Using Mutli-Agent Systems (MAS) and Evolutionary Computation (P,C), we explore the creation of stable societies and social structures with applications in the computer games domain. Genetic programming techniques are used to automatically generate robust populations of ambient characters. We evolve scripts for a group of Non Player Characters (NPCs) to solve the collective problem of subsistence by way, of a. common pool. By considering the group of NPCs in this way, we create implicit relationships and varying behaviours for the members. The defined problem requires coordination among the different NPCs and their roles, while containing features of many social dilemmas. We investigate the effect of varying the problem so that. it, ranges from a relatively easy evolutionary problem to a. difficult one.