ABSTRACT
Several years ago (Doran 1982), I suggested that multiple agent systems (MAS) theory could form the basis of models of socio-cultural dynamics including the growth of social complexity. Since then MAS theory and distributed artificial intelligence (DAI) generally have developed substantially (Bond and Gasser 1988; Gasser and Huhns 1989; Demazeau and Muller 1990) and now the idea of studying ‘societies’ on computers is becoming not just tenable but fashionable – although the emphasis is as yet largely on studying the properties of systems of abstract rather than realistic agents. In spite of this limitation, it now looks possible to develop my original suggestion in a more serious way, and briefly to compare it with the more prominent alternatives.
