Agent-based modelling: past, present and future

It is about fifteen years* since social scientists first started talking about ‘agent-based models’. In this talk, I’ll take a look at what has been achieved during those years, try to summarise what I see as the current ‘state of the art’, and list some challenges for the future. Most of the talk will be based on personal reminiscences (it is the last seminar of the academic year!), but I’ll also consult some presentations I gave in the late 1990s, co-citation analyses of JASSS and CMOT, and abstracts of recently funded European research projects. I shall conclude that agent-based simulation has so far avoided becoming a ‘normal’ science (Kuhn 1970) but may not do so for much longer.

* the earliest reference to multi-agent model I can find is 1994.