Votes Can Be Confidently Bought in Instant Runoff Elections, and What to Do About It

Abstract

We show that in Instant Runoff Voting (IRV) elections it is often possible to violate the secret ballot. There are so many ways to rank even a handful of candidates that most possible votes will not be cast in a typical IRV election, so a vote buyer could purchase these votes, and then use the announced election results to verify that they were successfully cast. We examine the feasibility of this attack both theoretically and empirically. Theoretically, we show that when voters can rank a reasonably large number of candidates, most possible votes in an IRV election will be unique. Under very conservative assumptions, we show that vote buyers should expect a very low probability that a vote they purchase will be coincidentally cast by someone other than the colluding voter. We then verify this finding using empirical data from 36 IRV contests, including Cast Vote Records from San Francisco. While there is no evidence that this vote-buying scheme has ever been used, its existence has immediate implications for the administration and security of IRV elections: this scheme is more feasible when more candidates can be ranked in the election, and when election results are reported at lower levels of aggregation.

Keywords: Instant Runoff Voting, Ranked-Choice Voting, Election Security, Vote Buying