Optimal Voting Rules

Abstract

Modern social choice theory, following Kenneth Arrow, treats voting as a method for aggregating diverse preferences and values. An earlier view, initiated by Marquis de Condorcet, is that voting is a method for aggregating information. Voters’ opinions differ because they make errors of judgment; absent these errors they would all agree on the best choice. The goal is to design a voting rule that identifies the best choice with highest probability. This paper examines maximum likelihood estimation. Surprisingly, the optimal rule can also be axiomatized by variations of Arrow's axioms.