8 September 2006, Computational Social Choice Seminar, Eric Pacuit
Whether made explicit or implicit, knowledge theoretic properties such as common knowledge of rationality are important in understanding and modeling game-theoretic, or strategic, situations. There is a large literature devoted to exploring these and other issues related to the epistemic foundations of game theory. Much of the literature focuses on what the agents need to know about the other agents' strategies, rationality or knowledge in order to guarantee that a particular solution concept, such as the Nash equilibrium, is realized.
This paper develops a framework that looks at similar issues relevant to the field of voting theory. Our analysis suggests that an agent must possess information about the other agents' preferences in order for the agent to decide to vote strategically. In a sense, our claim is that the agents need a certain amount of information in order for the Gibbard-Satterthwaite theorem to be 'effective'.
This is joint work with Rohit Parikh.