Nash Equilibrium Explained: The Logic of

There's a scene that game theorists love to hate in A Beautiful Mind. Russell Crowe's John Nash has his famous bar epiphany—staring at a blonde woman surrounded by friends, he declares that if everyone goes for the blonde, they'll block each other, so they should all pursue the friends instead. The crowd goes: genius. The actual game theorists go: that's... not Nash equilibrium. That's not even close.

MIT economist Ian Ball opens his Lecture 5 on Nash equilibrium with a gentle version of this observation—"their story is kind of wrong in the movie, it doesn't make mathematical sense"—and then spends the better part of 79 minutes explaining what the concept actually says. It's worth paying attention, because the real version is both more interesting and more useful than the Hollywood one.

Rationality Isn't the Problem. Correctness Is.

Before Ball gets to Nash equilibrium proper, he takes a detour through rationalizability—a concept that sounds niche but turns out to be the key to understanding why Nash equilibrium matters at all.

The setup: in any strategic game, we can make increasingly demanding assumptions about how players behave and get increasingly precise predictions in return. Start with nothing—assume nothing about players—and anything is possible. Assume everyone is rational (meaning they're each playing a best response to some belief about others), and you can rule some things out. Assume everyone knows everyone else is rational, and you can rule out more. Keep stacking these "knowledge of rationality" layers indefinitely, and you arrive at the set of rationalizable strategies—everything consistent with rationality and arbitrarily deep mutual knowledge of it.

Here's the problem Ball puts his finger on: even with all that, the prediction can still be lousy.

He uses a coordination game he calls the "Boston game"—two friends separately deciding whether to attend a Celtics game or a Red Sox game. They'd both rather be together than apart, but one is a Celtics fan and one prefers the Red Sox. Under rationalizability, every possible outcome qualifies. Including the one where they split: one goes to the Celtics game, one goes to the Red Sox, and they both sit alone, miserable, getting nothing.

That outcome is technically rationalizable. Player one can explain their choice—"I thought my friend was going to the Red Sox game"—and back it up with a chain of nested beliefs that never hits a logical contradiction. But something is clearly off. Ball names the problem directly: "Rationalizability means you have an explanation, you're behaving optimally given this explanation or given these beliefs, but there's no requirement that the explanation is correct."

That's the gap. Rationalizability demands a coherent story. Nash equilibrium demands that the story also be true.

What Nash Actually Requires

A Nash equilibrium is a strategy profile—a complete specification of what every player does—where no single player can improve their outcome by changing their strategy alone, given what everyone else is doing.

The "given what everyone else is doing" part is load-bearing. Ball is emphatic about this: "The most important idea that Nash introduced was that when I'm reasoning about other people, I should take their behavior and their strategies as fixed because I can't directly control what they do." You can only pull your own lever. Everyone else's choices are, from your perspective, fixed inputs.

What this demands from beliefs is striking. In a Nash equilibrium, your belief about what other players will do has to match what they actually do. The equilibrium closes the loop that rationalizability leaves open. You're not just behaving optimally given your theory of the world—you're behaving optimally given a theory that happens to be correct.

This is why Ball calls it a matter of consistent beliefs. In a Nash equilibrium, beliefs and reality are the same thing, so the distinction collapses. If you believe your friend is going to the Red Sox game, they are going to the Red Sox game. If you believe they're going to the Celtics game, same deal. The only equilibria in the Boston game are the two coordinated outcomes: both Celtics, or both Red Sox. The miserable split—alone at different stadiums—doesn't survive, because each stranded person would rather switch and join their friend. They have a profitable unilateral deviation. That kills it as a Nash equilibrium.

The Underline Algorithm and What It Reveals

For anyone who's tried to calculate Nash equilibria in matrix games and gotten lost, Ball offers a simple mechanical procedure: go through the payoff matrix and underline each player's best responses. Any cell where both players' payoffs are underlined is a Nash equilibrium.

The logic is transparent once you see it. If player one's payoff is underlined in a given cell, player one is playing a best response to whatever player two is doing. If player two's is underlined too, the reverse holds. A cell where both are underlined is exactly a strategy profile where both players are simultaneously best-responding to each other—the definition, rendered visually.

It's a small thing, but I find the elegance here genuinely satisfying. The algorithm isn't a shortcut that obscures the math; it's just the definition expressed with a marker instead of notation.

The Honest Limitations

Ball is careful not to oversell the concept, and that carefulness is what makes the lecture worth engaging with beyond a standard textbook treatment.

Nash equilibrium doesn't tell you which equilibrium will be reached when multiple exist. In the Boston game, there are two pure strategy equilibria—both go Celtics, or both go Red Sox—plus a third involving mixed strategies (randomizing between the two options). The theory establishes that these are the stable points. It doesn't tell you how two friends, without communication, converge on one of them. That's a separate question about coordination, social norms, and what economists call "focal points"—a Schelling-flavored problem that Nash equilibrium alone can't resolve.

There's also the question of whether real people actually play Nash equilibria. The concept is a prediction—a statement about where rational, correctly-informed players would end up. Behavioral economics has spent decades documenting the many ways humans deviate from this benchmark. Nash equilibrium isn't a description of human behavior; it's a description of behavior under a specific set of idealized conditions. Whether those conditions approximate reality well or poorly depends on the context.

And then there's the subtler issue Ball raises but doesn't dwell on: Nash equilibrium is a stability concept, not a welfare concept. A Nash equilibrium can be collectively terrible. The canonical example—the Prisoner's Dilemma, where both players defecting is the unique Nash equilibrium even though mutual cooperation would make both better off—is left offscreen here but haunts the whole discussion. Stability and optimality are different things. Equilibria can be stable traps.

Why the Two-Paragraph Nobel Prize Matters

Ball mentions, almost as an aside, that Nash's original paper introducing this concept was roughly two paragraphs long. "If you have good ideas, you don't need to go on and on and on."

Nash published his foundational results on equilibrium in 1950 and 1951—he was in his early twenties—and received the Nobel Prize in Economics in 1994, more than forty years later. The concept spread slowly at first, then became the central organizing idea of modern economic theory, political science, evolutionary biology, and computer science. Arms races. Auctions. Evolutionary stable strategies in animal behavior. Mechanism design for kidney exchanges. The same core idea—mutual best response under correct beliefs—shows up everywhere rational actors interact strategically.

What's interesting about Ball's lecture is how much work it takes to properly distinguish Nash equilibrium from weaker notions of rationality. The payoff of that work is a sharper concept. Rationalizability tells you a player has a story. Nash equilibrium tells you the story is real.

Whether that additional precision is always worth the additional assumptions—correct beliefs, specifically—is a question the field keeps wrestling with. But it's the right question to be asking.

Ellis Redmond is the Personal Development & Productivity Correspondent at Buzzrag.