New model army

There may be sound practical reasons for modelling human behaviour and human associations, for example, to develop a heuristic to use in designing a computer system. Here, epistemological and theoretical questions about the relationship between the model and the modelled may be less relevant to the process of modelling than the model's ability to lever practical problems.

The editors of Simulating Societies, however, subordinate practical considerations to epistemological, theoretical and methodological questions involved in modelling and simulating "society". They argue that modelling is a methodological device for the study of society which they champion because it can provide precise, objective and repeatable findings, and thereby claim that modelling and simulation provide scientific access to social structure. In this vein their introductory chapter, which discusses the concepts of modelling and simulation, emphasises questions concerning the relationship between the model and "reality", verification methods, procedures and processes of modelling, theory selection, and the problems involved in defining society.

Their claimed scientific rationale for modelling social structure is continued in Ann Seror's chapter where the epistemological underpinnings of a number of simulation models of social systems are discussed and their "contribution to scientific enquiry" emphasised.

The remaining chapters are given over to studies which use models and simulations to examine a range of traditional issues in the social sciences such as social change; the relationship between individual behaviour and social order; urban movement; hunter-gatherer societies; cognition and social interaction; the evolution of technology; the adaptation of social structure to changes in space and resources, and the emergence of so-called societal processes (the "society" under question here being an ant colony). Although these chapters are presented as studies of particular societal phenomenon, they also carry over the epistemological and theoretical themes of the first two chapters and endorse the scientific claims of modelling and simulation, giving as much, if not more weight to their models and simulations as they do to the topics they are investigating.

To a non-social science readership the attempts to model and simulate societal relationships and processes on computers may appear to be a novel and innovative step in the use of computer technology in scientific investigations. However, the emphasis upon the use of models and simulations in the study of society targets the book on the traditional social science disciplines and many, especially in sociology, will find the epistemological, theoretical and methodological foundations of Simulating Societies to be neither innovative nor novel but depressingly familiar. The idea that human activity can be accounted for in terms of mathematical formulations will be viewed by many as the rankest version of the traditional scientistic/positivistic character of social science enquiry that, from their point of view, has confused sociological thinking from its very inception.

The development of a mathematical sociology was once considered to be a high priority in establishing the scientific credentials of social enquiry. By claiming a scientific status for modelling, Simulating Societies picks up this baton. Previous attempts to mathematicalise the study of social structure were, however, undertaken at the zenith of a positivistic tradition in modern sociology. Currently, and partly in reaction to this tradition, sociology has, across a number of fronts, attempted to purge itself (often unsuccessfully) of its scientistic protestations. For example, the assumption that there is an isomorphism between mathematical tokens and the observed world has been challenged because it ignores the ways in which mathematical measurement is entwined with the social order it purportedly measures and thus ignores the way in which measurement systems constitute the very social order they are characterised as objectively representing. This challenge has a poignant relevance for the present collection for it has been articulated by the Sociology of Scientific Knowledge movement with which one of the editors of the present collection (Nigel Gilbert) has been associated.

To many in the social sciences the extensively articulated epistemological, theoretical, and methodological discussions about how to build models of social processes, select, and apply appropriate programmes will thus seem both naive and arcane. While the legitimisation of a mathematical account of social structure that it is "scientific" may, in itself, have satisfied many sociological criteria at the time when mathematical sociology was first broached, it will be now viewed in many quarters as simply no longer sufficient. Far from being seen as a novel turn in the study of society, the emphasis upon an anachronistic positivist conception of social structure will mean, for many, that the collection is becalmed in an ill-conceived epistemology that it does little to move forward. The opportunity to think innovatively about the ways in which social processes may be simulated on the computer for other than claimed scientific purposes is another matter, but then that would be different book from Simulating Societies.

Graham Button is senior scientist at the Rank Xerox Research Centre, Cambridge Laboratory.