Agent-based Modeling as Quintessential Tool for Open-ended Social Theorizing

Abstract

Thought experiments have long been employed as a technique of analytical thinking where the theorist poses “what if” questions to some articulated model. The entities in those models are typically aggregate variables, though sometimes those variables masquerade as individuals by adopting such analytical devices as representative agent models. While this procedure has tractability on its side, it also prevents analysis of the myriad ways in which social interaction can generate new phenomena. For instance, a representative agent does not generate prices or business organizations, nor engage in conflict and develop procedures for settling conflict. To understand the place of such practices and institutions within a social economy requires a generative and not a stipulative mode of analysis. This paper explains how agent-based computational modeling can serve as a valuable modeling platform for theorizing about societies in an open-ended manner where internally-generated change is a property of social interaction.

Keywords: agent-based modeling; open-ended theorizing; mechanical vs. creative social systems; generation vs. stipulation; debate vs. discussion in political economy

JEL Classification: B40, B55, C90, P16