Definition Condorcet Winner
definition formalized

Condorcet Winner

An alternative $a \in A$ is a Condorcet winner for a preference profile $P$ if it beats every other alternative in pairwise strict-majority comparison ([[social_choice.voting.majority_prefers]]): $$ \forall b \in A,\ b \ne a,\ \mathrm{MajorityPrefers}(P, a, b). $$

A profile has a Condorcet winner if some such $a$ exists.

In Lean: SocialChoice.Voting.CondorcetWinner and HasCondorcetWinner.

A Condorcet winner — when one exists — is the canonical "majority-acceptable" choice. But Condorcet winners need not exist: cyclic pairwise majorities ("Condorcet paradox", [[social_choice.voting.condorcet_paradox]]) are possible already with 3 voters and 3 alternatives.

Concrete rules that always elect the Condorcet winner when it exists are called Condorcet methods (Copeland, Schulze, Ranked Pairs, …). Borda ([[social_choice.voting.borda_score]]) and plurality ([[social_choice.voting.plurality_score]]) are not Condorcet methods.

References

  • [MSZ, Chapter 21] Maschler, Solan, and Zamir, Game Theory. Condorcet winner; see the source block above for the precise locator.

Used by

Also in