Oskar Morgenstern’s conference proceedings contribution, originally published in 1967 and republished in 2012, presents game theory as a conceptual reconstruction of social science. Its eight sections move from concept formation and rational behavior through probabilistic strategies, rules, and solution concepts to applications and an experimental research agenda. The central claim is that social interaction requires mathematics suited to purposeful agents who partly control—and deliberately influence—one another’s opportunities. Neither imitation of physics nor conventional economic maximization adequately captures this structure.
Morgenstern begins by making conceptual precision, rather than the accumulation of observations, the condition of scientific advance:
Only proper, sharply defined concepts can give exactness; data, no matter how good, never can do this.
The distinctive problem is antagonism combined with possible cooperation. Unlike physical particles, people pursue advantages, form coalitions, and dispute the division of their joint gains. Mathematics must therefore express strategic interdependence. Morgenstern presents the minimax theorem and the subsequent development of game theory as a break with approaches that treat economic actors as isolated maximizers confronting fixed conditions.
This shift changes what rationality means. An individual can maximize directly when all relevant conditions are given, even if some are given probabilistically. In social situations, however, other people control consequential variables in pursuit of their own interests. Optimal conduct cannot be defined independently of that interaction:
Game Theory does not assume rational behavior but wishes to explain it, to give it precise mathematical meaning.
The theory assumes the pursuit of advantage within the rules, but seeks to establish what conduct actually serves that pursuit. It is consequently both descriptive and normative: it identifies the structure of social situations while advising participants how to act. Morgenstern’s criticism of conventional economics is correspondingly fundamental. Treating interaction as a collection of ordinary maximum problems mistakes the logical character of the object being studied.
The discussion of mixed strategies makes this reconstruction concrete. In two-person zero-sum games without a saddle point in pure strategies, minimax establishes optimal probabilistic mixtures. Randomization does not mean treating an intelligent opponent as a natural event with statistically predictable behavior. Rather, the player deliberately generates uncertainty to prevent exploitation:
The safest way not to pass on information is not to have any.
Matching pennies illustrates why a chance device can implement rational strategic conduct. The appropriate probabilities depend on the payoffs, not on a general preference for randomness. Morgenstern also distinguishes game theory from linear programming: their mathematical duality does not erase the difference between strategic interaction and resource allocation under unified control.
Randomization raises empirical and psychological questions. People desire definite courses of action, yet secrecy, surprise, and delayed decisions suggest practical recognition of strategic uncertainty. Whether actual decision-makers employ suitable mixtures remains a research problem. Morgenstern also recognizes that dissemination of the theory could transform its subject: widespread adoption might change markets and society, without necessarily producing greater stability or justice.
The treatment of rules qualifies the analogy between parlor games and social life. Rules are physical, legal, and moral constraints, not simply customary behavior. Knowledge of them may be unequal, while legal and moral restrictions can be violated. Such violations introduce uncertainty beyond ordinary randomness. Morgenstern identifies this as an insufficiently explored problem, while defending initial abstraction as necessary for constructing a tractable theory.
His account of many-person games then replaces the expectation of a unique answer with a structured multiplicity of acceptable distributions. Coalitions make some divisions of proceeds enforceable; relations of domination organize these “imputations” into solution sets:
The solution set represents a "standard of behavior."
A solution is thus not necessarily one allocation. Its alternatives can sustain an accepted outcome through their availability as counterpossibilities. Indeterminacy, on this account, expresses a property of social relations rather than a defect in analysis. Morgenstern acknowledges computational difficulties and the absence, at the time of writing, of a general existence proof for the solution concept he favors. Increasing the number of players can also introduce new structures, making aggregation from firms to economies a substantive problem rather than an innocent change of scale.
The applications section illustrates the reach of this approach through bargaining, discrimination, coalition power, bluffing, and voting. Symmetric positions need not yield symmetric solutions; substitutable coalition members need not receive equal valuations; formal voting weight can differ from effective influence. These results challenge intuitive equations between equality of conditions, justice, and outcomes. Morgenstern presents them as mathematical discoveries, while acknowledging that practical applications still concern relatively simple cases. His confidence in game theory’s scientific necessity is stronger than his claims about its immediate operational reach.
The concluding outlook turns toward experiments on utility, coalition formation, learning, trust, and betrayal. Experiments can test, correct, and extend theory, but cannot prove it. The major unfinished task is the “integral game”: individuals participate simultaneously in several interdependent games and allocate resources among their strategies. The essay’s enduring methodological relevance lies in this conjunction of ambition and incompleteness. Game theory promises precise accounts of previously obscure relationships, but connecting mathematical structures with social evidence remains a difficult research undertaking, not something achieved merely by adopting the vocabulary of strategy.
This work was divided into 8 sections when it entered the library's research corpus—an apparatus for search and citation, not necessarily the author's own table of contents. Each title opens its summary.
Put a question to this work; the Librarian answers from its 8 sections and cites the passage.
Ask the Librarian