A Quantitative and Qualitative Analysis of the Robustness of (Real-World) Election Winners

TL;DR

This paper assesses the robustness of election winners to noise, comparing voting rules and analyzing real-world elections like Formula 1 and political contests, revealing many non-robust winners and complex robustness patterns.

Contribution

It introduces a comprehensive framework for evaluating election robustness, applying it to real-world cases and highlighting limitations of traditional analysis methods.

Findings

Many election winners are highly non-robust to noise

Complex robustness patterns are common and often undetectable by classical methods

Real-world elections exhibit delicate robustness characteristics

Abstract

Contributing to the toolbox for interpreting election results, we evaluate the robustness of election winners to random noise. We compare the robustness of different voting rules and evaluate the robustness of real-world election winners from the Formula 1 World Championship and some variant of political elections. We find many instances of elections that have very non-robust winners and numerous delicate robustness patterns that cannot be identified using classical and simpler approaches.