Game Theory · GT-01
A stable standoff, not a good outcome — and almost never the only one.
A state where no player can improve their own outcome by unilaterally changing strategy, given what everyone else is doing. It says nothing about whether the outcome is good, fair, or efficient — only that nobody, acting alone, can do better.
Formalized by John Nash in his 1950 paper "Equilibrium Points in N-Person Games" (Proceedings of the National Academy of Sciences), work that earned him a share of the 1994 Nobel Memorial Prize in Economic Sciences alongside John Harsanyi and Reinhard Selten.
The Mechanism
Click a cell — see why only the diagonal holds
Both driving-left and both driving-right are Nash equilibria — once everyone has converged on one convention, no single driver benefits from switching alone. The mismatched cells aren't equilibria: whoever is out of step has an immediate, unilateral incentive to conform.
01 · BEST RESPONSE
Not the best outcome — the best response
Each player asks only one question: "given what the other player is doing, can I do better by switching?" An equilibrium is the point where every player's answer is no, simultaneously. It's a fixed point of mutual best responses, not a search for the best possible joint outcome.
02 · MULTIPLICITY IS NORMAL
Most games have more than one equilibrium
The driving-side game above has two: both-left and both-right. Nash's theorem guarantees at least one equilibrium exists in any finite game — it does not promise there's only one, and it says nothing about which one a real population will land on.
03 · EXISTENCE, NOT NICENESS
Nash proved existence — not quality
Nash's actual theorem is narrower than people assume: every finite game has at least one equilibrium (allowing for randomized, "mixed" strategies if needed). It is silent on whether that equilibrium is efficient, fair, or even remotely desirable — see Prisoner's Dilemma for the sharpest case where it's neither.
Where It Fails / Inversion
The trap: mistaking "stable" for "good"
A Nash equilibrium can be a genuinely bad outcome for everyone involved — stability is not the same claim as desirability. The Prisoner's Dilemma is the canonical case: mutual defection is the unique Nash equilibrium, yet it is worse for both players than mutual cooperation. Equilibrium tells you where a rational, self-interested population gets stuck — not where it should be. The second trap is the multiplicity problem itself: when a game has several equilibria (as above), the concept alone cannot tell you which one will actually emerge. That requires something outside the payoff matrix — a focal point, a historical convention, a regulator's decree (see Schelling Points).
How To Use It
Worked example · Sweden's "Dagen H," September 3, 1967
Sweden drove on the left for centuries, like Britain. But every neighboring country — Norway, Finland, Denmark — drove on the right, and most Swedish-owned cars were already left-hand-drive imports built for right-side traffic. Two Nash equilibria existed: universal left, or universal right. Sweden was stuck in the worse one purely by historical accident, not because left-hand driving was actually better for Sweden's circumstances.
No individual driver could unilaterally switch sides without causing a collision — exactly the "no incentive to deviate alone" signature of equilibrium. The only way out was a coordinated, government-mandated switch of the entire population at a single instant (4:50 a.m. that Sunday, all traffic stopped, switched sides, and resumed). That's the practical lesson: if you're stuck in a bad-but-stable equilibrium, look for a mechanism that can move everyone at once — individual defection alone won't work, and usually makes things briefly worse.
How to use it
When you're stuck in an arrangement that feels suboptimal but nobody individually wants to be the first to change (a shared tool everyone tolerates, a norm nobody likes but nobody breaks alone), check whether you're actually in a stable equilibrium. If so, don't waste energy asking any single party to defect unilaterally — design a coordinated switch instead, the way Dagen H did.
See Also