Voices of Justice: Finding Consensus in the Multitude of Claims

A well-known aphorism in English law can be summarised as: "not only must justice be done, it must also be seen to be done". In this paper, we consider the implications of this statement for a policy of distributed distributive justice intended to deliver 'fair' resource allocation in open multi-agent systems, in particular, where one designated agent is responsible for computing and declaring the outcome of the resource allocation process. The proposed solution uses an information dissemination framework that is based on trust, social influence and opinion formation, allowing multiple local processes to aggregate and converge on an overall feedback measure for the effectiveness of the system in delivering fair outcomes. We consider a scenario where agents are subject to different resource allocation policies, and show how the system is able to deal with diverse and divergent accounts and reach consensus on its general fairness. Experiments show that our strategy is able to distinguish between fair and unfair allocations, is robust to malicious agents by preventing propagation of false impressions, and can operate successfully in different topologies.