Bridging game theory and cryptography: recent results and future directions

Motivated by the desire to develop more realistic models of, and protocols for, interactions between mutually distrusting parties, there has recently been significant interest in combining the approaches and techniques of game theory with those of cryptographic protocol design. Broadly speaking, two directions are currently being pursued: Applying cryptography to game theory: Certain game-theoretic equilibria are achievable if a trusted mediator is available. The question here is: to what extent can this mediator be replaced by a distributed cryptographic protocol run by the parties themselves? Applying game-theory to cryptography: Traditional cryptographic models assume some honest parties who faithfully follow the protocol, and some arbitrarily malicious players against whom the honest players must be protected. Game-theoretic models propose instead that all players are simply self-interested (i.e., rational), and the question then is: how can we model and design meaningful protocols for such a setting? In addition to surveying known results in each of the above areas, I suggest some new definitions along with avenues for future research.