Harsanyi, Nash & Selten (1994): The Mathematics of Strategic Thinking

The Science of Strategy

Before game theory, economics had a blind spot. It could analyze a single consumer choosing among goods, or a single firm choosing output levels. But it couldn’t rigorously analyze situations where the outcome depends on what everyone else does — where your best move depends on my best move, which depends on your best move, in an infinite loop of strategic reasoning.

Three scholars broke this loop. John Nash defined what it means for a strategic interaction to reach equilibrium. Reinhard Selten refined this concept for games that unfold over time. John Harsanyi extended it to situations where players don’t know each other’s motivations. Together, they gave economics — and much of social science — a rigorous framework for analyzing conflict, cooperation, and competition.

Nash: The Equilibrium That Changed Everything

In 1950, a 21-year-old Princeton mathematics student named John Nash wrote a 27-page doctoral dissertation that would become one of the most cited works in all of social science.

The problem: in a game with multiple players, each choosing a strategy, what outcome should we predict? Von Neumann and Morgenstern had solved two-person zero-sum games (where one player’s gain is the other’s loss). But most real situations aren’t zero-sum — trade, negotiation, arms races, market competition all involve mixtures of conflict and cooperation.

Nash’s solution — the Nash equilibrium: a set of strategies, one for each player, where no player can improve their outcome by unilaterally changing their strategy, given what everyone else is doing.

Key properties:

• Self-enforcing: No one has an incentive to deviate. If everyone expects the Nash equilibrium to be played, everyone’s best response is to play their part of it
• Existence: Nash proved that every finite game has at least one Nash equilibrium (possibly in mixed strategies — randomizing between options). This was a stunning mathematical result, using Kakutani’s fixed-point theorem
• Generality: Unlike von Neumann’s solution, Nash equilibrium applies to any number of players, any payoff structure, and any combination of conflict and cooperation

The concept is everywhere:

• Prisoner’s dilemma: Two suspects each have an incentive to confess, even though both staying silent would be better for both. The Nash equilibrium (both confess) is individually rational but collectively disastrous
• Market competition: Firms choosing prices or quantities reach a Nash equilibrium where no firm can profit by changing its strategy alone — the Cournot and Bertrand models are Nash equilibria
• Arms races: Nations build weapons because the other side is building weapons — a Nash equilibrium that leaves everyone worse off but that no one can unilaterally escape
• Traffic: Drivers choosing routes reach a Nash equilibrium where no one can shorten their commute by switching roads — even though the overall traffic pattern may be inefficient

Selten: Credible Threats and Subgame Perfection

Nash equilibrium has a problem: some equilibria rely on threats that no rational player would actually carry out. A monopolist might threaten to start a price war if a competitor enters the market. If the threat is believed, the competitor stays out — and this is a Nash equilibrium. But would the monopolist really destroy their own profits just to punish an entrant? If not, the threat isn’t credible, and the equilibrium is unreasonable.

Reinhard Selten solved this with subgame perfect equilibrium (1965): a strategy must be optimal not just at the start of the game, but at every point within the game — including points that are reached only if someone deviates from the expected path.

This eliminates incredible threats:

• The chain store paradox: A chain store facing sequential entry in 20 markets might threaten to fight every entrant. But in the last market, fighting is irrational (the game is over). Knowing this, the entrant in market 19 enters. By backward induction, the threat unravels — the chain store accommodates in every market
• Bargaining: Selten’s refinement predicts specific outcomes in bargaining games — the first mover advantage, the effect of patience, and why delays in negotiation are costly
• Commitment: The theory explains why players sometimes benefit from limiting their own options — burning bridges, making public promises, or signing binding contracts — to make their threats or promises credible

Harsanyi: Games of Incomplete Information

Real strategic interactions involve uncertainty — you don’t know your opponent’s costs, preferences, or information. A firm doesn’t know its competitor’s production costs. A buyer doesn’t know a seller’s reservation price. A country doesn’t know whether its adversary is bluffing.

Before Harsanyi, game theory couldn’t handle this. John Harsanyi’s breakthrough (1967-68) was the concept of Bayesian games: games where players have private information about their own “type” (preferences, costs, capabilities), and each player holds probabilistic beliefs about others’ types.

The key innovation:

• Nature moves first: Harsanyi introduced a fictional player — “Nature” — who randomly assigns types to players at the start of the game. Each player knows their own type but not others’
• Bayesian Nash equilibrium: Players choose strategies that maximize their expected payoff given their beliefs about others’ types. Beliefs are updated using Bayes’ rule as the game unfolds
• Unifying framework: This transformed incomplete information from an intractable problem into a standard game — just with more players and more strategies. Every game of incomplete information can be analyzed using the same tools as complete information games

Applications transformed multiple fields: auction design, mechanism design, signaling in labor markets, insurance markets with adverse selection, and international diplomacy.

Their 1994 Nobel Prize was awarded “for their pioneering analysis of equilibria in the theory of non-cooperative games.”

Explain It to a Child

Imagine you and your friend are both choosing what to wear to school — but you can’t call each other. You both want to match. If you both pick blue, great. If you both pick red, great. But if one picks blue and the other picks red, you’re mismatched. A Nash equilibrium is when you both pick the same color — neither of you would want to switch once you see what the other chose. Now Selten added: what if someone threatens “I’ll wear green and ruin everything if you don’t pick blue”? Would they really do that? If not, it’s an empty threat and we should ignore it. And Harsanyi asked: what if you don’t even know your friend’s favorite color? He figured out how to play the game smartly even when you’re not sure what the other person wants.