Applications-centred multi-agent systems design (with special reference to markets and rational agency)

This paper addresses the problem of the trade-off between system scale and complexity and agent functionality. The central argument is that the adoption of market and agent analogies from economics limits both agent functionality and system scale and generality. The feasibility of shifting out the trade-off between system scale and agent functionality is demonstrated by comparing an economics inspired specification of an information market and rational agents due to Kephart et al. (1998) with a market specification informed by the historical analysis and agent specification developed in social simulation research to represent real actors. These agents are able to function in a more complex software environment than are rational agents and the results of simulation experiments verify the system with respect to a historically based theory of markets. Moreover, the agent specification supports clear indications of the further requirements for agent functionality