A cardinal approach to straightforward probabilistic mechanisms (Q761227): Difference between revisions

This paper extends Gibbard's theorem on the manipulation of schemes that mix voting with chance, to the domain of cardinal utilities. Let there be a finite set of voters N (at least two), and a finite set of alternative S (at least three). Each player i has a von Neumann-Morgenstern utility \(u_ i\) defined over S. A probabilistic mechanism is a map from the vector of reported utilities \((u'_ i)\) to the space of probability measures on S. Such a mechanism is straightforward if, for every player i, \(u_ i\) is a dominant strategy for expected utility maximization. The major result of the paper gives necessary and sufficient conditions for a probabilistic mechanism to be straightforward. The class of such mechanisms includes Gibbard's unilateral and duple mechanisms as special cases.