Axiomatizations of a simple Condorcet voting method for Final Four and Final Five elections

Wesley H. HollidayarXiv 2025 · DOI · arXiv

Introduction

As political reforms in the United States move toward "Final Four" and "Final Five" systems—where the top candidates from a primary advance to a ranked-choice general election—the search for an ideal voting method intensifies. Condorcet methods are often hailed as the "gold standard" because they ensure that a candidate who can defeat every other opponent in a head-to-head matchup wins the election. However, these methods often suffer from high mathematical complexity and the "cycle problem," where no single candidate defeats all others. This paper provides a rigorous mathematical foundation for a simple Condorcet method designed specifically for these small-field elections, bridging the gap between social choice theory and practical political application.

The Complexity and Cycle Problems

Condorcet methods face two primary hurdles. First, sophisticated rules like Schulze or Kemeny-Young are difficult to explain to the public, often appearing as a "black box." Second, in rare instances, a "Condorcet cycle" occurs (e.g., A beats B, B beats C, and C beats A), requiring a tie-breaking mechanism. Empirical data from over 300 U.S. ranked-ballot elections suggests that while Condorcet winners almost always exist, a robust system must handle cycles fairly and transparently.

The Most Wins, Smallest Loss (MWSL) Method

The author proposes the Most Wins, Smallest Loss (MWSL) method. This rule is designed to be both mathematically sound and intuitively explainable:

  1. Copeland Score: Count the number of head-to-head matchups each candidate wins.
  2. Selection: The candidate(s) with the highest number of wins are considered.
  3. Tie-Breaking: If multiple candidates have the same number of wins, the winner is the one whose smallest head-to-head loss (by vote margin) is the smallest.

This "smallest loss" logic identifies the candidate "closest" to being an outright winner; a minor shift in votes in their worst matchup would have secured them the Condorcet position.

Tournament scenarios illustrating Condorcet winners and various cycle types.

Axiomatic Characterization

The paper employs axiomatic characterization to evaluate MWSL. By defining a set of desirable fairness principles (axioms), the author proves that MWSL is the unique method satisfying them for small candidate pools. Key axioms include:

  • Proximity to Condorcet/Copeland: A candidate "closer" to winning should not be passed over for one further away.
  • Independence of Irrelevant Defeats (IID): The outcome between candidates A and B should not be affected by the margins of matchups between other candidates.
  • Win Monotonicity: Increasing a winner's margin should never cause them to lose.
  • Win Dominance: A candidate who is dominated in total wins by another should not be elected.
  • Rare Ties: The method should rarely result in an unbreakable tie in large elections.

Key Results

The core contribution of the paper is the unique characterization of MWSL for specific election sizes. Theorem 4.1 proves that for four-candidate elections, MWSL is the only method satisfying the core axioms. Theorem 5.3 extends this to five-candidate elections by adding the Immunity to Spoilers axiom.

A critical finding is the comparison between MWSL and the popular Minimax method. While Minimax focuses on minimizing the worst loss, it can violate "Win Dominance" by electing a candidate with fewer total wins than their peers. MWSL maintains win dominance while remaining simple to calculate.

Comparison of voting methods against key axiomatic properties.

Conclusion and Future Work

By proving that MWSL is the unique method satisfying intuitive fairness principles for Final Four and Final Five elections, this research provides a "seal of approval" for a practical, explainable Condorcet rule. While the proofs currently target 4 and 5 candidates, future work may extend these characterizations to larger fields. Additionally, while the method is normatively fair, its resistance to strategic voting remains an open area for game-theoretic analysis.

Summary updated

Abstract

Proponents of Condorcet voting face the question of what to do in the rare case when no Condorcet winner exists. Recent work provides compelling arguments for the rule that should be applied in three-candidate elections, but already with four candidates, many rules appear reasonable. In this paper, we consider a recent proposal of a simple Condorcet voting method for Final Four political elections. Our question is what normative principles could support this simple form of Condorcet voting. When there is no Condorcet winner, one natural principle is to pick the candidate who is closest to being a Condorcet winner. Yet there are multiple plausible ways to define closeness, leading to different results. Here we take the following approach: identify a relatively uncontroversial sufficient condition for one candidate to be closer than another to being a Condorcet winner; then use other principles to help settle who wins in cases when that condition alone does not. We prove that our principles uniquely characterize the simple Condorcet voting method for Final Four elections. This analysis also points to a new way of extending the method to elections with five or more candidates that is simpler than an extension previously considered. The new proposal is to elect the candidate with the most head-to-head wins, and if multiple candidates tie for the most wins, then elect the one who has the smallest head-to-head loss. We provide additional principles sufficient to characterize this simple method for Final Five elections.
BibTeX
@misc{holliday2025axiomatizations,
  author = {Wesley H. Holliday},
  title = {{Axiomatizations of a simple Condorcet voting method for Final Four and Final Five elections}},
  year = {2025},
  eprint = {2508.17095},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2508.17095},
}
Figure