Agent-Based Simulation as a Method for International Political Science: A Way of Expressing Diversity

Abstract

This chapter tried to demonstrate utility of agent-based simulation (ABS) for the studies of international relations (IR). Traditionally, IR studies have been heavily dependent on historical approaches emphasizing detailed empirical analyses. Although more “scientific” approaches such as mathematical methods and game theories were introduced, empirical approaches are still much more dominant than economics.

However, it is impossibility of verification through experiments in this field, because IR studies have an inherent difficulty. That is, elements of the events discussed in this field are usually diverse and complicated which forces researcher to shed lights on a variety of elements of specific events. To discuss specific events in the detailed way within the framework of a general theory, one would need to make more presumptions, or they have to be replaced with more concrete alternatives. But how can one select appropriate presumption?

ABS model, we are proposing in this chapter, has advantages in overcoming the difficulties we discussed above. We create agents in computer and let them interact with each other within the virtual world setup in a computer. By “cultivating” nations this way, we can reduce, if not completely eliminate, risks of introducing arbitrary presumptions by the model builder. It also allows us to observe dynamic process of interactions of nations or to conduct an experiment of virtual history by obtaining path-dependent properties of agents.