Election Manipulation 100

Abstract

The true election margin for an Instant Runoff Voting (IRV) election can be hard to compute, because a small modification early in the elimination sequence can alter the outcome and result in a candidate winning the last round by a large margin. It is often assumed that the true margin is the last-round margin, that is half the difference between the two candidates who remain when everyone else is eliminated, though it is well known that this need not be the case. Perceptions of confidence in the outcome, and even formal policies about recounts, often depend on the last-round margin. There is already some prior work on how to compute the true election margin efficiently for IRV, and hence how to find the minimal manipulation. In this work we show how to manipulate an election efficiently while also producing a large last-round margin. This would allow a successful manipulation to evade detection against naive methods of assessing the confidence of the election result. This serves as further evidence for accurate computations of the exact margin, or for rigorous Risk Limiting Audits which would detect a close or wrong election result (respectively) regardless of the last-round margin.