JOVANA
Explore Library Glossary Getting Started Three Levels Fields How it works Mission
Join the mission
Back to the library
Economics 1962

College Admissions and the Stability of Marriage

David Gale & Lloyd Shapley

A matching no pair would walk out of — and a simple algorithm that always finds one.

Choose your version
In depth · the introduction

Pair up a room of people two by two so that no secret couple would both rather ditch their partners for each other — Gale and Shapley proved you always can.

The big idea

Imagine pairing people off — job-seekers with employers, students with schools, partners with partners. A pairing is stable if there is no pair of people who aren't together but would both rather be: no temptation strong enough to blow it up. In 1962 two mathematicians, David Gale and Lloyd Shapley, proved that such a stable arrangement always exists, however tangled everyone's preferences are — and they gave a simple recipe to find it.

How it came about

Gale and Shapley wrote their seven-page paper for a teaching journal, framing it as the 'stable marriage problem': n men and n women, each with a ranked wish-list. Their recipe is a courtship in rounds. Everyone proposes to their favourite; each person on the receiving end holds on to the best offer so far and turns the rest down; the rejected try again, one rank lower, the next round. Nobody is ever permanently stuck — the receiving side keeps trading up — and the music stops at a stable match.

What they didn't know is that a U.S. clearinghouse for placing new doctors had stumbled onto essentially the same procedure a decade earlier, without ever proving it worked. Gale and Shapley supplied the proof the practice had been missing.

Why it mattered

The result says a fair, sustainable arrangement isn't merely a hope but a guarantee — reachable by a fast and honest procedure. Decades later the economist Alvin Roth used it to redesign how American doctors are matched to hospital residencies, how children are assigned to public schools in some big cities, and how living kidney donors are paired with patients. Shapley and Roth shared the 2012 Nobel Memorial Prize in Economics for it; Gale had died in 2008 and could not be named.

An everyday picture

Think of musical chairs played politely. Each round you walk toward your favourite chair; whoever is sitting there compares you with themselves, keeps the better fit, and nudges the other to keep looking. Because a chair only ever swaps its sitter for someone it likes more, the game can't loop forever — and when it ends, no two people would both rather trade seats. There's a catch the widget makes visible: whoever does the walking gets the better deal. The proposers end up with their best possible stable partner; the proposed-to, with their worst.

An interactive matching diagram: suitors A–D on the left, reviewers 1–4 on the right. Drag a slider to play the deferred-acceptance rounds; offers appear as lines, the best is held (solid), the rest are rejected (faint dashes), until a stable match with no blocking pair settles. A toggle switches which side proposes and changes the result.

Where it sits

This paper sits in the Library's game-theory line, beside von Neumann and Morgenstern (1944), John Nash (1950), and Kenneth Arrow (1951). Arrow had just proved a famous impossibility — that no voting rule can fairly distil everyone's preferences into a single social ranking. Gale and Shapley answer with a possibility: a different, humbler goal — a stable match — can always be achieved, and built by hand. Read side by side, they trace the boundary between the collective arrangements we can guarantee and the ones we cannot.

The original document
Original source text
D. Gale & L. S. Shapley · The American Mathematical Monthly 69(1): 9–15 · January 1962
The problem
Gale and Shapley open with college admissions: applicants rank colleges, colleges rank applicants, and quotas must be respected. To strip the problem to its core they restate it as the marriage of n men and n women, each holding a strict ranking of everyone on the other side; a set of marriages pairs each person with one partner.
Stability
A set of marriages is called unstable if there are a man and a woman, not married to each other, who each prefer the other to their actual partner — a pair that would break away. A set with no such pair is stable. The whole paper turns on showing such a set can always be found.
Theorem 1
There always exists a stable set of marriages.
The proof is the deferred-acceptance procedure. Each man proposes to his favourite woman; each woman who holds offers keeps, tentatively, the suitor she ranks highest and rejects the others; rejected men propose to their next choice, round after round. Because a woman only ever trades up, the process must stop, and at the end no breakaway pair can remain.
[ … ]
A closing note
any argument which is carried out with sufficient precision is mathematical, and the reason that your friends and ours cannot understand mathematics is not because they have no head for figures, but because they are unable to achieve the degree of concentration required to follow a moderately involved sequence of inferences.
The American Mathematical Monthly · January 1962