Reputation-based Trust in Decentralized Markets

Trust is an important factor that facilitates our social life. Since more and more social interactions are occurring over electronic channels supporting trust in such an environment is a fundamental issue, both for private and professional aspects of daily life. Traditional security approaches can assure that information originates from the right party and is not tampered with, but do not help to answer the question whether the party can be trusted. Exclusively relying on trusted intermediaries or trusted third parties implies to sacrifice the own autonomy. Decentralized ways for establishing trust correspond to a fundamental need and desire of humans to reduce risks in daily life and to preserve autonomy at the same time.

Reputation mechanisms offer a promising alternative for building trust in online markets. The reputation of an agent (i.e. information about past behavior) directly influences the revenues of that agent. Present misbehavior attracts bad reputation which in turn decreases the future income of the agent. When carefully designed, reputation mechanisms make cheating economically unattractive, since the present gain obtained from cheating is outweighed by future losses caused by bad reputation.

The goal of this project is to study robust and scalable computational mechanisms for implementing reputation-based trust in decentralized markets. The project brings together competences from the three most relevant areas: experimental economics, software agents, and distributed information systems. As results, we expect new mechanisms for exploiting the reputation effect in decentralized system, as well as sofware implementations, and experimental evaluations of their properties.

The project is a collaboration between three research laboratories: Artificial Intelligence Laboratory, the Institute of Research in Management and the Distributed Information Systems Laboratory coming from two universities: Ecole Polytechnique Fédérale de Lausanne and Université de Lausanne

This project is financed by the Swiss National Science Foundation.