Whoever asks first wins

A hundred people looking for work and a hundred jobs, or students and schools, or anyone and anyone. Everyone has their own private order of preference. One side does the asking. It turns out to matter a great deal which.

The rules, all of them. Everyone ranks everyone on the other side, at random. An asker with no partner asks the best choice they have not yet tried. Someone who is asked says "maybe" to the best offer so far and turns the rest away for good. When every asker is held by someone, the maybes become yeses. This is the method David Gale and Lloyd Shapley published in 1962.

The result is always stable: no two people would both rather have each other than the partners they got. But among all stable results it is the best possible one for every asker and the worst possible one for everyone asked (checked here against every possible pairing for groups of six). With 100 a side, askers end up with their 5th choice on average; the others with their 20th. Press "Swap who asks" and the same lists give the other side the advantage. What it leaves out: real preferences are not random, and people agree far more about who is desirable, which shrinks the gap.