The power of parallel thinking

Abstract

A small computer model demonstrates that an appropriate organization of boundedly rational individuals can find optimal policies in an environment that is overwhelmingly complex for unorganized decision makers. The model is also used to identify conditions under which optimal -- or even good -- policies are not found. The demonstrated adaptive power of the model is interpreted in light of recent developments in the theory of computational complexity that place new stress on powerful methods of search, and of new models from computer science which markedly advance search effectiveness by harnessing parallel structures of information processing.