False-name-proof and strategy-proof voting rules under separable preferences

Abstract

We consider the problem of a society that uses a voting rule to choose a subset from a given set of objects (candidates, binary issues, or alike). We assume that voters’ preferences over subsets of objects are separable: adding an object to a set leads to a better set if and only if the object is good (as a singleton set, the object is better than the empty set). A voting rule is strategy-proof if no voter benefits by not revealing its preferences truthfully and it is false-name-proof if no voter benefits by submitting several votes under other identities. We characterize all voting rules that satisfy false-name-proofness, strategy-proofness, and ontoness as the class of voting rules in which an object is chosen if it has either at least one vote in every society or a unanimous vote in every society. To do this, we first prove that if a voting rule is false-name-proof, strategy-proof, and onto, then the identities of the voters are not important.