The archive.

Why should interchangeable factors guarantee determinate values? In this brief conference abstract, Oskar Morgenstern challenges that familiar economic assumption through game theory. He treats the extra value created by combining factors not as an awkward exception, but as something the characteristic function of an n-person game can describe. His sharper challenge concerns substitution: he points to games involving six players or factors in which substitutability does not permit values to be assigned. The examples are announced rather than demonstrated here, but the distinction is clear: interchangeability need not secure the determinate equilibrium economists seek. This compact statement shows why Morgenstern redirects attention toward the different, more complex game-theoretic notion of a “solution.”

A military program can demand more tanks and ships yet undermine itself by diverting steel from the railroads needed to move them. This concrete tension anchors Oskar Morgenstern’s 1950 article on economic computation: requirements cannot be assessed separately when industries compete for resources and depend on one another’s output. He argues for mathematical analysis while confronting its limits—delayed statistics, information lost through aggregation, and an economy coordinated by decentralized decisions rather than commands. An interindustry calculation traces how an order for chemicals and munitions generates still larger demands elsewhere. The reader discovers both what computation can expose beyond administrative intuition and why proving a program feasible does not establish that it is the best available choice.

Economics had long borrowed its self-image from mechanics, treating prices and quantities as passive magnitudes tending toward equilibrium. Against that inheritance, this programmatic essay introduces the theory of games that Morgenstern developed with John von Neumann, insisting the mathematics is essential rather than decorative. Isolated maximization may suffice for a Robinson Crusoe, or for the limiting cases of monopoly and pure competition; it fails wherever each agent's best move depends on what rivals conceal, threaten, or choose. Two-person zero-sum games yield saddle points and mixed strategies, making bluff and secrecy formal elements of rational conduct; games of three or more introduce coalitions, compensations, and solution sets rather than single equilibria — the natural language of cartels, unions, and bilateral monopoly. Rationality itself, he argues, cannot be defined before the strategic situation has been analyzed.

Jeder einzelne strebt nach seinem maximalen Vorteil, und die Interessen aller oder der meisten stehen miteinander in Widerspruch.

English translation: “Each individual strives for his maximum advantage, and the interests of all, or of most, stand in conflict with one another.”

A textbook can explain established theory clearly while concealing where that theory fails. This tension drives Oskar Morgenstern’s 1950 review of the 1948 German edition of Heinrich von Stackelberg’s Grundlagen der theoretischen Volkswirtschaftslehre. Morgenstern praises its teaching value and treatment of market organization, yet questions what equilibrium means between competition and monopoly and whether indifference curves can accommodate time and expectations. His concern is not simply technical: presenting economics as settled knowledge, he argues, may repel the students most capable of advancing it. This compact review offers a pointed distinction between command of analytical techniques and willingness to expose their limits, ending with a separate, unsettling observation about Stackelberg’s theoretical competence and Nazi commitments.

This work, too, is written as if there were no unsolved problems in economics and as if what is presented were the final answer.

Economic figures acquire authority because they look precise, and that appearance, Morgenstern argues, is often typographical rather than epistemic. His revised methodological study insists economics learn what the natural sciences long ago accepted: every observation carries an error, and no serious use of data can ignore its size. Ranging across foreign trade and gold-movement statistics, business accounting, national income, and growth-rate computation, he shows how paired export and import reports fail to match, how revisions never settle, and how deliberate concealment distinguishes economic data from a nature that does not lie. The demand is not for less quantification but for more honesty about it: reliability estimates, stated error ranges, and a refusal to let refined econometric operations amplify the illusion of knowledge. Data, on his account, become scientific information only when integrated with theory.

We must carefully distinguish between what we think we know and what we really do and can know.

A stable calculation need not describe a stable economy—or rest on accurate observations. In this jointly authored conference-paper abstract, Oskar Morgenstern and Max A. Woodbury examine how errors in input-output coefficients propagate into matrix inverses. Two 18-by-18 matrices, altered in only 42 entries by at most two per cent, yield remarkably similar inverses. Yet the authors question whether this reassuring result reflects economic interdependence or merely the matrices’ proximity to the identity: better data might reveal connections currently recorded as zeros. Woodbury’s mathematical results add bounds on error transmission and an exact inverse-update formula for a change in one row. The abstract offers a compact encounter with the distinction between computational robustness and empirical adequacy in economic modelling.

The stability of inverses must not be interpreted as proving stability of the economy. It is merely an indication how, and to what extent, errors in the given matrix carry over into its inverse.

A military supply operation can succeed without showing whether it could have used fewer resources or responded faster. In this 1955 article, revised and extended from a RAND memorandum, Oskar Morgenstern asks what must be understood before logistical performance can be judged or optimized. His economic and organizational perspective makes a shipment more than a quantity moved: personnel and machinery create continuing demands, complementary goods must arrive together, and requests may conceal inflated expectations. These dependencies make uniform cuts and rigid supply packages unreliable. Readers can discover why sources must exercise judgment rather than merely fill orders, and why military command needs broadly circulating information as well as directed instructions. Morgenstern’s proposed groundwork for theory exposes the practical choices hidden inside apparently straightforward calculations of requirements.

Schumpeter inspired lasting intellectual loyalty without founding a distinct school of economics. In this 1951 obituary, Oskar Morgenstern explores that contrast as both a former student and a critical fellow economist. He recalls the revelatory force of Schumpeter’s early theory while questioning the empirical conclusions of Business Cycles and the kind of capitalism whose demise Schumpeter predicted. Admiration sharpens rather than suspends his judgment: an economist receptive to new scientific methods remained attached to a vanished social world. The result is a compact portrait that helps readers distinguish Schumpeter’s theoretical achievements from his historical expectations—and understand how his generosity toward younger scholars fostered independent work rather than doctrinal allegiance.

A firm, an army, a university, a logistics system: none can be understood as a mere sum of its inputs, because their capacities depend on arrangements of authority, information, timing, and control. Morgenstern's unfinished RAND memorandum clears the ground for a future quantitative theory of organization, arguing that economics has treated the firm as a black box and lacked even a descriptive language for what happens inside it. He builds one, distinguishing inner from outer activity, defining operations as chosen sequences of acts, and replacing organic metaphors with "competences," units delegated authority to set variables within domains. Signals activate them, memory sustains them, control counters the drift toward disorder. Written in 1951, the memorandum anticipates cybernetics, information theory, operations research, and bounded rationality, insisting that careful phenomenological description must precede any premature axiomatization.

Organization in itself, wherever it occurs, whether in social or natural life, is something exceptional and extraordinary.

Faster calculation does not make an economic model true. In this 1952 article, Oskar Morgenstern asks what electronic computers can genuinely add to econometrics when observations are unreliable and the equations themselves may misrepresent economic life. His distinctive proposal is to treat computation not merely as the execution of established theory, but as a form of experiment: models can be tested against later production figures, temporal relationships explored across many series, and strategic choices calculated for competing firms. He couples this exploratory ambition with exacting demands for meaningful solutions and attention to propagated error. The result offers a concrete encounter with the promise—and limits—of computational economics before larger calculations could be mistaken for better knowledge.

Welche Gleichungen die Wirklichkeit tatsächlich löst, kann heute niemand angeben.

English translation: “No one today can specify which equations reality actually solves.”

Even a fully developed economic theory, Morgenstern contends, would leave many policy problems unsolvable — because policy adds demands theory never faces: aims stated quantitatively, timing, tolerated side-effects, feasible computation, and, decisively, a list of admissible means. A problem impossible under one set of permitted operations may become trivial under another, as squaring the circle, Columbus's egg, and the Gordian knot each illustrate. Unemployment could be 'solved' by conscription, work camps, or forced relocation; ruling those out changes the problem itself. Invoking Gödel as a reminder that decidability runs deep, he separates the theoretical solution from the policy solution and warns that civilization, by narrowing the morally acceptable instruments of action, tends to make its economic problems harder even as economic knowledge advances.

The statement that a problem has no solution for a given set of means is exactly equivalent to stating that a contradiction prevails.

Computers can solve economic models faster—but can they also disclose relations that existing theory has not taught economists to seek? In this 1954 essay, Oskar Morgenstern treats computation as an experimental instrument, not merely a shortcut to familiar answers. He exposes the mismatch between equilibrium theory’s prodigiously calculating individuals and economists’ difficulty in computing their own models, then explores how business experiments and repeated numerical trials might give theory firmer empirical footing. His most speculative proposal is also carefully qualified: large-scale analysis of economic time series might uncover unexpected patterns, but numerical output does not automatically constitute meaningful data. The essay offers a concrete encounter with computational ambition disciplined by questions of measurement error, interpretation, and what an economic model must demonstrate before its results deserve confidence.