Optimised Reputation-Based Adaptive Punishment for Limited Observability

The use of social norms has proven to be effective in the self-governance of decentralised systems in which there is no central authority. Axelrod´s seminal model of norm establishment in populations of self-interested individuals provides some insight into the mechanisms needed to support this through the use of metanorms, but is not directly applicable to real world scenarios such as online peer-to-peer communities, for example. In particular, it does not reflect different topological arrangements of interactions. While some recent efforts have sought to address these limitations, they are also limited in not considering the point-to-point interactions between agents that arise in real systems, but only interactions that are visible to an entire neighbourhood. The objective of this paper is twofold: firstly to incorporate these realistic adaptations to the original model, and secondly, to provide agents with reputation based mechanisms that allow them to dynamically optimise the intensity of punishment ensuring norm establishment in exactly these limited observation conditions.