Arrow's Impossibility Theorem: Why No Voting System Is Perfect

simulator advanced ~12 min
Loading simulation...
Candidate A wins under plurality voting

With default parameters (5 voters, 3 candidates, plurality voting, random preferences), the candidate with the most first-place votes wins. Switch voting methods to see how the same preferences can produce different winners.

Formula

Borda score for candidate i: sum over all voters of (num_candidates - rank_of_i)
Condorcet winner: candidate c where for all j != c, |{voters: c > j}| > n/2
IIA test: winner(full ballot) vs winner(ballot without losing candidate k)

The Impossibility of Fair Voting

In 1951, economist Kenneth Arrow proved something astonishing: no voting system for three or more candidates can simultaneously satisfy a small set of basic fairness requirements. This isn't a limitation of current methods — it's a mathematical impossibility. Any ranked voting system must violate at least one of: unanimity (respect unanimous preferences), independence of irrelevant alternatives (removing a non-winner shouldn't change the result), or non-dictatorship.

Same Voters, Different Winners

This simulation makes Arrow's theorem visceral. Enter a set of voter preferences, then switch between five different voting methods: plurality, Borda count, Condorcet pairwise comparison, instant-runoff ranked choice, and approval voting. Watch as the same ballots produce different winners under different counting rules. The 'will of the people' depends entirely on how you aggregate their preferences — there is no neutral, objectively correct method.

Condorcet Cycles and Intransitivity

Perhaps the deepest paradox is the Condorcet cycle. Set the preference scenario to 'cycle' and observe: candidate A beats B in a head-to-head vote, B beats C, but C beats A. The group's preferences form a rock-paper-scissors loop with no coherent 'best' choice. This is not a flaw in any particular voting system — it's a fundamental property of aggregating individual preferences into group decisions. The Marquis de Condorcet discovered this paradox in 1785, and it remains one of the deepest results in social choice theory.

Implications for Real Democracy

Arrow's theorem doesn't mean democracy is futile — it means we must choose which fairness criteria to prioritize. Plurality is simple but ignores depth of preference. Ranked choice prevents spoilers but can violate monotonicity. Approval voting escapes Arrow's framework (it uses cardinal, not ordinal input) but introduces strategic threshold decisions. Understanding these tradeoffs is essential for informed democratic design. Every voting reform debate is, at its core, a debate about which of Arrow's axioms we're willing to sacrifice.

FAQ

What is Arrow's Impossibility Theorem?

Arrow's theorem (1951) proves that no ranked voting system with 3+ candidates can simultaneously satisfy three reasonable fairness criteria: unanimity (if all prefer A to B, so should the group), independence of irrelevant alternatives (removing a non-winner shouldn't change the winner), and non-dictatorship (no single voter always determines the outcome).

What is a Condorcet cycle?

A Condorcet cycle occurs when majority preferences are intransitive: A beats B, B beats C, but C beats A in pairwise comparisons. This means there is no 'true' majority winner — the group's preferences are inherently contradictory, making 'the will of the people' incoherent.

How do different voting methods produce different winners?

Plurality counts only first-place votes; Borda assigns points by rank; Condorcet checks pairwise matchups; ranked-choice eliminates candidates iteratively; approval lets voters approve multiple candidates. Each method weighs preference information differently, so the same ballots can yield different winners.

Is there any way around Arrow's theorem?

Cardinal voting systems like approval voting and score voting technically escape Arrow's framework (which applies only to ordinal/ranked systems). However, the Gibbard-Satterthwaite theorem shows that all non-dictatorial voting methods are susceptible to strategic voting — there is no perfect system.

Sources

Embed

<iframe src="https://homo-deus.com/lab/voting-democracy/arrows-impossibility/embed" width="100%" height="400" frameborder="0"></iframe>
View source on GitHub