Abstract urbanism

Matthew Fuller/Texts/Essays/Abstract urbanism.pdf

Abstract urbanismMatthew Fuller / text
P. 1
5 Abstract urbanism Downloaded by [University of California, San Diego] at 18:41 10 May 2017 Matthew Fuller and Graham Harwood Introduction One of the first computational models of a city was set out in Thomas Schelling’s paper Models of Segregation (1969). In this and related papers, he attempted to provide a logical model of the dynamics of racial segregation in North American cities and laid groundwork for what later became known as agent-based modelling (Schelling 1969). Although Models of Segregation did not at first use a computer, it sets up some of the basic characteristics and problems of the field. Such work is also expressed contemporarily, for instance in the work of J. M. Epstein (2002) and others in the area of computational social modelling.1 We use this work as a starting point to think about the relationship between urban morphologies and the politics of models and with the increasing and multiform kinds of merger between computational systems, models and city forms, what it means to inhabit different scales of abstract structure. It is in this juncture that abstract urbanism arises. This chapter examines the ways in which logical forms are positioned in relation to urban life as a means of discussing the relations between the city and software and will develop a discussion of such logics in relation to questions of abstraction, reduction and empiricism. By working with the ­materiality of computational systems, especially as they unfold into the urban – and the urban in a full sense, as something involving complex comings into being of desire, imagination, technologies and forms of power – we can at the same time recognise an art of working with the tendency to reductionism through which modes of abstraction may operate and also work with the highly and complexly empirical. As social simulations are increasingly embedded in, or cleave close to, lived social forms, the texture and reality-forming capacities of these logics and the fantasies they inspire and live by need to be examined. Development of simulation as a scientific practice One attractive aspect of modelling as a means of experimental understanding is that it offers a science of behaviours rather than of essences. It is peculiar,
Abstract urbanismMatthew Fuller / text
P. 2
Downloaded by [University of California, San Diego] at 18:41 10 May 2017 62 Matthew Fuller and Graham Harwood therefore, that one of the earliest examples of social simulation derives from a highly essentialist ontology. Perhaps this might be seen as an example of a new epistemic form emerging out of a prior set of commitments that it has yet to break. Models of Segregation builds on the game theory established by Morgenstern and von Neumann (1947). Schelling’s (1960) earlier gametheoretic book, The Strategy of Conflict, can be seen as a presiding spirit in Models of Segregation’s attempt (Schelling 1969) to map and rationalise options in the decisions around actions in the schematised space of non-zerosum conflict.2 The opening stages of the paper set up segregation as a fundamental axiom of great applicability. Schelling mentions men and women, Catholics and Protestants, boys and girls, and officers and enlisted men in an army. Not all types of segregation necessarily tend towards dichotomous formation. People are also sorted by ‘sex, age, income, language, colour, taste, comparative advantage and the accidents of historical location’, amongst other factors. It is assumed that the sorting behaviour for each of these is the same. In the model, a two-dimensional line is drawn (it is important to note that this is a line, not a grid) with equal divisions of space along its axis. The line is populated with an equal number of ‘black’s and ‘white’s. While the distribution looks even on the macroscopic level, at the microscopic level it is uneven. Maybe three blacks are conjunct with one white, then a black and then three whites. If the whites and blacks are content with a 50% split between the colour of their neighbours then those who have a white neighbour on one side and a black neighbour on the other reach the contentment threshold and stand still if the neighbourhood to be considered has a radius of one. Those with ‘too many’ black neighbours or white neighbours will move in order to achieve contentment. In a neighbourhood with a radius of one, the line BBBWBWWW would, several iterations later, become BWBWBWBW. If the neighbourhood extends to two houses, then the B and W in the middle of BBBWBWWW would be looking for new neighbourhoods. To summarise, in Schelling’s model, each agent is ‘black’ or ‘white’ and aims to reside in a neighbourhood where the fraction of blacks or whites is above a predefined tolerance. Schelling’s algorithm for determining the pattern of residence either creates complete integration or complete segregation. Curiously, there is no reflection on the constitution of racial sorting even in excusatory fig-leaf terms. Like the stories of house-hunting amongst ‘professors and their wives’ that Schelling (1971; 1978: Chapter 4) describes elsewhere, the specific categories upon and through which segregation operates are described as if natural, not even worthy of equivocation as to their relation to social structure. The racism of the work is both that it operates by means of racial demarcation as an autocatalytic ideological given and second that it provides a means of organising racial division at a higher level of abstraction. To say that Schelling operates within an ideologically racialised frame is not to claim either way as to whether Schelling as a person is or was consciously racist, but that, in these papers, racial division is an uncontested,
Abstract urbanismMatthew Fuller / text
P. 3
Abstract urbanism 63 Downloaded by [University of California, San Diego] at 18:41 10 May 2017 ‘obvious’ social phenomenon that can be reduced in terms of its operation to a precise set of identifiers and operations. Goldberg’s formulation of the problem of racism is useful here: The mark of racist expression or belief, then, is not simply the claim of inferiority of the racially different. It is more broadly that racial difference warrants exclusion of those so characterized from elevation into the realm of protection, privilege, property, or profit. Racism, in short, is about exclusion through depreciation, intrinsic or instrumental, timeless or time-bound. (Goldberg 2009: 5) The naturalisation of such a situation of depreciation by at-a-distance means in which entities kindly self-organise into ghettos out of their own otherwise unlimited choice must have been a marvellous boon to someone. What these papers offer is the construction of a machine for the operation of binary categorisation that in turn becomes an engine for spatial organisation, of preference-based segregation, as if the provision of housing in the form of a market is entirely smooth and demand driven, as if there are no variations in housing kinds and qualities, geographical features, cultural variations in population, of wealth, and so on. What Schelling’s work allows for is an operation of governance beyond that of direct sorting and selection, the direct command and control of populations, but rather by eliciting and installing an action grammar in which people ‘spontaneously’ recognise, in the words of Nina Simone’s Mississippi Goddam (1964), ‘I don’t belong here, I don’t belong there’. Schelling offers the image of urban form being operated upon by an ‘invisible hand’, emerging at a higher level in social and material channelling. There is a tension, then, between the figure of this invisible hand and the view of the agent. The hand operates in an ostensibly emergent or natural way, arising out of the conditions of the situation as they are, beyond how they are seen by individual actants. Abstraction as urbanism Schelling’s abstract machine is one for the bipolar reduction of variation. One of the advantages of such an abstraction is that it requires no specific material form, simply logical equivalence. As recounted in a glowing festschrift chapter, Schelling initially used pennies, heads or tails up, on a draughts board to simulate, ‘what sort of segregation patterns develop given various types of preferences and alternative definitions of neighbourhood’ (Zeckhauser 2006: x). The scale of the board becomes the limit factor of the diagrams published in a later paper, Dynamic Models of Segregation (Schelling 1971). One can imagine a media-archaeological analysis of the history of simulation starting with such boards. John Conway, in developing
Abstract urbanismMatthew Fuller / text
P. 4
Downloaded by [University of California, San Diego] at 18:41 10 May 2017 64 Matthew Fuller and Graham Harwood the Game of Life, famously extended his to cover most surfaces of his office (Gardner 1970). Equally, only having four significant neighbours, termed ‘Neumann neighbours’,3 draws a simplifying factor from the board, the constraints of which may in turn be surpassed by the volume of processing offered by electronic computing. Indeed, a media analysis of the field can divulge a number of aspects of its material practice that are often rendered conceptually and procedurally invisible. One such is that models tend to be bound by the temporal constraints of ‘turns’ in which all agents shift at the same time. Most models need to have all variables change at the same time – but models of sociality need to vary the periodicity of change for individual agents. Equally, in the model’s interaction with hardware, the need to represent data to human users renders the allocation of CPU cycles to drawing graphical representations something of an interference when compared with how many agents could be processed instead. A few years after Schelling’s work was published, Ted Nelson (1987: 149) stated in Computer Lib that simulation is always political.4 Computers, as an abstract machine for the integration of all symbol systems – those operated upon by discrete values, or values that can be rendered as digital – provide a great degree of plasticity in the social forms they might potentiate: hence the significance of Nelson’s formulation. But the specific kind of politics simulated is also articulated by the qualities of the mathematical structures they come into composition with (rule sets; systems of four or eight neighbours; bounded, unlimited or wrapping grids; and so on). It is a rare case in which there is a direct correlation between the various scales of model, media, mathematics, the social form modelled, the ideological commitments specified as politics in such simulations and the actual politics of the material operations of such systems in use. Each of these scales is active. Diagram city Epstein and Axtell’s (1994) Growing Artificial Societies: Social Science from the Bottom Up drew on Models of Segregation and from Conway’s Game of Life. In Conway’s cellular automata and Schelling’s space of segregation, the environment has no active properties, something that has consequences when these models carry over into urban planning and cityscape modelling. Epstein and Axtell’s innovation was to place agents in an active environment and to programme them to explore for simple codifications of basic resources to keep their metabolisms alive. Agents and environment have internal states and behavioural rules that are fixed at the start or that can inherit change in interaction with each other. This is a model as a form of regression analysis, or rather of using regression as a form of proposition-making mechanism, where the relations between entities are fixed but variable. The environment is a lattice of resource-bearing sites in a medium that is separate from the agents, but on which they operate and with which they interact.5
Abstract urbanismMatthew Fuller / text
P. 5
Downloaded by [University of California, San Diego] at 18:41 10 May 2017 Abstract urbanism 65 Epstein has produced a body of work discussing the ethics around agentbased modelling that seek to affect US governmental policy by creating explicit models that can be used to explain social phenomena; something he is careful to distinguish from prediction. In Why Model?, Epstein (2008) challenges the assumption that scientific theories are created from the study of data. He asserts that without a good theory, it is not clear what data should be collected. Modelling requires theorisation and so creates enquiring habits of mind that he posits as essential to freedom. By contrast, agent-based models have been eagerly taken up as objective explanations of conflictual social forms. The capacity to express forms of emergence, with the invisible hand effectively rationalising commonsensical observations of the inevitability of such phenomena as racial segregation excites dreams of implementation. As such, this aspect of this work evinces a fascination with finding fundamental laws of social aggregation, rhetorically building on those found in natural sciences, in turn afforded by those historically associated with mathematics. Such kinds of model and associated discourse still act in a representational mode, rather than one of enquiry. Simulations now operate in a wide range of cases and kinds. They act as a form of prognosis and forecasting, of pre-emption and the maintenance of irresolvability as well as having the ability to formulate an explanation with empirical traction without having to be true. Simulations also develop specific kinds of techniques and vocabularies, as well as the software to handle and interpret them – object-oriented programming being one such example (see Fuller and Goffey 2014). Object-oriented programming is fundamental to how agent-based modelling conceives of itself, as it allows objects to hold data and functions in internal states. The object exports a limited set of methods with which to interact with it – and the data, rather than being globally accessible, are held privately to the object. This is why the behaviour of objects comes to the surface – rather than the data that underlie them. Functions or methods are the agents’ rules of behaviour. One novelty in this kind of work is the way particular forms of computational abstraction themselves become operative elements in social and urban formations. Computation becomes folded into the operations of societies, and social forms become computational problems. As the programmable city begins to incorporate models, such systems become more than representational. Here, there is a correlation between formulations, such as those of Epstein and Schelling (and those that followed in developing simulated societies), and the social sorting by software described by Steven Graham (2005) in his noted article Software-Sorted Geographies. In agent-based modelling, by contrast, there is an interplay between the schema of sorting and the actions of individuals and social formations, without engaging with the level of implementation. Where there is a difference is that Graham describes a disciplinary sorting on the social. There are kinds of sorting occurring, but these are more adequately expressed as a multiscalar, multivariable sorting enacted by agents bearing seemingly lucid and operable preference lists arrayed in
Abstract urbanismMatthew Fuller / text
P. 6
Downloaded by [University of California, San Diego] at 18:41 10 May 2017 66 Matthew Fuller and Graham Harwood relation to the behaviour and imagined preferences of others, apparently reducible to hard and fast organisation. A particularly interesting moment to watch for is when the two merge to some extent, either in actual implementation, or in the seductive idea that such reductions are fully adequate explanations of specific slices of reality. In the case of the racism of Schelling’s Models of Segregation, the categories pre-exist the machine. The machine is there to sort them, to anticipate their actualisation, to provide a degree of abstraction in which they can be reckoned, and by which the abstraction too can be worked up into an actor of a kind in itself. This operation of abstraction is crucial to understanding software as a cultural, city-making force. Logics The use of computers implies the interrelation of different forms of logic, at the levels, for instance, of programming the machine to perform calculations and of regulating the behaviour of users in pushing around mice and navigating menu systems to produce desired results. One way to think about how the mass adoption of these forms of logic effects society is in the mode Foucault described as discipline, one that analyses and breaks down a phenomenon through modelling it to produce a kind of remote control. Computation disciplines the way a phenomenon is approached and analysed so that when it becomes visible again from within the computer it makes the phenomenon materially available for comparison and modification. As users participate in the flows of power created by the comparison of information, they become normalised to its process and are themselves enrolled in the interrelation of logics. Computational forms of normalisation establish the configuration of logics needed to make the materiality of the phenomenon available for modification via abstraction, verification and reward. The repeated construction and use of these forms of logic provides a form of progressive training for those that model, feed, collect, process and react to such logics, as well as those objects that are the subject of its calculations. Logics decompose processes and the entities, including people, that are aggregate with them. The routine processing of or interaction with such models provides a collective logic to be applied to all areas of society and the natural world. The move beyond discipline, however, is characterised by the absence, further withdrawal or multiplicity and duplicity of the ultimately reliable, central control that discipline implies as a structuring principle. Logic gates Part of the legacy of Schelling’s and Epstein’s work is in the police, academic and intelligence projects aiming to predict riots via sentiment analysis. ‘Negative words’, ‘hate speech’, ‘positivity’ and expressions of anger stand
Abstract urbanismMatthew Fuller / text
P. 7
Downloaded by [University of California, San Diego] at 18:41 10 May 2017 Abstract urbanism 67 in for a population of shifting emotional registers, moving from stable states to those that can be used to require the maintenance of policing budgets, harsher policies and sudden rashes of inflamed and excited research budgets. The operators of such machines sell their technical fixes as providing a neutral oversight, in which the free expression of populations and individuals can be mapped and cross-checked for ‘naughtiness’. What happens is rather more complex; social forms are interwoven with those of the state, which itself attempts to follow too many filiations and clusterings. In the meantime, academic chancers position themselves as dubious mediators, able to appear to delve into the firehouse of text produced by a population mapped according to weightings assigned to strings of characters. We enter into a condition of a generalised politics of experimental control without controlled experiment. Claus Pias suggests that recent theories such as actor–network theory and radical constructivism come from the same stock of ideas as simulation, since for both: Their knowledge is consciously – and as a matter of course – furnished with a hypothetical index, they admit to their fictional components, they position themselves within their conceptual frame of reference, they thematize their performance, they are aware of their problematic genesis, and they specify their limited application. (Pias 2011: 54) A useful provocation following such a proposition is to be found in Latour and Lépinay’s reading of Tarde (2009: 19): ‘If you really want to quantify – which is after all the foundation of all sciences – you should try to find all the available types of quantum, instead of just using one to analyse all the others.’ This premise underlies some of the enthusiasm for big data analysis at present. It also perhaps implies that social reality is a simplified model of more adequately complex modelling schema. But we can also suggest that William Bunge’s (1971) later mode of maximalist empiricism coupled with high degrees of statistical abstraction is of great relevance here. The proposition is that to study is to become actively involved, to observe is to change, but also to recognise, that though such change may be reciprocal, it may not be symmetrical and equivalent. These, now, are the stakes of watching and participating, since in the city understood as a platform for self-organisation, algorithms, rule sets, data structures, interfaces and procedures have highly and perhaps questionably promising agency (Chopra 2014). The recent scandal of researchers from Facebook and the Universities of California and Cornell using Facebook’s news feed to operate and experiment on whether people responded to the filtering of what appeared in their news feed on the basis of whether it was associated with emotional ‘negativity’ or ‘positivity’ should be seen as a part of this tendency. The researchers note that Facebook constantly experiments with the algorithm to fine-tune this aspect of their
Abstract urbanismMatthew Fuller / text
P. 8
Downloaded by [University of California, San Diego] at 18:41 10 May 2017 68 Matthew Fuller and Graham Harwood ‘product’ (Kramer, Guillory and Hancock 2014). It is this state of perpetual experiment, linking different scales of realities, that is characteristic of the condition of abstract urbanism and the kinds of operation that the integration of modelling with cities encourages. This operation of the city as open experiment is, of course, one subject to the analysis of power. For Epstein and Axtell, agent-based modelling enforces habits of mind that are essential to intellectual and democratic freedom. An agent-based model must be explicit and open and be able to be examined and doubted, reconfigured and rerun. Epstein aligns agent-based modelling to scientific modes of inquiry that he sees as antithetical to established discursive intellectual systems. Agent-based modelling provides a freedom to doubt large monolithic and deductive forms of knowledge. Epstein and Axtell propose that we are on the edge of a new enlightenment based on the ubiquity of computing. One in which, for Epstein (2008), ‘intellectuals have a solemn duty to doubt, and to teach doubt. Education, in its truest sense, is not about “a saleable skill set”. It is about freedom, from inherited prejudice and argument by authority.’ The question of whether the enlightenment can be fully called upon in this way is in turn open to doubt, but there is something here that suggests some possibilities in that it is a science that explicitly calls subjects into being. This proposed new mode of science of active abstractions involves cities and social forms in what Stuart Kauffman calls (2000) ‘the physics of semantics’, logics that have effects in the organisation of conjunction, calculation, control and communication. Such a physics of semantics can be seen, at other scales, in the way that the agent-based model is involved in the specific forms of hardware and software development that conjoin both meaningmaking scaffolds and physical properties. Object orientation in programming is seen as a cogent worldview, capable of answering difficult questions about behaviour that emerge from complex subjects in the social or in economics, where, ‘[it] facilitates essentially any interaction structure (social network) and activation regime’ (Axtell 2003). In contemporary accounts, agentbased modelling also links its ambition to the growth of CPU processing and the availability of hard-disk space and network processors assumed under Moore’s law. The ‘promising’ nature of abstract cities is thus also woven into multiple scales of their materiality. This suggests that there is the possibility for a mode of experimentation, and of experimental politics and urban living that moves from the logics of theorems or axioms to an abstract empiricism. Historically, software-based simulations essentially replaced the kinds of hardware-based simulations or analogues of biological, cognitive and social systems developed in places such as Heinz von Foerster’s Biological Computer Laboratory in Illinois (Müller and Müller 2007). The questions posed change in this transition, becoming allied less with the philosophical concerns characteristic of the Biological Computer Laboratory, with its emphasis on epistemology and the question of abstraction and reduction from material empirical conditions. We are now
Abstract urbanismMatthew Fuller / text
P. 9
Downloaded by [University of California, San Diego] at 18:41 10 May 2017 Abstract urbanism 69 well into another similar transition where, instead of moving from hardware to software (with hardware becoming less experimental and idiosyncratic as it is rendered in the form of commodity electronics), social and urban forms become places of computational interoperation and experiment. Urban space is increasingly produced in the production, circulation and analysis of large volumes of structured and unstructured data. Models and modelisations are being integrated into the design of such spatial forms as stadia, streets and stations at conceptual and pre-emptive stages for the purposes of safety, transit design and revenue protection. In such cases, agents become active as urban entities installed and active in the symbolic and material orders of the city. Just as computational forms structure reality, so do other kinds of model. Abstract urbanism is hypothetical, fictional, maximally empirical and, of course, abstract. This means that the way in which abstractions become materially operative has to operate through these conditions, and also – under certain regimes of rhetoric – to shield them, as simply fact-based extrapolations. To recognise that they are imaginary, as models, without being merely false or simple reifications, is part of the art of abstract urbanism. To see agent-based modelling in such a way is to recognise that models are also partial cities operating like partial objects, formalised slivers of an urban configuration taken for a whole and working their drives into active diagrams. Such a condition, in which the possibilities of social fractures being triggered in the models and then implemented are manifest, cannot but add an ambiguous potency to the operations such an art promises. To work abstract urbanism in the condition of models becoming cities then is also to open the possibility of operating with a maximalist empiricism. It is to operate with delicacy and attentiveness in the design of models, but also to the arguments, spaces and politics that they bear, that they determine and into which they are smuggled, driven and suffused, and which in turn they rely on to sustain themselves. It is to saturate models with variables, and to open abstraction to social disruption rather than to prepare the abstract retrenchment of urban injustice. But to recognise abstract urbanism is not solely to postulate an interesting set of potential political practice, but more, to come to terms with a fundamental change in the consistency of cities today – they are suffused with logics. This is not simply to say that streets are data structures, people are variables and the city is a grid laced with numerical nutrients, which in their interaction produce an adequate if simplified mimicry of urban life; but that the city, the exemplary space of modernity in all its complexity of desire, violence, multiscalar layering, imagination, invention and struggle is also a place of experiment with modes of composition and of self-emergence at multiple scales of abstraction. Such a space is one where fantasies of control, of understanding, of ordering, of establishing implicit and explicit coordination and pre-emption coexist with their enactment, their failure, their use as excuse, and as a space where logic coexists with the surprise of the unforeseen.
Abstract urbanismMatthew Fuller / text
P. 10
Downloaded by [University of California, San Diego] at 18:41 10 May 2017 70 Matthew Fuller and Graham Harwood Agent-based modelling provides a means for the phantasmatic appearance of logics as an always present compliment of logic itself in that it mobilises means by which things occur in and for themselves in the mode of emergence, and for a space for arranging the coming into being of ideas of the city that are beyond the habitual means of interrogating existing coordinates. Here, in the state of being promising, logics both pre-empt surprise and rely upon it to provide a gateway to emergence understood as the self-constitution of reality; a reality that is, on the one hand, seemingly unblemished by mess, or, on the other, one forged in the full ongoing complications of the cityscape in which it becomes manifest. This is a deeply ambivalent position. The physics of semantics in which such emergence is made is therefore worthy of attention with all the precision and inventiveness that can be mustered, as it too becomes a space in which the city occurs. Notes 1 Exemplary of work building on Epstein’s model includes Casilli and Tubaro (2011) and Davies et al. (2013). 2 The paper is revised in Schelling (1978). 3 Edward F. Moore gives his name to systems of eight neighbours, expanded to include those at the corners in a rectilinear grid. 4 ‘All simulation is political. Every simulation program, and thus every simulation, has a point of view. Just like a statement in words about the world, it is a model of how things are, with its own implicit emphases: it highlights some things, omits others and always simplifies.’ (Nelson 1987: 149). 5 See Axtell and Epstein’s Sugarscape software. References Axtell, R. (2003) ‘Economics as distributed computation’, in T. Terano, H. Deguchi and K. Takadama (eds) Meeting the Challenge of Social Problems via Agent-Based Simulation, Heidelberg: Springer, pp. 3–23. Bunge, W. (1971) Fitzgerald: Geography of a Revolution, Cambridge, MA: Schenkman Press, reprint University of Georgia Press, 2011. Casilli, A.A. and Tubaro, P. (2011) ‘Why net censorship in times of political unrest results in more violent uprisings: a social simulation experiment on the UK riots’, Social Science Research Network, available from http://papers.ssrn.com/sol3/ papers.cfm?abstract_id=1909467 [accessed 18 November 2015]. Chopra, S. (2014) ‘Computer programs are people too’, The Nation, 29 May 2014, available from http://www.thenation.com/article/computer-programs-are-peopletoo/ [accessed 29 August 2015]. Davies, T.P., Fry, H.M., Wilson, A.G. and Bishop, S.R. (2013) ‘A mathematical model of the London riots and their policing’, Scientific Reports, 3(1303): 1–9. Epstein, J.M. (2002) ‘Modeling civil violence: an agent-based computational approach’, Proceedings of the National Academy of Sciences, 99(Supplement 3): 7243–50. Epstein, J.M. (2008) ‘Why model?’,    Journal of Artificial Societies and Social Simulation, 11(4): 12.
Abstract urbanismMatthew Fuller / text
P. 11
Downloaded by [University of California, San Diego] at 18:41 10 May 2017 Abstract urbanism 71 Epstein, J.M. and Axtell, R. (1994) Growing Artificial Societies: Social Science from the Bottom Up, Cambridge, MA: MIT Press. Fuller, M. and Goffey, A. (2014) ‘The unknown objects of object orientation’, in P. Harvey, E.C. Casella, G. Evans, H. Knox, C. McLean, E.B. Silva, N. Thoburn and K. Woodward (eds) Objects and Materials, London: Routledge. Gardner, M. (1970) ‘The fantastic combinations of John Conway’s new solitaire game “life”’, Scientific American, no 223, October 1970: 120–3. Goldberg, D.T. (2009) The Threat of Race: Reflections on Racial Neoliberalism, Oxford: Wiley Blackwell. Graham, S.D.M. (2005) ‘Software-sorted geographies’, Progress in Human Geography, 29(5): 562–80. Kauffman, S. (2000) Investigations, Oxford: Oxford University Press. Kramer, A.D.I, Guillory, J.E. and Hancock, J.T. (2014) ‘Experimental evidence of massive-scale emotional contagion through social networks’, Proceedings of the National Academy of Sciences of the United States of America, 111(24): 8788–90. Latour, B. and Lépinay, V. (2009) The Science of Passionate Interests: An Introduction to Gabriel Tarde’s Economic Anthropology, Chicago: Prickly Paradigm Press. Morgenstern, O. and von Neumann, J. (1947) A Theory of Games and Economic Behaviour, Princeton: Princeton University Press. Müller, A. and Müller, K.H. (eds) (2007) An Unfinished Revolution, Heinz von Foerster and the Biological Computer Laboratory (BCL) 1958–1976, Vienna: Edition Echoraum. Nelson, T. (1987) Computer Lib: Dream Machines, 2nd edn, Redmond: Tempus Books of Microsoft. Pias, C. (2011) ‘On the epistemology of computer simulation’, Zeitschrift für Medienund Kulturforschung, 1: 29–54. Schelling, T. (1960) The Strategy of Conflict, Cambridge, MA: Harvard University Press, 2nd edn, 1980. Schelling, T. (1969) ‘Models of segregation’, The American Economic Review, 59(2): 488–93. Schelling, T. (1971) ‘Dynamic models of segregation’, Journal of Mathematical Sociology, 1: 143–86. Schelling, T. (1978) Micromotives and Macrobehaviour, New York: W.W. Norton and Co. Simone, N. (1964) ‘Mississippi Goddam’, in Nina Simone in Concert, Eindhoven: Philips. Zeckhauser, R. (2006) ‘Thomas Schelling, ricochet thinker’, in R. Dodge (ed.) The Strategist: The Life and Times of Thomas Schelling, Hollis, NH: Hollis Publishing.