Robust Perfect Revenue From Perfectly Informed Players

Jing Chen, A. Hassidim, Silvio Micali

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Maximizing revenue in the presence of perfectly informed players is a well known goal in mechanism design. Yet, all current mechanisms for this goal are vulnerable to equilibrium selection, collusion, privacy and complexity problems, and therefore far from guaranteeing that maximum revenue will be obtained. In this paper we both clarify and rectify this situation by • Proving that no weakly dominant-strategy mechanism (traditionally considered immune to equilibrium selection) guarantees an arbitrarily small fraction of the maximum possible revenue; and, more importantly, • Constructing a new mechanism, of extensive-form and with a unique sub-game-perfect equilibrium, which (a) guarantees a fraction arbitrarily close to 1 of the maximum possible revenue; (b) is provably robust against equilibrium selection, collusion, complexity, and privacy problems; and (c) works for any number of players n > 1, and without relying on special conditions for the players utilities.
Original languageAmerican English
Title of host publicationICS
StatePublished - 2010

Bibliographical note

Place of conference:China

Fingerprint

Dive into the research topics of 'Robust Perfect Revenue From Perfectly Informed Players'. Together they form a unique fingerprint.

Cite this