Evolutionary paths to a corrupt society of artificial agents
Engineering Management Program, American University of Beiru
Last modified: June 13, 2006
A computer-generated analogue of societies with a high level of corruption is presented based in iterated prisonsers' dilemma (IPD) interactions among spatially mobile self-reproducing agents. The society does not naturally die out as the corrupt class siphons off resources, as habitually found in "evolution of altruism" models. Instead, agents survive by ignoring or avoiding non-cooperating individulas, allowing a majority of unethical (non-cooperative) individuals to survive off the efforts of a cooperative, hence productive, minority.
The limits of this phenomenon in parameter space are explored, and the evolutionary path traced by a society as it spirals into a corrupt state is described for the different types of circumstances that lead to this fate. Speculation about possible lessons to draw for actual human societies in which similar conditions exist is then presented, for example for an international infrastructure contractor or aid agency trying to do a project in a third-world country where cooperation is not to be taken for granted.