The social life of computer simulations

Abstract

Computer simulations of social dynamics have been used in social science for over 50 years. Their impact on society is, however, not very well studied. While the public discourse around large simulation efforts in the climate sciences receives academic attention, the mundane use of computer-simulated social dynamics, for instance in banks or insurance companies, has increased over the last years, albeit “under the radar” of academic inquiry. My paper seeks to shed some light on the use of computer simulations for forecasting social dynamics in the field of finance. The main argument is that numerically produced forecasts are contested forms of predictive knowledge. They compete with alternative forms of knowledge which are based on personal experience or statistical procedures. Thus, the paper sketches out the social life of social simulations as they are produced, discussed, legitimised, and challenged in financial institutions. The main research question is derived from pragmatist reasoning by looking at how the predictions drawn from computer simulations are “made true”. The empirical material provides insights into the procedures of creating and running simulations, into the practices of calibrating and adjusting the models with available data and into possible interpretations thereof. The ensuing analysis draws in particular from research on epistemic instruments and computer simulations in science and technology studies.