Game Theory · GT-08
In a strict two-player, zero-sum world, minimizing your maximum possible loss is the same as maximizing your guaranteed minimum gain.
Proves that in any finite two-player zero-sum game, there exists a value V and a strategy for each player such that Player 1 can guarantee at least V and Player 2 can guarantee at least -V — the game has a determinate, calculable value.
Proved by John von Neumann in 1928 ('Zur Theorie der Gesellschaftsspiele'), predating and later folding into the broader Nash equilibrium framework as its zero-sum special case.
The Mechanism
Guaranteed value — the floor a rational player can always secure
The two bounds meet exactly at V — this is the theorem's core claim: what the maximizing player can guarantee from below equals what the minimizing player can guarantee from above. No gap remains for either side to exploit through cleverness alone.
01 · A GUARANTEE, NOT A HOPE
The strategy works even against a perfect opponent
The minimax strategy doesn't assume the opponent will make a mistake — it's calculated to secure at least V even against the best possible counter-play. This is what makes it fundamentally different from a strategy that merely performs well 'on average' against typical opponents.
02 · ZERO-SUM IS THE LOAD-BEARING ASSUMPTION
The theorem doesn't generalize past it
The clean equivalence between minimizing max loss and maximizing min gain relies entirely on the fact that one player's gain is exactly the other's loss. In non-zero-sum games (most real strategic situations), there's no single 'value' of the game in this sense — this is precisely why Nash's later, broader equilibrium concept was needed.
03 · IT PREDATES AND MOTIVATED THE ENTIRE FIELD
Von Neumann's 1928 result came before game theory had a name
The minimax theorem is often considered the founding result of game theory, developed 16 years before von Neumann and Morgenstern's 1944 book and 22 years before Nash's general equilibrium concept — it solved the zero-sum case completely before anyone had tools for the broader problem.
Where It Fails / Inversion
Where it fails / inversion
Applying minimax reasoning to a genuinely non-zero-sum situation systematically produces overly defensive, needlessly adversarial strategies — treating a negotiation or partnership as if any gain to the other side is a loss to you (when it may not be) forecloses cooperative moves that could make both sides better off. The theorem's power is inseparable from its zero-sum scope.
How To Use It
Worked example · security and adversarial planning
Minimax logic is the right tool specifically when facing a genuinely adversarial, zero-sum-like opponent — cybersecurity defense against an attacker, for instance, where the attacker's gain (a successful breach) is essentially the defender's loss. Building defenses around 'what is my guaranteed floor even against the smartest attacker' is minimax reasoning applied correctly, precisely because the interests are truly opposed.
How to use it
Before applying worst-case defensive thinking to a strategic problem, check whether it's actually zero-sum. If the other side's gain isn't strictly your loss, minimax reasoning will make you needlessly guarded — reach for Nash equilibrium reasoning (or cooperative/positive-sum framing) instead.
See Also