What Counts as the Best Way to Count Votes?
July 12, 2024
What Counts as the Best Way to Count Votes?
-Allen Stairs
It's easy to think that in most cases–choosing a mayor or a member of Congress–the question of who wins an election has a simple answer. Eric Pacuit (pacuit.org), who has been a member of the Philosophy Department since 2012, would beg to differ. The most common rule for vote-counting is that the candidate with the most votes wins; but since a plurality isn't a majority, that can misrepresent voter preferences. Some elections (in Alaska and Maine, for instance) ask voters to rank their choices. If there's no majority winner, that triggers an instant runoff. But as we'll see, instant runoff rules can produce perverse results.
In Eric's view, if there's a candidate who wins against everyone else one-on-one, a so-called Condorcet winner (after the Marquis of Condorcet, an 18th-century mathematician and philosopher) then that candidate should win the election. Real world voting systems don't always work that way. Eric points to Alaska in 2022. There were three candidates in a special-election race for the US House: Mary Peltola, Nick Begich and Sarah Palin. Voters ranked their choices, but no one got a majority of first place votes. That election used instant run-off rules, and at each stage in the counting, the candidate with the fewest first-place votes was eliminated. In this election, Nick Begich was eliminated in the first round. After deleting Begich's name, Peltola had a majority of first-place votes against Palin and won the seat. But in fact, more voters preferred Begich to Palin than vice-versa and more voters preferred Begich to Peltola than vice-versa. Begich was the Condorcet winner. On Eric's view, respect for majority rule demands that the seat should have gone to Begich. He and his collaborator Wesley Holliday from UC Berkeley are writing a book defending a voting method called Stable Voting that addresses this and other problems with common vote-counting methods.
In addition to picking the head-to-head winner if there is one, Stable Voting addresses another serious issue: the spoiler. Roughly, Y spoils the election for X if X would win if Y weren't in the race, but with Y in the race, X and Y both lose. On voting systems like plurality voting ("first past the post") and instant run-off, voting for the candidate you most prefer may help elect your least-favored candidate. Stable Voting is designed to minimize the chance that voting your heart will have that consequence. Eric believes that this feature of Stable Voting could help overcome the deep polarization that besets American politics. On Stable Voting, if X would win an election against Y, adding a candidate who wouldn't beat X one-on-one can't change that. The same holds if the field is larger. But the best strategy for being a candidate like X is to be the kind of candidate who might well beat all other candidates one-on-one. That calls for broadening your appeal and being willing to compromise. Some voters clearly thrive on extreme partisanship, but Eric and his collaborator find it hard to believe that that's good for democracy in the long run. They argue that for many kinds of elections, Stable Voting is the best way to count the votes.
Perhaps someday you'll have a chance to cast a vote in a civic election that uses Stable Voting rules. In the meantime, you can find out more about Stable Voting by going to
There you can read explanations, play with hypothetical scenarios, and even use the program to conduct real polls and elections.