Complexity Theory and the Social Sciences

What then, can the concepts behind chaotic and complex systems tell us about the artistic net- works in the East End? First of all, it is worth making the point that these concepts are of lim- ited use at a strictly quantitative level. The social sciences are notoriously intractable in terms of predictive and mathematical modelling, largely by dint of the fact that they are inherently non- linear, and therefore intrinsically unpredictable (Kiel & Elliott, 1996:2), and Ruelle argues that although chaos theory has often been successfully applied, there are also many situations where it has not (Ruelle, 1997). Admittedly, this is in strictly mathematical terms. Ruelle’s diagram, reproduced overleaf, “indicates the position of dynamical systems pertaining to various areas of science with respect to uncertainty in the basic equations and complication of the dynamics.

Author Approach Focus Relevant Key Concept

Wolfram Taxonomic Cellular Automata Universality Classes Langton Taxonomic Cellular Automata Classes !Edge of Chaos"

Gell-Mann Dynamic Complex Adaptive Systems Feedback Loop Arthur Dynamics Economic Systems Increasing Returns Prigogine & Stengers Self-organised Systems Bifurcation Bifurcation

Bak Dynamic Complex Adaptive Systems Self-organised criticality Kauffman Dynamic/Modelling Evolutionary Behaviour Fitness Landscapes Holland Modelling Complex Adaptive Systems Seven !Building Blocks"

Only below the uppermost curved line may we say that we have a satisfactory understanding of the dynamics, and useful applications of the ideas of chaos” (ibid). What the diagram makes clear is that the social sciences—and that is where this study lies—are too complicated to be un- derstood or explained through mathematics alone. Paradoxically perhaps, neoclassical econom- ics has relied heavily on mathematical modelling, but the acceptance in the early 1980s that “year after year economic theorists continue to produce scores of mathematical models… …without being able to advance, in any perceptible way, a systematic understanding of the structure and the operations of a real economic system” (Leontief, 1982, quoted in Hodgson, 1993:5) combined with increasing interest in non-linear dynamical systems and economics’s mathematical tradition to make it the first social science to be explored in terms of both evolu- tionary theory and complexity theory. Hodgson (1993) put evolutionary economics in an his-

torical context, and edited collections such as Nonlinear Dynamics and Evolutionary Econom-

ics (Day & Chen, 1993) attempted to develop the mathematical modelling of such systems.

Social network analysis aside, there is no substantive mathematical modelling in this pro- ject, but this does not preclude the use of the concepts of chaos and complexity as effective tools in the understanding of social systems such as the East End artists’ agglomeration, even if we cannot explain them through formal mathematical language. In fact the title for this section is stolen from David Byrne’s 1998 book of the same name. Of the few texts—actually three—which deal directly with the application of chaos and complexity theory to the social sci- ences (as opposed to economics), Byrne’s is at the time of writing the most complete explora- tion of the topic, albeit from a strongly epistemological point of view. And although the main

Finance Economics Social Sciences Biology Climate Chemical kinetics Astronomy of the Solar System Hydrodynamic Turbulence

Complication of the System Uncertainty of

Basic Equations

Applied

focus of this project is empirical rather than epistemological there are still useful lessons to be learned here. Byrne is less willing than Ruelle to accept that the inherent unpredictability of so- cial systems limits the usefulness of complexity theory in the social sciences. Rather, he argues that there are two approaches, the weak and the strong. The weak approach is primarily taxonomic—our understanding of social systems is enhanced by knowing that they are chaotic, but we not necessarily in a position to predict outcomes (Byrne, 1998:41). The strong position, by contrast, is more optimistic in its outlook. Thus we know that a system will reach a bifur- cation point, and that there are axiomatically two possible outcomes, one of which is better than the other. We may, argues Byrne, be able to nudge the system towards the better of the two (ibid). An example of “exploiting” the chaotic nature of a system was given by Ian Stewart in New Scientist magazine (Stewart, 1999), and although it is to do with spaceships rather than so- cial systems, it makes its point well. The story goes like this.

About ten years ago, a group of NASA engineers decided that they wanted to recycle one of their satellites which was drifting aimlessly in space. Its fuel supply was far too low to propel it the millions of kilometres into the desired new orbit, which would enable it to gather informa- tion about a comet which was flying into the inner solar system. However, the NASA engineers saw that by exploiting the Butterfly Effect they could still use the spacecraft. It was simply a case of getting the butterfly to make the right flap at the right time, to get the response they de- sired from the satellite. They chose to exploit the gravitational pull of three objects in space, the satellite itself, the earth and the moon, to set the spacecraft on a chaotic orbit. Such an orbit is fundamentally unpredictable, but there are neutral points in the orbit, where no one body has a greater gravitational influence on the spacecraft than the others. Here, a slight nudge in the form of a short, carefully calculated blast from the depleted fuel stocks will have a big effect, ena- bling the space craft to be flown repeatedly past the desired observation point, at which time data could be collected. Thus was the theory that comets are large dirty snowballs confirmed, with a satellite that should by rights have been out to pasture.

The point here of course is that we are not helpless in the face of chaotic or complex adaptive systems, but we do need to treat them differently. Harvey and Reed (1996) are also concerned with limits of predictability, and they argue that there are different levels of ontologi- cal complexity in social systems, and that appropriate strategies for modelling particular social systems must therefore be adopted (Harvey & Reed, 1996:307). Their point is that predictability is not a property that a system either does or does not have, but is a property which a system possesses to a limited and effectively finite extent.

Reed and Harvey, who adopt as their theoretical foundation Bhaskar’s notion of scientific realism, a position endorsed and adopted by Byrne (1998), argue that social systems are a sub- set of dissipative systems, a concept which has its origins in thermodynamics (Reed & Harvey, 1996:302). Dissipative systems are natural systems characterised by the fact that they exhibit negative entropy as well as positive entropy. In other words they can dissipate positive entropy to their environment, and channel their negative entropy into the development over time of an increasingly complex internal structure (ibid). In short, they can evolve. A “dissipative social

system” then is:

an inherently historical entity whose evolution is driven as much by internal instability as by external perturbation. Moreover, the grounding of dissipative social systems in nature and in the dynamics of deterministic chaos demands a materialist interpretation of dissipa- tive social systems not unlike that developed by critical Marxism.

Despite their commonalities, however, there are important differences separating dissi- pative social systems from their physically constituted counterparts. Most of these differ- ences hinge on the fact that societies and their institutional activities are constructed by the collective action of human beings, and, thus, are profoundly influenced by the way in which humans subjectively define themselves and their actions. This fundamental difference has al- ready been expressed in Bhaskar’s critical naturalist paradigm, for when he describes society and its functions he underscores the ‘wild card’ nature of human beings and their innovative abilities. This same exceptionality has long been recognized in dissipative systems theory, and can be neatly inserted into the paradigm advocated by Prigogine and the Brussels School (Reed & Harvey, 1996:306).

As we have seen in the previous section, Holland (1995), Kauffman (1995), and Gell-Mann (1994) would all recognise in a “dissipative social system” what they define as a complex adap- tive system, and this does of course allow for human agency. This disparity in terminology may simply be a function of the vagaries of getting a book published, for although Kiel and Elliott’s collection, in which Reed and Harvey are published, has a publication date of 1996, none of the sources mentioned above is referred to in it. We shall therefore stay with the more common term of complex adaptive system in this project, and briefly turn our attention to the ways in which chaos and complexity theory have been used in social science.

Initial attempts to apply chaos theory to social systems, such as those described in Kiel and Elliott (1996) were of a taxonomical nature (table 9.3 overleaf). Thus Richards (1996:89–116) argued that the “aggregation of individual preferences into group choices” is chaotic, while Brown (1996:119–137) seeks to demonstrate that the political process is itself nonlinear. Berry and Kim (1996: 215–236) analyse economic “long waves” from the late 18th to the late 20th centuries, and find that such “long waves” are constrained by limit cycles. Den- drinos (1996:237–269) argues that cities can be viewed as “spatial chaotic attractors”. In other words, many social systems can be described as chaotic, although modelling them mathemati- cally proves quite problematic, as Ruelle, ironically perhaps, predicts it will be.

Attempts to apply complexity theory to social systems are less limited in scope. Thus Al- isch et al (1997) modelled the dynamics of children’s friendships making extensive use of mathematical tools. Dissatisfied with existing models, which they saw as too static, they argued that “the process of friendship can be modelled as a change in commitment and described by a vector with three components: intensity1 _{[sic], exclusivity, and intensiveness” (Alisch et al,}

1997:174). Horsfall and Maret (1997:182–196) measured the changes in the domestic division of labour from 1974 to 1978 and argue that “chaos and complexity theories provide a larger and more fruitful framework for analyzing the domestic division of labor than the current change- limited methodologies and theories” (ibid:196). Dooley et al (1997:243–268) studied the rates for adolescent childbearing Texas from 1964 to 1990 and their conclusions regarding chaos the- ory are equivocal to say the least—their data may be chaotic, the system may be sensitive to ini- tial conditions, the fractal dimension of the data may change over time (Dooley et al, 1997:265). Here then, we find no analogues for the evolution of a social system in an urban context, al- though we do at least find a general acceptance of the assertion that complexity theory is better able to deal with the dynamics of such systems.

There has been more focused study of the evolution of the urban context itself, that is of the physical form of the city, and how it changes over time (Batty and Longley,1994). Xie and Batty (1997) and Batty & Xie (1999; 1996) argue with the use of empirical data that the physi- cal form of cities, and the way in which that form has evolved, may be described using concepts of fractal dimension (Batty & Xie, 1996), cellular automata (Xie & Batty, 1997), and most re- cently self-organised criticality (Batty & Xie, 1999). Examples of both cellular automata and Batty and Longley’s simulations may be found in the figures overleaf. These models however, do not address the evolutionary nature of the social systems which exist within that urban form, and this is clearly a gap in the work that attempts to apply complexity theory to the social sci- ences. The next section is an attempt to fill that gap.

9.4

The Evolution of a Phenomenon

9.4.1 Introduction

We saw above in arguments from Ruelle (1997) and Reed and Harvey (1996) that social sys- tems lie somewhat beyond the possibilities of predictive mathematical modelling. Here, I shall

Table 9.3 CHAOS & COMPLEXITY THEORY IN THE SOCIAL SCIENCES

Author/year Approach Focus

Hodgson/1993 Discursive/historical Evolutionary Theory & Economics

Day & Chen/1994 Mathematical Non-linear dynamics & evolutionary economics

Batty & Longley/1994 Taxonomic Fractal Nature of Cites

Batty & Xie/1996 Taxonomic Fractal Dimension of Cities

Harvey & Reed/1996 Ontological Limits of Predictability

Reed & Harvey/1996 Taxonomic Dissipative Systems

Kiel & Elliott/1996 Taxonomic Chaos in Social Systems

Xie & Batty/1997 Taxonomic Modelling Urban Growth with CA

Ruelle/1997 Taxonomic Limits of Predictability

Eve, Horsfall& Lee/1998 Taxonomic Complexity in Social Systems

Byrne/1998 Epistemological Complexity and Social Sciences

explore the hypothesis that the basic tenets of complexity theory which we discussed above can function as powerful tools of conceptualisation which will extend our understanding of the way in which the East End artists’ agglomeration has evolved over the last three decades. A less re- fined version of this hypothesis may be found in Green (1999), appended to this thesis.

We know from chapter four that the East End artists’ agglomeration has grown “from the ground up”, that it has been an artist-led phenomenon driven by individuals responding to their own personal circumstances, and seeking to fulfil their professional needs within a certain con- text. In chapter eight we learned that the social networks at an organisational level are rather less significant than the networks which exist at an individual level, and that it is at this level that we should look if we wish to get glimpses of the underlying dynamics of this phenomenon.

If we define the East End artists’ agglomeration as a system, then, we can propose five “indicators” for its being a complex adaptive system. They are phase transition or an “edge of chaos” urban context, non-linearity, sensitive dependence upon initial conditions, adaptiveness and emergence. We shall go through these one by one, testing each to establish whether the ob- servations fit the hypothesis. We shall then examine the evidence in terms of Holland’s “seven basics”. Next, we develop the theory further through the concept of fitness landscapes, and then propose a theory comprising a simple set of rules and corresponding set of assumptions, which, it is argued, will generate the behaviour observed in the system.