Two groups bargain over the same thing, across the line between them. Same strategies, same payoffs, same learning, no difference in skill or preference or patience. The only difference is how many of them there are, and the smaller group ends up asking for less.
Fair Shares asks what a single population converges on when it divides something and every split that exhausts the pie is an equilibrium. This is the same game with a visible group label attached to everybody, and bargaining that happens between the groups rather than inside them.
The mechanism is arithmetic, and once seen it is hard to unsee. If a group is a fifth of the population, each of its members bargains with the other group four times as often as each member of the other group bargains with them. More encounters means faster learning.
And in a bargaining problem, unlike almost anywhere else, adapting faster is a disadvantage. Whoever adjusts first is whoever accommodates. Bruner (2017) named the effect after the Red King, the counterpart to the Red Queen: here it pays to evolve slowly.
This is the most uncomfortable page on the site, because it offers an account of where discrimination could come from that requires nobody to be prejudiced, and no group to differ in any respect that anybody would defend.
Persistent inequity between groups is usually explained by one of two things: a real difference between them, or a preference on somebody's part for one over the other. The first is empirically contestable and the second is morally straightforward to condemn. Both put the explanation somewhere identifiable.
O'Connor's project in The Origins of Unfairness is to show how far you can get with neither. Take a social category that carries no information — one that nobody chose and that predicts nothing about ability — and let conventions of resource division evolve on top of it. Inequitable conventions arrive on their own.
In biology the Red Queen describes an arms race where you must keep evolving to stay level. Bruner's observation is that bargaining inverts it. If two parties are converging on a division and one adjusts faster, the fast one ends up on the accommodating side, because by the time the slow one has moved at all the fast one has already settled into something that works against what it faced.
A group that is a twentieth of the population deals with outsiders nineteen times as often per person. That is the whole of it.
Group size sets the adaptation rate directly, with no further assumptions. Being small means dealing with the other group constantly; being large means dealing with them rarely. So the small group learns the terms of the encounter first, and learning first is losing.
Move the bargaining inside the groups and the effect largely goes. That locates the mechanism precisely: it is not smallness, and it is not visibility. It is the rate at which you are obliged to negotiate with people outside your group. Which suggests, uncomfortably, that the thing driving the disadvantage is integration rather than separation — and O'Connor is careful about what does and does not follow from that, because the obvious policy reading is one nobody should want.
Not much on its own, and the literature is arguing about exactly this. A recent line of criticism holds that none of the standard defences of these models — structural similarity to the real case, empirical confirmation, or how-possibly explanation — actually establishes what they are used to establish. On that reading, the model shows something is possible and is routinely cited as though it showed something is happening.
The narrow claim survives that criticism and is still worth having: a difference in group size is sufficient to produce inequity, so observing inequity does not license inferring a difference in ability or an act of prejudice. That is a claim about what evidence supports what, and it does not need the model to be a picture of any actual society.
Two groups, one resource, no history, no institutions, no power differences other than the one the model generates, and no way for anybody to notice what is happening and object. Real discrimination involves law, wealth, violence and belief, none of which appear here. Nothing on this page should be read as an account of any particular inequity, and O'Connor is explicit about that too.