Self-Uncertainty in Mechanism Design

Mechanism design aims to make the best decisions when the necessary data is in the hands of selfish players with different motives and incentives. A mechanism specifies what each player may report and a way to select the outcome from the players' reports. Although mechanism design was originally developed to study auctions in the field of finance, interest in mechanism design arose from its potential in the design of protocols for large distributed heterogeneous systems where the preferences of people and organizations influenced the outcome of the protocol. The proper and computationally efficient working of such systems does not solely rely on fast algorithms and cryptographically secure protocols but also on a correct incentive analysis.

This project investigates more general models of mechanism design. The players are assumed to be uncertain of their own true valuations, knowing only a set of candidate valuations. The notion of utility by a player may also be broadened. A player may maximize their utility while using regret minimization to refine their set of non-dominated strategies if needed. A player may seek to decrease their uncertainty and sharpen their understanding about their own utility by investing in the means to increase their knowledge about their own valuations before acting in a mechanism. These extensions will allow more robust and realistic mechanism design of cryptographic protocols.