The Mathematics Behind Voting Systems and Their Inherent Flaws

The Impossibility at the Heart of Democracy

In 1951, economist Kenneth Arrow proved a theorem so unexpected that it won him the Nobel Prize and has never been refuted: no voting system that converts individual ranked preferences into a collective ranking can simultaneously satisfy a small set of seemingly reasonable fairness requirements. Every voting method in use — plurality, ranked-choice, proportional representation, Condorcet — violates at least one of Arrow's criteria. There is no perfect democratic procedure, not as a political failure but as a mathematical theorem.

This is not a counsel of despair. Different voting systems make different tradeoffs, producing systematically different outcomes even from identical voter preferences. The choice of electoral method determines not just who wins but what kind of governance emerges — a fact well understood by the politicians who choose or refuse to change the rules. Arrow's theorem makes this explicit: any choice of voting system is a choice of which fairness criteria to prioritize and which to sacrifice.

What Voting Systems Are Trying to Do

Social choice theory — the mathematical study of collective decision-making — frames the problem precisely. An election involves N voters, each with a preference ordering over M candidates. The voting system is a function that takes these N individual orderings as input and produces a single collective ordering (or a winner) as output. What properties should this function satisfy?

Arrow identified four seemingly minimal requirements:

• Unanimity (Pareto efficiency): if every voter prefers candidate A over candidate B, the collective outcome should also prefer A over B.
• Independence of Irrelevant Alternatives (IIA): the collective ranking of A vs. B should depend only on voters' preferences between A and B — not on their attitudes toward a third candidate C who is not being chosen.
• Non-dictatorship: no single voter should always determine the collective outcome regardless of others' preferences.
• Unrestricted domain: the system must work for all possible combinations of voter preferences.

Arrow's theorem: no voting function satisfying unrestricted domain can satisfy all of unanimity, IIA, and non-dictatorship simultaneously. One of these must be violated.

Plurality Voting: Simple, Widely Used, Deeply Flawed

The plurality system — used in US presidential primaries, UK general elections, and most local elections — is simplest: each voter picks one candidate; the most votes wins. It violates IIA dramatically, producing the spoiler effect.

In the 2000 US presidential election in Florida, Al Gore received 2,912,790 votes, George W. Bush received 2,912,537 votes (winning by 537), and Ralph Nader received 97,421 votes. Most Nader voters' second preference was Gore. Under plurality, Nader's presence changed the outcome — removing him likely would have given Gore the state and the presidency. Nader was, in Arrow's terminology, an irrelevant alternative: he didn't win, but his presence changed who did.

Plurality also systematically disadvantages third parties through Duverger's Law — the empirical observation that plurality elections tend toward two-party systems. Voters who prefer a third party rationally abandon their preference to avoid wasting their vote on a likely loser, producing preference falsification at scale and reinforcing the two-party equilibrium.

Common Voting Systems Compared

The Condorcet Paradox

The Marquis de Condorcet identified the fundamental cyclical problem in 1785. Three voters have preferences:

Voter 1: A > B > C. Voter 2: B > C > A. Voter 3: C > A > B.

Pairwise comparisons: A beats B (voters 1 and 3 prefer A over B). B beats C (voters 1 and 2 prefer B over C). C beats A (voters 2 and 3 prefer C over A). The collective preference is A > B > C > A — a cycle with no winner. This intransitivity — rational individual preferences producing an irrational collective preference — is why no perfect aggregation rule exists.

Condorcet cycles are not theoretical abstractions. Research by William Riker and others has argued that several historical legislative outcomes, including the passage of certain US constitutional amendments, involved majority cycles that were resolved only by agenda manipulation — the order in which alternatives were voted on determined the outcome.

Gibbard-Satterthwaite: The Manipulation Theorem

Arrow's theorem applies to ranked systems. Allan Gibbard (1973) and Mark Satterthwaite (1975) proved an equally disquieting result for voting rules that produce a single winner: any deterministic voting rule with more than two candidates that is not a dictatorship can be strategically manipulated by some voter misrepresenting their true preferences. Honest voting is not always optimal under any non-dictatorial system.

This explains the ubiquity of strategic voting. Under plurality, genuine supporters of a third-party candidate often vote for their less-preferred major-party candidate to avoid the worst outcome. Under ranked-choice voting, voters sometimes have incentive to rank a strong second-choice lower than truthful preference to avoid triggering elimination dynamics. The specific manipulation differs by system, but some manipulation is always possible.

Approval Voting and Score Voting: Breaking Free of Rankings

Some alternatives step outside the ranked-choice framework entirely. Approval voting lets each voter approve any number of candidates; the most approved wins. It satisfies many desirable properties and dramatically reduces spoiler effects — approving a sincere favorite never hurts your preferred major candidate. It's used by the Mathematical Association of America, American Statistical Association, and several other professional organizations.

• Score voting (Range voting): voters give each candidate a score on a numerical scale (0–10, say); highest average wins. More expressive than approval voting; allows intensity of preference to register. Satisfies IIA (as a non-ranking system, Arrow's theorem doesn't apply in the same form), but faces criticism that voters will strategically give extreme scores rather than honest ratings.
• Quadratic voting: voters receive a budget of "voice credits" and can allocate votes to issues proportional to the square root of credits spent. Allows intensity of preference to count; reduces the power of concentrated minorities to dominate over diffuse majorities. Developed by Glen Weyl and Eric Posner; used experimentally in Colorado legislative priority-setting in 2019.

Every electoral choice is a tradeoff embedded in mathematics. Plurality is simple and decisive but strategically toxic. Ranked-choice reduces spoilers but introduces non-monotonicity. Proportional representation accurately mirrors voters but complicates governance. Condorcet methods are theoretically compelling but can cycle. Arrow proved that no system avoids all problems — but that doesn't mean the choice is arbitrary. It means the choice must be made explicitly, with clear understanding of what each system maximizes and what it sacrifices. Democratic design is applied mathematics.