Game theory turns confrontation into a payoff matrix
Game theory models multi-agent strategy as games among players who choose actions to raise payoffs. It spans biology, computing, economics, law, and philosophy, and has earned many economics Nobels since John Nash, John Harsanyi, and Reinhard Selten shared the 1994 prize. The 1944 von Neumann–Morgenstern book launched the modern field.
A mathematical game abstracts players’ available actions, the payoffs for strategy combinations, and the information held when choosing—Rasmusen’s PAPI essentials. Early economic theory assumed perfect rationality, self-interest, complete information, and a fixed game. Theory of Games and Economic Behavior (1944) by John von Neumann and Oskar Morgenstern built on von Neumann’s fixed-point methods that later became standard. Later Nobels went to Schelling and Aumann (2005), Hurwicz, Maskin, and Myerson (2007), Roth and Shapley (2012), Tirole (2014), and Milgrom and Wilson (2020).
Cooperative games allow externally enforced binding deals; non-cooperative ones demand self-enforcing agreements. Symmetric games look the same to every player; zero-sum contests redistribute a fixed pie. Simultaneous moves hide others’ choices; sequential moves reveal history. Complete information means everyone knows rules and payoffs; Bayesian games encode incomplete information. Perfect information further requires seeing the whole past—chess and Go qualify, poker and bridge do not. Repeated play enables reputation and tit-for-tat punishment after defection.
Standard models treat players as rational maximisers of expected utility with common knowledge of that rationality. Utility in the von Neumann–Morgenstern sense ranks lotteries linearly in probabilities. Decision theory handles single-agent preference; mechanism design flips the script by designing rules rather than taking them as given—auctions and incentive schemes included. General-equilibrium work with huge numbers of agents may still borrow game-theoretic tools, including dynamic stochastic general equilibrium models for policy questions.
Applications now range from international relations and voting systems to network science, linguistics, military technology, and ethical theory. Simulation and agent-based models use reward–punishment, resource contests, and coalitions. Subfields include algorithmic, behavioral, combinatorial, evolutionary, and quantum game theory. The method’s edge over static optimisation is explicit strategic interdependence among agents.
Source: Game theory
