.

144

THE UNPREDICTABILITY OF GAMEPLAY

21 For a comprehensive examination, see Jon Peterson, Playing at the World: A History of Simulating Wars, People and Fantastic Adventures, from Chess to Role-Playing Games (San Diego, CA: Unreason Press, 2012). 22 Levi R. Bryant, ‘What are singularities?’. Larval Subjects, 14 June 2012. Available at https://larvalsubjects.wordpress.com/2012/06/14/wh at-are-singularities/. 23 With this said, it is not impossible to construct an argument that, in fact, ensuring that external instability cannot occur is an element of good game design. One might argue that a well-designed online game should contain ways of working around potential instabilities, such as the ability to use one server if another server goes down, the ability to use multiple kinds of hardware and so forth; a deck of cards should perhaps be difficult to bend (and thereby mark) and one might even argue they should be visually distinct and striking to reduce the chance of them being lost or misplaced; magnetic chess boards exist precisely to reduce the chance of pieces being dislodged from the board and therefore upsetting play; and so forth. Although it is of course impossible to anticipate every external intrusion that might render a game unstable, it is not impossible to anticipate some of the more obvious potential intrusions, given the material specifics of the game and its context(s) of play, and take steps in the game’s design to prevent or mitigate these. 24 ‘Code wheels’ were circular items usually made out of paper or cardboard, which consisted of several moving parts in the form of overlapping circles, some of which would have holes in them allowing the user to look through one layer onto the next. In order to access a game, the game would pose a question on screen, of the sort ‘When X, Y and Z are aligned, what does A read?’. The player would have to align the X, Y and Z elements on different parts of the code wheel – which might be shapes, numbers, colours, faces, phrases, visual representations of in-game elements – and then look at what appears in the ‘A’ slot of the code wheel; this would usually be a number or a word. Inputting that number or word into the game would then allow the player access to the game; failing to input, or inputting an incorrect response, would prevent the game from loading any further. For a deeper historical look at the code wheel phenomenon, see Nerdlypleasures, ‘Exposing the code wheels – PC game document copy protection at its “most advanced”’. Nerdlypleasures, 29 March 2013. Available at http://nerdlypleasures.blogspot. co.uk/2013/03/exposing-code-wheels-pc-game-document.html. 25 Bainbridge and Bainbridge, ‘Software errors’, 61. 26 Seb Franklin, ‘On game art, circuit bending and speedrunning as counter practice: “Hard” and “soft” nonexistence’. CTheory (2009). 27 Gabriel Menotti, ‘Videorec as gameplay: Recording playthroughs and video game engagement’. G|A|M|E 1, no. 3 (2014), 81–92. 28 Thomas Smith, Marianna Obrist and Peter Wright, ‘Live-streaming changes the (video) game’. In Proceedings of the 11th European Conference on Interactive TV and Video (ACM, 2013).

GAMES WITH INSTABILITY

145

29 Rainforest Scully-Blaker, ‘A practiced practice: Speedrunning through space with de Certeau and Virilio’. Game Studies 14, no. 1 (2014). 30 James Newman, Best Before: Videogames, Supersession and Obsolescence (Oxford and New York: Routledge, 2012), 150. 31 Scully-Blaker, ‘Re-curating the accident’. 32 Ibid., 83. 33 Johan Redströ m, ‘RE: Definitions of use’. Design Studies 29, no. 4 (2008), 410–23. 34 Dean Nieusma, ‘Alternative design scholarship: Working toward appropriate design’. Design Issues 20, no. 3 (2004), 14. 35 Nieusma, ‘Alternative design’, 17. 36 Wiebe E. Bijker, Of Bicycles, Bakelites, and Bulbs: Toward a Theory of Sociotechnical Change (Cambridge, MA: MIT Press, 1995), 123. 37 Wiebe E. Bijker, ‘Do not despair: There is life after constructivism’. Science, Technology & Human Values 18, no. 1 (1993), 117. 38 Redströ m, ‘RE: Definitions of use’, 410–23.

146

PART TWO

148

6 Procedural Generation

We now proceed to three case studies of the three main elements of the typology presented here – randomness, chance and luck. Each of these case studies is designed to apply in practice the theoretical framework developed here, to show the wide variety of forms that player experience of unpredictability can take depending on the implementation of unpredictability, and therefore demonstrate the importance of understanding the different forms of unpredictability outlined here, and consequently the applicability of this theorization to a wide range of ludic domains. The first chapter in this part explores the concept of procedural generation, with the goal of examining some of the completely (or mostly) unique experiences that playing PCG games can bring to a player. Procedural (content) generation, also known as PCG for short, is the overall term for a suite of techniques by which video games are able to highly vary the actualized elements of a game – levels and areas, the placement of enemies, the placement of items and so forth – that a player experiences. Long a staple of the ‘roguelike’ genre of games, which we explore in detail as the exemplary case study of this chapter, PCG has recently become used by a greater range of game designs and game developers than ever before, enticed by its potential for creating endlessly new scenarios and keeping players playing these scenarios over long periods of time. The next case study chapter applies the theorizations of randomness and chance to unpicking the concepts of replay value and grinding, while the final chapter in this part applies luck to help us study a collection of player practices focused on the negation of unpredictability – which is to say, playing games with unpredictability in their designs, and then seeking to remove or overcome that same unpredictability. Taken collectively, these chapters are designed to both offer new analyses of these three sets of design elements and the player practices they can inculcate, and demonstrate the analytic and conceptual value in separating intended unpredictability into these four discrete components – randomness, chance, luck and instability – or as I term them, locations.

150

THE UNPREDICTABILITY OF GAMEPLAY

PCG has been an element of game design for decades. In brief, it involves the creation of game content or assets through the use of an algorithm or a set of algorithms, instead of being handmade by a human designer. For example, ordinarily a game designer might place a range of trees as part of an outdoor environment. By using PCG, they would instead create a set of rules – no two trees can be closer than A, trees cannot be close to B, trees tend to grow near C, the total volume of trees should be no more than D and so forth – and then allow a computer system to select appropriate locations for the placement of trees, and consequently insert trees in those locations. This can take place either when the game is being designed – in which case all players experience the same content, but that content was originally created by algorithm rather than by hand – or (in almost all cases) anew while the game is being played, in which case certain elements of the game vary every time a player plays through them. From the player’s perspective, therefore, the first use of PCG is not actually a form of unpredictability; it is instead just a different process for producing game content which is then distributed, identically and unchanged, to all players. In the second and far more common case, however, the player experiences the unpredictability because the placement of the trees (or whatever else) varies every time the game is played, rather than taking place once before any players get access to the game, and it is naturally this implementation on which we focus. PCG of this sort is used for a range of design goals, three of the most prominent of which are ‘ensuring replay value and prolonging novelty’, maximizing the possibility for the consistent production of challenging and interesting gameplay mechanics and creating game mechanics focused on the concept of exploration.1 PCG is able to provide a new or semi-new experience each time a player begins a new session of play, keeping games – particularly singleplayer games without a multiplayer component pitching players against other unpredictable humans – fresh for longer; these games constantly pose new and never-before-seen gameplay challenges, encouraging players to develop what Daniel Cook calls mastery over the game’s systems and sets of variables no matter how they come out in a particular permutation; and it can create unpredictable worlds of tremendous size and variety, leading to the creation of gameplay focused on the experience of simply exploring these worlds, and the sometimes-unusual and ‘unrealistic’ procedural architectures worldbuilding algorithms can sometimes spawn. Given both these range of purposes and the growth in use of the technique (as we explore in this chapter), PCG is therefore a central tool for the implementation of unpredictability in specifically video games, creating unprecedented levels of unpredicted novelty, unpredicted challenges and unpredicted game spaces to experience, which are of a sort found nowhere else. It also allows us to develop new terminology for considering in greater depth how players see unpredictable systems, a body of experiences which take on distinct forms in games with procedural generation.

PROCEDURAL GENERATION

151

The purpose of this chapter is therefore to address in detail the unique experiences produced by PCG, and develop a vocabulary with which to talk about them. The chapter begins with an exploration of the roguelike genre, the game genre that, by quite a margin, has made the most out of procedural generation systems. I briefly summarize the history of roguelikes and their milestones where relevant – especially in terms of the first games to procedurally generate game elements that had never, or incompletely, been generated before – and focus particularly upon the diverse range of gameplay purposes to which procedural generation and its attendant unpredictability has been put. The chapter then uses roguelikes and their procedural generation systems to explore three interrelated concepts in PCG games I term ‘depth’ (how many steps it takes to generate something), ‘length’ (the length of a ‘chain’ of generated elements in a game, which is to say, elements that affect each other or are contingent on each other), and ‘co-dependence’ (the degree of causal relationship between elements in these chains). These are valuable tools for thinking through the generative systems of unpredictable games per se, but most clearly so in PCG games where many of the most profound or extreme instances of depth, length and co-dependence have been designed to give distinct experiences to the player. I then proceed to explore how players identify procedural elements of a game in contrast to handmade elements of a game, drawing both on my previous work in this area and developing the arguments made in Chapter 2 regarding how players perceive and develop mental models of the systems and variables that create game randomness. This develops a notion of a ‘procedural aesthetic’, a style of experience created by PCG, which marks an unusual experience of unpredictability in gameplay contexts. All of these elements make PCG a distinctive platform for tremendous variety in game unpredictability.

Roguelikes The genre of game most closely associated with PCG is the ‘roguelike’. As noted earlier, the definition of this term remains contentious, but perhaps the only two universally agreed elements of the roguelike are the central role of PCG in the game’s design, and the presence of permadeath. This is reflected in the name ‘roguelike’ being a reference to seminal 1980 procedurally generated dungeon-crawling permadeath game ‘Rogue’. In this game the player descends through a dungeon of levels whose spatial layouts are procedurally generated; they usually take the form of a three-by-three grid of nine rooms, but these rooms are connected in different ways by corridors, and the precise width and height of each room varies on each playthrough. Sometimes a corridor hits a dead end without a room, and sometimes a room is replaced by a small maze, but this basic grid structure

152

THE UNPREDICTABILITY OF GAMEPLAY

is generally repeated throughout the game. Upon reaching the lowest level of the dungeon, the player must collect the fabled ‘Amulet of Yendor’, and then return to the surface in order to win. Throughout the dungeon the player will encounter and be able to collect various items which will aid them in victory against the denizens of the dungeon; however, the relationship between the names of some of these items (such as scrolls and potions) and their effects were also unpredictably assigned each time a new game world was generated, meaning that the result of using any of these for the first time was unknown, and the potential benefits of a potentially desirable item had to be weighed tactically and strategically against the risks of a potentially undesirable item. In turn, the denizens the player encounters get gradually more challenging as the player descends, during which the player also gradually collects a superior set of equipment with which to do battle. Both the items and the enemies are also distributed unpredictably across the levels of the dungeon, encouraging the player to use what they find and respond to what the game offers them, and to always think on their feet when new challenges arise. The game is turn-based, allowing for a potentially infinite length of time with which to make decisions. Alongside the obviously central role of the PCG aspect, we must not overlook the role of permadeath and its intersection with PCG in the original Rogue (1980) and the later genre. As mentioned earlier, permadeath is the practice of preventing a player from reloading their character upon death; combined with PCG, it meant that players only ever had one chance to deal with any unpredicted scenario the game might throw at them. As Jason Hawreliak argues, whereas most games follow what he calls a loop of ‘life-death-rebirth’ – in which the player’s character (or equivalent) acts, dies if unsuccessful, but then returns to a near-previous saved state – dying in a game with permadeath instead ‘closes that loop completely’.2 The consequent focusing of the player on the need to navigate one particular permutation, a navigation with such high stakes, is an important part of understanding the appeal of the roguelike genre, and how – albeit through very different means – it reproduces to some extent the nature of the experiences found in other games of extensive unpredictability, such as card games. In all of these cases a particular identity emerges in a playthrough through the guarantee of differences between one permutation and another, and it is a unique game state, which will only ever occur once, that a player must navigate. Although Rogue was not the first game to procedurally generate dungeon levels – an honour which seems to go to Beneath Apple Manor (1978) – the Rogue formula rapidly proved strikingly popular, and its presence on university campuses allowed it to rapidly spread to communities of computer-literate students and professionals who were taken by the concept, and started to iterate on it themselves. We can reasonably split roguelikes into three ‘generations’, each of which has changed the way PCG was used, the context of other gameplay elements that were built around (or within)

PROCEDURAL GENERATION

153

the procedural elements, and what elements were considered by designers to be elements that could be procedurally generated, and therefore could be made unpredictable. As we shall see in this brief history, whereas Rogue and its early descendants focused on the generation of maps and physical spaces, over time this expanded to procedurally generating monsters, items and much else besides (not just their placement), and even more recently, roguelikes now generate entire world histories, aesthetic styles for generated cultures and much else besides. These all led to new experiences for players, showing a greater range of purposes to which unpredictability can be put than we have previously seen in this work. Considering these three generations of roguelikes allow us to trace a genealogy of unpredictability as implemented through PCG, how PCG shifted and evolved as a method to create particular experiences in the game player and how it has been used to create entirely new species of unpredictable game experiences for video game players which merit experiential consideration. The first generation of roguelikes were those most explicitly based on Rogue, iterating on the established patterns of gameplay by adding a range of new elements, and broadening the variety of existing elements. Hack (1982) introduced a role for shops and buying items in the roguelike formula, expanded substantially the variety in existing gameplay elements such as items and monsters. Moria (1983) was the first to feature a ‘town’ level where one could buy a range of items from a wide variety of shops; although the name’s derivation is from Tolkien, the game itself shares relatively little in common with Tolkienian mythos. Larn (1986) saw the inclusion of multiple dungeons all connected through an above-ground region which gave the player access to these areas; the dungeons were in turn quite varied in their spatial orientation and layout when compared to the earlier Rogue, for these subterranean labyrinths included rooms of various shapes, mazes, open and closed spaces, and more besides. Omega (1988) included large environmental or countryside areas alongside the now-traditional ‘dungeon’ archetype, and like Larn, was arguably one of the first roguelikes to implement any substantial story elements, beyond the central quest for a valuable item at the bottom of a dungeon. All of these games iterated on standard medieval fantasy tropes, settings, monsters, items and overall thematic elements, and all remained within the broad constraints laid out by Rogue’s original structure. This first generation of roguelikes established a particular set of norms, demonstrated the appeal of the roguelike model, and began to explore what procedural generation could do to create a compelling and endlessly changing game world. They made clear the differentiation between traditional RPGs – more focused on story, more focused on graphics, less focused on replay, less focused on challenge – and the newly emerging roguelikes, while the genre itself began yield differenciation through a selection of games that both defined and represented the variation within the genre. In turn, it was through these

154

THE UNPREDICTABILITY OF GAMEPLAY

games as well as the original Rogue that players began to experience video games where each playthrough seemed as profoundly different as, say, dealing a deck of cards, or potentially even more so. The second generation of roguelikes maintained the aesthetic style of Rogue – which is to say, in almost all cases the use of ASCII characters to depict the game world – while branching further out in terms of gameplay, and more crucially for our discussion here, branching out further in terms of what elements could be procedurally generated, and how those elements might be distributed by procedural generation. In NetHack (1987), procedural generation was extended to include a range of different level types, including swamps and abandoned towns. In turn, NetHack introduced a tremendous range of systems which interact across a range of scales and places in the game world. The ‘polymorph’ system allows for the transformation of items and enemies alike, with categories of items being transformed into other items of that category, and with equivalent effects for monsters, forging virtual and potentially actual relationships between otherwise diverse elements; the ‘genocide’ system allows for the player to permanently kill all of a species or a group of species that are represented ingame by a particular ‘glyph’, which when performed reverberates through areas of the dungeon already explored and future areas not yet even generated. The procedural generation in Angband (1990) was implemented not just as the creation of spaces, but as something to be actively used and manipulated by the player. Each time the player travels from one level to another in Morgoth’s fortress (this being the second game to explicitly draw from Tolkien) that level is regenerated, encouraging players to consider their movement throughout the game world as a central strategic element to their play. In Ancient Domains of Mystery (1994) roguelikes for the first time found themselves presenting a detailed story to the player, but one whose elements – crucial characters, crucial items, the layouts of crucial regions, and so forth – all intersected with the game’s PCG systems to produce a world in which completing the same story meant a very different set of intersections each time. In Linley’s Dungeon Crawl (1997) and its close successor Dungeon Crawl: Stone Soup (2006–present), the player is offered the opportunity to play as any one of a diverse range of species with diverse abilities, and to commit to worshipping a number of virtual gods, each of whom offer the player particular benefits and disadvantages. What makes these systems noteworthy is the extensive number of ways that species and gods interweave with other forms of PCG in the game world, whether through particular items one god forbids, or particular kinds of terrain only one species can traverse, or particular enemies which become far more or far less deadly; the generative system which determines the placement of these other elements never shifts, but their significance emerges each playthrough in a synergy with the choices made by the player. However, perhaps the most important element across all the games of this second generation

PROCEDURAL GENERATION

155

of roguelikes was the newfound belief that the complexity of the game’s unpredictability, as well as challenge posed by that same unpredictability, should be a major value of roguelikes, and that the greater the number of interacting systems, the better. What I will term the third generation of roguelikes are those games which, perhaps surprisingly, have come to challenge what precisely the roguelike actually is. In doing so, they have arguably taken procedural generation to a previously unknown level of complexity and variety. The third generation of roguelike saw the emergence of a range of games that kept the central elements of PCG, permadeath (or some close variation thereof) and complexity (understood here as a wide range of differential relations and differential elements in procedural generation systems), but moved beyond the standard aesthetic style of roguelikes and some of the high-level superstructures of roguelikes, especially in terms of the length of gameplay. These new roguelikes were classed together under the slightly derogatory label of roguelites or the tongue-in-cheek label of roguelikelikes. They tended to be much shorter than traditional roguelikes, offer a range of different aesthetic styles and feature systems – sometimes called ‘metagame’ elements, sometimes more generally described as ‘unlock systems’ – whereby achieving certain things in one playthrough will affect procedural generation systems or lexicons in later playthroughs. At the same time, a range of ‘classic’ roguelikes which maintained aesthetic styles focused around ASCII or ANSI graphics continued development, while a set of ‘new classic’ roguelikes – such as Infra Arcana (2011), my own Ultima Ratio Regum (2012), Caves of Qud (2010) and Cogmind (2015) – iterated on the aesthetics of the original while developing new forms of gameplay3 and PCG quite unlike the original Rogue or its immediate descendants. In roguelikes, procedural generation and therefore unpredictability have been put to a tremendous range of gameplay purposes, creating a wider range of player experiences via unpredictability than in arguably any other game genre. PCG in roguelikes is used to create spaces of play, to distribute creatures and items within those spaces, to create the creatures and items themselves, to create the thematic or aesthetic elements of those games, to create quests and objectives and missions for players to complete, to create the histories and cultures of virtual worlds, to create diverse species and societies and much else besides. This therefore makes roguelikes, and their use of PCG, a noteworthy case study that moves us beyond the case studies of randomness and chance explored previously in the work, and towards some of the greater breadth to which these can be put (and consequently which players can experience). However, the complexity of some of these systems means that they require additional terminology to understand how they are implemented and how players perceive them. Just as much of this work is predicated on unpicking ‘games of chance’ to reveal the different locations of unpredictability, we now need to unpick in greater detail the

156

THE UNPREDICTABILITY OF GAMEPLAY

tremendous variety of generative systems and inter-system relationships present in PCG systems, and how each of these in turn affects a player’s potential experiences of randomness and chance.

Depth, length and co-dependence Having established roguelikes as the central case study for PCG as a form of game unpredictability, in this section I will develop three concepts that help us to address the experience of playing a game with extensive PCG. I term these depth, length and co-dependence. By depth, I refer to a measure of the number of steps between the start of a moment of unpredictability and its conclusion, even if the ‘thing’ it produces might end up being quite simple. This is akin to a tree where the bottom of the trunk is the origin of the unpredictability, and each leaf represents a different possible outcome, being the result of following dozens or potentially hundreds of bifurcating or fragmenting processes until their eventual conclusion. This is distinct from the randomness of, for example, a world map which might be created through PCG; this might involve a large number of unpredictable elements, but they might each have a simple rule with only a few steps for generating; deserts must be placed in certain locations, mountains in others, tundra and jungle cannot be adjacent and so forth. By length, I am talking here about the length of a ‘chain’ of connections that can be drawn across unpredictable systems that are contingent on or in some way affect each other. For example, a game which is otherwise deterministic but then procedurally generates a single room on a single level would have an extremely small length of procedural generation, depending on the number of elements in that room; by contrast, some games which generate entire worlds can display very long procedural generation chains. By co-dependence, I mean the extent to which elements in a PCG chain are contingent on each other; in some cases we might see two elements of PCG that intersect with each other but have no influence upon the other, while in other cases we might see one generated element which is required for another element to be generated, a situation where the second element is fully co-dependent on the first. Each of these, I will argue, can create a different experience in the player. A game with a lot of depth in its procedural systems generates confusion and obfuscation; a game with a lot of length in its procedurally generated content can induce a sense of scale and scope; and a game with a lot of co-dependence will generate a sense of realism, and a sense of a grounded and consistent game world. Each of these is a distinct experience generated by the play of unpredictability via PCG which we have not addressed in previous chapters, which hopefully collectively demonstrate the analytic value of considering roguelikes and their attendant use of PCG in the consideration of gameplay

PROCEDURAL GENERATION

157

experience through unpredictability. Naturally other unpredictable games can also display these three traits to a far lesser extent, but the computational power that most procedurally generated games take advantage of allows these to exist in a far greater capacity in video games than anywhere else, making the study of PCG inescapably connected to the study of these three concepts.

Depth Depth refers to the number of steps required to actualize an element of a PCG system. The more steps required and the more complex those steps, the greater the confusion it generates in a player, and the harder it is for a player to synthesize their experiences into a clear model of how that aspect of a procedurally generated game is actualized. When it comes to the contrivances of unpredictability in games, Malaby notes, there has been a turn from ‘explicit’ means such as dice rolling or deck shuffling to the ‘implicit’ means of computer programming,4 a shift which has enabled the emergence of deeper unpredictable systems that become increasingly challenging for a player to divine. A prime example of this can be found in Dungeon Crawl Stone Soup (2006–present). Some of the most powerful enemies in the game are known as ‘Pandemonium Lords’ – these rule over an infinitely large area of the game world called Pandemonium, a challenging late-game part of the game world. Although interesting enemies in numerous ways, what matters for our discussion is how the game procedurally generates their properties and abilities. First, all the basic statistics of creatures in the game – their ability to dodge, their resistance to damage, their health, their speed and so forth – are generated within wide confines through the Degrees of Outcome model of chance, or through the Different Outcomes model of chance; for example, when selecting a new Pandemonium Lord’s fire resistance, there is a two-in-three chance of having either one or two levels of fire resistance; if it has no levels, then it was a one-in-ten chance of being actively vulnerable to fire (taking extra damage). However, the real complexity of these chances appears for Pandemonium Lords which are ‘spellcasters’ (itself a two-in-three chance). Each Lord has six ‘slots’ that magical spells can be placed in, which will be the spells they are able to use in battle with the player and the player’s allies. In the first slot they get one unpredictable spell from a particular list of spells; in the second they get another from the same list; in the third they have a one-in-four chance of nothing, otherwise a spell from another list; the fourth slot has a one-intwo chance of being empty, otherwise it picks from another list; the fifth slot also has a one-in-two chance of being empty, otherwise it is filled in the same way as the fourth slot; and the final slot has a one-in-four chance of being empty, a one-in-four chance of a particular spell and a one-in-two

158

THE UNPREDICTABILITY OF GAMEPLAY

chance of another spell. After that, certain spells are replaced by other spells or empty slots. Then, the game iteratively tries to replace spells in each slot with other spells; in slot one five spells are cycled through, each with a onein-twenty-five chance; in slot two the same takes place with four instead of five spells; in slot three two spells are considered with one-in-twenty-five chance, then another with a one-in-twenty-two chance, then another with one-in-twenty, then two with one-in-twelve chance, then a final one-in-ten chance, leading to roughly an overall one-in-three chance of the previous spell in slot three being replaced; slot four has three one-in-twenty chances, then if still empty at the end it is filled with a particular spell; the fifth slot has a one-in-fifteen change of being replaced with a particular spell; and slot six is unchanged. With this complete, the player then faces the finished Pandemonium Lord. If the reader has now quite lost track of the range of forms a Pandemonium Lord might adopt and the rhizomes selected in order to actualize this form, they are not the only one. It is clear such a complex system will produce outcomes unlike an equally weighted die roll for each one of the elements described above, yet a significant knowledge of probability theory and some time spent with a calculator is required to figure out what precisely those chances actually are, and what exactly the difference between this system and some die rolls actually is. Are some elements more or less likely than they would be in a fair roll? Will some elements often be accompanied by others due to the network of dependencies identified above, or will they not? A deep unpredictable system makes it harder and harder for players to judge the nature of the generative system, and therefore promotes either divergence between a multiplicity and a generative system (if the player begins to comprehend the system), or simply entire gaps in the player’s multiplicity where a particular segment of the generative system remains profoundly opaque. Making generative rules deeper can also make it less clear what is actually repeating – which elements are fixed, and which elements are actually shifting, and where do the contingencies lie? In addressing depth, length and co-dependence in this section, I propose that each generates a distinct experience for the player, these experiences being much of what makes procedural generation an important case study for unpredictability. In the case of depth, it is confusion and puzzlement that deep unpredictable systems create in the player. For some players this will mark a game which is compelling precisely because it remains unknown and because the player cannot divine the mathematical rules hidden behind the graphical or narrative game world; for others, it will be a source of stress and annoyance. Understanding depth therefore helps us understand the process by which multiplicities are constructed, and most crucially, how different kinds of unpredictable systems can significantly impact on the progression, and indeed the very success, of that process.

PROCEDURAL GENERATION

159

Length Within the broad scope of ‘roguelikes’, although the use of the term for these games might be contested, these are several games which are known for being focused on the procedural generation of entire worlds. Of all these games, Dwarf Fortress (2006–present) is the most comprehensive, the most detailed, the one with by far the greatest length of procedural generation, and the most popularly visible. When one begins a game of Dwarf Fortress an entire world with hundreds or thousands of years of history is procedurally generated, and created in such a way to ensure tremendously long chains of procedural content which both interact with each other in the unpredictable generation process and continue to interact once the player starts actually exploring that world. Long chains of relationships can emerge where one character moves from one town to another, interacts with new characters, who then themselves move, and carry something of that interaction to their new location, and might produce a piece of artwork, which another character might then take and transport to yet another location. Long chains of history connect abstract historical stories, such as the battles between great heroes or the raiding of towns or the emergence of deadly monsters from beneath the ground, to things and places and artefacts that the player can actually encounter, or which might be reproduced once more; an ancient battle might be all but forgotten until one of the player’s dwarves decides to engrave it on a wall or create a sculpture that depicts its most important character. Items and characters, no matter how obscure or unimportant, are tracked by the game even when the player is nowhere near them, and might never actually physically encounter those game elements. In all of these ways the game is able to generate, maintain and let the player interact with tremendously long chains of unpredictable generated content, and allows for the emergence of new chains and the continuing growth of existing games as play progresses. Whereas depth generates a sense of confusion and puzzlement the deeper a procedural generation system is (the more steps it has from initiation to conclusion), a game with procedural generation of great length instead generates a sense of scale and scope. It suggests the existence of a wide range of independent or semi-dependent elements of the game world that are all unpredictable, and in some way operating outside of the player’s actions, or around those actions. A game with long chains of procedural generation also points players towards things they might not have previously considered as potential actions they could take: one could seek out the origins of the elements in a chain, or seek out other examples of such elements, or the settings that such elements were actualized in or through and so forth. In turn, having seen one long chain of connections, it is reasonable to assume that other long chains of connections exist, and that much of the game world one is experiencing still lies out there rather than within the boundaries of

160

THE UNPREDICTABILITY OF GAMEPLAY

the player’s present experiences. This is not a feeling that can necessarily be created with any kind of ease in analogue games, given the extremely low computational power offered by the use of dice and other mechanical unpredictable systems, or by the distribution of pre-made elements on a game board. Any kind of length in unpredictable systems beyond a short chain can only truly be found in PCG, and it is here that such chains of unpredictability allow players to experience something of the scope and scale of the unpredictable systems they are playing within, and consequently shape their perspectives on those systems, and their future actions within the game space.

Co-dependence The third aspect of this terminological toolkit for considering the experiences created by procedural generation is what I term ‘co-dependence’, which is related to the concept of length. Whereas length refers to the potential size of chains of related PCG systems, co-dependence refers to how closely related each element of those chains are, which is to say, are there contingencies and causalities at play, or are these just unrelated elements which can nevertheless interact? Although the two are often interwoven, even long chains of procedurally generated elements can be quite profoundly disconnected, with incidental links and passing associations standing in for deeper codependence between the elements. Equally, it can be hard to unpick the relationships in long chains of elements, for two elements together might appear connected and thus become such in a player’s multiplicity, without being so in the generative system; equally, two aspects might appear unconnected but have emerged from the same source. As such, in keeping with our primary focus on subjectivity, examining co-dependence means examining a player’s experience of finding links between unpredictable elements in these kinds of games, and the overall sense that a game with strong links, and strong commonalities between its unpredictable elements, generates in a player. We begin with a case of high co-dependence. If the reader will pardon the use of this case study, although it is inevitably one with which I am highly familiar: in my spare hours for the last few years I have been developing a roguelike game of my own, called Ultima Ratio Regum (2012–present). Although in many ways a divergence from the gameplay of more traditional roguelikes – there is little if any combat, there are no ‘dungeons’, and it plays in some ways more like a classic adventure game than a hack-and-slash RPG – it, like several others in the genre and its nearby genre cousins, generates a ‘complete’ world at the start of a new playthrough. Alongside the obvious geographical and topographical features and climatological variation one might expect, the game also procedurally generates nations, cultures and

PROCEDURAL GENERATION

161

belief systems, which spread in what I hope is a reasonably organic way across the terrain. This is one of the most detailed aspects of the game with almost every cultural or social element (that might be relevant to the game’s exploration-focused gameplay) being algorithmically actualized during this process. These PCG elements include national flags, architectural styles, aesthetic styles in artwork, the names and cultural associations of plants, thousands of years of history and much else besides. The crucial point here is that these are all designed to have a high, and sometimes complete, codependence on each other, and for this co-dependence to take place at all possible scales and in all possible directions. Length is shorter than Dwarf Fortress, but co-dependence somewhat higher. A species of plant might be generated, which then becomes sacred, which is then taken as a symbol by a ruler, which is then referenced in a novel; or a great conflict might take place, which is then immortalized in a painting, which is then discussed in polite society and so forth. Rather than these connections emerging from a single point of origin, they emerge and converge and intersect from many points of origin and towards many points of conclusion. In doing so, they are designed to give the sense of a world which is very physical, very grounded: one should be able to read a book and see a painting and talk to someone about the same event, and it should be difficult for the player to unpick where this sequence began or ended. The co-dependence of elements in the game is therefore extremely high, deliberately so, with the goal of creating a world which, in a manner of speaking, actually appeals to be handmade, even while being heavily procedurally generated. For an example of low co-dependence, we can look once more to Dungeon Crawl Stone Soup (2006–present), a roguelike which consists of a number of highly visually distinct regions the player can visit. In the ‘Dungeon’ proper, we find dark stone walls, sometimes partly covered by a banner, or a blood stain, or marked by cracks, or with an unfortunate skeleton bricked up; in the ‘Elven Halls’ we explore hallways with walls marked in intricate silver or mercury designs, and walk over floor constructed of shining azure tiles; in the ‘Lair’, a realm of real and fantastical beasts, we find blocks of perhaps sandstone or clay, something more in keeping with its wild and natural (or faux-natural) denizens, and walk over grass; in the ‘Vaults’ we explore a domain of grey metal walls and regular rectangular floor tiles, all intricately and meticulously meshed into a region that resembles the guts of a machine as much as it does a level of a dungeon; in the ‘Crypt’ we wander over dull, dry stone flooring and between walls that depict dramatic archways, cobwebs and candles, all constructing the sense of an archetypal haunted castle; in the idiosyncratically named ‘Realm of Zot’ we find shining marmoreal hallways, a cleanliness and a smoothness unseen anywhere else in the dungeon, evoking our proximity to power and the game’s resolution. In each case the gameplay mechanics of the floors and the walls – what can be walked on, what cannot be walked on, what can be destroyed, what can

162

THE UNPREDICTABILITY OF GAMEPLAY

be changed by magical means and how and so forth – do not shift. All that shifts is the aesthetic overlay played atop of the areas the player explores, designed to denote the move to a new region of the game, and thereby also overall progress through the game. Although in recent versions of the game the generation algorithms for the layouts of these branches normally differs significantly, in older iterations this was not always the case, and the visual style of the tiles, rather than the layouts of those tiles, was a crucial visual aid. Whatever orientation of other elements might be placed, the textures adapt; equally, dungeon levels will be built and monsters and items will be placed regardless of how the area actually looks. There is therefore no co-dependence between the layouts and the appearance of dungeon levels in DCSS; both systems can be easily intermeshed without needing to take account of the other. Depth generates a sense of confusion and puzzlement in players; length generates a sense of scope and scale. Co-dependence, by contrast, generates a higher sense of grounding and realism the higher the co-dependence is between game elements. This is qualitatively a very different player experience to the kinds of experiences we have considered before; this is an experience with a sense of worldbuilding, a sense of narrative, a sense of being within a game rather than simply playing that game, and that feeling is inevitably contingent on the construction of a detailed web of elements that can interact and intersect with each other in numerous ways. With low co-dependence, aspects of the game come to take on what, in the following section of this chapter, I term a ‘procedural aesthetic’. Whereas a high codependence game actually comes to look remarkably real and remarkably convincing, due to the deep interweaving of its elements, a game with low co-dependence looks more like a game which is procedurally generated, which is to say, where elements were distributed by algorithm rather than by human intention. We now shift to exploring this, and what it means for players attempting to develop their multiplicities of games in a context of the sort of profound unpredictability that PCG allows for.

Identifying procedurality We now come to one of the most intriguing elements of procedural generation, especially when understood as a subset of randomness and in light of our previous discussion about the identification of systems and their components: how does one identify procedural generation, and what experiences of play suggest that procedural content is present? The overwhelming majority of those who come to play a roguelike, or a game with numerous and extensive components of PCG, will likely know before starting the game that these elements are present. However, that does not make it immediately clear what elements one sees have been procedurally

PROCEDURAL GENERATION

163

produced or procedurally distributed, as opposed to the elements which were handmade, or hand-placed. In the earlier discussion of this topic in Chapter 2, we assumed that the player was able to identify what was unpredictable and what was not, although we did acknowledge that even that process might not necessarily be trivial or unproblematic. In some games, however, this is not always obvious, and we might think this would be hardest in games of procedural generation due to the complexity of the systems that can create the game world players experience. In this section I aim to demonstrate that procedurally generated content can often display a particular style, which I will term the procedural aesthetic; far from making the identification of unpredictable elements harder, this can actually make it significantly easier, once a player knows what to look for. So what does a player seeking to identify a procedurally generated element look for? There are two ‘clues’ in particular which I believe are illustrative of a procedural aesthetic. The first is aspects of a game world which are clearly designed to make sense, and adopt written or aesthetic or spatial structures which are supposed to be recognizable, and which in the real world are always occupied by things which do make sense, but in the context of the game they do not make sense. An example would be the construction of a building from constituent rooms, but the rooms leave a lot of space between them, and that space is occupied only by bricks; buildings in the real world do indeed consist of rooms, but rarely contain large swathes of empty, unused space.5 By spotting these aspects – where something has been generated according to a certain set of rules meant to produce structures of a category players have in the real world – players may recognize the gap between real-world examples of that thing and the examples of that thing they are now faced with in-game. The second illustration of the procedural aesthetic is the generation of gameplay challenges or situations which, although certainly within the rules laid out by the game, brush up against the concept of the exploit: a human designer would not have put these in the game, they appear to be against the spirit of the game, and in one extreme example I will consider here, they can even prevent the game from being won at all. Examining these will allow us to both develop a concept of a procedural aesthetic and, more importantly, assess how that procedural aesthetic – predicted on randomness and chance as subsets of unpredictability – is experienced by the player. What does the player see, how do they make sense of these things, and what tell-tale markers allow a player to distinguish procedural from non-procedural elements of these games? A strong case study of the procedural aesthetic as reflected in a game world is Caves of Qud, a post-apocalyptic open-world roguelike game inspired by diverse sources such as classic science fiction, cosmic horror, cyberpunk and retrofuturism. As one explores the game’s generated world one comes across a tremendous range of textual markers that reorder

164

THE UNPREDICTABILITY OF GAMEPLAY

concepts and places and people in profoundly strange ways, having been procedurally generated within broad rhizomatic constraints and then allowed to adopt whatever shapes they will. For example, one might come across a shrine depicting an ancient ruler, who assassinated another ruler over a dispute about stained glass, all while being only six years old. One might encounter a piece of poetry which does not rhyme and lacks any clear meaning, even while grammatical rules are observed, and speaks of worshipping a metal, consuming a certain kind of flesh, or speaking to a particular kind of monster. You might be informed that a particular species is fond of you, such as dogs or tortoises, and that one is welcome in their ‘holy places’ (one can only imagine what such places might be). While all formal rules of language are observed and these are often written or presented in an appropriate tone for their status as literature, as histories, as information being given to the player, such examples cut across and upset our expectations of logic and order. This feeling is akin to Michel Foucault’s reflection on a famous Jorge Luis Borges quote, describing the ontological rules by which a fictional encyclopaedia was constructed. This reordering, Foucault suggests, was striking for ‘breaking up all the ordered surfaces and all the planes with which we are accustomed to tame the wild profusion of existing things, and continuing long afterwards to disturb and threaten with collapse our age-old distinction between the Same and the Other’.6 In Caves of Qud, too, these things cross semiotic boundaries, yield confusions between classes and categories, raise the possibility of impossible things, of self-referential things. This, in a worldbuilding sense, is the procedural aesthetic. What makes this especially noteworthy is that these forms are actively embraced, considered desirable, and it is clear from any observance of the game’s community that its players actively want and seek these kinds of strange constructions, which speak to implied worlds that function on laws different to our own. Spotting these moments of procedural aesthetic in the world the player explores is a clear signifier of PCG, and in this case, the deep semantic unpredictability it can herald. Moving now from the game world to the game play, we can also look for the procedural aesthetic by spotting challenges or gameplay situations that a human designer would never (we assume) deliberately throw at a player. We have already seen this in the unusually powerful item combinations in The Binding of Isaac that constitute part of the game’s procedural aesthetic, and here we will consider two other examples, arguably even more extreme examples, which further demonstrate this concept. Classic roguelike NetHack (1987–present) offers two excellent examples of this, beginning with the concept of the ‘Gnome with a Wand of Death’. NetHack contains a wide variety of monsters, and one of the hallmarks of the game is that monsters have abilities, properties or what DeLanda would call capabilities7 which are logical to their physical structure. Monsters with

PROCEDURAL GENERATION

165

hands can pick up and use items; only creatures with heads (and therefore necks) can potentially don amulets; the game differentiates between most monsters having two eyes, a small number having only one (such as a cyclops), eyeless monsters possessing none and being therefore blind, ‘intelligent’ monsters can use tools and so forth. In turn, almost any item has a possibility, however small, of appearing in almost any part of the dungeon; it is possible to find an end-game item quite early in the game, if the unpredictable conditions of randomness or chance happen to be favourable. One such item is a ‘Wand of Death’, which is a wand that, when cast on the overwhelming majority of enemies including the player, will kill them outright. It has a chance, although extremely small, to appear anywhere, and therefore potentially be picked up and used by an intelligent monster against the player. Gnomes, a class of monster in NetHack, are extremely weak early-game enemies; the Wand of Death is an immensely powerful end-game item. It is nevertheless possible for a gnome on an early floor of the dungeon, even the first floor, to stumble by tremendous good fortune across a Wand of Death, pick it up and blast the defenceless player with it, immediately ending their playthrough. There are ways to survive being shot with a Wand of Death, but it is extremely unlikely that any player character in the early game will possess these; and by the time the player character does, they will have generally progressed to a point in the game where gnomes, being weak early-game enemies, no longer appear. As a result, it is possible for the player to be slain in a manner which is truly unavoidable: if confronted by a gnome with a Wand of Death, when unable to defend oneself, and unable to flee to safety, if the blast from the Wand hits, the player will die. In almost all cases it is impossible for the player to defend themselves from this attack; it is a guaranteed death. Roguelikes pride themselves on being beatable by a highly skilled player no matter what is thrown at them, and yet the existence of such an enemy vitiates that claim. A human deliberately placing enemies and items would surely never place such an enemy;8 the potential existence of such an enemy is therefore an indicative element of the procedural aesthetic. Another equally striking aspect of the procedural aesthetic in NetHack was the possibility (present in the game for decades until recently) for certain playthroughs to be unwinnable. This is not because of the gnome explored above, but rather takes place in a context where the player does everything right, survives through almost 100 per cent of the game, reaches the final sequence of gameplay, has a full-strength character they are ready to win with, yet remains unable to seize victory. This is because in order to complete a game of NetHack, the player must acquire a range of specific items. These are not for making one’s player character stronger or more resilient to damage, but are rather essential ‘key’ items required to reach the game’s final sequence. These items include seven candles. These seven candles must be attached to another item, known as the ‘Candelabrum of Invocation’.9

166

THE UNPREDICTABILITY OF GAMEPLAY

The problem is that although the Candelabrum is a guaranteed item which will always appear in the game no matter the whims of the procedural generation systems, and the same can be said of all other essential items in the game, seven candles were not guaranteed to exist. A lighting shop from which the player might buy candles was guaranteed, but the number of valid candles is unpredictable. As a result, the placement of candles through algorithm rather than through designer means that, as ordinarily played, certain permutations of NetHack cannot actually be completed  – and the player could not know if they are in such a permutation until the entire game except for the ‘Ascension Run’ has been completed, every part of the dungeon mapped, every shop examined for potential candles and so on. In principle, for a perfectly stable game, this would mean that a player would have exhausted all possibilities within that generated game world, and, unable to step outside the confines of intended gameplay (the same intended gameplay which had failed to generate sufficient candles), the game would be unwinnable. Once more, a human designer addressing this should – we would hope – remove such a possibility, especially as playthroughs of NetHack can readily take upwards of dozens of hours. Its presence, however, is once more illustrative of the procedural aesthetic: something a human would not do, but something we can readily imagine a less-discerning algorithm might have placed. In summary, the repetition of difference, and often quite profound, strange, jarring difference, lies at the core of the player’s experience of spotting and identifying PCG. In this case another form of difference is implicated in player experience, which is the difference between what players expect to be human-crafted and what players believe to be machinecrafted, and how these two notions are gradually differentiated (made conceptually distinct) from one another by the ongoing differenciation (emergence and subsequent perception) of elements which strike the player as ‘looking’ human or ‘looking’ machinic. Via the factors explored in this section players give elements of games with PCG identities as human-made or machine-made, which changes our expectations of their semantic or mechanical sense. In turn, we can also note that the relationship between the procedural aesthetic and instability is consequently quite highly contested. On the one hand, the gameplay examples from NetHack outlined above certainly seem to be outside the spirit of the game, for surely that spirit includes the ability to actually complete the game, and to always have a chance to complete and progress in the game? On the other hand, the lack of a developer-led fix for these ‘issues’ signals an embracing of a procedural aesthetic, comparable to that found in Caves of Qud, except in this case on a gameplay level rather than a worldbuilding level. The role of the procedural aesthetic (both gameplay and world) complicates and problematizes our notions of the game being played and what the player expects to be offered or shown by that game. As well as offering the potential for depth, length

PROCEDURAL GENERATION

167

and co-dependence on a level rarely if ever exhibited by other kinds of game unpredictability, PCG also allows for the creation of distinct moments which constitute a procedural aesthetic. These moments generate a spirit of the game very unlike any other, where concepts that might ordinarily break the player’s immersion in a seamless game world instead construct that world, and gameplay challenges which ‘should’ be impossible can instead stand out as one of these games’ most trying, yet repeatable and infamous, moments. In this chapter we have examined the notion of PCG, a set of programming tools and techniques designed to offer significant variation in the actualized elements of games. Most clearly epitomized by the roguelike genre, PCG techniques have been applied to game elements ranging from enemies and items to books and poetry, world maps and towns to social and cultural norms, and dialects and phrases to histories and sagas. We first examined roguelikes, exploring how the complexity and variation of game elements created via procedural generation grew over time and became increasingly important to the identity of the genre, in the process examining some of the new game experiences these implementations offered to their players. The chapter then proposed the terms depth, length and co-dependence, to represent the number of steps in the actualization of an unpredictable element, the size of the chain of connected unpredictable elements and the strength of the causal or contingent relationships between those elements. Each of these, I argued, generates a different feeling in the player when depth, length and co-dependence are high, that is confusion, scale and realism, respectively. Lastly we considered what I called the procedural aesthetic, a disruption of our expectations for both the game world and the play of the game set within that world, which is central to how players of these games identify its unpredictable elements. For many players, this aesthetic is an enjoyable and desired element of the gameplay experience. As I hope to have demonstrated, PCG is integral to any discussion of game unpredictability; it advances trends from earlier in the book into more dramatic manifestations, enables a range of experiences difficult or impossible to find in analogue games and is a form of unpredictability only growing in importance with the continued expansion of PCG into the mainstream of game design.

Notes 1 Mark R. Johnson, ‘Integrating procedural and handmade level design’. In Level Design: Processes and Experiences, edited by Christopher W. Totten (Boca Raton, FL: CRC Press, 2017). 2 Jason Hawreliak, ‘Heroism, gaming, and the rhetoric of immortality’ (PhD diss., University of Waterloo, 2013), 340.

168

THE UNPREDICTABILITY OF GAMEPLAY

3 However, it should be noted that the ‘surprise factor’ in these games – seeing something generated which one has only previously experienced as a piece of handmade game content – only applies to experienced roguelike players, for whom the core genre has been sufficiently differentiated. Only these players have an idea of what a ‘roguelike’ traditionally encloses in its concept. 4 Thomas M. Malaby, ‘Beyond play: A new approach to games’. Games and Culture 2, no. 2 (2007), 108. 5 This is comparable to what Loren Schmidt calls the ‘perception of intent’, which is to say the overall sense or rationale the player experiences when engaging with a generated element. See Loren Schmidt, ‘Generative artwork’. In Procedural Generation in Game Design, edited by Tanya X. Short and Tarn Adams (Boca Raton, FL: CRC Press, 2017). 6 Michel Foucault, The Order of Things: Archaeology of the Human Sciences (London and New York: Routledge, 2001 [1966]), xvi. 7 DeLanda, Intensive Science, vii–viii. 8 One clear exception is the ‘Medusa’ enemy in NetHack, a unique foe who can immediately turn the player’s character to stone, and is famous for slaying new players. There are, however, some (obscure) clues to her presence before she is encountered. 9 In its portrayals of demonic and divine creatures, items and themes, NetHack draws heavily upon Judeo–Christian mythology. The Candelabrum of Invocation and the other two items – the ‘Bell of Opening’ and the ‘Book of the Dead’ – required to begin the Ascension Run are references to the Catholic excommunication ritual known as the ‘Bell, Book and Candle’, whose origin is unclear but may go back as far as the ninth century. Many designers of the second generation of roguelikes were known for including a diverse range of mythological and historical components in their games, and this ritual is no exception. For more on these thematic elements, see Mark R. Johnson, ‘The demonic properties of an ampersand’. Kill Screen, 2 July 2015. Available at https://killscreen.com/articles/demonic-properties-ampersand/.

7 Replay Value and Grinding

In this chapter we will examine two practices that involve players performing the same in-game tasks, more than once, in order to obtain or experience a greater percentage of a game’s possible elements than would be possible by doing that task only once, or to achieve something in-game that will not always occur. These are replay value and grinding. These are distinct in their scope and scale, for the former takes place across an entire (or most of a) ‘playthrough’ of a game from start to finish, while the latter takes place when a player performs a single task within a game over and over, rather than performing much of the entire game over and over. They therefore vary according to the relative length of time one repetitive action – the replay, or the grinding action – takes to play out before being done again, and consequently the pace and cadence of feedback given to the player, and players’ traditional emotional responses to carrying out these actions. We begin by defining our terms. ‘Replay value’ is the value assigned to playing a game more than once – which is to say, the proposition that playing the ‘same game’ more than once will be worth doing. This is generally understood in three ways. In the first case, for deterministic games (or other media items whose material form and process of ‘reading’ does not shift between one ‘read’ and the next) a replay will yield the same experience on multiple playthroughs, but this experience is thought to be desirable enough that it is worth having a second time, unchanged. Many players carry out this kind of replay for games they have not played in many years,1 where the distance of time is seen as lending particular value to revisiting an old game. In the second case, replay value is an important concept in games with multiple mutually exclusive branching paths (even if fully deterministic), such as the majority of contemporary role-playing video games, or board games. The replay value of a game in this context denotes the breadth of possibilities that can occur in the game, the variation between those possibilities (they are different enough to be worth pursuing) and the extent to which players can continue to play the same game, take different paths and not yet exhaust all the possible paths of the game. In the third

170

THE UNPREDICTABILITY OF GAMEPLAY

case, replay value is understood in the context of a game that contains some unpredictable elements, and therefore an additional playthrough will likely allow a player to discover other elements of that game – which are possible but not yet real, but with the potential to be actualized – that could not have been discovered or experienced the first time around, due to the fall of the dice, the deal of the cards, the outcome of the procedural generator and so forth. This is different from the second kind, where all possibilities could have been experienced had the player(s) taken a particular path; in this case, however, the game content offered to the player is contingent on processes of unpredictability, rather than (or possibly as well as on) player action. Recall the discussion in Chapter 2 on the set of possible relationships in a given permutation: in this third case of replay value, playing again enables the emergence of new elements and relations which can then give rise to new singularities. It is the third of these we will focus on here – the concept of ‘replay value’ when used in the context of a game that contains unpredictable elements that cannot all necessarily be seen or experienced on a single playthrough, no matter the player’s actions. Games with unpredictability of this sort are regularly touted as being particularly ‘replayable’, for they should offer a different experience, and a different set of challenges, to each player each time. One will explore different areas of the game, potentially encounter different characters or different events, all of which might have been experienced on previous playthroughs, but were not, or could not have been. For a sufficiently complicated unpredictable game, there might be – depending on one’s definition of replay value, as we will shortly see – an effectively endless volume of replay enjoyment that can be extracted from the game, so varied are its elements and so numerous are the elements that vary. This is certainly also the case for non-digital games of unpredictability, such as card games, some board games (where dice or other unpredictable elements are used), and the understanding of the potential for a generative system (machine or human) to keep games continually interesting and replayable can also be found in tabletop RPGs.2 Replay value is consequently a rich case study for deepening an understanding of games with unpredictability: it is where players seek out difference and repetition in their play experiences and truly focus on the uniqueness and specificity of each iteration, rather than looking to the long term or, indeed, necessarily thinking about the game’s unpredictable system; it is enough to have the new experience, even if that newness is still contingent on past knowledge and future expectations of difference. For a player ‘replay value’ promises as close as possible to playing a game for the first time once again, while bringing with it the advantages of having some idea about the game’s structure, systems of play, strategies and so forth. Alongside replay value, this chapter will also consider the practice of grinding. This is the process of continually performing a given in-game

REPLAY VALUE AND GRINDING

171

action until a particular outcome is achieved.3 This might be because the player has to collect a large number of a given item, for example, and must perform the same item-giving in-game act repeatedly until enough of those items are acquired; or it might involve performing an action that offers only a small chance of the desired outcome over and over until the game’s chance systems select the outcome the player wants. In both cases, the same action is done repeatedly until the desired endpoint is achieved, but the action does not change; the player therefore does not encounter or experience anything new or stimulating during this process. Although this might seem less than compelling, the incentives for players to grind in many game environments are strong: ‘Because [grinding] theoretically allows players to obtain all of the thousands of rewards in a video game …  players may be motivated indefatigably to acquire every single reward’.4 Rather than offering rewards to players through the completion of meaningful tasks which might require the player to experience something new in the process of acquiring something new, grinding asks them to do the same thing again in order to get a different result, without ever promising when that different result will manifest. Grinding therefore takes place in a context of perpetual anticipation of the eventual reward of the repetitive actions,5 where the act itself is meaningless except as a means to an end. To put it another way, grinding entails ‘performing an easy and repetitive series of actions in a game in order to maximise reward payout’,6 which is to say extracting maximum ‘value’ from one moment of a game before then moving onto the next. Understood in this way, the player sees each point of unpredictability in a game where grinding is possible as offering the player an experience but withholding it until an arbitrary volume of effort is invested. When the reward requires no particular ability or thought or strategy, but merely the investment of time in order to achieve what is, over the long term, a guaranteed reward, it begins to look very promising; why would one move on when the parts of the game already explored have not been ‘completed’? As Daniel Cook somewhat bleakly puts it, ‘If there is an exploit that ruins the game for [your players], most will happily go for it.’7 Jesper Juul echoes this observation, noting that optimal play, although optimal in terms of pure rationality, can be deeply tiresome and uninteresting.8 Although my intention here is not to explore in detail the player psychology which leads to the willingness of players to grind,9 it is apparent that situations which encourage grinding are ubiquitous in games, and large numbers of players are willing to accept this and grind for the outcomes they desire. Whereas replay value is a case study of players seeking out the broadest possible difference they can, with repetition as unlike any they have seen before and specifically, ideally, impossible to predict, grinding involves the pursuit of a single difference and a single repetition of a sort known in advance, and primarily involves the deliberate and continuing recreation of generality until the game’s system offers the desired outcome. Both, nevertheless,

172

THE UNPREDICTABILITY OF GAMEPLAY

involve players selecting a ‘do over’ of a game or some segment of a game, and are consequently interesting to compare against one another. Unsurprising given their deep differences, the reception of these concepts or practices by players is starkly divided. When it comes to games with high replay value we are told that ‘people love to play them’;10 equally, given the desire of players to often want more of their favourite games, replay value is seen as a way to continually provide value from a game that, although perhaps ‘completed’ once, still offers much to the committed player.11 PCG, the focus of the previous chapter, has been particularly noted as being highly beneficial for replay,12 owing to its ability to vary games outside of the player’s actions and present unique challenges and experiences on each playthrough; equally, games that give the player tremendous options for constructing and sharing their own play challenges, such as the LittleBigPlanet (2008– 14) series, have been lauded for ‘redefin[ing] replay value’.13 Indeed, so committed are some players to extracting replay value that third-party tools exist to generate ‘challenges’ that will refresh a player’s experience of a game, such as playing within certain constraints or pursuing new objectives.14 By contrast, however, few players report an enjoyment of grinding;15 it can be ‘boring and repetitive to accomplish’,16 and is consequently often described by players in ‘less than enthusiastic terms’.17 Grinding is seen often as a failure of game design because a well-designed game should not just continually offer the player new and interesting things to do, but actively encourage them to pursue those by making them more appealing than a tiresome alternative. There is therefore a stark contrast between the value assigned to experiencing an entire game with unpredictable outcomes once more and the experience of replaying portions of a game with unpredictable outcomes.18 With these terms established, the chapter now proceeds to develop our conceptual understanding of replay value and grinding, both predicated on unpredictability, and to consider the different but related player experiences they both create. It first identifies a clear contrast in the placement of rewards for players in replay value and grinding in games with unpredictability. Replay value is about the unpredictability of experience throughout the repeated content; grinding is focused on the unpredictability of outcome throughout the repeated content, and performing the same input until the desired output is, through chance, produced by the game. This means the values and the motivations behind these practices are very different: the former seeks enjoyment in the guarantee of an unknown outcome, while the latter seeks enjoyment by acquiring an outcome which, although it might be known in advance, is not guaranteed by any interaction; it is only a possibility. The subsequent two sections then each focus in detail on grinding and replay value, respectively. In the first, drawing on the Soulsborne (a conflation of Demon’s Souls, Dark Souls and Bloodborne) series of games (2009–16) the chapter focuses in more closely on grinding

REPLAY VALUE AND GRINDING

173

and explores the curious tension present in games which give the player a way to reduce the experiential tedium of grinding – thereby tacitly acknowledging that it produces these feelings – and yet still retain grinding as an intended gameplay element. In the second, drawing primarily on No Man’s Sky (2016) the chapter offers a detailed exploration of the notion of the ‘completionist’ as a form of player who seeks to exhaust the multiplicity of a game, and an approach to playing games with high replay value. I propose three forms of completionism, and examining these will highlight – as throughout this work – another distinctive set of player practices, all united by their relationship to unpredictability and the experience of playing an unpredictable game with particular objectives in mind.

The value of returning We have now outlined the concept of replay value and the practice of replay, and the practice of grinding. Replay entails playing through an entire game or a portion of a game more than once, with the goal of experiencing something which was not experienced on the first playthrough, seeing elements of the game which were not actualized first time but might be actualized this time, making different choices at forking paths in a game’s narrative or quests, and the like. Grinding entails doing the same segment of a game over and over in order to achieve a specific instrumental objective, such as the acquisition of an item which appears only unpredictably when certain in-game actions are performed. In this section we will look more closely at the differences between the two, with a particular emphasis on what value players are getting out of these returns, these repetitions and generalities, when they make the deliberate choice to replay a game or to grind a particular moment of a game. Given that replay value and grinding both entail the player explicitly and deliberately instigating a return of a given sort, thinking about these practices in terms of the player’s experience and synthesis of time appears to be a valuable place to start. What creates the inclination to return to previous experiences? What do players know, or think, they will get out of this deliberate return to unpredictability? What aspects of unpredictability are players focusing on and wishing to experience again, but also, anew? In the case of replay value, the value of returning to the site of previous gameplay lies in the experience that the player gets when an unexpected set of differential elements and differential relations are actualized. Replay value is understood as playing all or large portions of a game once more, and as wanting to have a significant portion of the singularities one encounters in that period be new and distinctive, and unlike what one has encountered before. In replay value, the past is therefore identified not just as having been a compelling or enjoyable gameplay experience, but also one which –

174

THE UNPREDICTABILITY OF GAMEPLAY

the player understands – might have played out other that it did. The past therefore serves as a marker of experienced gameplay experiences, but also marks out other experiences by their absence; it points towards the future through both what it is (which is not desired again) and what it is not (which is desired). The present is the moment where one of those desired, previously absent experiences is hopefully played out, and becomes an enjoyable repetition rather than an identical generality through its difference from what has come before, while nevertheless remaining within the same broad constraints, namely being the same game. A specific outcome is not desired; indeed, we might say that a specific outcome is actively that which is to be avoided, since the player knows only of specific outcomes they have previously experienced (in the past) which represent the body of replays they have no interest in repeating. This forms a two-way relationship, where replay value is predicated on the unpredictability of the player’s experience, but also the player’s experience of unpredictability; so Deleuze’s conception of the syntheses of past events in the anticipation and recognition of future ones offers us a valuable theoretical hook to explore this notion in greater depth. In replay value the newness of the second playthrough is not just new in its own right through its actualization of elements previously unseen, but it is also new in relation to the previous experience, whereby the apprehended and reproduced experiences differ sufficiently to create a sense of Deleuzean repetition, and therefore to inculcate a feeling that the replay was worth pursuing. Replay value is enjoyed in the moment of replay, in the process of replaying which might take moments or potentially days or longer, not in any clearly defined end point. In summary, the reward of replay lies in the experience of finding new ways to experience a game previously experienced, and in the case of unpredictable games this new experience is predicated on a different set of permutations being ‘selected’ by a game’s unpredictability systems to be offered to the player. However, in the case of grinding, the value in returning lies in the acquisition of items, or in-game goods, or access to in-game areas, that are sometimes accessible as a result of performing a particular action, but are not always available as a result of performing that action. The process by which grinding takes place is only understood as a means to an end; it might or might not be inherently enjoyable in its own right, but the motivation towards the continued performance of that action lies in the desired outcome, not a desire to experience the same process or carry out the same action over and over again. The reward, therefore, lies in the outcome of the grinding process, and the change in the player or player character’s options, abilities or experiences, as a result of successfully grinding an unpredictable scenario enough times to reach the desired outcome. The unpredictability of a particular action–outcome connection creates the need for grinding in players who desire the positive outcome that might arise; and yet, the corollary to this observation is that grinding is often deliberately

REPLAY VALUE AND GRINDING

175

implemented as a way for dedicated players to achieve difficult outcomes or acquire rare items. As such, both replay value and grinding are, in their respective ways, a matter of rewarding the player with singularities previously unseen and unexperienced. Replay value rewards the player when entire possibilities which were foreclosed on in a first playthrough are now, instead, opened up. In this regard the player is oriented specifically and deliberately towards the creation of the new: Deleuze notes that ‘actualisation or differenciation is always a genuine creation’.19 Pursuing replay means that players force a differenciation in an unpredictable game, and in doing so create an unpredicted experience, an experience which did not exist prior and was likely not known to exist prior, rather than the pursuit of a specific desired outcome. In replay value we find a(nother) way by which unpredictability serves the purposes of creation: through enabling the creation of not just new world maps or new constellations of enemies and items, but potentially new narratives and the actualization of experiences previously unrealized, with which the player is rewarded in exchange for throwing themselves once more unto the unpredictable breach. Grinding, meanwhile, rewards the player willing to redo a single segment of a game over and over, with a simple and one-sided reward. This reward is one where only one element of the game varies, shifting from being generality into being repetition. Yet it is not new, but rather expected – it represents the actualization of game elements based on anticipated or expected virtuality that had been established previously in the mind of the player (through in-game information, knowledge gained from a source such as a walkthrough or guide, or a logical deduction that if items A, B, C, E and F are available, item D is probably available too), but it does not, of course, resemble the game’s virtuality, rather being produced through the rules that virtuality offers. Deleuze is highly critical of any kind of practice we, as games scholars, might define as grinding: he states that for the average person ‘even when [one] is given a situation of chance or multiplicity, he understands his affirmations as destined to impose limits upon it, his decisions as destined to ward off its effects, his reproductions as destined to bring about the return of the same, given a winning hypothesis’.20 We have previously seen this understanding in our examination of mastery as opposed to skill for games of unpredictability, and the opposite pursuit of a ‘lucky’ playthrough in which one might win. In grinding we see the same concept on a smaller scale. In replay, repetition’s value is the creation of a new identity for a new playthrough; but doing something over and over in order to get a specific identity which a player believes or knows is ‘out there’ is not a true repetition, for it is the forced enacting of something already known. This is as opposed to the open, unexpected, surprising creation of something new (even if that newness might contain elements a player might have guessed at from previous experiences, the specific newness is not anticipated nor desired in replay, whereas it is in the case of grinding).

176

THE UNPREDICTABILITY OF GAMEPLAY

Games are not designed to prevent a player from getting replay value when they want it. Designers recognize the value of making it as easy as possible for players to pursue replay value, and thus it is a rare game indeed which would prevent a player from starting anew, or following a different path in the game world. There is therefore little put in the way of a player achieving what they want from a replay, which is to say, the experience of new transit through semi-familiar territory. However, games are designed to stop players getting what they want as the outcome of grinding; the very need to grind demonstrates that there is something – often arbitrary, often time-consuming – between the player and their desired outcome. This form of repeating is therefore not contingent on giving players what they desire within the game space, but is actually contingent on withholding that from players until sufficient grinding has been done that the desired occurrence is actualized. As such, whereas the act of repeating inherently yields reward when replaying an unpredictable game, the act of repeating does not inherently yield the reward when grinding. Nevertheless, numerous games have implemented ways to make grinding less tedious or less trying, while still allowing for grinding to take place as an ‘accepted’ gameplay mechanic. Allowing players to shorten the grinding period we might take as a tacit acknowledgement that many dislike grinding, and yet grinding is maintained as a legitimate method for the acquisition of some items, or other beneficial in-game options.21 It is to these games, and what we can learn from them about how players approach unpredictability, we now turn.

Ease and difficulty in grinding Many games have ways to increase – but not necessarily to a probability of 1 – the chance of drops from enemies. For example, in the Soulsborne (2009–16) series of RPGs there are numerous items (with permanent effect while they remain worn by the player character), consumables (used once, lasting for a period of time, and then wearing off) and traits (possessed internally by the player character which can nevertheless sometimes be reduced by death or other events), which increase the rate at which all enemies drop items upon death. In a similar vein, many games – including this series – often give player characters a ‘luck’ trait. This will likely affect all forms of chance, but certainly normally includes drop rates of the sort described here. The presence of luck as a trait, and one which in many cases the player can actively affect and improve, shapes player experiences of grinding while also reflecting older cultural notions of fate, fortune and destiny we have not yet considered. Specifically, three cultural notions, that of luck as an inherent trait, something that can be improved, and something which is transient and passing but can be taken advantage of

REPLAY VALUE AND GRINDING

177

when present, all appear present in games that allow a character to affect their ‘fortune’ in such ways. A brief summary of these cultural forms will lay the groundwork for examining how player experience of unpredictability is affected by the in-game ability to change the positivity or negativity of that game’s unpredictably, which is to say to briefly or gradually take command of its unpredictability and shift it to one’s will, and it is to this we now proceed. Luck as an inherent trait has been a global cultural constant for much of history. Lisa Ann Raphals shows that in ancient Chinese culture, the power of the divine managed the ‘division and allotment of shares according to the commands of the gods (or ancestors), whose orders had the force of fate’. In the ancient Greek case, we see the spinning of thread apportioning good and ill fortune or the lengths or life or the natures of death, much akin to the Chinese.22 In both cases, she notes, we are confronted with the idea of fate as an endowment,23 as something given to us, to possess as a property of our selves. In terms of luck as something to be improved, meanwhile, in modern-day Vietnam, one’s success in trade and the accumulation of wealth are seen by many as part of their fate as decreed by heaven. Yet rather than promoting a fatalistic attitude and a disengagement from life, cultural tropes of hard work and discipline – increasingly universal around the world with the spread of neoliberal subjectivities, although still, as here, shaped by deep national and cultural specificities – remain strong, and working hard in one’s trade is seen as a way to pursue moral virtue,24 which in turn improves the propensity towards trade and prosperity handed down from above. Lastly, in terms of luck as something to be used, Machiavelli’s writing on Fortuna emphasizes the ability for good fortune to be used and turned to the ends of the skilled person, and indeed that one of the skills of that individual must be identifying when moments of chance have occurred, and acting accordingly.25 Similarly, in the contemporary Mongolian context, ‘fortune’ is once more a ‘bodily affect’ which can change and fluctuate; ‘heightened fortune allows a player to perceive what had previously been imperceptible to them – including a winning move which, ordinarily, only that player’s more virtuosic opponent would have detected’.26 Here fortune is seen as something which comes and goes, something which ebbs and flows, rather than an absolute; it is also something which can be used by the player of a game (or an actor in any other space) to potential benefit. Equally, this same intermingling of fortune and character is also apparent in the Arabic world, where wisdom entails turning ‘chance’ into ‘good fortune’ – which is to say, transforming an unpredictable event of any potential moral worth into a specific outcome of positive moral worth and positive long-term impact.27 These are only a minute portion of the examples one might draw on to illustrate these three concepts, but all three of these ideas – inherent luck, improving luck and taking advantage of luck – are all globally visible cultural forms.

178

THE UNPREDICTABILITY OF GAMEPLAY

As such, I propose that luck traits or stats or abilities in games are indicative of these three cultural notions of fortune. In the first case they reflect the notion that luck can be something inherently possessed by an individual, and which in turn will then shape their life from birth until death. The presence of a luck ‘trait’ or ‘stat’ perceptually removes unpredictable outcomes from the game’s systems, from some unknown calculation taking place somewhere within the game’s systems, and instead position these as being present within the player’s character. A poor outcome can now come to be seen as a reflection that a player character has unusually low luck, rather than a reflection of the vicissitudes of an unseen system. This is something internalized within the player’s avatar (or equivalent), not externalized and manifested in their environment. In both cases these bring some of the agency or control over unpredictable outcomes into the player’s character, and as I will argue in this section, this is best understood as an attempt to give players a sense of agency over grinding, and consequently make the practice of grinding more palatable or acceptable. In the second case, in games where one’s luck stat can be changed, the presence of such stats and the ability to improve them reflect the notion that luck is something that can be changed and improved with appropriate actions, appropriate lifestyles or appropriate religious, philosophical or moral behaviours. This, again, makes it appear to be something in the player’s control, something which can be worked on, improved. We can understand this in greater detail through the distinction Gerda Reith makes between ‘magical’ and ‘religious’ thinking vis-à -vis unpredictability. A magical belief places the power to affect unpredictable outcomes within the player, represented by ‘power or powers possessed by the individual’; by contrast, a religious belief places this outside the player in the purview of ‘some kind of providential authority’.28 Luck stats change a player’s perspective on the game’s unpredictability from one which might be (albeit implicitly and subconsciously) religious into one which might be better understood as magical in nature. This, in turn, brings deciding when to grind into what we might call the player’s locus of control, which is the third cultural form: rather simply doing the same task until the desired solution is created, players can also make strategic choice to enhance their grinding ability, which is to say, to shorten (on average) the length of time it will take to grind for a particular objective. It appears this shortening – even if it only shortens the grinding period from a ‘very long’ time to a ‘long’ time – is intended, and perhaps correctly, to make grinding seem more predictable, more manageable, more within the player’s control. Throughout this work we have seen many ways of players seeking to regain control over unpredictable games, either in a mechanical sense or in a personal, emotional sense through the development of particular discourses. Although replay value is about ‘giving up’ control over unpredictability in an experiential sense, grinding is about gaining

REPLAY VALUE AND GRINDING

179

control over unpredictability, about forcing the game to (eventually) offer up what the player desires, and these items appear designed to feed into this need to control the act of grinding, although even with them (and luck stats) it remains fundamentally outside of the player’s agency. It is also interesting to note the two narrative discourses that surround these items in the Soulsborne series. These narratives explore different attitudes both to grinding as a whole, but also to the reduction of grinding, and the change in the player’s experience when they make an active choice to reduce their time spent in the practice. In the Souls games (2011–16), the in-game descriptions of the ‘Covetous Serpent Rings’ – a pair of items, one of which increases item discovery, the other of which increases the acquisition of souls, the ‘currency’ in the game – mention ‘gluttony’ and ‘greed’ as the associations with acquiring increased volumes of these rewards. By contrast, in Bloodborne (2015) the ‘Eye Rune’ boosts item discovery and states that it allows the player to make ‘additional discoveries’, just as another discovery-boosting item from Dark Souls 3 (2016) mentions this bonus as the ‘fruit of [a] lifetime of research’. These item descriptions therefore put forward two notions. In the first case, this is that the acquisition of greater rewards is almost morally undesirable, something inappropriate, indecent, and that perhaps the player should be implicitly satisfied with what they already get, and pursuing greater acquisition reflects, perhaps, a particular set of motives on the part of the player. These discourses suggest – albeit gently, and perhaps tongue-in-cheek – that the player should be satisfied by what is actualized when playing normally, and perhaps recognize that players who use these items to boost their item discovery values will likely be completionists, or pursuing the game’s most difficult achievements, or some personal hard-to-achieve metric of success. In the second case these are framed in a far more positive way, as something to be discovered (with the positivistic ideological framing of the notion of discovery, something found anew, something previously hidden by nature but uncovered by the skilled seeker) and as something which is a reasonable reward for the player who elects to use them. In this regard grinding is presented as something which leads to justified rewards for the committed player; the desired outcome is ‘out there’, in some sense, and waiting to be acquired. Although these are undoubtedly passing comments, the overall Soulsborne series is notorious for the meticulous use of text and language, thematic depth and meticulous attention to detail,29 so it seems unlikely that this phrasing is accidental or unintended. They take the opportunity to deploy two contrasting discourses about the appropriateness of grinding as a behaviour, acknowledging that such behaviour exists (otherwise such items would not exist) but presenting the behaviour in two very different lights. Even more interesting is the pair of descriptions from the ‘Rusted Coin’ and ‘Rusted Gold Coin’ items in Dark Souls 3, which are one-use consumables that can be crushed by the player character for a temporary

180

THE UNPREDICTABILITY OF GAMEPLAY

boost to item discovery: the former states that ‘those who have lost their fortunes rely upon this superstitious practice, hoping to retrieve what was once theirs, and more’, while the latter states that ‘a rich man lost his fortune, but it returned to him twofold. He was swift to squander his retrieved fortune, smugly confident that it was bound to double once again’. I would suggest that each of these exists in dialogue with a number of broader discourses surrounding game unpredictability which we have explored in this work. In the case of the Rusted Coin, we see a nod towards the gambler’s fallacy,30 the role superstition plays in assessing and perceiving game unpredictability – such as item discovery and drop rates – and, I would suggest, the relationship between player actions and game outcomes of the sort explored extensively in Chapter 2. This is an item, as discussed earlier, designed to shorten or mitigate traditional grinding behaviour in order to acquire items they lack or items the player believes they deserve; the item description, however, counters such a notion, implicitly noting that even with the use of the Rusted Coin the player still might not be successful in acquiring the item they desire. In the case of the superior Rusted Gold Coin, which grants a larger boost to item discovery, a different set of ideas are mobilized. This is what is known as the ‘hot hand’ fallacy, which is in many ways the inverse of the gambler’s fallacy. This is the belief that a particular player is ‘running hot’ as a result of unusually positive results in the past, and is therefore more likely to experience success: ‘If an individual has won in the past, whatever numbers they choose to bet on are likely to win in the future, not just the numbers they’ve won with previously.’31 In the case of this item, we might see this manifested if we consider the relationship between what (incomplete) knowledge the player might have about the game’s unpredictability and their previous engagements with that unpredictability. Having appraised one pattern of unpredictability, a player might readily come to believe that particular set of actualized elements or distribution of differenciated parts is more common, regular and possible to anticipate than, in a mathematical or systemic sense, it actually is. These item descriptions reflect these two significant gambling fallacies, and are interesting both for this reason, and for the implicit understanding lying beneath this: that grinding is gambling, and one is ultimately ‘paying’ with one’s time, in pursuit of only a single desired outcome. There is therefore a curious paradox here in the presence of these items, their functions within the game world and the fact that they nevertheless stop short of fully achieving their intended purpose. Drop rate modifiers of this sort can reasonably be seen as an acceptance that the practice of grinding is not necessarily especially exciting, and should therefore be accelerated; and yet they stop short of completely removing the need to grind, despite this being the implicit gameplay direction all such options suggest. Looking at these systems as a whole, the claim is therefore as follows: grinding can take less time for the items you want, but you should still grind. Admittedly the

REPLAY VALUE AND GRINDING

181

use of items, abilities and consumables of this sort do lend a small strategic element to grinding gameplay, such as electing when and where to use them, dealing with the potentially negative effects of these items, and building one’s character in such a way that ‘item discovery’ is as high as possible, but grinding nevertheless remains present as a tacitly accepted way to engage with unpredictability, even when items exist that covertly acknowledge the design flaws and tedium of grinding as a way to achieve in-game progress. Luck stats and luck-boosting items are consequently all designed to return to players a sense of agency. Whereas one would traditionally recognize that unpredictability is out of the player’s control – note in this case the use of the term ‘luck’ for most of these stats, given the associations with luck we have explored elsewhere in this work – the presence of these stats or these abilities in the player’s character relocates the ability to determine the outcomes. No longer are these entirely in the domain of the generative system, but shift to becoming something that the player has some control over. By raising a luck stat over the duration of a game, one gradually gains more and more control over the outcomes yielded from moments of chance; by temporarily boosting a luck stat, the player is able to choose when the possible outcomes are especially important to them, and make an active choice to reshape a particular sequence of unpredictability in the manner they desire. The use of these is especially noteworthy when we consider that any player of these games is likely to experience long chains of grinding experiences without the desired outcome. As Peter L Bernstein argues, ‘The loser wants a short run to look like a long run, so that the odds will prevail’32: if one has thus far been unable to secure the grinding outcome one desires (and thus one is a ‘loser’) boosting one’s luck temporarily increases the chance of the immediately subsequent outcomes pushing the overall set of outcomes closer to the average. As such, allowing the player to reduce grinding yet still including grinding in one’s game design is not, in fact, the paradox it first appears, despite seemingly acknowledging the undesirability of the practice while allowing the practice to remain. Allowing the player to reduce grinding is a method for maintaining the gameplay objectives of systems that encourage grinding behaviours, such as extending the play of a game or having moments of chance which can significantly shape the player’s experience (but can potentially be grinded), while allowing them to take a greater sense of control over those unpredictable gameplay systems by defining the conquest of chance as a form of skill in its own right.

Completionism and exhaustive play Moving now from grinding into replay, in this section I want to explore this synthesis of past representation with future anticipation by considering the following question, worded in two different ways: When does a player

182

THE UNPREDICTABILITY OF GAMEPLAY

feel a game is ‘complete’? Or, alternatively, when does a player feel they have ‘seen it all’, and there is nothing left to do, and their multiplicity of the game is complete and seems to match the generative system at a resolution they consider acceptable? In almost all cases answering such questions necessitates extensive amounts of replay; most games close off certain paths while others open, with the mutual exclusivity of potentially parallel arborescence lines preventing a player experiencing everything in one go. Completing a game in this replay-heavy manner, which is to say completing sufficient variations of a game to be satisfied, is on one level another method for a game-playing subject to individually handle, accept and gain some feeling of agency over the vicissitudes of unpredictability as we have seen throughout this work. Yet completionism is perhaps also the most extreme expression of game unpredictability, and the most extreme embracing of that unpredictability, for players want to see all repetitions and experience all the unpredicted differences across the repetitions that a game can give them. For both of these reasons, a study of game unpredictability and its effects on players cannot be complete without an analysis of the completionist phenomenon, the practice of which is a subset of more general practices of replay. The concept of ‘completionism’ and the behaviour of the ‘completionist’ player represent a particular attitude towards games and an attendant structuring of a player’s priorities in playing an unpredictable game, navigating that game world and understanding the worth, value and longterm meaning of their own play activities. We previously noted that replay value is, in a sense, about giving up control and not wanting a specific permutation of the game, so long as that permutation is a genuine Deleuzean repetition and therefore new. This is true, but through completionism, this can shift from wanting something which is new but not specific, to wanting specific forms of newness which become increasingly specific as more and more are actualized. The completionist, I will seek to demonstrate here, is a particular kind of gaming subject, one who addresses the fundamental differences in an unpredictable game in a unique manner, and consequently shapes their play in a way designed to actively generate and seek out singularities. We can usefully begin by summarizing extant literature which gives us an initial handle on the completionist as a kind or style of player. Attempts to categorize players and their motivations have been present in game studies for decades: perhaps most famous are the categories of Richard Bartle,33 who proposed grouping players according to what aspects of play they most enjoy into ‘Killers’ (competition with other players or AI opponents), ‘Achievers’ (pursuit of achievements, points or other markers of accomplishment), ‘Socializers’ (interacting with other people or virtual avatars) and ‘Explorers’ (discovery, finding secrets) according to the forms of gameplay they preferred, and the objectives and forms

REPLAY VALUE AND GRINDING

183

of satisfaction and self-actualization they pursued within spaces of play. In these models and those which followed it, the ‘Achiever’ seems closest to the ‘Completionist’, but does not tell the entire story. More recently, as Chris Vammen and Jennifer Perkins put it, in what appears to be the first academic use of the term in a video game context, ‘A completionist is one that has to accomplish everything that the game has to offer’.34 Olli Sotamaa defines35 this slightly differently, as signifying the need to ‘perform all possible tasks’ in a game; where the first evokes a sense of mastering a game’s challenges, the latter suggests a more exhaustive set of activities, and a total appraisal of the options presented by a game’s systems and content. As we will see, these distinctions are important: There are many ways we might understand a completionist and their approaches to exhausting a game’s unpredictability, and each of these has repercussions for our understanding of completionism in games that contain any kind of unpredictable element. The drive for completionism and the existence of the completionist player have also been mentioned in passing by other game scholars,36 but it has not been a focus of study to date, nor unpicked in detail with regard to what it actually means to pursue ‘completion’, especially in the context of an unpredictable game with a potentially nearinfinite number of permutations. I propose that the notion of completionism in unpredictable games can potentially take three different forms. All entail the repetition of randomness, chance and luck, but differ both according to how much the player must do in order to ‘complete’ all the experiences an unpredictable game can offer – I relate them here in descending order, beginning with the most exhaustive, and finishing with the least exhaustive – and how commonly players actually attempt to ‘complete’ an unpredictable game according to these frameworks. In this case, the first is the least common, and the last is the most common. Despite the drive for completionism as a concept seemingly felt by many players, the observation that many players seem to work within the least demanding form of completionism, with perhaps no player adopting as their mental model the most demanding form of completionism, suggests that the feeling of completing everything in an unpredictable game (in a psychological sense) is of more importance and consequence to many than a stricter, more formal, mathematical exhaustion of all a game’s possibilities. As such, this shows us some of what players of unpredictable games are interested in when they continue to play the same game or elements of that game repeatedly; they perhaps acknowledge tacitly that the possibility spaces of these games are too broad to be ‘truly’ completed, but nevertheless long for the feeling that if they cease playing the game, they will not be ‘missing out’ on anything they might otherwise have experienced. We now examine these three kinds of completionists in detail, and what each can teach us about players and their attitudes towards navigating the sometimes vast possibility spaces of game unpredictability.

184

THE UNPREDICTABILITY OF GAMEPLAY

Completionism of all possibilities The most exacting completionist we might dream up wishes to see everything possible within a game. They want to ponder every possible position in chess, from the perspectives of both white and black; they want to visit every possible dungeon level that might arise in a roguelike, and navigate every possible orientation of ‘game pieces’ (creatures, items, terrain features) arrayed within; they want to play out every possible deal of bridge, from all four positions, with all possible successful bids. All rhizomes must be brought into play, and tethered together in every possible structure. The ambitions of this first completionist are somewhat hindered by the impossibility of doing any one of these in a human life time, and – assuming a normal human memory – the difficulty of (for example) remembering the one PCG terrain tile in this permutation which is slightly different to the terrain tile in that almost-identical permutation millions of permutations ago. Nevertheless, this would appear to be the most comprehensive rationale for the exhaustive ‘completion’ of an unpredictable game and the attendant exhausting of its replay value; the generative system is exhausted, the multiplicity is comprehensive, all aleatory points seen and singularities navigated, and the difference between each permutation – whether large or small – witnessed. Such a player would then become one who sees the entirety of the multiplicity of an unpredictable game, and would thereby come to understand the entirety of what might be actualized from that game, all which is virtual in that game, and therefore every possible aleatory point and resultant singularity which might be experienced by a player, or rather, by all players. We can see a literary model of such a dedicated completionist in the Jorge Luis Borges short story ‘The Aleph’, in which the narrator – a fictionalized version of Borges himself, often the case in his work – meets another character who, with an over-inflated sense of his own poetic ability, wishes to write a masterwork that describes every part of the earth in great detail. This would be impossible were it not for this character’s discovery of an indescribable something, a crack in the universe which he terms the ‘Aleph’. Those who gaze into the Aleph are able to simultaneously perceive everything in reality from every perspective – quite the useful aide for such an undertaking. Borges’s narrator-self later describes the range of experiences he sees gazing into this spot in the universe (which is, he stresses, simultaneously all spots in the universe). Some of the things seen he talks over at length, in detail, as if seeing a person, and then their hair, and then their clothes, and seeing multiple elements of the same thing over and over; then it pans out and back across a dizzying range of other qualia. Given that we have looked at the rhythms and cadences of unpredictability through this work, we can propose the emergence of another sensation of the patterns of unpredictability here. This player would likely spend some time ranging across a set of totally different permutations, and then spend

REPLAY VALUE AND GRINDING

185

some time in dozens or hundreds or thousands or similar ones, due simply to the unpredictable order with which the unpredictable permutations are actualized, and indeed perhaps spend time in some permutations where one could not even distinguish the difference, even with some extrahuman method of knowing these are different when one cannot see the nature of that difference (they resemble generality). Nevertheless, once a game’s generative system is exhausted, this player would certainly have extracted maximum replay value from the game, with every permutation of the formal properties of the game having been played through from every possible angle. Assuming a perfectly accurate transmission and cognition of information from the game to the player, their multiplicity of the game should match perfectly the generative system itself. However, as above, such a completionist could never actually exist in reality, except for an unpredictable game so simple it is unlikely anybody would feel especially inclined to experience it in this kind of depth. One has constraints of time, which limit how many permutations can be experienced in a lifetime; and one has constraints inherent to the unpredictability, which is to say that one cannot guarantee that permutation x+1 is actually unlike x and all the other preceding iterations. The closer one becomes to having seen every permutation, the less and less likely a new permutation becomes each time a permutation is generated, slowing down the accomplishment of their perverse objective even further. As such, to meet the desire for completionism, compromises must be made, and players have found two ways to make such a compromise: by experiencing every significantly (to a given player) distinct permutations (with permutations which are too similar to others being passed over); and by looking to an outside source, generally a game’s designer, for a ‘list’ of permutations or scenarios deemed to be worth experiencing or pursuing – once all those are completed, the player’s process of experiencing what the game has to offer is considered ‘complete’. In both of these forthcoming cases compromises are made which are seen to be in keeping with the spirit of the completionist, even if not, in a strict Deleuzean sense, doing what would be necessary to call oneself a truly exhaustive completionist. Examining these two models and the underlying implicit assumptions they build upon will help us to understand what precisely it is players hope to achieve in pursuing a completionist ethos, and the ways completionism in practice (rather than, as with this first model in theory) interacts with unpredictable games. We begin with the completion of all discrete experiences.

Completionism of all discrete experiences The second form of completionist we can imagine differs in two ways from the first: they need to see fewer of a game’s total possible permutations in order

186

THE UNPREDICTABILITY OF GAMEPLAY

to be ‘satisfied’; and they actually exist. As we noted in the above section, many permutations in most unpredictable games, when taking a broad view across all possible permutations, are extremely similar in practice. Any given shuffled deck of cards will always be only one card position away from thousands of other shuffles; any given generated dungeon will always be one tile feature or property away from thousands of other almost-identical permutations. As such, this more pragmatic completionist appreciates that the difference between two of these permutations will, in many cases, be effectively zero. Instead of attempting to see every permutation, they instead only look for every permutation which is significantly distinct from other permutations they have encountered, an approach which will cut the number of permutations in any unpredictable game that need to be seen for the game to be ‘completed’ down to a fraction of its original size. However, given the complexity of many unpredictable games, this might still mean an unrealistic number of permutations to be experienced before this completionist will be satisfied. Although witnessing every discrete experience appears at first glance to be such a bold request that it is all but indistinguishable from the first, it is therefore a question of – to return to a concept explored earlier in this work – the granularity of interest. What does one player define as a discrete experience? For a player of an unpredictable game, this means the player must perceive a significant difference, and must see repetition rather than generality. Recall the discussion earlier in this work about the central importance of subjectivity to Deleuze’s work, and in turn, to this present project: the first version of the completionist takes the formal generative system of an unpredictable game as the ‘benchmark’ for completion, whereas this second version takes the mental model of that generative system, which is to say a multiplicity, as the benchmark for completion. Therefore, this completionist is focused on the divergent experiences they get when they see something as repetition (the repeating of difference each time they play, the actualizing of the new) and when they see something as generality (they find themselves playing through something which they feel they have played through before, even if that game scenario generated differently or might contain, in a formal sense, significant divergence from the other that the player believes resembles this one just a little too closer). These are, of course, personal metrics. As DeLanda states, ‘Deleuze agrees that things resemble each other, of course, but these are the results of processes, not the foundation of a complete ontology’ (33) – which is to say, any commonality sighted between replays will be a matter of the process by which a player regards and assesses the commonality (or lack thereof) of seen permutations, not through any deeper process. For some players, therefore, witnessing every discrete experience, by their own faculties of judgement or what experiences are and are not similar enough to be comparable, might take hundreds of permutations, or even merely dozens. Where one player might see a subtle

REPLAY VALUE AND GRINDING

187

distinction that yields new tactical possibilities, another sees only the same scenario with an enemy in a slightly different location; where one sees a subtly intriguing repetition and remains interested in what the game can offer, the other sees generality, and loses interest. We can of course never say that things are truly the same, ‘so when we say that they are the same, we are saying that they are sufficiently alike, that they resemble one another enough for a given purpose’.37 The depth of player understanding is therefore an essential component of replay value; the deeper a player’s understanding of the game, the finer a granularity of difference they perceive, and the greater the number of permutations they find distinctive, the longer it will take to exhaust replay value. With this in mind, let us now consider an illustrative case study which demonstrates how this kind of completionist addresses a game, the challenges they may face and how the experience of one such completionist varies from another. In the months and now years since its release, much has been written of open-world space-exploration game No Man’s Sky (2016); few games have captured both the fascination and ire of such a wide swath of the gaming public, and few games have met with as much passionate debate as to their merits and demerits. Here I will focus on critical and popular commentary regarding the relationship between three elements, which are relevant to exploring the notion of completionism, the role of unpredictability and the attendant player experience. These are the portrayal of the game’s procedural generation systems, players’ immediate experiences of those systems and players’ wider reflections on the difference between one region of the game and the next, and therefore how close to completing the content of the game it is possible to come. No Man’s Sky depicted its procedural universe using a particular set of both linguistic terms and rhetorical devices, and a particular set of visual offerings to the player through adverts and other promotional media. From the very first, the game was ‘marketed as a game that defined itself in terms of its procedural generation’.38 The initial trailer is noteworthy for making two particular claims: that the game offers a world with Every Atom Procedural and a universe with Every Planet Unique. These implied two things. First, that even down to the smallest scale every world and everything on that world would be unique, and we would truly get procedural generation and the kind of length and co-dependence noticeable in real-life ecosystems and environments. Secondly, more obviously, that no two worlds would be alike. In these trailers it was made clear the diversity that players could encounter across the planets in the universe; one might be a stormy land of mountains inhabited by quadrupeds, another might be an ocean world with gigantic fish, another might be occupied primarily by flying creatures soaring over grassland, and that not just did these ecosystems vary wildly, but the colour palettes, topography (and therefore navigational gameplay), and resources of these systems varied to the same

188

THE UNPREDICTABILITY OF GAMEPLAY

(significant) extent. As games researcher and designer Michael Cook notes in his article ‘How We Talk About Procedural Generation’ (2016), popular and designer-led discussion around PCG ‘uses words like discover, unique, endless, forever, replayable. It talks in numbers and powers of ten, and bigger is always better’. In this vein, No Man’s Sky showed off its PCG systems, promised infinite variation, and therefore infinite difference and infinite repetition; and, crucially, that things would always be different enough across worlds that one would not end up routinely experiencing the same ones. In turn, the game’s trailers – quite naturally – portrayed some especially interesting worlds, having little to explicitly say about the risk of less interesting planets, but presenting a universe where not just were all worlds distinct, but all worlds were interesting, compelling and worthy of the player’s time. Coming now to the experiences of players, it soon became apparent that many players’ experiences of the game (whose experiences came to dominate the overall discourse surrounding the game) were not in line with the experiences implied by these same marketing techniques. Especially telling in this regard is a well-shared video entitled ‘No Man’s Sky Release Trailer’. It begins by showing one of the game’s most famous early trailers: towering dinosaur-like creatures roaming a verdant land, brilliant white and red foliage, ibex-like creatures jumping through a rich grassland. This section is labelled, in this fan-made video, ‘Gameplay Trailer E3 2014’, noting that this trailer was released at the ‘E3’ gaming event in 2014, and presented itself as showing accurate gameplay footage. The video then cuts to showing a barren brown landscape, devoid of any especially compelling terrain features, with an unusual creature passing in front of the virtual camera: a top-heavy dinosaur teetering on spindly legs, something like a Tyrannosaurus heavily weighted towards the upper half of its body, and unable to achieve any kind of smooth or continuous locomotion. Intentionally comedic and jarring background harmonica music plays, and the overall image is one of absurdity; an evolutionary dead-end stumbling around a deserted world, and a far cry from the towering megafauna and multi-coloured alien flora of the game’s promotional material shown at the start. At time of writing this video has been viewed two and a half million times, has been overwhelmingly ‘Liked’ by those who view it, and is filled with comments from viewers expressing their agreement with the claim being made here. Which is to say, first that the experiences of players did not match how the game was originally marketed; and secondly, that the manner in which experiences did not match is unusually striking, and perhaps for some players even seemed actively deceptive. The game’s unpredictability might have yielded some of the experiences the game publicly presented, but also yielded a range of deeply uninteresting experiences alongside it. As a result, unhappy with what completionism would actually entail, and how fun and interesting it would (not) be, players begin to feel that

REPLAY VALUE AND GRINDING

189

completionism was no longer something exciting centred on the exploration of a complex universe, but something that would mean a lot of tedium and a form of tedium that they been inadequately prepared for in their play and observation up to that point. In turn, the reflections of players and commentators on how close they were to ‘completing’ the range of possibilities offered by the game are telling. Having been promised such a vast and varied universe, players found it challenging to assess what portion of it they had actually seen; and with the marketing of the game having focused so heavily on this aspect, a significantly greater percentage of players than in most games became interested in the question of completionism. Too many planets were similar, producing a sense of what completionism would entail that was quite different from that shown in the trailers. Other players felt it should be simpler to find other planets, struggling to break out of cycles of worlds seen as having too much in common. As result, players were uncertain how many planets they should see, and whether the kinds of planets they had thus far encountered were the kinds of (and variety of) planets they were supposed to encounter, or whether they had simply proved unusually unlucky. The experience of playing No Man’s Sky consequently came to be understood as consisting not so much in the gameplay one experiences on one planet, necessarily, but rather on the wider experience of gameplay that takes place across and between planets, foregrounding the roles of difference and repetitions in what players expected. When these expectations gave way to a quite different set of experiences, and a set which did not seem indicative of the experiences promised, player reception becoming rapidly, and sharply, negative. We can now note the following. The initial portrayal of No Man’s Sky showed a universe of widely divergent experiences, each interesting and worth visiting. However, players instead reported experiences that were in the first case less compelling than promised, and in the second case too similar to other experiences they had encountered. As such, the experience of No Man’s Sky and its players allows us to consider the completionism of discrete experiences as a model for players who seek the feeling of having ‘completed’ all significant repetitions within a game’s multiplicity. To understand the perception of discrete experiences for these players who found that the game had let them down, consider a square grid, with x and y coordinates both running from 0 to 100. This represents, if you will, the complete possibility space of a game of randomness (of course a full representation might have a dozen axes or a hundred axes or more, but two is somewhat simpler to visualize). Now, we start using a random number generator to pick pairs of coordinates and place circles with a diameter of 20 centred on those coordinates – these are the gameplay scenarios chosen by the game, and the ‘area’ of the actualized possibilities they represent. The number 20 might be larger or smaller, depending on whether the player sees only the surface level of the game or understands the game in great depth,

190

THE UNPREDICTABILITY OF GAMEPLAY

respectively. In each gameplay possibility one sees, therefore, the player also sees all the very similar scenarios (according to their understanding of similarity) which are no longer needed in order for that player to feel they have seen the entire grid – (20,56) and (20,57) are extremely similar and it probably isn’t necessary to view them both, except for the most exacting and meticulous player, while (20,56) and (91,12) are very different and should offer quite distinct vistas.39 By continually placing circles of that same diameter, which each cover the same potential volume within the grid, the grid will eventually be filled and all coordinates, or what the individual player takes to be their perceptual equivalents, will have been experienced. However, sometimes this will take perhaps a dozen circles, when the random number generator selects coordinates well-spaced from one another; sometimes it might take hundreds, as circles are continually placed next to their neighbours; and, hypothetically, it might take a potentially infinite number of circles until the entire grid has been covered at least once. In the first case, the game will feel continually fresh, new and varied throughout; in the second case the game will ebb and flow between providing new experiences and rehashing old ones; in the third case the player might, simply through the game’s uncaring unpredictability, find themselves ‘trapped’ in a part of the grid they are forced to experience over and over, until they potentially abandon the game for good. By this point the player will be deeply bored of green, jungle, bright, large-animal, minimal-ocean, planets (or whatever the specific kind of planet might be), and all those future planets which intersect on any of these axes will be less interesting than they otherwise might have been. Manuel DeLanda encourages us to ask whether ‘the same individual [can] exist, slightly altered, in other worlds? Can he or she maintain this identity across many worlds, after several slight alterations have accumulated? Could we identify him or her after all these changes?’40 Studying the completionist who seeks to exhaust the space of discrete (to them) experiences helps to answer these questions; the maintaining or not of identity across small differences varies from subject to subject, through a gradual chain of differences that accumulate until a player is convinced they have shifted into something sufficiently new to merit a different identity. The completionism of discrete experiences, therefore, is a question of how one recognizes difference and the resolution in which one recognizes it, and how ‘broad’ the identities one perceives are (how wide a block of permutations does one see as resembling each other and consequently possessing the same identity).

Completionism as written into the game Our third imaginary completionist is the one with the most lax and most realistic set of requirements for satisfaction, and arguably the most common

REPLAY VALUE AND GRINDING

191

– the majority of ‘completionists’ seem to fall into this category. For this completionist, completing a game means completing everything that the game’s designers consider to be worth completing. How – without direct one-to-one contact between a player and the developer of a particular game, which is extremely rare – might any understanding of what the developer(s) consider(s) to be significant be acquired? This knowledge is acquired through players referring to an increasingly wide range of systems that video games have acquired in the last ten years: these are variously known as ‘Achievements’, ‘Trophies’, ‘Challenges’ and the like. ‘Achievements’ (I will use this term to refer to all such concepts) are meta-goals outside the objectives written into the game’s rules, and are akin to a list produced by a game’s developers of things a player might want to accomplish for amusement, edification or interest. These are present across a wide range of games and play experiences. In video games with detailed multiplayer modes through which players can ‘unlock’ new items and maps and ‘ranks’, we find players seeking to unlock everything that might be unlocked through their play; in games where players are rewarded with ‘Achievements’ or ‘Trophies’ for completing certain parts of the game, players pursue the acquisition of all these extra-diegetic markers of ability and perseverance; in many unpredictable games, as we have seen, the generative system is constrained when play begins and slowly grows in size and complexity as the player achieves various objectives. These are also comparable to the lists of potential objectives in player ‘guides’ for video games, produced sometimes by first-party developers and sometimes fans, which are designed to encourage players to ‘eke out every last gameplay opportunity’41 – as defined, of course, by that guide. The ‘ethics’ of Achievements, however, are contested. Some consider them to be amusing or compelling alternative ways to play a game, and can encourage players towards play regions or play experiences they might otherwise overlook. By contrast, others42 consider them to be arbitrary, distracting, uninteresting, and present only to lengthen the appeal of a game without actually adding any new compelling actions for players to engage with (as would be the case with additional Downloadable Content, for example). Nevertheless, despite both scholarly and popular debate, such elements are central to contemporary games. They do, however, vary in the kinds of actions they require from the player. Some Achievements will be reached simply by completing a game – there might, for example, be an Achievement for defeating a mandatory enemy for the first time. Other Achievements, however, demand actions of the player that the game itself does not: perhaps the player must complete the game only using a single item, or discover a particular secret, or complete the game a number of times, or talk to an arbitrarily large number of in-game characters. Although analogue games do not have Achievements, hypothetical examples will further elucidate the nature of the Achievement: in chess, for example, an

192

THE UNPREDICTABILITY OF GAMEPLAY

Achievement could entail winning without using any bishops or rooks; in Go, an Achievement might entail winning before the hundredth move; in poker, an Achievement might entail winning five hands in a row. These do not therefore affect the core play of a game, but are external goals that offer a sense of accomplishment without affecting, in a pragmatic or instrumental sense, player success or failure within a game world. Let us call these usual and unusual Achievements; the former are normally accomplished simply by playing within the spirit of the game, while the latter require the player to step outside those traditional bounds a little, but not so much that it becomes an exploit, and perform some other action instead. These latter recommendations to the player ‘signal and map out pleasures unassociated with the apparent objective of gameplay, frequently setting new ludic challenges and establishing supersets and subsets of rules within which to perform’.43 Achievements, therefore, represent a form of ordering: how to categorize what one has done in a game, and therefore how to assess how much one has done. They serve to take the vast range of play experiences possible within a game and order them into a smaller, orthodox selection of experiences, leaving surprisingly little room for truly heterodox play, because many Achievements capture what might otherwise be unusual ways of playing in the service of codifying supposedly significant play experiences. Where developers have not just predicted an unusual style of play, but actively put in place a system that rewards the player for that style, it ceases to be an unusual style and becomes something captured and standard. In this way we might propose that the player is shown what some of the most unusual experiences the game can offer might be, and gently encouraged towards performing them, widening their horizons of game possibility and what they might be interested in doing. Equally, many unpredictable video games are large in scope (especially if randomness is high) and as such potentially daunting to the new player; Achievements for these games allow players to gain a foothold on the breadth and scope of their unpredictability. In doing so, they help people to define the expected virtuality; of course, some of that virtuality is defined in a player’s mind simply through the play of the game, but Achievements,44 in a sense, allow a player to ‘skip ahead’ and gain a more complete mental model of the game’s virtuality than would otherwise be possible for the number of repetitions seen. This could therefore reasonably be seen as a method for helping players through the unpredictability of such games. Alternatively, we might also propose that such players want the feeling of ‘having done it all’, having ‘completed’ all the game has to offer, without having to work towards and go through the experience of actually ‘doing it all’. Following the achievements, trophies, unlocks or other metrics written into the game (or associated with the game on a particular content distribution platform) allows players to – once all are done – feel that everything of

REPLAY VALUE AND GRINDING

193

significance within that game has been accomplished, experienced. In the case of unpredictable games, this is, as with many other design and social phenomena described in this work, effectively an attempt to control the tremendous scope of unpredictability in some of these games and parcel down completionist practice into something smaller, more practical, more achievable. It also enables players to have closure which is reached not from the player’s reflection on their own practices and experiences and an informed decision that the game has nothing left to offer them, but rather from an external source which circumscribes the experiential boundaries of worthwhile play. Also, Achievements in some games require replaying the entire game multiple times which inevitably becomes akin to grinding; the experience of play, ordinarily central to replaying an entire game, becomes superseded by the outcome of play of the sort we see in grinding. In the pursuit of completionism, replaying an entire game is no longer about the new experience acquired, but rather ‘ticking off’ the captured experiences from a list of possible experiences. For most completionists, therefore, acquiring all Achievements is the point at which a game’s possibilities appear exhausted, losing the excitement of an unknown permutation of a game, a true repetition, and replacing it with the pursuit of specific and predicted permutations the player knew about, in some sense, beforehand. Ordering one’s actions though Achievements means reducing the unpredictability of the game to its smallest, in terms of what is unexpected when it appears, in terms of what must be done to feel all unpredictability has been seen, and in terms of the openness and variability of the completionist experience. There are therefore three models of completionism, through which a player attempts to define and subsequently experience the sum total of experiences offered by a game’s multiplicity. In the first case, a player wishes to see every permutation which possesses any formal variation in its actualization, however small. In the second case the player wishes to see every permutation in which they can discern a difference, which depending on the game might be anything from the same set as the first to a set infinitesimally smaller. In the third case the player looks to exterior sources, such as ‘Achievements’, to frame and structure what experiences they pursue and remove the onus on the player to establish and construct their own multiplicity, and decide on the detail of that multiplicity which in turn determines the choice of completionist rationale. Completionism takes replay to its most extreme manifestation, and entails the exhausting of a generative system and the construction of what appears to be a comprehensive multiplicity for that game, predicated on a desired to know everything that system can yield. This shifts replay from being ‘something new’ into a form of replay which requires the player to see ‘everything new’ when considered over a sufficiently long timeframe; a significant shift in player understanding of the replay experience. The pursuit of completionism thereby entails what we can term ‘exhaustive play’ – playing a game with the objective of undergoing all possible experiences

194

THE UNPREDICTABILITY OF GAMEPLAY

that, according to one’s preferred personal metric, together form a complete summary of what the game can offer the player. When faced with opaque unpredictable game systems, completionist players do not just tackle this opacity through more traditional forms of play which implicitly and tacitly construct knowledge, but instead take an active approach to unpicking the game’s generative system into their own multiplicity, and consequently deliberately seeking to exhaust what that multiplicity contains. In this chapter I have examined the concept of replay value (and the practice of replaying) and the practice of grinding. Both of these entail players doing a certain portion of, in our case, an unpredictable game numerous times in order to pursue something that they want to emerge from the game’s multiplicity. The first entails playing all or a significant portion of a game again in order to pursue a new experience, which could take any form so long as it does not adopt a form previously experienced; the second entails playing a small portion of a game over and over again in order to pursue a specific outcome, which can only have one acceptable form and might or might not have been previously experienced. The chapter argued that replay value is a question of experience while grinding is a question of outcome, and both represent quite different player attitudes to game unpredictability, what the purpose of unpredictability is and the appropriate way to act in order to gain what one desires. The chapter then explored both grinding and replay value in greater depth, beginning with grinding. We considered the concept of the ‘luck stat’ in games, which is to say a number embodied in a player’s character which affects the positivity of chance outcomes, and the existence of games which allow players to mitigate, although never entirely, the presence of grinding in their gameplay. In turn, we considered replay value through the sense of the ‘completionist’ impulse of gameplay, which is to say the exhausting of all the interesting elements a game has to offer. I identified three different completionist rationales which vary in their implicit attitudes towards both a game’s underlying mathematics and the emergence they generate, and a player’s experience of those aspects. In both of these cases the chapter demonstrated that grinding and replay value are both core player responses to unpredictability, both attempts to (re)gain a sense of control over unpredictable game systems and push those systems towards offering the player particular permutations, particular manifestations and particular experiences and outcomes.

Notes 1 This is most common with ‘retro games’, such as games from the 2000s or 1990s, now replayed by those in their late twenties or early thirties; many livestreamers on the website Twitch.tv are particularly noted for their dedication to (re-)exploring these games.

REPLAY VALUE AND GRINDING

195

2 David M. Ewalt, Of Dice and Men: The Story of Dungeons & Dragons and the People Who Play It (New York, NY: Simon and Schuster, 2014), 6. 3 Daniel L. King and Paul H. Delfabbro, ‘Understanding and assisting excessive players of video games: A community psychology perspective’. Australian Community Psychologist 21, no. 1 (2009), 62–74. 4 Daniel K. King, Paul H. Delfabbro and Mark D. Griffiths, ‘Cognitive behavioral therapy for problematic video game players: Conceptual considerations and practice issues’. Journal of CyberTherapy and Rehabilitation 3, no. 3 (2010), 268. 5 Frans Mä yrä , An Introduction to Game Studies (London, UK and Thousand Oaks, CA: SAGE, 2008), 132. 6 King and Delfabbro, ‘Excessive players of video games’, 70. 7 Daniel Cook, ‘Loot drop tables’. Lost Garden, 7 December 2014. Available at http://www.lostgarden.com/2014/12/loot-drop-tables.html. 8 Juul, Art of Failure, 60–1. 9 Grinding is not the only context where a form of gameplay that seems deeply unappealing can, when in concert with other contextual elements, become quite compelling. For example, it has been noted that players can lose tremendous amounts of money to a casino institution, and rather than expressing anger, sadness or making excuses, do so happily and express their contentment and appreciation of the opportunity to lose money and still have a good time. In this way ‘such a gambler is the epitome of the ideal player in terms of system integration’ – Vicki Abt., James F. Smith and Martin C. McGurrin, ‘Ritual, risk, and reward: A role analysis of race track and casino encounters’. Journal of Gambling Studies 1, no. 1 (1985), 73. Having been fully socialized by the game system, they accept their role as long-term losers, derive enjoyment from the experience, and become ‘integrated’ into a system designed for such a player. Just as in grinding, a play experience that might seem bleak and undesirable in the cold light of day comes, within the context of its presentation and the structuring of rewards and payouts it offers, to seem quite acceptable. 10 Damian Hughes, The Winning Mindset: What Sport Can Teach Us About Great Leadership (London: Macmillan 2016), 183. 11 Kevin Pocock, ‘Battlefield, and the importance of replay value’. Den of Geek, 30 October 2013. Available at http://www.denofgeek.com/games/ battlefield/27922/battlefield-and-the-importance-of-replay-value. 12 For example, see Erin J. Hastings, Ratan K. Guha and Kenneth O. Stanley. ‘Evolving content in the galactic arms race video game’. In 2009 IEEE Symposium on Computational Intelligence and Games (CIG, 2009), and Calvin Ashmore and Michael Nitsche, ‘The quest in a generated world’. In Situated Play, Proceedings of DiGRA 2007 Conference (2007). 13 Jeff Bakalar, Dan Ackerman and Scott Stein, ‘LittleBigPlanet 2: Redefining replay value’. Cnet, 18 January 2012. Available at https://www.cnet.com/uk/ news/littlebigplanet-2-redefining-replay-value/.

196

THE UNPREDICTABILITY OF GAMEPLAY

14 Luke Plunkett, ‘Your games have more replay value than you think’. Kotaku, 22 February 2015. Available at http://kotaku.com/your-games-have-morereplay-value-than-you-think-1687359958. 15 King and Delfabbro, ‘Excessive players of video games’, 70. 16 Henrik Schoenau-Fog, ‘The player engagement process–an exploration of continuation desire in digital games’. Think Design Play: Digital Games Research Conference (2011). 17 Faltin Karlsen, ‘Entrapment and near miss: A comparative analysis of psychostructural elements in gambling games and massively multiplayer online roleplaying games’. International Journal of Mental Health and Addiction 9, no. 2 (2011), 199. 18 Although the terms grinding and farming are traditionally used interchangeably, for the sake of discussion here they can usefully be distinguished. I propose that whereas grinding entails repetitive actions with uncertain outcomes, farming, although very similar, is distinct due to the rhythm of rewards given to the player: which is to say, they are constant and guaranteed, if often of minimal value. Grinding is therefore still a question of unpredictability at heart: one does not know which ‘instance’ of grinding will yield the outcome one desires, the outcome which is new and might transform the experience into a question of repetition rather than mere generality. 19 Deleuze, Difference and Repetition, 276. 20 Ibid., 147–8. 21 We might argue that grinding simply emerges from unpredictable drop rates which can be repeated, but it’s more complex than this; one can easily implement a way to guarantee having those items, or attaching non-repeating drops for repeating enemies or the like; there are many design choices that prevent grinding. We therefore cannot say that games which allow grinding merely ‘happen’, by interactional fluke, to allow grinding; the possibility of grinding requires a specific set of differential relations and elements, and the absence of others (listed above), which makes for a quite distinctive gameplay design unlikely to arise by accident, and certainly all but impossible to not be noticed by designers. 22 Raphals, Fate, Fortune, Chance and Luck, 560. 23 These differ in complex ways in terms of ‘what was apportioned, by whom, to whom, and to what end’, but this commonality of the giving out of fate, and therefore implicitly that fate is something received and internalized by an individual which becomes part of their nature, is clear. 24 Kirsten W. Endres, ‘“Lộ c bestowed by heaven”: Fate, fortune, and morality in the Vietnamese marketplace’. The Asia Pacific Journal of Anthropology 16, no. 3 (2015), 227–43. 25 De Goede, Virtue, Fortune and Faith, 32. 26 Swancutt, ‘The ontological spiral’, 244. 27 Dalya Cohen-Mor, A Matter of Fate: The Concept of Fate in the Arab World as Reflected in Modern Arabic Literature (New York: Oxford University Press, 2001), 190.

REPLAY VALUE AND GRINDING

197

28 Reith, Age of Chance, 159. 29 For example, see David Chandler, ‘Bloodborne and the history of horror’. Kill Screen, 30 April 2015. Available at https://killscreen.com/articles/bloodborneand-history-horror/; Andreas Inderwildi, ‘The artists and paintings that bleed through bloodborne’. Kotaku, 9 December 2016. Available at http:// www.kotaku.co.uk/2016/12/09/the-artists-and-paintings-that-bleed-throughbloodborne; and Chloi Rad, ‘8 dark souls details you probably never noticed’. IGN, 21 March 2016. Available at http://uk.ign.com/articles/2016/03/21/8dark-souls-details-you-probably-never-noticed. 30 James Sundali and Rachel Croson, ‘Biases in casino betting: The hot hand and the gambler’s fallacy’. Judgment and Decision Making 1, no. 1 (2006); Daniel M. Oppenheimer and Benoî t Monin, ‘The retrospective gambler’s fallacy: Unlikely events, constructing the past, and multiple universes’. Judgment and Decision Making 4, no. 5 (2009), 326–34. 31 Sundali and Croson, ‘Biases in casino betting’, 3. 32 Bernstein, Against the Gods, 14. 33 Richard Bartle, ‘Hearts, clubs, diamonds, spades: Players who suit MUDs’. Journal of MUD Research 1, no. 1 (1996). 34 Chris Vammen and Jennifer Perkins, ‘An unobserved societal issue, video game addiction in the 21st century’. In 2007 IEEE International Symposium on Technology and Society (IEEE, 2007). 35 Olli Sotamaa, ‘Achievement unlocked: Rethinking gaming capital’. Games as Services Final Report, 2010. Available at http://uta32-kk.lib.helsinki.fi/ bitstream/handle/10024/65772/978-951-44-8167-3.pdf?sequence=1, 73. 36 Niklas Nylund, ‘Walkthrough and let’s play: Evaluating preservation methods for digital games’. In Proceedings of the 19th International Academic Mindtrek Conference (ACM, 2015); Janne Paavilainen, Elina Koskinen, Hannu Korhonen and Kati Alha, ‘Exploring playful experiences in social network games’. In Proceedings of DiGRA 2015: Diversity of play: Games – Cultures – Identities (2015); Victor Cheung and James R. Wallace, ‘Felines, foragers, and physicists: Supporting scientific outreach with multi-surface and multi-space games’. In Proceedings of the 2016 ACM on Interactive Surfaces and Spaces (ACM, 2016). 37 Williams, Critical Introduction, 35. 38 Michael Cook, ‘More unpredictable stuff’. Gamasutra, 15 October 2015. Available at https://www.gamasutra.com/blogs/MichaelCook/201510 15/256421/More_Unpredictable_Stuff.php. 39 The ‘Danesh’ tool, developed by Michael Cook, is specifically designed to let the player generate ‘grids’ of this sort to examine the scope, range and similarities or procedural outcomes. See Michael Cook, ‘Introducing Danesh (Part 1!)’. Games By Angelina, 10 February 2016. Available at www.gamesbyangelina.org/2016/02/introducing-danesh-part-1/. For further discussion of how one might address the range of procedural generations, see Gillian Smith, ‘Understanding the generated’. In Procedural Generation in Game Design, edited by Tanya X. Short and Tarn Adams (Boca Raton, FL: CRC Press, 2017).

198

THE UNPREDICTABILITY OF GAMEPLAY

40 DeLanda, Intensive Science, 30. 41 Newman, Best Before, 79. 42 For example, see Keith Burgun, ‘An alternative to achievements’. Gamasutra, 25 January 2013. Available at https://www.gamasutra.com/view/ feature/185436/an_alternative_to_achievements.php, and Nathan Grayson, ‘“Achievement spam” games are causing controversy on steam’. Kotaku, 29 June 2017. Available at http://steamed.kotaku.com/achievement-spam-gamesare-causing-controversy-on-steam-1796528445. 43 Newman, Best Before, 78. 44 The nature of some Achievements is kept secret to prevent ‘spoiling’ the player. If there is an Achievement for ‘Kill X, the final boss’, then naturally reading that Achievement would give away the nature of the final boss. In these cases Achievements cannot function to give the player information about later parts of the game, although secret Achievements remain rare compared to their overt cousins.

8 The Negation of Unpredictability

The final set of phenomena I wish to examine in the third of these case study chapters are a selection of practices which, collectively, I term actions designed to negate unpredictability, in games designed to have unpredictability. By partaking in any of these practices, players continue to play games which feature designed and intended systemic unpredictability, whether randomness, chance, luck or some combination, but act in such a way – either within the game fabric or by using tools or methods which exist outside the intended fabric of the game itself – as to reduce or entirely mitigate some of all of the forms of that game’s unpredictability. As I explore in this chapter, there are a range of motivations and justifications behind these practices, ranging from a desire to ensure multiple players experience the same version of a game to a desire to practise and improve one’s skill in challenging and punishing ‘permadeath’ games – although these both, and the other practices explored here, tend to come with unintended consequences that shape gameplay in unusual ways. The very existence of these practices in turn tells us much about players’ orientations towards unpredictability: when it is deemed desirable, when it is deemed undesirable, what players consider to be an acceptable mitigation of unpredictability and why, and how entire cultures and practices have arisen surrounding the desire to make unpredictable games at least a little more predictable, whether entirely, or within particular spaces and contexts of play. Exploring these practices allows us to conceptualize some more of the potential conflicts between designer intentions and player desires which play out on the field of unpredictability, beyond questions of instability previously examined; where these can sometimes be unified in new and unexpected ways; and why the context of different elements of in-game unpredictability is always of the utmost analytic importance. After this introductory section the chapter shifts to several case studies which illustrate this desire to negate unpredictability, how it plays out across a diverse range of game-playing contexts, and the impact of these on our understanding of the unpredictability of gameplay. I first consider the

200

THE UNPREDICTABILITY OF GAMEPLAY

practice of ‘duplicate deals’ – ensuring all competitors are dealt the same hands of playing cards – in the game of bridge, the cultural and historical impetuses behind it, and implicit understandings of the relationships between randomness, luck and skill it represents. I secondly consider the practice of ‘save scumming’, involving backing up and reloading saved game files in order to get around the strictures of permadeath, how it shapes and changes gameplay, and the paradox through which save scumming, meant in significant part to bolster player skill, can often have the opposite effect. The third and final instance of the negation of unpredictability the chapter considers is then the concept of the ‘gameplay seed’ – a value or file or set of numbers which ensures that the unpredictable systems in one playthrough will play out exactly the same for someone else using that same gameplay seed. The gameplay seed sets the RNG in one instance of a video game to the same starting value as in a second instance of that video game, meaning that the unpredictable game generated remains deeply unpredictable before being seen, but is then predictable in all further uses of that gameplay seed. Across these case studies it must be noted that the practices here are not like the other ways of ‘handling’ or ‘navigating’ unpredictability explored earlier in this work. These have thus far fallen into two categories: navigating unpredictability through gameplay, such as pursuing a completionist agenda or playing a level a thousand times in pursuit of the desired permutation, and navigating unpredictability through discourse, which might emphasize the centrality of the long term to unpredictable games, or present the role of luck as actually being emblematic of skill and so forth. Instead, the negation of unpredictability explored in this chapter refers to a negation in the sense of the formal characteristics of the game being played, which is to say altering the ‘code’ of the game in question in order to remove all, or some, of the unpredictability that would ordinarily be part of those games. This makes them distinct from other examples we have seen in this work, and important to study as yet another response that players can make to the challenge of playing games of unpredictability. The chapter concludes with a theoretical re-engagement vis-à -vis the negation of unpredictability and a summary of what these practices show us about game unpredictability, the contexts in which it manifests and how players respond to it. As such, whereas the previous two case study chapters have explored contexts of randomness and chance and (briefly) instability, this chapter is focused primarily on luck.

Duplicate deals Games played competitively, it seems at first glance, would require a level playing field in which skill can emerge victorious. If not, there is no competition, for being ‘the luckiest’ is a form of quasi-competition no person can exercise true agency over, transforming a competition between

THE NEGATION OF UNPREDICTABILITY

201

actors into simply a comparison of desirable or undesirable flukes. However, many unpredictable games are still played at a competitive level. In all of these cases, it is scores (understood broadly) which are examined to weigh up who wins, the ability of the player to achieve within the framework of the game, to do more within that framework, accrue more wins or more kills or more points or more melds or more hands or more currency. As Gary Smith astutely notes, in any competitive endeavour, we ‘observe scores, not ability. If [unpredictability] is involved, what, if anything, can we infer from …  scores?’.1 Similarly, Michael and Ellen Kaplan, in their examination of chance, argue that a game ‘must have rules. How can we try our skill or strength against each other if every trial is different? …  We must show we were always playing the same game for the scores to count’.2 Although these are relatively passing comment in these works, this is an important observation. We are unable to ever witness the ‘truth’ of game-playing skill for it is multifaceted, embodied and often obscure; thus instead skill must be measured through its enacted manifestations in the world, each of which are inevitably contingent on a tremendous range of other factors, because as Reith notes, ‘no amount of skill can ever eliminate uncertainty and confer absolute control, for chance is an ontological feature of the world’.3 How, therefore, do some competitive game players navigate this fact? To understand the desire to reduce luck by reducing randomness, we must look in more detail at how competitive (i.e. skill-based) events across all kinds of games – card games, board games, video games played at an Esports level and video games played competitively without financial import – are structured. In the case of card games, although some games such as bridge have made deliberate attempts to reduce short-term luck as much as possible, other players of card games, even those in the highest level of competitive or professional play, accept the presence of short-term luck. There are no attempts in the competitive or professional communities of poker and rummy to reduce luck by managing how cards are dealt, reproducing deals or doing anything of that sort; instead it is understood that the skilled player is always playing for the long term (as explored earlier) and that responding to short-term luck is simply a part of the game. Indeed, in games with global economic ecosystems such as poker, short-term luck is embraced as a crucial part of what allows professional players to make a living. When anyone can win in the short term, weaker players are enticed into the game and get to experience the feeling of victory every now and then, even if in the long term they will be giving money to the more skilled players over the course of thousands or millions of hands. Without this short-term luck, so the argument goes, many such players would lose interest in poker; consequently, defeats even after one makes the correct play should be embraced, not loathed, for they enable the poker ecosystem to exist.4 Short-term luck is therefore not a distraction from competitive play, nor seen as something which prevents the most skilled players from

202

THE UNPREDICTABILITY OF GAMEPLAY

winning, but is instead conceptualized as something which is at the very least acceptable, and at most deeply necessary, for competitive play itself, or for the upward flow of money from less skilled to more skilled players, and therefore the creation of professional careers in these games. However, the superstructures of poker and bridge, which is to say the economic ecosystems of play, technology, players, money and competition that have been built up around them, are profoundly different. bridge events tend to be smaller than poker tournaments, whether live or online, making duplicate deals a more realistic and practical notion. Equally, the play structure, with a greater focus on each hand and no inter-hand elements such as poker chips, lends itself more to duplicate deals; in poker, it is more difficult to see when periods of duplicate dealing would or could start or end. It must also be noted that in poker, income can be found at all levels of competition, all stakes of wager, whether in analogue play or upon tens of thousands (at least) virtual poker tables. In bridge, by contrast, one must earn incomes primarily from tournaments and from clubs, which allow for competition in small enough number that a greater focus on the short-term seems logical. Additionally, each hand of bridge takes much longer than one hand of poker, meaning that the time it would take to reach the volume of hands required for luck to balance out, both in terms of time per hand and the total availability of hands per se, is far greater than in a game like poker. This means that if a bridge player had to wait for the same number of deals to be played out as a poker player, given both the relative scarcity of profitable games and the speed per hand, they would be waiting many times as long. The time required to amass sufficient data to make informed judgements about one’s play in the domain of poker is already high, but in bridge this period would have to be expanded significantly, making this a far less realistic prospect for the accurate arbitration of playing ability. No one of these factors appears to have been the core determinant but brought together, the gradual impetus towards developing and maintaining a duplicate deal system in bridge is quite clear. In duplicate bridge the same bridge deal is played at each table within a tournament or club. When the hand is completed, a special card holder is used to transfer that table to the next table along; or, sometimes, players shift tables and leave the cards where they are. There are numerous different methods of scoring and different methods of moving from one hand to another, which vary by organizer preference, the size of the tournament and so forth; at time of writing this has become the standard competitive format for the game’s play. In this way duplicate dealing has been widely accepted as a reasonable way to take the unpredictability in any card game and reduce it to the smaller possible level, in the process maximizing – so the argument goes – the chance for skill to dominate and come to the fore. Such a negation of unpredictability is entangled with numerous other factors, primarily pre-formed notions of skill being understood as a counterpoint to

THE NEGATION OF UNPREDICTABILITY

203

luck, and the wider ecosystem of bridge play, competition and profitability, showing us that unpredictability (and the desire for its negation) always exists within the social contexts of play. This is not only the case for instability, which is the most transparently socially determined of the four locations of unpredictability described in this work, but for the others as well: games are in dialogue with other games, with broader cultural forms, which inform their play and how we look at their unpredictability. These are themes we unpicked by considering the Random Number God, the concept of expected value and much else besides, and it can be seen once again here. It is also worth noting that playing in this manner also leads to a change in tactics for players seeking success, because scoring is no longer a question of absolute scoring, but rather a question of superior scoring to other pairs using the same decks of cards. Consequently, what emerged as an attempt to preserve an existing form of gameplay, which is to say bridge, but in a formal competitive context where the role of luck is reduced to the smallest level possible, has actually come to affect the tactics of that same game. In this regard the negation of unpredictability fundamentally alters play, creating a new rhythm of unpredictability, changing the locations that unpredictability can be found in (adding or removing) and changing how players address themselves to navigating unpredictability as well as the wider ecosystem of play. It does not just allow the same game to be played in a manner which is ‘fair’, but transforms that play, for skill-centric notions of fairness and luck-heavy games are difficult, or perhaps impossible, to reconcile in the short term. What is nominally the ‘same’ game is in fact changed quite significantly by this player choice; the negation of unpredictability is therefore contingent on a wide range of social and ideological factors, and the very process of negation can, in its own right, introduce new kinds of unpredictability (instability, a new spirit of the game) into the act of play.

Save scumming and permadeath Roguelikes, and games inspired by roguelikes that contain procedural generation and permadeath, will – when well-designed – have game designs that are centred on the experience (and evading the experience) of permadeath. Permadeath, explored previously in this work, is a system of game design whereby the player is denied the extra-diegetic5 affordance of reloading one’s game after one’s character reaches a fail state (a death, a critical objective not completed, a task that must be completed within a restricted time is not). Permadeath games still generally allow for the saving and loading of a particular playthrough, so that players are able to spread a playthrough across multiple sessions of play, but they cannot reload a save. When a player’s character in a permadeath game expires, that character remains dead; the game will wipe the character from its memory, generally

204

THE UNPREDICTABILITY OF GAMEPLAY

also wipe the world generated around that character, and a player wishing to play the game again will have to create a new character, and begin once more from the very start of the game. Such permanent death is, needless to say, not universally popular. Many take issue with the punishing repercussions of failure such a system represents, and for games without unpredictability, it is easy to see why. Were permadeath attached to a game with no or low unpredictability, it is easy to understand why that game might feel reduced to nothing but a memorization exercise, with the player arbitrarily forced to replay the same experiences at the start of the game after each (perma)death. The purposes and contextual deployments of permadeath are numerous. Although currently experiencing something of a ‘renaissance’ in its use (and the use of comparable but slightly ‘milder’ systems) in video games, permadeath was actually a staple of the earliest days of home console games and arcade games, where technical constraints might have made saving difficult, or social constraints (such as playing in an arcade context) made saving and loading impractical,6 or games were sufficiently short that there was little motivation to implement saving and loading. More recently, however, games (both roguelikes and others) with none of these technical or social constraints have implemented permadeath systems, bringing with them a set of distinctive decision justifications. Game designer Jim Rossignol, one of the developers of Sir, You Are Being Hunted (2014) suggests that permadeath ‘heightens excitement. If the player is risking something – in the case of most games the time you have invested to reach a certain point – then taking risks with that investment is more thrilling. If you can just hit a key and get everything back, it’s less of a thrill to succeed, and less of a horror to lose.’7 Similarly, Jake Solomon, one of the designers of XCOM: Enemy Unknown (2012), argues that it ‘brings real consequences to the games we play’, and that consequently ‘emotions are real [and] the loss is real’ upon a death, which ‘in some way ... makes the game real’.8 This added ‘sense of realism and reward’9 is seen as a method for giving deep meaning to the unpredictable game conditions that arise – for, like real life, one has but a single chance to negotiate unfamiliar terrain – and the psychological rewards are thereby perceived to be far greater. From this understanding permadeath does not serve to alienate and dispirit players who are sent back to the start of the game with each defeat, as we might expect, but instead functions as a shared ‘metric of game difficulty and player mastery’10 and what Rob Parker calls a ‘significant motivating and unifying factor’ for roguelike players11 (and for players of comparable games). Permadeath is designed to fundamentally shift the gameplay experience towards something more exciting, more meaningful, more consequential, if also potentially more devastating and dispiriting if the game’s conclusion is not a victory. Most important for our discussion of the experience of unpredictability, and how players are using the practice of save scumming in permadeath games to route around that experience, is the insight that permadeath

THE NEGATION OF UNPREDICTABILITY

205

can also create what Walter Benjamin would call an ‘aura’ – a unique or distinctive sense of something – around a character12. This emerges when the player knows that character’s life (like their own) is unique, limited and finite,13 and they exist and navigate a completely unique game space. This uniqueness of character is present in playing non-unpredictable games with a permadeath setting as in the examples listed above, but only in mechanically unpredictable games has it led to a cultural form as widespread as YASD player-created narratives. Death is of course an underlying part of many games, but as we have seen with permadeath the ‘underlying intention is to make the experience of death more meaningful and memorable for the player’14 – with unpredictability this comes from the navigation of a unique game world, and with permadeath with a unique game character. As such, Thomas Rousse15 valuably suggests that ‘permadeath’ might be something of a misnomer, for it focuses on the irrevocability of death rather than the changes made to the player character’s life (which is to say, how that life is played) as a result of the resultant ‘high-stakes’ nature of gameplay16 and the fact that ‘no choice can be taken back or tried again’.17 Games journalist and critic Ben Griffin suggests narrative is impossible alongside permadeath18 because of its fundamental disruption of any story a game seeks to tell, both through sending the player back to the start of that story each time they die and by simultaneously ‘cutting’ the story at an arbitrary, rather than deliberately-paced, point. However, we might instead propose that it produces different narratives: ones focused on the player character and their unpredictable adventure, rather than a fixed story based on the world they move through. A second important element is cross-playthrough learning, a factor in unpredictable games examined many times already in different contexts in this work. When coupled with permadeath, in unpredictable games permadeath serves another purpose, which is to push players towards learning the rules of these complex and challenging games. In a blog post that generated some debate in the roguelike community, I have previously argued that permadeath is designed to produce better players by forcing players to confront their mistakes if they wish to progress, although other elements of contemporary roguelikes can sometimes act to mask that intended moment of learning.19 Similarly, noted roguelike developer Andrew Doull states that ‘the process of learning the emergent rules of a game with permadeath and randomised levels transforms the game play from a fixed author led narrative into a meta narrative about the experience of the player learning through repeated and hopefully interesting and unique failure’.20 As Jesper Juul puts it, ‘Once we accept responsibility, failure also concretely pushes us to search for new strategies and learning opportunities in a game. Failure reveals strategic depth to us,’21 which can then be pursued and mastered, thereby lessening the future risk and consequently the future sting of permadeath. The only case where such trans-playthrough learning does not take place is

206

THE UNPREDICTABILITY OF GAMEPLAY

in ‘permapermadeath’,22 in which one can only play the entire game once, but this remains a deeply niche off-shoot of the increasingly mainstream use of ‘standard’ permadeath. Just as permadeath can create new forms and new strengths of narrative generally unavailable in other games, it can also serve as a strong motivator towards the learning of the unpredictable systems underlying a game with permadeath, often essential to future success in subsequent playthroughs of the same game. One can ‘lose’ the game and still achieve the play of the game and the learning that comes with it, even if one did not achieve a desired victory.23 Despite the benefits to game design of coupling permadeath with unpredictability I have sought to outline above, permadeath in unpredictable games remains, for some, as unpopular or as unnecessarily punishing as it would be in a game with fixed content. As a result, it has become a chief site for the negation of unpredictability. There are several reasons for this continued unpopularity. First, and most obviously, as noted above, players don’t ordinarily enjoy playing the same early game content over and over in order to get to something new and more interesting. Although games with permadeath often work hard to ensure significant variation in the actualized permutations of the game presented to the player when they start a new game, it is not difficult to appreciate why the psychological impact of constantly restarting from scratch is hard to stomach for many players. Secondly, and relatedly, the loss of progression, distinct from having to replay the early elements of a game once more, is equally undesirable. Everything a player has achieved within a playthrough will be discarded when permadeath is triggered; crucially, this means not just the loss of their progress through the game, but also the feeling that the most compelling or fortunate elements of the game were for nothing. This means that every impressive move, every close escape, every clever strategic choice, is all for nothing. Of course, in the long term these do not count for nothing: a close escape might have entailed the player learning a new way to perform a close escape, which will help them again later; a new strategy might also help once more in a similar future scenario; and so on. However, in the heat of the emotional moment that takes place when a permadeath game is lost, especially one where one had been playing with a particular character or a particular playthrough for quite some time,24 it is difficult to take oneself out of the moment and reflect on the benefits gained from the lost playthrough; instead, it is a lost playthrough, and such moments appear to have been for naught. In turn, the loss of a playthrough in which one particular unpredictable system was unusually kind to the player might also generate the feeling of a ‘lost opportunity’ – the player found an immensely powerful item on only the second floor of the dungeon, for example, and now that moment of great fortune has been lost for good, and wasted. In permadeath games with a tremendous number of system and tactical and strategic possibilities, such as most roguelikes, it is extremely common that something which seems

THE NEGATION OF UNPREDICTABILITY

207

unusually fortunate will occur, maximizing the chance that a player will often, or almost always, feel that a death has cost them a ‘promising’ playthrough. Thirdly, as well as denying access to later parts of games until players’ skill improves significantly, permadeath also makes practising or learning those later parts of a game harder and harder, and therefore appears to gate content in an unusually hard way. If we postulate that any given player will struggle equally at all parts of a game when each part is unknown, then their speed of learning about the game overall (and therefore improving at the game) will decrease as time goes by. The better they get at the game, and therefore the further through the game they progress with each attempt, the longer it takes on each playthrough to reach a new part of the game. Of course, they should also be improving at playing the game in general with each playthrough, and so the number of visits to each new area of the game needed for mastery should decrease the later in the game a given area is; but these new areas still take longer to reach. As such, to deal with these issues,25 some players ‘save scum’, to which we can now fully turn having established the interwoven relationships between permadeath, failure, learning and now save scumming. ‘Save scumming’ entails backing up one’s character in a permadeath game, and then restoring that save if or when one’s character perishes. This can be used in two ways: to play until one’s death or until a critically difficult part of the game has arisen, and then to reload to a previous point and play through that section of the game again, which will be different on this second attempt due to the game’s unpredictability; or to deliberately play the same section of a game repeatedly, even if it is not especially or at all challenging, until an unusually positive (as opposed to simply ‘not fatal’) permutation is created, whose benefits the player accepts before continuing the rest of the play of the game (although they might return to such a practice once or many times in the future). In the first case, a player might elect to ‘back up’ their save file every five floors, for example, in a traditional dungeon-descending roguelike. If they make it to the end of those five floors, then they create a new save; if they don’t, then although their save – the copy of the save within the game itself – will have immediately been deleted, they move their backed-up save back into the appropriate place on their computer system for it to once more register as an active saved player character, and start from the last level they saved it. Save scumming in this way is understood as a method to make the game a little easier, for now ‘permadeath’ only applies to a group of levels instead of the entire game. This also disrupts the intended gameplay somewhat, but not immensely; the larger the number of levels one plays before backing up, of course, the less intended gameplay will be disrupted. However, at the other end of the spectrum, players can also utilize save scumming not to explore the game and backup their progress, but rather to actively generate easier progress on the road ahead. For example, a player

208

THE UNPREDICTABILITY OF GAMEPLAY

of NetHack might elect to save their game every time they are about to descend to – and therefore generate – a new level. They then backup their save, reopen the game and progress into the subsequent level. They then play through that level; if it is to their liking, they repeat the process, saving their game before the next level and backing up that save. If they do not like it, however, or if their character is killed, they then restore their previous save, generate that floor again, and see if they like it this time; if not, then this loop repeats until there is a level they are content with (and survive), before then starting the cycle again. Taken to its extreme, and given that most roguelikes only generate a new level when the player’s character first steps into it, a player could continually respawn a floor until that floor is replete with money, or filled with coveted wands of wishing,26 or some other extraordinarily unlikely but extraordinarily beneficial permutation.27 Save scumming thereby enables players to mitigate, to a greater or lesser extent, the role of permadeath in games of this sort, or simply the role of unpredictability in games where this can be done per se, and can be put to the purpose of either being able to try again after a failure (but continually trying to succeed with what the game offers), or to actively make the game easier by only progressing gameplay when extremely positive permutations of elements are offered to the player. Most obviously, therefore, save scumming trivializes a game. When different solutions to a problem can be attempted over and over, it is only a matter of time until the player – so long as they meet any physical reflex requirements of the game, although all classic roguelikes used in the examples above are turn-based – is victorious. With a finite set of possibilities, and the time to try every possibility, the (or a) correct possibility will eventually be located. By save scumming, players are granted this kind of time; they do not have to be correct the first time, as is the intention within a permadeath game design paradigm, but they simply have to be successful once. As soon as this success is achieved, they move on from that part of the game as if they had found the correct solution on their first try. This means a game that might be intended to take months or even years to complete for the first time – some players speak of playing a roguelike such as NetHack for decades without a playthrough that ends in an ascension (victory)28 – can be completed in a far shorter space of the time, of the order of weeks, days or even hours with sufficient dedication. With the ability to try the problems the game sets the player until a solution is uncovered, the central challenge of any unpredictable game – thinking on one’s feet, and responding to that unpredictability correctly, without the ability to try again – is almost completely removed. This can be seen in Figure 8.1; in a save-scummed game, the outcomes the player continues to play with are unusually good (with the precise metric of ‘good’ depending on the player’s preferences), resulting in an unusually easy play experience. In a standard playthrough (squares) this means whatever moments they are given, which range from good to ill; in a

THE NEGATION OF UNPREDICTABILITY

209

FIGURE 8.1 Progression in a game over time shown against the ‘benefit’ of unpredictable moments the player encounters or accepts.

save-scummed playthrough (diamonds), it means the moments they choose to keep. Whereas the normal playthrough shows a standard distribution, the save-scummed playthrough shows an unusually high number of highquality unpredictable moments, and no moments which seriously threaten the player (for these are decided against in the act of save scumming). There is therefore a second crucial element to save scumming, if done in this way: it makes players believe the game is easier than it actually is, and in doing so reduces player learning, and generates fewer advantages for the aspirant player than they might think. Consider the player who save scums and only advances through the game when a level is generated with relatively simple enemies, or a very positive distribution of items, or without notoriously challenging segments of gameplay or some equivalent. Over time, this player only comes to actually play levels whose random elements have fallen in an unusually generous way, and which represent the lower end of the difficulty bell curve for possible generated levels in that game. By contrast, a player not save scumming is forced to play both the ‘easy’ and the ‘difficult’ levels the game in question throws at them, and is unable to affect how many of each kind of level they encounter in one playthrough. A save-scumming bias towards simpler challenges leads to a situation where, for the player who only elects to play such levels, the entire game appears significantly easier than it is. Such a player will inevitably gain a warped view of the game’s procedural generation systems and content, and may come to believe they are skilled enough to tackle the game without save scumming before they are, and then find themselves without the skills needed to handle the full spectrum of the game’s permutations, having trained themselves only on the easiest iterations. In the first case, the game’s challenge is reduced through the ability to ‘fish’ for a correct solution; in the second case, the game’s challenge is reduced through a selection bias tilted heavily towards easier actualized permutations which then form the player’s

210

THE UNPREDICTABILITY OF GAMEPLAY

multiplicity of the game, which would not be the case were save scumming not practised. These factors lead to the situation shown in Figure 8.2: the player’s character in a save-scummed game becomes unusually powerful, trivializing much of a game’s challenge. It shows the ‘power curve’ – which is to say the rate at which the player’s character (or nation, etc.) gains strength. Save scumming (dotted line) leads to a situation where because the player’s character is faced only with easy challenges or finds desirable upgrades at an unnaturally fast pace, their power increases much faster than it should (unbroken line), simplifying later stages of play. We can therefore propose a rather curious paradox which emerges out of the practice of save scumming in order to develop the skills needed to then play the game as intended. Save scumming is about negating unpredictability, but if people return to normal play it leaves them less equipped to negate unpredictability, which is to say, to navigate the game and not be affected by unpredictability. When a game is played according to the spirit of that game, it is the concept of mastery we explored earlier that allows players to negate unpredictability, by exerting the strongest possible agential force on the game and being able to secure victorious outcomes regardless of the particular unpredictable elements a game might choose to throw at them. Save scumming slows or potentially entirely prevents the acquisition of mastery because, depending on how the practice is performed, one either never encounters the more challenging permutations of a game that might arise, or one’s character is unusually strong and thus challenging permutations are made unusually easy, or one is never forced to develop the skills required to make strong strategic choices in pressured situations, or a combination of all of the above. Nevertheless, save scumming is indicative of much about unpredictability and the player experience of unpredictability we have seen throughout this work: the desire that many players have to make the game appear ‘fair’, the desire of players to learn about the unpredictable games

FIGURE 8.2 The ‘power curve’ when playing such a game with (dotted line) or without (unbroken line) save scumming.

THE NEGATION OF UNPREDICTABILITY

211

they play, the desires of players to gain or regain a sense of agency in these games and so forth. As such, it is a practice which merits a more committed critical engagement, the beginnings of which I have sought to outline in this section. Save scumming is not just a response to the player experience of unpredictability, but is also an experience of unpredictability in its own right – a biased, tilted view of a generative system – and as such merits inclusion in any comprehensive study of the subjectivity of game unpredictability.

Gameplay seeds We turn now to the third and final player practice designed with the negation of unpredictability in mind: the notion of the gameplay seed. In order to generate an unpredictable set of elements, any game must have a starting point: a number which is the ‘seed’ for the game’s RNG, and thereby whatever complex calculations take place and lead to the creation of that particular game world. A game’s RNG might take its seed from the seconds on the player’s clock, from the micro-movements of a computer mouse, from the number of programs open on that machine, from the exact amount of memory currently available, or any other source of data that changes rapidly and is either beyond the control of the player, or nominally within their control but extremely difficult or near impossible to manipulate. Doing so means that from the player’s perspective the outcomes are entirely unpredictable, although from a computational perspective, ‘true’ randomness is extraordinarily difficult to achieve, hence the need for a seed. However, this also means that if the precise seed could be recreated, which would ordinarily be impossible, then the precise same experience could be generated a second time. In recent years, some games with high levels of randomness and to a lesser extent chance have increasingly offered the ability for the player (or another system) to save and share, or input, a particular gameplay seed; this has numerous effects. Most obviously, the ability to capture a gameplay seed and then put that seed in place again allows a player to experience the same generation over and over. This is not the same as save scumming, for this player must start from the start of the seed each time, but the levels do not change, because the rhizomatic elements actualized do not shift; this transforms a normally unpredictable game into a game very similar to a ‘handmade’ game with no initial-condition randomness, and allows players to rote-learn elements of this specific generation in order to complete it. Utilizing a gameplay seed means players can play the same generation over and over, secure in the fore-knowledge of what comes next; players can enjoy a particular generation which might have a unique or highly unusual challenge, an amusing segment of gameplay, an extremely rare enemy or something of that sort; and lastly, a potentially infinite number of players

212

THE UNPREDICTABILITY OF GAMEPLAY

can compete in the same generation – for score, or for time or for some other metric – without any possibility for luck to play a role in their competition. When all players are addressing the precise same generation of an otherwise unpredictable game, there is no way for one player’s generated world or generated challenge more broadly to give an advantage over that of another player, which is to say via luck. In some games, this has led to the notion of the ‘Daily Challenge’, or an equivalent term: a particular gameplay seed which all players are offered on the same day, and can only play once. For example, the ‘Daily Challenge’ in modern roguelike Spelunky is designed to construct a ‘truly competitive, high-stakes Spelunky game that removes most of the elements of randomness that make it difficult to accurately compare the skill of two different players’29 by making the caves explored consistent across all players. In this case the use of the gameplay seed to rote-learn a challenge is abrogated, but the use of the gameplay seed to offer the precise same (but still unpredictable) challenge is elevated to a central part of gameplay. Therefore, the individual experience of players within purely the technical and mechanical confines of the game is unchanged: they begin a new game, they attempt that permutation once, and regardless of whether the eventual outcome is success or death, they cannot attempt that permutation again. However, in the broader sense of inter-player competition, this is quite different from competitive play in traditionally unpredictable games. Just as duplicate deals make sure that unpredictability does not shape the outcome of play by actualizing different virtual elements for different players, and just as save scumming allows players access to parts of the game world at different rates which are determined through factors beyond the unpredictability of a particular generation, the sharing of gameplay seeds creates an experience which is unlike any other form of unpredictability, where player subjective experience of unpredictability actually becomes shared. All players share in the same moment of unpredictability and the same outcome of unpredictability, shifting the unpredictability of gameplay into a group experience which enables new forms of gameplay, such as direct competition within a competitive framework where luck is considered unacceptable. As we have seen several times in this work, unpredictability is often treated as an undesirable element in highly competitive games. The reason is easy to appreciate: skilled game players want to always win when their skill outweighs that of their opponents, and do not want to experience losing a bout of competitive gameplay due to the luck of their foe (and, in many cases, professional players would not want to win through luck either30). Those who play competitive games generally wish to be rewarded (or not) appropriately according to the relative abilities of themselves and their opponents. However, in even the most competitive card games, unpredictability is, generally, considered entirely acceptable.

THE NEGATION OF UNPREDICTABILITY

213

By contrast, in video games it is rarely considered acceptable. A gameplay seed therefore allows for a kind of competition which is distinct from Esports (directly competitive zero-sum contests), speedrunning (contests over time in an unchanging game) and arcade games (contests over score in an unchanging game), which is to say a kind of competition where unpredictable games can be played by and across multiple players, places and contexts, keeping the unpredictability of what players will face while still ensuring a degree of skill-based fairness by ensuring nobody is (dis)advantanged by the permutation they happen to encounter. Much like duplicate deals, therefore, gameplay seeds are designed to match an unpredictable game to a certain model of a fair competition, but do so on a far greater scale. These kinds of competitions – given the success of games like Spelunky and the relative dearth of high-level bridge players – are certainly experienced by a far greater number of players as part of their overall experiences of game unpredictability. The sharing of gameplay seeds and duplicate deals both allow for a sharing of unpredictability which is unlike that found elsewhere in this work, creating a distinct social experience of randomness, chance and luck which complicates the elements we’ve often seen, such as skill, time and unpredictability, in new ways.

The negation of unpredictability The negation of unpredictability is at the centre of each of the practices mentioned in this chapter. Although unpredictability is at the core of the design of these games as intended by the games’ creators or developers, numerous reasons have arisen – ranging from the formalization of competitive play to the desire to gain skill at a high-difficulty and highly punishing unpredictable game – for wanting to remove or at least temporarily ignore that same intended unpredictability. These practices are all designed to negate both repetition and difference, and create a generality across gameplay experiences – they give each seed or deal a unique identity which is then reproduced for multiple players in multiple contexts, ensuring that the starting state of the game is identical and free of any randomness. We have explored duplicate deals in the game of bridge, the act of save scumming in games of permadeath with high levels of unpredictability, and the concept of the gameplay seed in video games whose players wish to compete collectively on an unpredictable level or challenge each day, transforming one individual’s subjective experience of unpredictability into something experienced on a potentially global scale, yet still maintaining its surprise and its novelty. In all of these examples players are playing games whose gameplay and game experiences are fundamentally predicated upon their unpredictability, and yet their players are seeking to negate that same

214

THE UNPREDICTABILITY OF GAMEPLAY

unpredictability; why not, therefore, simply play a game that has one fixed set of elements? In all of these cases, the impetus towards the negation of unpredictability lies within the expectations of players when coming to engage with these games. In the case of duplicate deals, players approach the play of bridge with a particular set of assumptions and associations about the nature of competitive gameplay, which I argue is influenced by the broader ecosystem of how competitive bridge play has evolved and developed, emphasizing for players the necessity of skill dominating in the short term, not just the long term. In the case of save scumming, players approach from a perspective that presents seemingly unfortunate moments of chance or randomness as incorrect or undesirable functioning of the game’s generative system (see Chapter 3’s discussion of the Random Number God), rather than integral part of both its gameplay and the development of skill in players. In the case of gameplay seeds, players again approach with a particular set of expectations about what makes for fairness and equity in a competitive game environment, and this consequently informs the desire to share unpredictability. This shows us that there will often be permutations people prefer, either as individuals (save scumming) or in the wider ecosystem of the game (gameplay seeds and duplicate deals), because of what they allow the player to do, to learn, what judgements the player can make, how the player can engage with others ‘fairly’ in an unpredictable game and so forth. Through these kinds of practices players aren’t changing how they play a game with a particular ruleset, nor are they generating new discourses to negotiate with or rationalize the game’s unpredictability, but rather they are changing the technical fabric of the games themselves, which is to say the distribution of unpredictable elements and the extent to which players control those same elements: deals are duplicated instead of unique, permadeath doesn’t count, and unpredictable levels are shared. A crucial part of thinking about how players deal with unpredictability, therefore, is thinking about how players actually reject it, yet still wish to the play games in which it is an inescapable constituent part. This final case study chapter has examined a set of related practices which all aim to reduce the regularity of, reduce the impact of, or entirely remove or cancel out, some or all of the unpredictability in certain kinds of unpredictable games – when played with certain rules, and in certain contexts. The main body of this chapter was the examination of three particular instances where players have decided to create various practices that achieve this negation of unpredictability. The first was the concept of the ‘duplicate deal’ in competitive bridge play, through which teams are dealt identical selections of cards in order to, its proponents argue, remove any luck from these contests. As that section argues, we can understand the impetus towards this choice by considering the broader political economy

THE NEGATION OF UNPREDICTABILITY

215

of bridge play and the structuring of potentially profitable contests in the game, which collectively form a context in which duplicate deals appear to be an appropriate response to the unpredictability of shuffling a deck. The chapter then considered the practice of ‘save scumming’ in video games that feature permadeath mechanics, which entails backing up and then reloading one’s game saves when undesired gameplay elements are actualized. I focused particularly on the interplay and potential tension between the motivations and impacts of this practice, unpicking both why players feel compelled to save scum and how deeply their experiences of unpredictability are shaped through the biasing of generative outcomes implicit within any kind of save-scumming practice. We then proceeded to consider the concept of the ‘gameplay seed’, which allows more than one player of an unpredictable video game, such as many modern roguelikes, to all play through the same generated world. This section highlighted, much like the exploration of duplicate deals, the desire for a ‘fair’ contest, which takes many of the more conceptual discussions earlier in this work and demonstrates how they manifest in practical contexts of play, showing the always-contested role of unpredictability when a priori notions of skill are mobilized. As such, playing unpredictable games and then seeking to negate that same unpredictability must be regarded as a central element of the play of such games, which although seemingly paradoxical shows us much about how players experience both the positives and the sometimes profound negatives of unpredictable play.

Notes 1 Smith, What the Luck, 35. 2 Kaplan and Kaplan, Chances Are, 32. 3 Reith, Age of Chance, 94; Caillois, Man, Play and Games, 172–3. 4 Nathan Williams, ‘How to deal with bad beats from bad poker players’. PokerNews, 10 May 2016. Available at https://www.pokernews.com/strategy/ how-to-deal-with-bad-beats-from-bad-poker-players-24734.htm. 5 Saving and loading are not, ordinarily, understood as elements of a fictional game world; they are elements of the simulation controlled by the player in their navigation of that world, and are not part of the suspension of disbelief of any fictional space. Even if a game is specifically designed to mirror the physical laws of reality, for example, the normally trivial act of reloading takes place outside the game’s suspension of disbelief. A small number of games play with this notion, by having characters reference the act of saving and loading – most notably Undertale (2015), but Eternal Darkness (2002) and Animal Crossing New Leaf (2012) are also noted for similarly entwining saving and loading with gameplay itself, while the Soulsborne series makes the death and reloading cycle a part of the games’ worldbuilding.

216

THE UNPREDICTABILITY OF GAMEPLAY

6 Amra Copcic, Sophie McKenzie and Michael Hobbs, ‘Permadeath: A review of literature’. In 2013 IEEE International Games Innovation Conference (IGIC) (IEEE, 2013), 40. 7 Ben Griffin, ‘Why permadeath is alive and well in video games’. Gamesradar, 7 March 2014. Available at http://www.gamesradar.com/why-permadeath-justwont-die-video-games/. 8 Andrew Groen, ‘In these games, death is forever, and that’s awesome’. WIRED, 27 November 2012. Available at https://www.wired.com/2012/11/ permadeath-dayz-xcom/. 9 Jake Magee, ‘Perma-death is the best way to enjoy strategy games’. GazetteXtra, 20 July 2016. Available at www.gazettextra.com/20160720/ press_start_perma_death_is_the_best_way_to_enjoy_strategy_games. 10 Rob Parker, ‘The culture of permadeath: Roguelikes and terror management theory’. Journal of Gaming & Virtual Worlds 9, no. 2+3 (2017), 123–41. 11 Parker, ‘Culture of permadeath’, 125. 12 Thomas Rousse, ‘On permadeath’ (2011). Available at http://dx.doi. org/10.2139/ssrn.2001550. 13 Rousse, ‘On permadeath’. 14 Copcic, McKenzie and Hobbs . ‘Permadeath: A review of literature’, 46. 15 Rousse, ‘On permadeath’, 6. 16 For the interested reader there is a rich body of science fiction literature which performs the ‘reverse’ of this discussion: instead of taking the ‘permadeath’ nature of real life into video games, numerous novels and short stories have taken the ‘respawn’ ability from video games and explored what ‘real life’ beyond game worlds would be like were that ability, through imagined future technological means, made manifest. Prominent examples include the Culture sequence of novels (1987–2012) by Iain M. Banks, the Takeshi Kovacs series by Richard K. Morgan (2002–5) and to an extent the Imperial Radch trilogy by Ann Leckie (2013–15). 17 Brian Moriarty, ‘Irrevocability in games’ (PhD diss., Worcester Polytechnic Institute, 2012), 7. 18 Griffin, ‘Why permadeath is alive’. 19 Mark R. Johnson, ‘The problem with the roguelike metagame’. Ultima Ratio Regum, 23 May 2015. Available at http://www.ultimaratioregum.co.uk/ game/2015/05/23/the-problem-with-the-roguelike-metagame/. 20 Andrew Doull, ‘Analysis: The game design lessons of permadeath’. Gamasutra, 27 July 2009. Available at https://www.gamasutra.com/view/news/115412/ Analysis_The_Game_Design_Lessons_Of_Permadeath.php. 21 Juul, Art of Failure, 116. 22 It is interesting to note the existing of a small number of games with ‘permapermadeath’ – not just can the player’s character not play the game twice, but the player themselves is limited to a single play; upon the death of their first character, the player is barred from continuing to ever play the game

THE NEGATION OF UNPREDICTABILITY

217

again. Some of these, it must be noted, are not free to play, and therefor entail the player paying for a game which can only be played until their first death. Although this new, seemingly innovative and even more ‘hardcore’ form of play gathered extensive visibility and debate when first announced, such games have not met with any significant success and have, at time of writing, all but faded from view. The author does not believe a thorough scholarly analysis is necessary to understand why. 23 Suits, The Grasshopper. 24 The time needed for a single playthrough of a permadeath game with substantial unpredictability varies significantly. At the low end, some of these games are designed to take between one and two hours for a playthrough; examples of this are The Binding of Isaac and FTL. At the other extreme, some classic roguelikes are designed to be played for tens of hours, just as many strategy games with high randomness in world map generation, such as Civilization, can certainly take that long. We might reasonably suggest that the longer a permadeath game playthrough, the greater the perception of risk and danger, and consequently the greater the inclination to save scum, or the lower the inclination to use a non-permadeath game mode. Losing one’s save after minutes, or losing one’s save after weeks, are profoundly different experiences. 25 From a developer’s perspective, there are other reasons not to want permadeath: most obviously, that a lot of players feel negatively about permadeath as noted in the paragraph which leads to this footnote, and developers might therefore be concerned that their potential player base, and therefore profit, and therefore sustainability and long-term income, will be threatened by such a ‘harsh’ system. 26 These allow the player, by typing something in which is then run through a reasonably sophisticated text parser, to ask the game for almost any possible item. They are, as one might expect, extremely sought after as a result of this ability. 27 This system – of continually re-generating a dungeon level until it is to one’s liking – is a crucial part of the current-ongoing ‘tool-assisted speedrun’ (TAS) of NetHack, being carried out by NetHack players ‘ais523’ and ‘dwangoAC’. Whereas save scumming is frowned upon by most players, the function of a ‘TAS’ is to push the limits of an optimal game playthrough by any means possible, and for this particular act save scumming is therefore considered acceptable, and indeed integral. For more information on this ongoing TAS, see the interview I conducted with these players for the Roguelike Radio podcast, see Mark R. Johnson, ‘Nethack tool-assisted speedruns’. Roguelike Radio Podcast, 12 June 2016. Available at www.roguelikeradio.com/2016/06/ episode-122-nethack-tool-assisted.html. 28 In the present day, such players are likely very rare, especially those who have been committed players for this long (as opposed to those who have played a few times a year, perhaps, or have played on-and-off across many years). Nevertheless, popular commentary around NetHack frequently has recourse to this pervasive notion of defining NetHack’s difficulty through stories of extreme temporal play practices.

218

THE UNPREDICTABILITY OF GAMEPLAY

29 Spelunky Wiki, ‘Daily challenge mode’. Spelunky Wiki, 12 April 2015. Available at http://spelunky.wikia.com/wiki/Daily_Challenge_Mode. 30 This is not always the case, however – many professional poker players seem to derive satisfaction and amusement when, having made a strategic or tactical error, they are saved by the good fortune of the deck. For Esports players, however, a willingness to embrace fortune is much rarer.

PART THREE

220

9 Conclusion: A Framework for Non-deterministic Games

Thomas M. Malaby, in his study of how Greek citizens in the town of Chania navigate unpredictability and complexity both through gambling and in areas of life that have apparent structural or epistemological commonalities with gambling, draws our attention back to one of Max Weber’s fundamental sociological questions. This is the question of, as Malaby puts it, how people ‘regard the distribution of fortunes in the world’.1 It is apparent that nobody likes to feel ‘cheated’ out of a victory or some other outcome that they had ‘rightfully earned’: when players invest hard work and skill to achieve a desired outcome, the feeling of having the outcome worked for denied due to some external factor, something that couldn’t be controlled, can feel galling, unfair and cosmically unjust. Our contemporary neoliberal milieu compels us to believe in the absolute triumph of individualistic hard work and ability, whether innate or cultivated, and a concurrent decline in thinking about the outcomes of one’s efforts as being outside of one’s control. In the world of games and competition more broadly, television and increasingly online broadcasting show us regular images of the world’s most skilled game players emerging victorious over their opponents, and winning fortunes on the back of not their fortune, but on their rational, decision-making abilities. In most competitive games the ability to profit and earn a professional living is predicated upon absolute game skill, which is to say being one of the best in the world at one’s chosen game, or ideally, the very best.2 Surrounding these players one hears much about the hours of practice they commit, the sacrifices they have made, the difficulties they have encountered, all of which have nevertheless been overcome through ability and effort, suggesting our ability to realize the best possible life through our own efforts and little else. As we noted in the introduction, games are routinely touted as emblematic of other domains of life, such as politics, war and finance; skill in games seems to translate to skill beyond the magic circle, while the same abilities

222

THE UNPREDICTABILITY OF GAMEPLAY

that ensure dominance in one – rational, objective, mathematical thinking – seem to ensure dominance in the other. On the other hand, we delight in the play of unpredictability and giving ourselves over to accepting whatever may come, whether it comes from deliberately getting lost in unknown countries with no idea of what lies around the corner, browsing through television channels without a specific programme in mind, or – as was the focus of this work – playing games where we cannot necessarily anticipate how they will start, how they will play out, how they will conclude and how these elements of unpredictability will intersect with our deliberate and determined actions (if any) throughout the play of that game. This fundamental tension, when manifested in play, is at the heart of what makes unpredictability in games so exciting. As Gerda Reith notes, games of unpredictability have a certain vertiginous thrill to them, can vacillate wildly from moment to moment between ‘excitement and boredom’, and lead to complex dialogues with fate, causation, free-will and agency3 of a sort not found in purely deterministic games (whose ideological baggage is clearer, and whose apportioning of outcomes is more transparent and more formal). These elements have been with us since the dawn of games themselves. Unpredictability was first ‘introduced [into games] before recorded time, and [was always] pursued with assiduous fanaticism’4 for its gameplay value, religious meaning and ideological connotations, all the way up to the present day. All kinds of games ‘are often ritualistic, repeatable, and laden with symbolic meaning’,5 and it seems this is all the more true when the game itself shifts and warps with each new instance of play, offering new possibilities, denying old ones and forcing players to create hypotheses and perspectives on the opacity and variability of such forms of play. It has been argued by Dan Ariely that many systems shape us into making irrational decisions, undermining our notions of being able to make objective and skilled judgements about our lives;6 by contrast, I suggest that people actually do much to engage with, make sense of, and attempt to shape systems that are designed to be ‘irrational’ (which in this context is to say non-deterministic and unpredictable and opaque) and thereby seem to prevent or hinder a rational, objective encounter. Players are not passive consumers of such systems, but do much both within and beyond the traditional boundaries of those systems in order to negotiate them. In this work I have sought to do two things – develop a typology and apply that typology to a number of case studies. First, the development of this typology focused on the different locations of unpredictability in a game, according primarily to what stage in a player’s experience unpredictability is encountered, or in the case of the fourth element (instability), how that unpredictability should be understood vis-à -vis the ludic boundaries of the game in question. Randomness, chance and luck allow us to think about unpredictability in the start, middle and end of gameplay, while instability exists outside traditional or otherwise prior boundaries of

CONCLUSION

223

play activity, being a distinct location of unpredictability which comes with its own particular set of entanglements. This typology is designed to move us beyond broad categorical phrases like ‘games of chance’ and in doing so instead acknowledge that, although slot machines and grand strategy games do both contain unpredictability, their implementations of unpredictability, their forms of unpredictability and the effects these have on players are utterly different. Any superficial commonality possessed by a shared stochasticity is insufficient to warrant shared terminology. This perspective and the attendant framework it has led to allows us to unpick this unexamined category, and appreciate both commonalities across many implementations of unpredictability at a given location, and the profound differences in how players regard, play with, think about, and negotiate their role within, games that portray these four kinds of unpredictability. I will say more on the typology, and its Deleuzean conceptual underpinnings, shortly. Secondly, in developing this typology, the work has throughout been deliberately replete (although inevitably not exhaustively so) on case studies and practical examples: my purpose has been both to explicitly relate the theoretical and conceptual to the empirical and practical, but also to demonstrate the tremendous range of games that utilize unpredictable systems, the purposes to which they are put and the player responses (and sometimes entirely new player-driven practices) which have emerged as a result of this incredibly wide diversity of unpredictability. There is tremendous ground to cover in such an examination, challenging even for a work of this length, and I have been keen to let case studies take over the heavy lifting, and the task of offering analytic clarity, wherever possible. Part One of the work related numerous case studies of randomness, chance, luck and instability, which aided in the development of these terms. Subsequently, Part Two of the work addressed itself to three particular case studies in greater detail. One was the variety of PCG found primarily in the roguelike genre, which I showed represented some of the most complex unpredictability systems to be found in games, and consequently some of the most varied experiences to be found in play, some of which can, at present, be experienced nowhere else. Another case study was the pair of concepts or practices we call grinding and replay (value), both of which entails the deliberate redoing of sections of gameplay in order to either secure an outcome or experience desired by the player. The last was the set of practices designed to actively mitigate or entirely negate unpredictability in games designed to contain significant unpredictability, designed to negotiate the challenges to agency, skill and self-worth posed by unpredictable games, primarily which have or are understood to have luck. Across these both small and large case studies the work has explored a number of second-order questions which must arise with any experiential consideration of unpredictability: Why do players enjoy such games? Why do

224

THE UNPREDICTABILITY OF GAMEPLAY

they accept some forms of unpredictability more than others? Why and how does the context of play matter, and why do players sometimes try to make unpredictable games, in some ways, predictable? Exploring these questions allows us not to just conceptualize unpredictability, but situate it practically within game design, gameplay experience and the actions of players when confronted by unpredictability in certain games, in certain social contexts and with certain stakes associated with the moment of play. I am hopeful that all of these case studies taken together have demonstrated the practical use of the proposed typology as well as its potential analytical clarity. Inevitably, however, the case studies explored even in a work of this length only scratch the surface of the rich variety of ways designers have implemented game unpredictability and players have experienced it. Although I have examined practices or experiences ranging from the implementation of global sharedunpredictability competition to the emergence of a contemporary deity with a surprising volume of adherents, from grinding the same two-minute level of a video game a thousand times to reading strange machinic poetry, and from walking around a room full of card tables to photocopying game security measures, these are only touching the surface of the vast ocean of game unpredictability waiting to be studied.

Deleuzean games In this analysis of unpredictable games and their wider personal and social connotations, I have drawn on the work of Gilles Deleuze as my primary theoretical orientation. As noted in the first chapter, this work is not the first time Deleuze’s work has been deployed in the service of studying games. However, Deleuze has not yet been applied to studying the element of gameplay which, I would suggest, he is actually best suited to: the consideration of unpredictability. This might be the first work to do so and – I hope – take a Deleuzean consideration of game unpredictability to its logical conclusion conceptually, illustrating this path with ample empirical case studies. The understanding of unpredictability I have sought to present here has been predicated on Deleuze’s understandings of difference, singularities, repetition and generality, and rhizomes and (more so) arborescence, transferred into a game context, or when already used in a game context in his work, developed further and applied across the range of contemporary game forms we see around us today. I would argue that all games with unpredictability are Deleuzean in two senses: in the sense of a generative system which repeats difference, and in the sense of the development of a multiplicity in the mind of a game-playing subject through the interplay of virtuality, actuality and synthesis. The exchange of information and the reciprocal determination between these two aspects is central to the analysis presented here, but it is the latter,

CONCLUSION

225

the subjective perception of game unpredictability, which has taken centre stage. Deleuze’s work gives us a framework for assessing how generative systems can produce a range of outcomes that players perceive and attempt to understand, how those perceptions differ depending on where in a game they are found, what moments of unpredictability players take as being significant and why, and how past experiences help to construct future expectations, and in doing so, shape gameplay in sometimes significant ways. I consequently proposed that Deleuze’s understanding of difference and repetition has much to offer any analysis of game unpredictability, yielding not just theoretical insights, but the potential to unpick the motivations for, and experiences from, a wide range of both personal and more broadly social practices that surround the different kinds of unpredictability to be found in games. On the back of this Deleuzean framework, the work developed a fourpart typology for thinking about the locations of game unpredictability: randomness, chance, luck and instability. We began by exploring ‘randomness’, which I defined as variation in the initial conditions of play. This is easiest to see in almost all card games, wherein a deck must be shuffled before play commences, and in a range of strategy, simulation and RPGs (and the blurred lines between the three) where a core part of the game’s challenge involves decision-making upon a long-lasting backdrop of often profound unpredictability. When unpredictability was positioned at the start of games as randomness, we see the fewest player responses designed to mitigate or evade that unpredictability; instead, randomness is a prime case study for understanding how players attribute significance and understanding to unpredictable games, and how their knowledge of these games builds up. We explored randomness further in Chapter 6’s examination of procedural generation, much of which often takes place in the location of randomness, Chapter 7’s analysis of completionism and the case study of No Man’s Sky (2016) and in Chapter 8 when considering duplicate deals and especially gameplay seeds. We then defined ‘chance’ as unpredictability which takes place during gameplay. Whereas randomness, when implemented, tends to be extensive in scope (but only takes place once), moments of chance tend to be far greater in number but normally far less in consequence. Experiencing chance means that things happen to a player, during their play, over which they have minimal control, but which must factor into their decision-making. As Katherine Isbister puts it, ‘People go through a rapid and automatic set of evaluations as things happen to them, about what each event might mean for their goals and plans’7 – in unpredictable games, and specifically games of chance, these experiences are more present than perhaps anywhere else in the field of play. Chance can serve a tremendous range of gameplay purposes, arguably far greater in number than randomness, and can be repeated infinitely or attempted a finite volume of times (whereas

226

THE UNPREDICTABILITY OF GAMEPLAY

randomness can take place only once). It can also be readily conflated with luck, as we saw in both Chapter 3 and Chapter 4, in the process causing a range of discourses, some faux-superstitious, some rationalistic and mechanical, to be deployed. We examined chance particularly in all of Chapter 7, especially in our consideration of grinding, and in Chapter 8’s exploration of save scumming and the biasing of chance towards positive outcomes. The third part of the typology, ‘luck’, refers to unpredictability which takes place at the end of gameplay. By this I meant the extent to which the outcome of play cannot be predicted based on player actions, which exists on a spectrum between games of complete systemic luck (e.g. slot machines) and games of no systemic luck whatsoever8 (e.g. Go). Luck, we saw, is implicated in notions of fairness, and also has a complex relationship with time and the concept of ‘skill’, all of which marked luck out as a highly distinct site for both gameplay unpredictability and player responses and attitudes towards it. Esther Eidinow argues that deploying or thinking through concepts such as ‘luck’ can ‘initiate associations with past, present and future events, invoke conceptions of the supernatural, raise ideas of causation and responsivity or praise and blame, or reinforce notions of identity’. The particular set of associations selected and circulated by (in this case) game players will ‘depend on the kind of story in which they appear, the audience for that story, and the circumstances of the story’s telling’.9 We explored this element of responses to fate and fortune, among other places, in both the study of the Random Number God in Chapter 3, the study of player responses to FTL (2012) in Chapter 4, and the attempts to create fair competitions in unpredictable games, such as bridge and roguelikes as a whole, in Chapter 8. The fourth location proposed for unpredictability was that of instability, which takes place not in a different temporal register (start, middle, end) but rather what we might think of as a spatial one: the problematization of whether or not a particular unpredicted element is ‘inside’ or ‘outside’ the confines of gameplay. We explored two kinds of unpredictability separated by an understanding of their Deleuzean origins, which is to say glitches and exploits, the causes of instabilities and also how instability can cease to be an undesired form of unpredictability and become reconstituted into accepted gameplay. Although quite different in nature from the other forms of unpredictability explored in the work, the presence of instability as being clearly within the magic circle of play and technical artefacts that construct a game, as well as the wide range of distinctive player responses to this form of unpredictability not found elsewhere, make it important to consider alongside the other, more visible, forms. We also explored instability in Chapter 6, looking at the potential instabilities that PCG can yield (which can also be reconstituted into their own aesthetic).

CONCLUSION

227

Each of these I call a ‘location’ of unpredictability. Each of them brings with it a wide and distinct set of broader cultural associations, both historically and in the present day, alongside the particular kinds of gameplay they enable, the responses players have had to those kinds of unpredictability, and the game genres that particularly select that location of unpredictability. Grand strategy games, for example, are dominated by their randomness; roguelikes by their chance; many casino games by their luck; a number of classic games by their instability, codified into appropriate contemporary practices such as speedrunning. By separating out unpredictability into these locations, we can expand previous broader labels for unpredictable games into a more detailed typology, and most centrally identify the differences in experience which these kinds of unpredictability yield. In keeping with Deleuze’s metaphysics, having reached the end of this work we find ourselves with new identities for randomness, chance, luck and instability. These were not pre-formed, but rather created out of the differences between the forms of unpredictability explored here, and how they repeat and differentiate themselves across a range of game contexts. In turn, given his focus on subjectivity, it is inevitable that my own differentiation of gameplay unpredictability might not mirror that of others; but I am confident the four locations outlined here are clear, highly distinct and give us a valuable conceptual window into the sometimes daunting, sometimes opaque and sometimes seemingly deeply unfair, world of game unpredictability.

The ludic experience of unpredictability In researching this book it became apparent that unpredictable games were fundamentally defined by a particular set of questions. What happened to me? What did not happen to me? What might have happened? How was my play experience contingent on the above? We have seen a number of ways players answer these questions. In some cases player responses involve the study of the game in question, or the development of skill or mastery, but more often than not a different set of behaviours can be seen. This second set either change the player’s perspective on the unpredictability they’re facing, through a range of discourses and perspectives that reframe unpredictability in particular ways, or by playing a game in a different manner than its rules would suggest, or even changing at the design level the very unpredictability they are encountering. I have sought to show that these are not ‘irrational’ responses, even when drawing recourse to fauxsuperstition or denying compelling evidence, but instead involve players finding new ways to resolve the challenges posed to play, to the self, and to one’s expectations, by unpredictable games. Malaby suggests that ‘to improve our treatment of indeterminacy we must cease to assume that

228

THE UNPREDICTABILITY OF GAMEPLAY

it is dangerous and that those who appeal to interpretations not based on risk assessment are stuck in the past’,10 a perspective Eidinow echoes, agreeing that for many individuals ‘modern theories of risk and trust may not have totally supplanted the older concepts of luck, fate and fortune’11 which must be approbatively addressed by any scholar of unpredictability. It is this rationale I follow in this work: it is relatively easy to offer mathematical explanations for the unpredictability encountered in most games, but offering explanations for what players might experience and how this might consequently change their interactions with those games is a more challenging, and I would argue more valuable, task. Players are not just passive consumers of unpredictable games, being pushed one way and then the other by the arbitrary vicissitudes of unpredictability: instead, through their play, their discourses and their acts of re-design, players tackle unpredictable games in a number of ways beyond a simple, positivistic metric of ‘getting better’ at the game in question (a notion which, itself, has been extensively problematized in this work). Unpredictability is central to the play of an overwhelming portion of all games, both contemporary and premodern, and its examination is an integral part of understanding play experience. Far from all being readily grouped into the traditional notions of ‘games of chance’ or ‘games of luck’, we have seen that games of unpredictability take many forms, with unpredictability implemented into many gameplay locations. These in turn give rise to the multiplicity of play experiences, reflections and reactions that unpredictability can, and does, provoke, and which this work has sought to examine. With the ever-growing expansion of games that use unpredictability, and the manner in which they are used, this is an excellent time to refocus our efforts on studying this absolutely central, yet sometimes profoundly infuriating (for player and scholar alike) element of play – and how it shapes almost every game there is.

Notes 1 Malaby, Gambling Life, xi. 2 Johnson, ‘Comparing the professionalization’. 3 Reith, Age of Chance, 131; 136; 179. 4 David, Games, Gods and Gambling, 21. 5 Juul, Art of Failure, 26. 6 Dan Ariely, Predictably Irrational: The Hidden Forces that Shape Our Decisions (London: Harper Collins, 2008). 7 Isbister, How Games Move Us, 2.

CONCLUSION

229

8 We did also, however, note the existence of the paradox of skill – which is to say, when two highly skilled players contest a purely deterministic game, it can reasonably be argued that the outcome is not determined by skill (for there is nothing between those players), but rather by external factors. That is to say: who is more well-rested, who makes the best move in the moment and so forth. 9 Eidinow, Luck, Fate and Fortune, 68. 10 Malaby, Gambling Life, 8. 11 Eidinow, Luck, Fate and Fortune, 4.

230

BIBLIOGRAPHY

Abarbanel, Brett L. and Bo J. Bernhard. ‘Chicks with decks: The female lived experience in poker’. International Gambling Studies 12, no. 3 (2012): 367–85. Abt, Vicki., James F. Smith and Martin C. McGurrin. ‘Ritual, risk, and reward: A role analysis of race track and casino encounters’. Journal of Gambling Studies 1, no. 1 (1985): 64–75. Apperley, Thomas H. ‘The body of the gamer: Game art and gestural excess’. Digital Creativity 24, no. 2 (2013): 145–56. Ariely, Dan. Predictably Irrational: The Hidden Forces that Shape Our Decisions. London: Harper Collins, 2008. Ashmore, Calvin, and Michael Nitsche. ‘The quest in a generated world’. In Situated Play, Proceedings of DiGRA 2007 Conference, 503–509, 2007. Bainbridge, Wilma Alice and William Sims Bainbridge. ‘Creative uses of software errors: Glitches and cheats’. Social Science Computer Review 25, no. 1 (2007): 61–77. Bakalar, Jeff, Dan Ackerman and Scott Stein. ‘LittleBigPlanet 2: Redefining replay value’. Cnet, 18 January 2012. Available at https://www.cnet.com/uk/news/ littlebigplanet-2-redefining-replay-value/, accessed on 31 August 2017. Bartle, Richard. ‘Hearts, clubs, diamonds, spades: Players who suit MUDs’. Journal of MUD Research 1, no. 1 (1996). Beckert, Jens, and Mark Lutter. ‘All you need is a dollar and a dream: Explaining the inverse relationship between socio-economic position and lottery play’. In Random Riches: Gambling Past & Present, edited by Manfred Zollinger, 143–59. Oxford and New York: Routledge, 2016. Bennett, Deborah J. Randomness. Cambridge, MA: Harvard University Press, 1998. Bernard, Suits. The Grasshopper: Games, Life and Utopia. Toronto, London: University of Toronto Press, 1978. Bernstein, Peter L. Against the Gods: The Remarkable Story of Risk. USA and Canada: John Wiley and Sons, 1998. Betts, Thomas. ‘An investigation of the digital sublime in video game production’. PhD diss. University of Huddersfield, 2014. Bijker, Wiebe E. ‘Do not despair: There is life after constructivism’. Science, Technology, & Human Values 18, no. 1 (1993): 113–38. Bijker, Wiebe E. Of Bicycles, Bakelites, and Bulbs: Toward a Theory of Sociotechnical Change. Cambridge, MA: MIT Press, 1995. Billings, Darse, Neil Burch, Aaron Davidson, Robert Holte, Jonathan Schaeffer, Terence Schauenberg and Duane Szafron. ‘Approximating game-theoretic

232

BIBLIOGRAPHY

optimal strategies for full-scale poker’. In International Joint Conference on Artificial Intelligence, 661–8, 2003. Binde, Per. ‘Gambling and religion: Histories of concord and conflict’. Journal of Gambling Issues 20 (2007): 145–65. Bogue, Ronald. Deleuze and Guattari. London: Routledge, 1989. Bowling, Michael, Neil Burch, Michael Johanson and Oskari Tammelin. ‘Heads-up limit hold’em poker is solved’. Science 347, no. 6218 (2015): 145–9. Bryant, Levi R. ‘What are singularities?’. Larval Subjects, 14 June 2012. Available at https://larvalsubjects.wordpress.com/2012/06/14/what-are-singularities/, accessed on 31 August 2017. Burger, Maya. ‘Drawing cards, playing destiny: Karma and play in new divinatory practices’. In Religions in Play: Games, Rituals, and Virtual Worlds, edited by Philippe Bornet and Maya Burger, 117–43. Zü rich: Theologischer Verlag Zü rich, 2012. Burgun, Keith. ‘An alternative to achievements’. Gamasutra, 25 January 2013. Available at https://www.gamasutra.com/view/feature/185436/an_alternative_ to_achievements.php, accessed on 31 August 2017. Caillois, Roger. Man, Play and Games. Champaign, USA: University of Illinois Press, 2001 [1958]. Callon, Michael. ‘Devices of attachment’. In Markets and the Arts of Attachment, edited by Franck Cochoy, Joe Deville and Liz McFall, 180–96. Oxford and New York: Routledge, 2017. Chandler, David. ‘Bloodborne and the history of horror’. Kill Screen, 30 April 2015. Available at https://killscreen.com/articles/bloodborne-and-his tory-horror/, accessed on 31 August 2017. Cheung, Victor and James R. Wallace. ‘Felines, foragers, and physicists: Supporting scientific outreach with multi-surface and multi-space games’. In Proceedings of the 2016 ACM on Interactive Surfaces and Spaces, 297–306. ACM, 2016. Childers, Jay P. ‘Going all in on the global market: The rhetorical performance of neoliberal capitalism on ESPN’s World Series of Poker’. In Sporting Rhetoric: Performance, Games and Politics, edited by Barry Brummett, 139–56. New York: Peter Lang, 2009. Cohen, John. Chance, Skill, and Luck. Wokin and London: Pelican, 1960. Cohen-Mor, Dalya. A Matter of Fate: The Concept of Fate in the Arab World as Reflected in Modern Arabic Literature. New York: Oxford University Press, 2001. Cook, Daniel. ‘Loot drop tables’. Lost Garden, 7 December 2014. Available at http://www.lostgarden.com/2014/12/loot-drop-tables.html, accessed on 31 August 2017. Cook, Daniel. ‘Understanding randomness in terms of mastery’. Lost Garden, 31 December 2012. Available at http://www.lostgarden.com/2012/12/ understanding-randomness-in-terms-of.html, accessed on 31 August 2017. Cook, Michael. ‘Introducing Danesh (Part 1!)’. Games By Angelina, 10 February 2016. Available at www.gamesbyangelina.org/2016/02/introducing-danesh -part-1/, accessed on 31 August 2017. Cook, Michael. ‘More unpredictable stuff’. Gamasutra, 15 October 2015. Available at .https://www.gamasutra.com/blogs/MichaelCook/20151015/256421/More_ Unpredictable_Stuff.php, accessed on 31 August 2017.

BIBLIOGRAPHY

233

Copcic, Amra, Sophie McKenzie and Michael Hobbs. ‘Permadeath: A review of literature’. In 2013 IEEE International Games Innovation Conference (IGIC), 40–47. IEEE, 2013. Costikyan, Greg. Uncertainty in Games. Cambridge, MA: MIT Press, 2013. Cremin, Colin. Exploring Videogames with Deleuze and Guattari: Towards an Affective Theory of Form. Oxford and New York: Routledge, 2015. Cremin, Colin. ‘The formal qualities of the video game: An exploration of Super Mario Galaxy with Gilles Deleuze’. Games and Culture 7, no. 1 (2012): 72–86. Crockett, Clayton. Deleuze Beyond Baidou: Ontology, Multiplicity, and Event. New York: Columbia University Press, 2013. Csikszentmihalyi, Mihaly and Stith Bennett. ‘An exploratory model of play’. American Anthropologist 73, no. 1 (1971): 45–58. Csikszentmihalyi, Mihaly, and Csikszentmihalyi, Isabella Selega, eds., Optimal Experience: Psychological Studies of Flow in Consciousness. Cambridge, UK: Cambridge University Press, 1992. Culin, Stewart. Games of the North American Indians. Courier Corporation, 1975 [1907]. Daston, Lorraine. Classical Probability in the Enlightenment. Princeton, NJ: Princeton University Press, 1998. David, Florence Nightingale. Games, Gods and Gambling: The Origins and History of Probability and Statistical Ideas from the Earliest Times to the Newtonian Era. New York: Hafner Publishing Company, 1962. De Goede, Marieke. Virtue, Fortune, and Faith: A Genealogy of Finance. Minneapolis, MN: University of Minnesota Press, 2005. DeLanda, Manuel. Intensive Science and Virtual Philosophy. London and New York: Bloomsbury Academic, 2013. Deleuze, Gilles. Difference and Repetition. London and New York: Bloomsbury Academic, 2014 [1968]. Deleuze, Gilles. Nietzsche and Philosophy. New York: Columbia University Press, 2006 [1962]. Deleuze, Gilles. The Fold. London and New York: Continuum, 2006 [1993]. Deleuze, Gilles. The Logic of Sense. London and New York: Continuum, 2004 [1969]. Deleuze, Gilles, and Felix Guattari. A Thousand Plateaus. New York and London: Continuum, 2004 [1974]. Deleuze, Gilles, and Charles J. Stivale. ‘Vincennes session of April 15, 1980, Leibniz seminar’. Discourse 20, no. 3 (1998): 77–97. Dotson, Brandon. ‘Divination and law in the Tibetan Empire: The role of dice in the legislation of loans, interest, marital law and troop conscription’. In Contributions to the Cultural History of Early Tibet, edited by Matthew Tom Kapstein and Brandon Dotson, 3–78. Leiden, The Netherlands: Brill, 2007. Doull, Andrew. ‘Analysis: The game design lessons of permadeath’. Gamasutra, 27 July 2009. Available at https://www.gamasutra.com/view/news/115412/Analysi s_The_Game_Design_Lessons_Of_Permadeath.php, accessed on 31 August 2017. Duncan, Aaron. ‘Reimagining the self-made man: Myth, risk, and the pokerization of America’. Western Journal of Communication 78, no. 1 (2014): 39–57.

234

BIBLIOGRAPHY

Dyer-Witheford, Nick, and Greig de Peuter. Games of Empire: Global Capitalism and Video Games. Minneapolis, MN: University of Minnesota, 2009. Eidinow, Esther. Luck, Fate and Fortune: Antiquity and Its Legacy. London and New York: IB Tauris, 2011. Endres, Kirsten W. ‘“Lộ c bestowed by heaven”: Fate, fortune, and morality in the Vietnamese marketplace’. The Asia Pacific Journal of Anthropology 16, no. 3 (2015): 227–43. Ewalt, David M. Of Dice and Men: The Story of Dungeons & Dragons and the People Who Play It. New York, NY: Simon and Schuster, 2014. Extra Credits. ‘Roguelike returns – how to revive a genre’. Extra Credits, 11 June 2014. Available at https://www.youtube.com/watch?v=gk8ZknjYmek, accessed on 31 August 2017. Feenberg, Andrew. ‘Alternative modernity? Playing the Japanese game of culture’. Cultural Critique 29 (1994): 107–38. Fine, Gary Alan. Players and Pawns: How Chess Builds Community and Culture. Chicago, USA: University of Chicago Press, 2015. Finkel, Irving L. ‘On the rules for the Royal Game of Ur’. In Ancient Board Games in Perspective: Papers from the 1990 British Museum Colloquium, with Additional Contributions, edited by Irving L. Finkel, 16–32. London: British Museum Publications Limited, 2007. Forde, Jack. ‘The evolution of the roguelike’. IGN, 21 January 2016. Available at http://za.ign.com/the-binding-of-isaac-rebirth/97192/feature/the-evolution-ofthe-roguelike, accessed on 31 August 2017. Foucault, Michel. The Order of Things: Archaeology of the Human Sciences. London and New York: Routledge, 2001 [1966]. Franklin, Seb. ‘On game art, circuit bending and speedrunning as counter-practice: “Hard” and “soft” nonexistence’. CTheory (2009). Giddens, Anthony and Christopher Pierson. Conversations with Anthony Giddens: Making Sense of Modernity. California: Stanford University Press, 1998. Girard, René . Violence and the Sacred. London and New York: Continuum, 2005. Glassco, Bruce. ‘Betrayal at house on the Hill’. Avalon Hill (2004). Grayson, Nathan. ‘“Achievement spam” games are causing controversy on steam’. Kotaku, 29 June 2017. Available at http://steamed.kotaku.com/achievementspam-games-are-causing-controversy-on-steam-1796528445, accessed on 31 August 2017. Griffin, Ben. ‘Why permadeath is alive and well in video games’. Gamesradar, 7 March 2014. Available at http://www.gamesradar.com/why-permadeath-justwont-die-video-games/, accessed on 31 August 2017. Groen, Andrew. ‘In these games, death is forever, and that’s awesome’. WIRED, 27 November 2012. Available at https://www.wired.com/2012/11/permadeathdayz-xcom/, accessed on 31 August 2017. Hacking, Ian. The Taming of Chance. Cambridge: Cambridge University Press, 1990. Haigh, John. Taking Chances: Winning with Probability. New York: Oxford University Press, 1999. Harper, Tauel. ‘The smooth spaces of play: Deleuze and the emancipative potential of games’. Symploke 17, no. 1 (2009): 129–42.

BIBLIOGRAPHY

235

Hastings, Erin J., Ratan K. Guha and Kenneth O. Stanley. ‘Evolving content in the galactic arms race video game’. In 2009 IEEE Symposium on Computational Intelligence and Games (CIG), 241–8, 2009. Hawreliak, Jason. ‘Heroism, gaming, and the rhetoric of immortality’. PhD diss., University of Waterloo, 2013. Hayano, David M. ‘The professional gambler: Fame, fortune, and failure’. The Annals of the American Academy of Political and Social Science 474, no. 1 (1984): 157–67. Holmes, Eben G. ‘Strange reality: On glitches and uncanny play’. Eludamos: Journal for Computer Game Culture 4, no. 2 (2010): 255–76. Hooper, David, and Ken Whyld. The Oxford Companion to Chess. Oxford: Oxford University Press, 1992. Hughes, Damian. The Winning Mindset: What Sport Can Teach Us About Great Leadership. London: Macmillan 2016. Hughes, Joe. Deleuze’s’ Difference and Repetition’: A Reader’s Guide. London and New York: Bloomsbury Publishing, 2009. Huizinga, Johan. Homo Ludens: A Study of the Play-Element in Culture. Kettering, OH: Angelico Press, 2016 [1938]. IEP. ‘Gilles Deleuze (1925–1995)’. Internet Encyclopaedia of Philosophy. Available at http://www.iep.utm.edu/deleuze/, accessed on 31 August 2017. Inderwildi, Andreas. ‘The artists and paintings that bleed through bloodborne’. Kotaku, 9 December 2016. Available at http://www.kotaku.co.uk/2016/12/09/ the-artists-and-paintings-that-bleed-through-bloodborne, accessed on 31 August 2017. Isbister, Katherine. How Games Move Us: Emotion by Design. Cambridge, MA: MIT Press, 2016. Johnson, Mark R. ‘Comparing the professionalization of pro gamblers and pro video game players’. Occasional Paper Series, 40. Las Vegas: Center for Gaming Research, UNLV University Libraries, 2017. Johnson, Mark R. ‘Integrating procedural and handmade level design’. In Level Design: Processes and Experiences, edited by Christopher W. Totten, 217–43. Boca Raton, Florida, USA: CRC Press, 2017. Johnson, Mark R. ‘Nethack tool-assisted speedruns’. Roguelike Radio Podcast, 12 June 2016. Available at www.roguelikeradio.com/2016/06/episode-122-nethack -tool-assisted.html, accessed on 31 August 2017. Johnson, Mark R. ‘The demonic properties of an ampersand’. Kill Screen, 2 July 2015. Available at https://killscreen.com/articles/demonic-propertiesampersand/, accessed on 31 August 2017. Johnson, Mark R. ‘The problem with the roguelike metagame’. Ultima Ratio Regum, 23 May 2015. Available at http://www.ultimaratioregum.co.uk/ game/2015/05/23/the-problem-with-the-roguelike-metagame/, accessed on 31 August 2017. Johnson, Mark R. ‘The use of ASCII graphics in roguelikes: Aesthetic nostalgia and semiotic difference’. Games and Culture 12, no. 2 (2015): 115–35. Jones, Rex L. ‘Poker and the American Dream’. In The American Dimension: Cultural Myths and Social Realities, edited by Willian Arens and Susan P. Montague, 170–80. Sherman Oaks, NY: Alfred Publishing Company, 1976.

236

BIBLIOGRAPHY

Juul, Jesper. The Art of Failure: An Essay on the Pain of Playing Video Games. Cambridge, MA: MIT Press, 2013. Kaplan, Michael, and Ellen Kaplan. Chances Are: Adventures in Probability. London, England: Penguin, 2007. Karlsen, Faltin. ‘Entrapment and near miss: A comparative analysis of psychostructural elements in gambling games and massively multiplayer online roleplaying games’. International Journal of Mental Health and Addiction 9, no. 2 (2011):193–207. Kattenbelt, Chiel, and Joost Raessens. ‘Computer games and the complexity of experience’. In Level Up: Digital Games Research Conference, edited by Marinka Copier and Joost Raessens, 420–5. Utrecht: Faculty of Arts, Utrecht University, 2003. Kern, Chris. ‘Re: Angband rule number one!’. games.roguelike.angband, 25 July 1998. Available at https://groups.google.com/forum/#!msg/rec.games.ro guelike.angband/MtMK_E__A4M/IOW_FdRdOfUJ, accessed on 31 August 2017. King, Daniel. L., and Paul Delfabbro. ‘Understanding and assisting excessive players of video games: A community psychology perspective’. Australian Community Psychologist 21, no. 1 (2009): 62–74. King, Daniel L., Paul H. Delfabbro and Mark D. Griffiths. ‘Cognitive behavioral therapy for problematic video game players: Conceptual considerations and practice issues’. Journal of CyberTherapy and Rehabilitation 3, no. 3 (2010): 261–73. LaTour, Kathryn A., Franck Sarrazit, Rom Hendler and Michael S. LaTour. ‘Cracking the cultural code of gambling’. Cornell Hospitality Quarterly 50, no. 4 (2009): 475–97. Lazardzig, Jan. ‘The machine as spectacle: Function and admiration in seventeenthcentury perspectives on machines’. In Instruments in Art and Science: On the Architectonics of Cultural Boundaries in the 17th Century, vol. 2, edited by Helmar Schramm, Ludger Schwarte and Jan Lazardzig. Berlin and Gö ttingen: Walter de Gruyter, 2008. Leibniz, Gottfried Wilhelm. New Essays on Human Understanding. Translated and edited by Peter Remnant and Jonathan Bennett. Cambridge: Cambridge University Press, 1981 [1765]. Lewis, Chris, Jim Whitehead and Noah Wardrip-Fruin. ‘What went wrong: A taxonomy of video game bugs’. In Proceedings of the Fifth International Conference on the Foundations of Digital Games, 108–15. ACM, 2010. Lock, Helen. ‘Transformations of the trickster’. Southern Cross Review 18 (2002): 1–7. MacIntyre, Alasdair. After Virtue: A Study in Moral Philosophy. Indiana, USA: University of Notre Dame Press, 2007. Magee, Jake. ‘Perma-death is the best way to enjoy strategy games’. GazetteXtra, 20 July 2016. Available at www.gazettextra.com/20160720/press_start_perma_ death_is_the_best_way_to_enjoy_strategy_games, accessed on 31 August 2017. Malaby, Thomas M. ‘Beyond play: A new approach to games’. Games and Culture 2, no. 2 (2007): 95–113. Malaby, Thomas M. Gambling Life: Dealing in Contingency in a Greek City. Champaign, IL: University of Illinois Press, 2010.

BIBLIOGRAPHY

237

Marshall, Emily Zoebel. ‘Anansi, Eshu, and Legba: Slave resistance and the West African trickster’. In Human Bondage in the Cultural Contact Zone: Transdisciplinary Perspectives on Slavery and Its Discourses, edited by Raphael Hö rmann and Gesa Mackenthun, 171–87. Mü nster: Waxmann Verlag, 2010. Mauboussin, Michael J. The Success Equation: Untangling Skill and Luck in Business, Sports, and Investing. Boston, MA: Harvard Business Press, 2012. May, Todd. Gilles Deleuze: An Introduction. Cambridge, UK: Cambridge University Press, 2005. May, Todd. Reconsidering Difference: Nancy, Derrida, Levinas, and Deleuze. University Park, PA: Penn State Press, 1997. Mä yrä , Frans. An Introduction to Game Studies. London, UK and Thousand Oaks CA: SAGE, 2008. McKinnon, Rachel. The Norms of Assertion. Basingstoke, UK: Palgrave Macmillan, 2015. Menkman, Rosa. The Glitch Moment (um). Amsterdam: Institute of Network Cultures, 2011. Menotti, Gabriel. ‘Videorec as gameplay: Recording playthroughs and video game engagement’. G|A|M|E 1, no. 3 (2014): 81–92. Moriarty, Brian. ‘Irrevocability in games’. PhD diss., Worcester Polytechnic Institute, 2012. MTGWiki. ‘Flipping a coin’. MTGWiki, 14 May 2016. Available at http://mtg. gamepedia.com/index.php?title=Flipping_a_coin, accessed on 31 August 2017. NerdlyPleasures. ‘Exposing the code wheels – PC game document copy protection at its “most advanced”’. Nerdlypleasures, 29 March 2013. Available at http:// nerdlypleasures.blogspot.co.uk/2013/03/exposing-code-wheels-pc-gamedocument.html, accessed on 31 August 2017. Newman, James. Best Before: Videogames, Supersession and Obsolescence. Oxford and New York: Routledge, 2012. Nieusma, Dean. ‘Alternative design scholarship: Working toward appropriate design’. Design Issues 20, no. 3 (2004):13–24. Nylund, Niklas. ‘Walkthrough and let’s play: Evaluating preservation methods for digital games’. In Proceedings of the 19th International Academic Mindtrek Conference, 55–62. ACM, 2015. Oppenheimer, Daniel M., and Benoî t Monin. ‘The retrospective gambler’s fallacy: Unlikely events, constructing the past, and multiple universes’. Judgment and Decision Making 4, no. 5 (2009): 326–34. Osborne, Harold. ‘Notes on the aesthetics of chess and the concept of intellectual beauty’. The British Journal of Aesthetics 4, no. 2 (1964): 160–3. Paavilainen, Janne, Elina Koskinen, Hannu Korhonen and Kati Alha. ‘Exploring playful experiences in social network games’. Proceedings of DiGRA 2015: Diversity of Play: Games – Cultures – Identities, 2015. Parker, Rob. ‘The culture of permadeath: Roguelikes and terror management theory’. Journal of Gaming & Virtual Worlds 9, no. 2+3 (2017): 123–41. Pelton, Robert D. The Trickster in West Africa: A Study of Mythic Irony and Sacred Delight, No. 8. Berkeley, USA: University of California Press, 1989. Peterson, Jon. Playing at the World: A History of Simulating Wars, People and Fantastic Adventures, from Chess to Role-Playing Games. San Diego, CA: Unreason Press, 2012.

238

BIBLIOGRAPHY

Piccione, Peter A. In Search of the Meaning of Senet. Boston and New York: Archaeological Institute of America, 1980. Plaidhatgames. ‘Dead of winter: Rulebook’. Plaid Hat Games. Available at https:// www.plaidhatgames.com/images/games/dead-of-winter/rules.pdf, accessed on 31 August 2017. Plunkett, Luke. ‘Your games have more replay value than you think’. Kotaku, 22 February 2015. Available at http://kotaku.com/your-games-have-more-replayvalue-than-you-think-1687359958, accessed on 31 August 2017. Pocock, Kevin. ‘Battlefield, and the importance of replay value’. Den of Geek, 30 October 2013. Available at http://www.denofgeek.com/games/battlefield/27922/ battlefield-and-the-importance-of-replay-value, accessed on 31 August 2017. Protevi, John. ‘Deleuze and life’. In The Cambridge Companion to Deleuze, edited by Daniel W. Smith and Henry Somers-Hall, 239–65. Cambridge: Cambridge University Press, 2012. Rad, Chloi. ‘8 dark souls details you probably never noticed’. IGN, 21 March 2016. Available at http://uk.ign.com/articles/2016/03/21/8-dark-souls-detailsyou-probably-never-noticed, accessed on 31 August 2017. Radburn, Breigh, and Rachel Horsley. ‘Gamblers, grinders, and mavericks: The use of membership categorisation to manage identity by professional poker players’. Journal of Gambling Issues 26 (2011): 30–50. Raphals, Lisa Ann. ‘Fate, fortune, chance, and luck in Chinese and Greek: A comparative semantic history’. Philosophy East and West 53, no. 4 (2003): 537–74. Redströ m, Johan. ‘RE: Definitions of use’. Design studies 29, no. 4 (2008): 410–23. Reith, Gerda. The Age of Chance: Gambling in Western Culture. London, USA and Canada: Routledge, 1999. Roberts, John M., Malcolm J. Arth and Robert R. Bush. ‘Games in culture’. American Anthropologist 61, no. 4 (1959): 597–605. Robertson, D. S. ‘The Delphian succession in the opening of the eumenides’. The Classical Review 55, no. 2 (1941): 69–70. Roll20. ‘Wand of wonder’. Roll20.net, 2017. Available at http://roll20.net/ compendium/dnd5e/Wand%20of%20Wonder, accessed on 31 August 2017. Romano, Alessandro. ‘God’s dice: The law in a probabilistic world’. University of Dayton Law Review 41 (2016): 57–88. Rousse, Thomas. ‘On permadeath’, 2011. Available at http://dx.doi.org/10.2139/ ssrn.2001550, accessed on 31 August 2017. Russo, Joseph. ‘A Jungian analysis of Homer’s Odysseus’. In The Cambridge Companion to Jung, edited by Polly Young-Eisendrath and Terence Dawson, 240–55. Cambridge: Cambridge University Press, 1997. RWhiteGoose. ‘Frigate 00 agent 1:06 and lower hostage completions repository’. The Elite, 8 February 2017. Available at https://forums.the-elite.net/index. php?topic=21774.0, accessed on 31 August 2017. Schmidt, Loren. ‘Generative artwork’. In Procedural Generation in Game Design, edited by Tanya X. Short and Tarn Adams, 153–61. Boca Raton, FL: CRC Press, 2017. Schoenau-Fog, Henrik. ‘The player engagement process–an exploration of continuation desire in digital games’. Think Design Play: Digital Games Research Conference, 2011. Schü ll, Natasha Dow. Addiction by Design: Machine Gambling in Las Vegas. Princeton, NJ: Princeton University Press, 2012.

BIBLIOGRAPHY

239

Scully-Blaker, Rainforest. ‘A practiced practice: Speedrunning through space with de Certeau and Virilio’. Game Studies 14, no. 1 (2014). Scully-Blaker, Rainforest. ‘Re-curating the accident: Speedrunning as community and practice’. M.A. diss., Concordia University, 2016. Shadowrun Wiki. ‘Shadowrun system’. Shadowrun Wiki, 1 May 2015. Available at http://shadowrun.wikia.com/wiki/Shadowrun_system, accessed on 31 August 2017 Shaw, Ian Graham Ronald and Barney Warf. ‘Worlds of affect: Virtual geographies of video games’. Environment and Planning A 41, no. 6 (2009):1332–43. Smith, Daniel and John Protevi. ‘Gilles Deleuze’. In The Stanford Encyclopedia of Philosophy, 2015. Available at http://plato.stanford.edu/archives/win2015/ entries/deleuze/, accessed on 31 August 2017. Smith, Gary. ‘What the luck?: The surprising role of chance in our everyday lives’. New York, NY The Overlook Press, 2016. Smith, Gillian. ‘Understanding the generated’. In Procedural Generation in Game Design, edited by Tanya X. Short and Tarn Adams, 231–45. Boca Raton, FL: CRC Press, 2017. Smith, Thomas., Marianna Obrist and Peter Wright. ‘Live-streaming changes the (video) game’. In Proceedings of the 11th European Conference on Interactive TV and Video, 131–8. ACM, 2013. Solarski, Chris. ‘The aesthetics of game art and game design’. Gamasutra, 30 January 2013. Available at https://www.gamasutra.com/view/feature/185676/ the_aesthetics_of_game_art_and_.php, accessed on 31 August 2017. Somers-Hall, Henry. Deleuze’s Difference and Repetition: An Edinburgh Philosophical Guide. Edinburgh: Edinburgh University Press, 2013. Somers-Hall, Henry. Hegel, Deleuze, and the Critique of Representation: Dialectics of Negation and Difference. New York: State of New York Press, 2012. Sotamaa, Olli. ‘Achievement unlocked: Rethinking gaming capital’. Games as Services Final Report, 2010. Available at http://uta32-kk.lib.helsinki.fi/ bitstream/handle/10024/65772/978-951-44-8167-3.pdf?sequence=1#page=73, accessed on 31 August 2017. Spariosu, Mihai. Dionysus Reborn: Play and the Aesthetic Dimension in Modern Philosophical and Scientific Discourse. New York: Cornell University Press, 1989. Spelunky Wiki. ‘Daily challenge mode’. Spelunky Wiki, 12 April 2015. Available at http://spelunky.wikia.com/wiki/Daily_Challenge_Mode, accessed on 31 August 2017. Stagoll, Cliff. ‘Difference’. In Deleuze Dictionary Revised Edition, edited by Adrian Parr, 74–6. Edinburgh: Edinburgh University Press, 2010. Stevenson, Matilda Coxe. ‘Zuñ i games’. American Anthropologist 5, no. 3 (1903): 468–97. Sundali, James and Rachel Croson. ‘Biases in casino betting: The hot hand and the gambler’s fallacy’. Judgment and Decision Making 1, no. 1 (2006): 1–12. Sutton, Damian and David Martin-Jones. Deleuze Reframed: Interpreting Key Thinkers for the Arts. London and New York: IB Tauris, 2008. Sutton-Smith, Brian and John M. Roberts. ‘Rubrics of competitive behaviour’. The Journal of Genetic Psychology 105, no. 1 (1964): 13–37. Swancutt, Katherine. ‘The ontological spiral: Virtuosity and transparency in Mongolian games’. Inner Asia 9, no. 2 (2007): 237–59.

240

BIBLIOGRAPHY

Sylla, Edith Dudley. ‘The emergence of mathematical probability from the perspective of the Leibniz-Jacob Bernoulli correspondence’. Perspectives on Science 6, no. 1 (1998): 41–76. Tait, W. John. ‘Were there gamesters in pharaonic Egypt?’ In Ancient Board Games in Perspective: Papers from the 1990 British Museum Colloquium, with Additional Contributions, edited by Irving L. Finkel, 46–53. London: British Museum Publications Limited, 2007. Taleb, Nassim Nicholas. Fooled by Randomness: The Hidden Role of Chance in Life and in the Markets. England: Penguin, 2007. Taleb, Nassim Nicholas. The Black Swan: The Impact of the Highly Improbable. England: Penguin, 2008. Talisman Wiki. ‘Grim reaper’. Talismanwiki, 10 February 2014. Available at www. talismanwiki.com/wiki/index.php?title=Grim_Reaper&action=history. Vammen, Chris and Jennifer Perkins. ‘An unobserved societal issue, video game addiction in the 21st century’. In 2007 IEEE International Symposium on Technology and Society, 1–4. IEEE, 2007. Van Binsbergen, Wim M. J. ‘Time, space and history in African divination and board-games’. In Time and Temporality in Intercultural Perspective, vol. 4, edited by Douwe Tiemersma and Henk Oosterling, 105–25. Amsterdam: Rodopi, 1996. Van der Klei, Alice. ‘Repeating the rhizome’. SubStance 31 no. 1 (2002): 48–55. Von Hilgers, Philipp and Ross Benjamin. War Games: A History of War on Paper. Cambridge, MA: MIT Press, 2012. Von Neumann, John. ‘Zur theorie der gesellschaftsspiele’. Mathematische annalen 100, no. 1 (1928): 295–320. Von Neumann, John and Oskar Morgenstern. Theory of Games and Economic Behavior. Princeton, NJ: Princeton University Press, 2007. Weinstock, Jeremiah, Carrie E. Massura and Nancy M. Petry. ‘Professional and pathological gamblers: Similarities and differences’. Journal of Gambling Studies 29, no. 2 (2013): 205–16. Wiener, Philip P. ‘Leibniz’s project of a public exhibition of scientific inventions’. Journal of the History of Ideas 1, no. 2 (1940): 232–40. Williams, James. Gilles Deleuze’s Difference and Repetition: A Critical Introduction and Guide. Edinburgh: Edinburgh University Press, 2013. Williams, Nathan. ‘How to deal with bad beats from bad poker players’. PokerNews, 10 May 2016, available at https://www.pokernews.com/strategy/ how-to-deal-with-bad-beats-from-bad-poker-players-24734.htm, accessed on 31 August 2017. Yee, Nick. ‘Motivations for play in online games’. CyberPsychology & Behavior 9, no. 6 (2006): 772–5. Zollinger, Manfred. ‘Dealing in chances’. In Random Riches: Gambling Past & Present, edited by Manfred Zollinger, 1–27. Oxford and New York: Routledge, 2016. Zysk, Kenneth G. Asceticism and Healing in Ancient India: Medicine in the Buddhist Monastery, vol. 2. New Delhi: Motilal Banarsidass, 1998.

INDEX

Abarbanel, Brett L.  116 n.8 Academy of Games  6 ‘accepted’ gameplay mechanic  176 achievements  191 ethics of  191 form of ordering  192 nature of  198 n.44 Ackerman, Dan  195 n.13 active random elements  42–4 Addiction by Design (2012)  10, 53 The Age of Chance (1999)  12 Ancient Domains of Mystery (1994)  154 Angband (1990)  154 Animal Crossing New Leaf (2012)  215 n.5 Anti-Oedipus (2004)  18 Apperley, Thomas  125, 143 n.12 arbitrary gating  76 Ariely, Dan  222, 228 n.6 Arth, Malcolm J.  26 n.2, 27 n.5 artificial intelligence (AI)  73 Bainbridge, William Sims  143 n.10, 144 n.25 Bainbridge, Wilma Alice  143 n.10, 144 n.25 Bakalar, Jeff  195 n.13 Banks, Iain M.  216 n.16 Bartle, Richard  182, 197 n.33 Beckert, Jens  116 n.16 Beneath Apple Manor (1978)  152 Benjamin, Walter  205 Bennett, Deborah J.  27 n.17, 95, 116 n.14 Bennett, Stith  27 n.14 Bernhard, Bo J.  115 n.8

Bernstein, Peter L.  99, 116 n.25, 118 n.48, 181, 197 n.32 Betrayal at House on the Hill (2004)  74 Betts, Tom  13, 30 n.48 Bijker, Wiebe E.  145 nn.36, 37 Binde, Per  27 n.15 The Binding of Isaac (2011)  39, 129–30, 164, 217 n.24 Binsbergen, Van  27 n.16 The Black Swan (2007)  11 Bloodborne (2015)  179 Bogue, Ronald  32 n.85 Borges, Jorge Luis  164, 184 Bowling, Michael  117 n.28 Bryant, Levi R.  133, 144 n.22 Burch, Neil  117 n.28 Burger, Maya  27 n.12 Burgun, Keith  198 n.42 Bush, Robert R.  26 n.2, 27 n.5 Caillois, Roger  10, 29 nn.32, 33, 116 nn.15, 17, 23 Callon, Michael  65 nn.13, 14 Candelabrum of Invocation  165–6, 168 n.9 casinos  5, 6, 36 Caves of Qud (2010)  155, 163–4 chance  2, 9, 21, 66–87, 225, see also games of chance divination and  4 forms of  70 purposes  70 repeating and individual  78–9 ‘to-hit’  76 Chandler, David  197 n.29 chess  1, 92–3, 184

242

INDEX

Childers, Jay P.  52, 64 n.11, 116 n.9 Civilization (1991)  2, 39 Civilization IV (2005)  40 Civilization V (2010)  64 n.3 Civilization VI (2016)  64 n.3 co-dependency  23 code wheels  144 n.24 Cogmind (2015)  155 Cohen, John  30 n.56 Cohen-Mor, Dalya  196 n.27 Command & Conquer 3 (2007)  128 commutable patterns  61 completionism  24 of all discrete experiences  185–90 of all possibilities  184–5 and exhaustive play  181–3 as written into game  190–4 confusion and puzzlement  159, 162 contingency  11 Cook, Daniel  108–9, 118 n.42, 119 n.51, 150, 171, 195 n.7 Cook, Michael  187, 197 nn.38, 35 Copcic, Amra  216 nn.6, 14 Costikyan, Greg  10 Cremin, Colin  13, 30 nn.45, 46 Crockett, Clayton  31 n.61 Croson, Rachel  197 nn.30, 31 Csikszentmihalyi, Isabella Selega  29 nn.35, 36 Csikszentmihalyi, Mihaly  11, 27 n. 14, 29 nn.35, 36 ‘Danesh’ tool  197 n.39 Dark Souls (2011)  75 Dark Souls 3 (2016)  179 Daston, Lorraine  82, 87 n.17 David, Florence Nightingale  27 n.9, 87 n.21, 228 n.4 De Goede, Marieke  28 n.22, 29 n.24, 87 n.18, 196 n.25 de Peuter, Greig  13 Dead of Winter (2014)  74 degree of consistency  121 Degrees of Outcome model  67, 71–3 DeLanda, Manuel  30 n.54, 57, 65 n.17, 168 n.7, 190, 198 n.40

Deleuze, Gilles  10, 13, 30 n.58, 31 nn.59, 60, 63, 65, 67, 69, 31 nn.71, 73, 83, 32 nn.88, 89, 64 nn.4, 7, 8, 65 n.20, 116 n.22, 118 n.40, 142 n.2, 143 nn.8, 9, 196 n.19, 224–7, see also game studies, Deleuze Deleuzean games  224–7 Delfabbro, Paul H.  195 nn.3, 4, 6, 196 n.15 depth  23 Difference and Repetition (2014)  14, 19 Different Outcomes model  67 artificial intelligence  73 damage system  74 monetary system  74 discovered virtuality  122, 136 Dotson, Brandon  27 n.11 Doull, Andrew  205, 216 n.20 Dungeon Crawl Stone Soup (2006)  154, 157, 161 Dungeons and Dragons (1974)  74 Dwarf Fortress (2006)  159, 161 Dyer-Witheford, Nick  13 Eidinow, Esther  118 n.47, 229 nn.9, 11 The Elite  99, 102–4 emotional gamble  12 Endres, Kirsten W.  196 n.24 equality of opportunity  96–7 equality of outcome  96–7 escape items  118 n.44 Eternal Darkness (2002)  215 n.5 Europa Universalis (2000–13) series  76–7 Ewalt, David M.  195 n.2 expected virtuality  122, 136 exterior system patterns  62 exterior variable patterns  62 external instability  133–6 Faster Than Light (FTL), see FTL (2012) Ferguson, Harvey  29 n.40 Fermat, Pierre de  5 Finkel, Irving L.  26 n.3

INDEX

first-person shooter (FPS) games  18, 140 Fooled by Randomness (2001)  11 Forde, Jack  118 n.49 Foucault, Michel  164, 168 n.6 Franklin, Seb  144 n.26 FTL (2012)  22, 39–40, 106–13, 217 n.24, 226 gambling games  5 cultural code of  94 dominant factor  99 finance and  6 institutionalization  6 public rise of  7 unproductive  10 Gambling In Everyday Life  11 Gambling Life (2003)  10 game studies, Deleuze differentiation and differenciation  17–18 equality  14 generative systems  13–14 ideas/multiplicities and singularities  16–18 identity and difference  15–16 possible, real, virtual and actual  16 repetition and generality  14–15 resemblances  14 rhizomes and arborescence  18–20 themes  13 game theory  7 games of chance  12–13, 55, 66–87, 223 codifying chance  68–71 Degrees of Outcome model  71–3 Different Outcomes model  73–5 different valence outcomes  77 ongoing system  66–8 Outcome or Failure model  75–7 Random Number God  79–86 repeating and individual chances  78–9 unique appeal  95 Games of Empire (2009)  13 generative systems  17 game studies, Deleuze  13–14 multiplicity  109, 113

243

Giddens, Anthony  87 n.23 Gigabet  100 Gigabet Dilemma  100 Girard, René   27 n.13 Glassco, Bruce  86 n.2 glitch hunters  126 glitches  123 and broken rules  125–7 exploring  124 zombie hover  123 Gnome with a Wand of Death concept  164–5 GoldenEye (1997)  102–4, 108 Gombaud, Antoine  5 granularity of interest  21, 38, 186 celebration of risk  52 false win  53–4 game of poker  51–2 interest  56 interplay  55 near misses  53–4 notion of  55 permutations  51 Planters Good Instinct Moment  52 reconfiguring loss  53 slot machines  53 true win  54 The Grasshopper (1978)  121 Griffin, Ben  205, 216 nn.7, 17 Griffiths, Mark D.  195 n.4 grinding  24, 149, see also replay value and grinding Groen, Andrew  216 n.8 Guattari, Felix  13–14, 18, 64 nn.4, 8 Guha, Ratan K.  195 n.12 Hack (1982)  153 Hacking, Ian  125, 143 n.13 Haigh, John  86 n.8, 99, 116 n.26 ‘handmade’ game  211 Harper, Tauel  13, 30 nn.44, 47 ‘harsh’ system  217 n.25 Hastings, Erin J.  195 n.12 Hawreliak, Jason  152, 167 n.2 Hayano, David M.  116 n.6 Hendler, Rom  116 n.10 Hobbs, Michael  216 nn.6, 14 Holmes, Eben G.  125, 143 n.11

244

INDEX

Homo Ludens (1938)  10 Hooper, David  143 n.19 ‘hot hand’ fallacy  180 Hughes, Damian  195 n.10 Hughes, Joe  30 n.50, 31 nn.66, 76, 32 n.91 Huizinga, Johan  10, 29 n.31, 32 n.91 in-game action  69 Indian shuffle  35 Infra Arcana (2011)  155 instability  20, 32 n.93 exploits and unforeseen scenarios  127–31 glitches and broken rules  125–7 internal and external  133–6 rocket jumping  140–2 speedrunning  137–9 subverting and reconstituting rules  136–7 tight and weak coupling  131–3 unplanned unpredictability  120–5 intellectual beauty  115 n.2 intended unpredictability  120 interior system patterns  62 interior variable patterns  62 internal instability  133–6 Isbister, Katherine  114, 119 n.53, 225, 228 n.7 Johanson, Michael  117 n.28 Johnson, Mark R.  167 n.1, 216 n.19, 217 n.27, 228 n.2 Jones, Rex L.  116 n.7 Juul, Jesper  12, 29 n.41, 86 n.9, 117 n.36, 171, 195 n.8, 205, 216 n.21, 228 n.5 Kaplan, Ellen  65 n.21, 116 n.20, 117 n.29, 201, 215 n.2 Kaplan, Michael  65 n.21, 116 n.20, 117 n.29, 201, 215 n.2 Karlsen, Faltin  196 n.17 Kattenbelt, Chiel  13, 30 n.49 Kern, Chris  81, 87 n.15 King, Daniel L.  195 nn.3, 4, 6, 196 n.15 knowledge of variables  58

Larn (1986)  153 LaTour, Kathryn A.  116 nn.10, 11 LaTour, Michael S.  116 n.10 Leckie, Ann  216 n.16 The Legend of Zelda: The Wind Waker (2002)  123, 138 Leibniz, Gottfried Wilhelm  5, 28 nn.18, 19, 28 n.19 Lewis, Chris  143 n.16 Linley’s Dungeon Crawl (1997)  154 LittleBigPlanet (2008–14) series  172 Lock, Helen  87 n.14 The Logic of Sense (2004)  14, 19 luck  2, 9, 22, 66, 88–119, 226 Deleuze and  90–2 levelling the playing field  95–7 location of  94 and luck-boosting items  181 and meaning of ending gameplay  92–3 mistaking randomness and chance for  106–15 skill and time  97–106 stat  178–9, 181, 194 Lutter, Mark  116 n.16 MacIntyre, Alasdair  11, 29 n.34 Magee, Jake  216 n.9 Magic: The Gathering (1993)  74 magic circle  20, 32 n.92 Malaby, Thomas M  10, 30 nn.42, 57, 111, 118 n.45, 168 n.4, 221, 227, 228 n.1, 229 n.10 Man, Play and Games (1958)  10 Marshall, Emily Zoebel  87 n.12 Martin-Jones, David  32 n.87 mastery concept  108–9, 114–15 Mauboussin, Michael J.  116 n.24 May, Todd  31 n.75, 108, 118 n.41, 119 n.52 Mä yrä , Frans  195 n.5 McKenzie, Sophie  216 nn.6, 14 McKinnon, Rachel  117 n.33 Menkman, Rosa  126, 143 n.14 Menotti, Gabriel  144 n.27 metagame  155 metastabilities  138

INDEX

Morgan, Richard K.  216 n.16 Moria (1983)  153 Moriarty, Brian  216 n.17 multiplicity  16–18 generative systems  109, 113, 136 mental model  84 notion of  57 rules  36 negation of unpredictability  24, 149 duplicate deals  200–3, 214–15 expectations of players  214 gameplay seeds  211–13 save scumming and permadeath  203–11 NetHack (1987)  154, 164–6, 168 n.9, 208, 217 nn.27, 28 Newman, James  145 n.30, 198 nn.41, 43 Nicoll, Fiona  11 Nieusma, Dean  145 nn.34, 35 No Man’s Sky (2016)  173, 187–9, 225 non-deterministic game  1, 5, 108, 221–8 normal functioning  120–1 null random elements  42–4 Nylund, Niklas  197 n.36 Obrist, Marianna  137, 144 n.28 Outcome or Failure model  67, 75–7 Pandemonium  157 Pandemonium Lords  157–8 Parker, Rob  204, 216 nn.10, 11 Pascal, Blaise  5 Patolli  3, 8 Pelton, Robert D.  87 n.13 Perfect Dark (2000)  102–4, 108 Perkins, Jennifer  183, 197 n 34 permadeath system  79, 203–11 coupling  206 permapermadeath  206, 216 n.22 permutations  208, 210 renaissance  204 save scumming  207 sense of realism and reward  204 Peterson, Jon  144 n.21 Pierson, Christopher  87 n.23

245

Planters Good Instinct Moment  52 player modelling of randomness  108–9 playing cards  35–6, 42, 92, 134–5, 200 Plunkett, Luke  196 n.14 Pocock, Kevin  195 n.11 poker  93–4, 116 n.27, 201, 218 n.30 luck and skill  99–101 superstructures of  202 ‘problem’ gambling  92 procedural aesthetic  162, 167 procedural (content) generation (PCG)  23, 36 co-dependence  151, 160–2 defining  149 depth  157–8 exploration  150 identifying procedurality  162–7 length  151, 159–60 roguelikes  151–6 (see roguelike games) Protevi, John  30 nn.53, 62 pseudo-chess game  41 Raessens, Joost  13, 30 n.49 random number generator (RNG)  79 Random Number God  22, 67, 79–86, 86 n.10, 203, 226 chance-necessity  82 chaos theory  83 coin-flip chance  80 fair play  82 invocation  80–1 notion of  80 personification  85 pre-rational values  82 RNGs  79, 84 trickster god  81–2 whims  82 randomness  2, 9, 18, 21, 35–65, 225 defining  37 gameplay, initial conditions  35–8 granularity of interest  51–7 initial conditions, lasting impact  38–44 significance  44–50 system and variable knowledge  57–64

246

INDEX

Raphals, Lisa Ann  115 n.1, 177, 196 n.22 real-time strategy (RTS) game  18, 121, 123 Reaper (2008)  74 Redströ m, Johan  142, 145 nn.33, 38 Reith, Gerda  12–13, 28 nn.20, 21, 29 nn.25, 37, 30 n.43, 116 nn.13, 19, 21, 178, 197 n.28, 215 n.3, 222, 228 n.3 replay value and grinding  24, 149 completionism (see completionism) defining  169 deterministic games  169 ease and difficulty in grinding  176–81 value of returning  173–6 responsible gambling  92 retro games  194 n.1 Roberts, John M.  26 n.2, 5, 93, 115 nn.4, 5 Robertson, D. S.  27 n.6 rocket jumping  137, 140–2 Rogue (1980)  152–4 roguelike games  39, 106, 110, 118 n.39 classic  155 generations  152–3 ‘genocide’ system  154 ‘polymorph’ system  154 superstructures of  155 role-playing game (RPG)  2 Romano, Alessandro  64 n.1 Rossignol, Jim  204 Rousse, Thomas  205, 216 nn.12, 13, 15 Royal Game of Ur  3 ruptures  135 Russo, Joseph  87 n.11 Sarrazit, Franck  116 n.10 save scumming  203–11, see also permadeath system power curve  210 practice  24 Schmidt, Loren  168 n.5 Schoenau-Fog, Henrik  196 n.16

Schü ll, Natasha Dow  10, 53, 64 n.12 Scully-Blaker, Rainforest  117 n.34, 126, 138–9, 143 nn.15, 17, 145 nn.29, 31 Senet  3 Shadowrun board game series (1989)  74 Shadowrun system  86 n.1 short-term luck  201 shuffling techniques  35 significance actions, outcomes and  47, 48, 49 externality  48 internality  48 milieu of exteriority  45 milieu of interiority  45 nomadic distribution  47 notion of  50 pieces  46–7 sedentary distribution  47 stability  48 singularities  16–18, 127, 130 Sir, You Are Being Hunted (2014)  204 skill and time, luck data mining  100 expected value  99–101 Gigabet Dilemma  100 reconciling or acceptance  99, 105 repetition  106 speedrunning  102 winning and losing parameters  98 Sklansky, Mike  101 Smith, Daniel  31 n.62 Smith, Gary  201, 215 n.1 Smith, Thomas  137, 144 n.28 Solarski, Chris  118 n.43 Solomon, Jake  204 Somers-Hall, Henry  30 n.52, 31 n.72, 65 n.19 Sotamaa, Olli  183, 197 n.35 Soulsborne (2009–16) series  176, 179 Spariosu, Mihai  27 n.10, 29 nn.23, 29, 87 n.16, 19, 22 ‘speech check’ mechanics  76 speedrunners  137 speedrunning practice  89, 102–4, 137–9

INDEX

Spelunky  212–13 Spelunky Wiki  218 n.29 Stagoll, Cliff  31 n.70 Stanley, Kenneth O.  195 n.12 StarCraft 2: Wings of Liberty (2010)  123 Stein, Scott  195 n.13 Suits, Bernard  121, 143 nn.3, 4, 217 n.23 Sundali, James  197 n.30, 31 super swim ability  139 Sutton, Damian  32 n.87 Sutton-Smith, Brian  93, 115 nn.4, 5 Swancutt, Katherine  17, 31 n.82, 196 n.26 Sylla, Edith Dudley  115 n.3 systematic unpredictability  11 systemic patterns  61 Taleb, Nassim Nicholas  11, 60, 116 n.18, 118 n.46 Talisman (1983)  74 Tammelin, Oskari  117 n.28 A Thousand Plateaus (2004)  14, 18, 46, 64 n.4 tight coupling  124, 131–3, 142 tool-assisted speedrun (TAS)  217 n.27 Ultima Ratio Regum (2012)  155, 160 uncertainty  10 Uncertainty in Games  10 Undertale (2015)  215 n.5 unlock systems  155 unpredictability Black Swan Events  12 defining  2 Deleuze and game studies (see game studies, Deleuze) diversity of  223 experience  9

247

historical epochs  8 intended  120 location of  9, 90, 222, 227 ludic experience  227–8 negation of (see negation of unpredictability) origins  96 outcomes and systems, era  3–9 ‘pre-modern’ orientation  5 structuring of  70 systematic  11 uncertainty  10 unplanned  120–5 Vammen, Chris  183, 197 n.34 Van der Klei, Alice  32 n.86 virtuality  122 von Neumann, John  7, 29 nn.27, 28 Wardrip-Fruin, Noah  143 n.16 weak coupling  124, 131–3, 142 Weber, Max  221 Whitehead, Jim  143 n.16 Whyld, Kenneth  143 n.19 Williams, James  14, 16, 30 n.51, 31 nn.64, 87, 49, 64 n.9, 105, 117 n.37, 120, 142 n.1, 197 n.37 Williams, Nathan  215 n.4 Wind Waker HD (2013)  137, 139 Wright, Peter  137, 144 n.28 XCOM: Enemy Unknown (2012)  204 Yet Another Stupid Death (YASD)  111–13 zombie hover  123 Zuñ i games  3–4, 27 n.7 Zysk, Kenneth G.  27 n.8

248

249

250

251

252

253

254