ON COMPUTATIONAL SOCIAL LAWS FOR DYNAMIC NONHOMOGENEOUS SOCIAL-STRUCTURES

Approaches to the coordination of artificial agents may be inspired by observations on human and other natural societies. On the other hand, the introduction and the analysis of general computational models and mechanisms of coordination may shed light on the general theory of coordination. This paper extends a fundamental approach to the coordination of artificial agent societies, the artificial social systems approach, in order to incorporate several features observed in human societies. As a result, the approach becomes more powerful, and illuminating computational results are obtained. An artificial social system is a basic mechanism of coordination. It decreases the need for both centralized control and on-line resolution df conflicts by introducing a set of social laws that enable agents to work individually in a mutually compatible manner. This work extends the existing work on artificial social systems in a variety of directions. (a) We present a model that refers explicitly to social laws in non-homogeneous societies. (b) We introduce the problem of dynamic change of the set of potential goals to the artificial social systems framework, and define and investigate the corresponding computational issues. (c) We address the design of societies taking into account their division into sub-societies. The computational problems we define and investigate in this paper extend in a precise sense the basic computational problems in the design of artificial societies.