Game theory

Managers sometimes predict rivals as if they were passive. A game model forces the team to ask how another party’s incentives change the outcome. It also helps distinguish a credible commitment from a public promise and explains why a jointly better outcome may not be stable without governance or repeated interaction.

MIT’s solution concepts distinguish a best response to a specified rival action from a dominant action that wins against every rival action. A Nash equilibrium is mutual best response, so a player cannot improve by changing alone while the others’ strategies stay fixed. That condition need not maximize joint payoff, select a unique outcome or predict how quickly real firms reach it.

Write down a reversal test before acting. If discounting is preferred only when the rival holds, a matching response can remove its attractiveness. If the rival’s capacity, contract or information changes, the feasible actions or payoff ranking can change too. Evidence about those constraints is more useful than a precise matrix filled with guessed profit numbers.

Models mislead when payoffs are guessed, players are omitted, or a simultaneous matrix is used for a sequential decision. Real organizations may not maximize a single measured payoff; they may have multiple decision-makers, incomplete information, and internal constraints.

A matrix can omit customers, capacity limits or decision timing that changes the game. Public announcements do not necessarily make a future action credible, and a competitor’s observed response does not reveal all its objectives. Treat the equilibrium as conditional on an explicit model and compare it with observed actions before using it for an independent business choice.