Robust Reputation in Decentralised Markets

Establishing cooperation and protecting individuals from selfish or malicious behaviour are key goals in open multi-agent systems. Incomplete information regarding potential interaction partners can undermine mechanisms such as trust and reputation, particularly in lightweight systems designed for individuals with significant resource constraints. In this paper, we (i) propose extending low cost reputation mechanisms with gossiping algorithms, (ii) introduce four simple aggregation strategies for incorporating gossiped information, and (iii) evaluate our model on a variety of synthetic and real-world topologies. We show that (i) gossiping can significantly reduce the potentially detrimental influence of incomplete information and network structure on lightweight reputation mechanisms, (ii) using the most recently received gossip to make decisions results in up to a 25% reduction in selfishness, and (iii) gossiping is particularly effective at aiding agents with little or no interaction history, such as when first entering a system.