“The time was ripe”

In December 1948, in one of the hotel rooms at the annual meeting of the American Economic Association, seven scholars with various backgrounds but similar interests conceived the idea of a conference: Tjalling Koopmans, Harold Kuhn, George Dantzig, Albert Tucker, Oskar Morgenstern, and Wassily Leontief (see Kuhn Reference Kuhn2008). One of them, Tjalling Koopmans, who in the summer of 1948 had become the Research Director at the Cowles Commission for Research in Economics, took the initiative. He did so in the context of a research contract with the RAND Corporation that he would sign in January 1949 titled “Theory of Resource Allocation.” The central idea of the conference was to develop the theory of linear programming, one of the successful methods of planning that had been developed in wartime research, and extend it to a more general economic theory of production. The contract called for Koopmans to bring together a small group of individuals who had been working on these kinds of projects for a conference on Activity Analysis of Production and Allocation. He scheduled it for June 20–24, 1949.

That conference defined, more than any other single event, the emergence of a new kind of economic theory growing from game theory, operations research, and linear programming and the related mathematical techniques of convex sets, separating hyperplanes, and fixed-point theory.Footnote ¹ The conference was the “coming out party” of the community that would transform the practices of academic economists for decades to come. It established the historical conditions for economics to become a modeling science. What later would be a commonplace for economists was discussed in nuce, in an early state when the boundaries between the theories were not yet drawn, their doxic elements not yet settled, and most importantly, their disciplinary commitments not yet formed.

Many of the young scholars who participated in the conference would later recall how the conference transformed their intellectual lives and heralded a new era. The 1972 Nobel Prize winner, Kenneth Arrow, for example, would call the conference a “key step in unifying and diffusing the developments in linear programming and relating them to the theory of general equilibrium” – the theory that would shape economic modeling for several generations of economists to come. He continued:

For Arrow, activity analysis framed the theory of production as it would later enter into general equilibrium analysis. In the aftermath of the conference he, simultaneously with Gerard Debreu and Lionel McKenzie, would begin working on an existence proof in general equilibrium theory that would be the “door that opens into the house of analysis” (Mas-Colell Reference Mas-Colell, Whinston and Green1995, 584). Indeed, nearly all of the ingredients of the future existence proof in general equilibrium theory were on the conference table, though it was too soon to put the separate pieces together. On the very first page of the conference volume's introduction Koopmans mentioned the early work on existence of equilibrium in Vienna by Karl Schlesinger and Abraham Wald in Menger's Mathematical Colloquium (Weintraub Reference Weintraub1983), and then referred to John von Neumann's seminal paper on dynamic equilibrium (Reference von Neumann and Menger1936). Koopmans noted that “even among mathematical economists their value seems to have been insufficiently realized” (Koopmans Reference Koopmans1951a, 2). Harvard's Robert Dorfman would later recall that:

While the conference did not address the integrative character of a general equilibrium model, it nevertheless created the conditions by which this integration would take place. Specifically the extended Cowles community recognized their shared belief in the importance of mathematical rigor (creating distance from Hicks’ Value and Capital [Reference Hicks1939]), and their need to de-politicize economic theory (creating distance from Lange's market socialism). What was crucial for the transfer of ideas occurring at the conference was a commitment to method and this weakened the commitments to beliefs that are formative for disciplines. As Chandler has argued: “One might even say that the stronger the emphasis on method as such, the further we are from doxa, from belief or opinion” (Reference ChandlerChandler 2009, 733).

Indeed, the very same conference would later be called the 0^th conference of a field (and society) unrelated to economics: mathematical programming. George Dantzig, who himself presented four papers, recalled:

The conference was the origin of what would within a decade provide closure to the discipline of economics. At the same time it created a field in applied mathematics in which hardly any economist would become active. When Koopmans received the Nobel Prize in 1975 he hesitated in accepting it as he thought that Dantzig and he should share it equally. Dantzig took linear programming into non-economics disciplinary waters using it to construct a new theory in organizational and management science that would come to be taught in engineering and business departments that were about to separate themselves from economics departments during those postwar years.

Those who came to the economics profession after the conference joined a community that had been freed from prewar debates about the right way to “do” economic analysis. They could simply assume, as their predecessors could not, the values of rigor shared at the conference. Gerard Debreu, who came to the US in the fall of 1949, recalled the shared enthusiasm for convexity analysis (in contrast to differential calculus, the mathematics previously used by economists) that the conference had introduced and that he would soon employ when using mathematical proof techniques in general equilibrium analysis.

Cowles liberated many mathematicians interested in the social sciences from the awkward feeling that they had to justify the use of mathematics in economics – instead, its importance could be taken for granted. “At Cowles I came to think, very quickly, that full understanding of a problem required no compromise whatsoever with rigor” (Debreu, quoted in Weintraub Reference Weintraub2002, 153).

As important as the conference appears in retrospect, it was not intended or planned as a path-breaking event, nor could any of the participants recognize at the time the effects it would have. Koopmans began planning the event only some months beforehand, and had sent out the invitation letters only one month before the conference while circulating drafts of papers only weeks before the meeting. The conference, as we will show, was meant to be a gathering of scholars pulling at the same string rather than the rally of a community proclaiming a new future of economic research. The Cowles Commission stood at the crossroads of various disciplines, as well as between university and national laboratories. Newly available archival material allows us to reconstruct what made this event so important.

Such an exercise is critical for understanding the “agency” of scientific change and disciplinary formation. Previous generations of historians have spoken of the change we consider as a “formalist revolution,” a phrase that suggests a shared intention of a small number of scholars who managed to impose their values on the profession (e.g. Blaug Reference Blaug, Backhouse and Creedy1999). Such was not the case. In fact, we show that the conference was unconcerned with the specific shape of the discipline of economics. Equally, one cannot provide a satisfactory explanation of this change by embedding it in a metanarrative concerning the postwar institutional structures of science. Narratives focused on the role of the military in postwar economics (e.g. Mirowski Reference Mirowski2002) are too often thin and unconvincing, seeing the “military-industrial complex” everywhere.

Our story begins with a “snapshot” of the conference. We will then locate it with respect both to the history of Cowles and the landscape of U.S. science and the RAND Corporation. Finally, after connecting the conference and its participants to the mathematics departments at Chicago and Princeton, we will revisit what has up to now been the canonical account of what transformed economics in the postwar years.

Snapshot

In the wake of World War II, it was institutions, perhaps more than individuals, that mattered in science. In his conference invitation letter Koopmans identified several “groups” of participants. The “Princeton University” group was represented by the mathematicians Albert Tucker, Harold Kuhn, and David Gale (who shortly before had left Princeton's mathematics department for Brown’s), and the economists such as Oskar Morgenstern and Ansley Coale. The “Cowles group” included Kenneth Arrow, Murray Gerstenhaber (a mathematics department PhD student of A. A. Albert), Cliff Hildreth, Tjalling Koopmans, Stanley Reiter, and Herbert Simon. Nicholas Georgescu-Roegen came from Wassily Leontief's “Harvard Economic Project.” Marshall Wood, Charles Hitch, Merrill Flood, Norman Kaplan, Paul Samuelson, Murray Geisler, George Brown, and a key figure of the conference, George Dantzig, were associated with RAND and the Air Force. In total, there were thirty-four papers and about fifty participants. Koopmans also invited John von Neumann, but he did not come.Footnote ²

This list constructs the world of mathematical social science in the United States in 1949. What stands out is the interrelatedness of the groups from both governmental and academic institutions. Marshall Wood, the named representative of the Air Force, moved back and forth to RAND which was doing contract work on game theory. Samuelson, the 1947 Clark Medal winner, was visiting RAND that spring from MIT.Footnote ³ Data issues confronted the Bureau of Labor Statistics, the Council of Economic Advisers, and Leontief's Harvard Economic Project as well as Cowles, where Lawrence Klein had from 1944 to1947 been constructing a national econometric forecasting model along lines that Jan Tinbergen had pioneered. Evsey Domar at Johns Hopkins and Tibor Scitovsky at Stanford were studying technological change. The mathematician Mina Rees, representing the Navy, had been supporting work in mathematical economics through her position in the Office of Naval Research: Morgenstern at Princeton as well as Arrow at Stanford had been receiving ONR funds as had John von Neumann as a consultant at RAND. In 1949 the ONR would fund the Logistic Branch of the Mathematical Sciences Division led by Fred Rigby: both Arrow and Koopmans collaborated there (Dalmedico Reference Dalmedico1996, 183). It was a very small world, a world that was both separate from the older generation of the Econometric Society and to a notable extent from the visible sites of academia, the economics departments of the Ivy League universities.

The theories discussed at the conference likewise stood “between” the concerns of academia and the national laboratories. As Koopmans described the purpose of the meeting to his invitees on May 11, 1949:

The theories central to the conference were thus Leontief's input-output model as restated by Samuelson, linear programming and its applications as shaped by Dantzig's simplex algorithm, and traditional welfare economics as developed by Bergson, Lange and Lerner in the 1930s.Footnote ⁴ Looking back from today, only a small number of the particular models would end up in economics textbooks. Though activity analysis was an important ingredient for reconstructing production theory in such way that it was amenable to an axiomatization, it did itself not survive this integration into a general equilibrium framework.Footnote ⁵ Similarly Leontief's input-output models were designed to examine inter-industry relationships, but the fixed nature of the production technology contrasted with traditional models in which inputs could be substitutes for one another as their relative prices changed. Leontief's model, in the years to come, would not contribute to the integration of economic theory as general equilibrium theory would do, even though, for example, Georgescu-Roegen thought of input-output analysis as “the first attempt to apply the general equilibrium theory to the analysis of an economic reality” (Koopmans Reference Koopmans1951a, 98).

Neither would Dantzig's simplex algorithm enter economics textbooks as it entered other textbooks.Footnote ⁶ Two years later, in 1951, Dantzig would organize what then would be called the 1^st of the conferences on mathematical programming, the activity analysis conference being the 0^th, jointly with Alex Orden and Leon Goldstein titled “Linear Inequalities and Programming” at the National Bureau for Standards. No one who would become relevant to Cowles or economics was present. The Mathematical Programming Society would not be related to the discipline of economics at all, but would instead come to be tagged as “mathematics and its applications in industry, business, and technology.”Footnote ⁷

The conference thus opened the gate allowing traffic in two directions: mathematical rigor flowed into economic theory, and economic theory flowed into mathematics, reinforcing its applications. This was made possible by the fact that among the participants there was no clear commitment to the discipline of economics, even as they agreed on a general notion of the economic relevance of their research. What united the conferees was their enthusiasm about the new techniques that had been developed in various institutions within and without academia in unstable political contexts. Understanding this trans-disciplinary format of the conference requires understanding the institutional milieu within which the Cowles Commission operated in this immediate post-war period.

The Cowles Commission between Academia and Laboratory

The Cowles Commission for Research and Economics had been founded in 1932 in Colorado, two years after the foundation of the Econometric Society. It was named after its founder Alfred Cowles, a wealthy investment adviser, who hoped for better predictions of stock market behavior by using mathematical and statistical tools. For this purpose he hired well-known economists such as Irving Fisher, Harold Hotelling, and Ragnar Frisch to work part time, or as consultants, to the Commission. Even though the research produced did not help his business affairs, he continued on as the patron of many of the members of the young Econometric Society. In 1939 both for family reasons and in order to avoid state taxes in Colorado, Cowles searched for a new home for his Foundation and found it at the University of Chicago.

This move was fortuitous. Chicago had been important in statistical economics, and Henry Schultz, one of the founding members of the Econometric Society, had attempted to unite theory and statistics in his work on estimating demand curves. Chicago's willingness to host Cowles was in part related to the fact that Schultz had recently died in an automobile accident. From the Department of Economics, only Oskar Lange was part of the Cowles group. Theodore Yntema, a former student of Schultz's became Research Director in 1939. As he became more involved in war related work over the next few years, some of his duties seem to have been handled by Lange: he for example wrote to Samuelson and others in 1940–41 asking about possible recruits for Cowles. Instead of seeking local faculty to hire, the Cowles Commission recruited from the pre-WWII European émigré community. As Roy Epstein wrote in his engaging history of econometrics: “It is also appropriate to record Cowles’ sponsorship of refugees from Nazism, in particular Abraham Wald and Horst Mendershausen. Perhaps owing to the liberal and internationalist outlook of the Cowles family, the Commission soon became a notable stopover point for many foreign economists visiting the United States” (Epstein Reference Epstein1987, 60–61).Footnote ⁸ Jacob Marschak had grown up in the Ukraine, was educated in Germany, and had headed the Oxford Statistics Institute. From Marschak's arrival at Cowles as Research Director (replacing Yntema) in 1943, from Oxford via the New School for Social Research, the Commission became increasingly attractive to technically trained European scholars most of them well-known to Marschak.Footnote ⁹ Herbert Scarf recalled:

The Cowles Commission was uniquely located at the crossroads of the sites that shaped postwar academia. The end of the WWII reconfigured the institutions of US science as it boosted optimism that the scientific achievements that had enabled victory could similarly enable a prosperous peacetime society.Footnote ¹⁰ The war-ending shock of the August 1945 atomic bombs, and the resulting Japanese surrender, supported the belief that planned government support of science would sustain western freedoms (Hollinger Reference Hollinger1996, 160–64). It also resulted in a re-evaluation of the meaning of the pure and the applied – they appeared to have more in common than expected (Dalmedico Reference Dalmedico, Bottazzini and Dalmedico2001).

The pressing question was how to organize science in the postwar period (see e.g. Dalmedico Reference Dalmedico1996; Leggon Reference Leggon and Graubard2001, 221–224). In 1944–45 the U.S. Congress commissioned a study of how scientific research should be funded. There appeared to be two possible models. The first was to have Congress set up a research agency and then fund or earmark projects prioritized by national needs, as had been done during the war. Alternatively Congress could give money to the scientists directly or through their employers, letting peer review and competition solve the allocation problem politically unencumbered by pork-barrel politics. The former model meant continuity with the wartime regime; the latter meant a return to the ideal vision of the autonomy of science. The negotiations between these two models required balancing maintenance of some elements of wartime scientific institutions and practices while simultaneously rejecting those elements that did not fit well with a democratic peacetime society.

And so universities were drawn into two not necessarily compatible roles: on the one hand, they were to host the “scientific community” which was supposed to exemplify the values of a free democratic society. The pursuit of truth among scientific peers was a model of the behavior that one would expect from a well-functioning democracy.Footnote ¹¹ On the other hand however, with the lifting of wartime secrecy, Americans learned that the success of scientists, and the technology they had brought to the war effort, had created a large government “owned” scientific community which had worked with a sense of national purpose: Oak Ridge, Hanford, and Los Alamos were the outward manifestations of this new scientific world. These institutions hosted science in ways quite apart from the democratic ideals of open universities – classified work remained taboo in universities. The continued success of science in the creation of the postwar society would thus depend upon a successful resolution of the tension between continuity and reform, between transparency and secrecy, between scientific control of society and traditional values like liberty, between the autonomy of open science and its emergent social role as preserver of democracy against its enemies – a real conflict of the norms of science, as Robert Merton had said (Reference Merton1973, 276), thus representing a full-blown Kulturkampf, as Hollinger had added (1996, 155ff).

In the end, Vannever Bush's report Science,the Endless Frontier (1945), presented Congress with a mixed model for support of postwar science.Footnote ¹² He linked the fortunes of science to those of the government: “Since health, well-being, and security are proper concerns of government, scientific progress is, and must be, of vital interest to the government” (Bush Reference Bush1945, chap. 2, sect. 2). By famously putting emphasis on the equal value of both “basic” and “applied” research (the former paying off in an unknown future, the latter in a near future), he muddled through the image conflict of older and newer forms of knowledge. Bush had been the Director of the Office of Science, Research, and Development (OSRD) which was to be closed in December 1947. In its place Congress would fund a National Science Foundation (NSF) (to be founded in 1950) with an annual budget appropriation, and the NSF would make grants to scientists through a peer review process organized by disciplinary scientific panels. At the same time government did not wish to give up the military infrastructure it had built up during the war. National security appeared to require that the government continue its direct funding of specific research projects on, for instance, nuclear and thermonuclear weaponry. Congress thus sponsored national laboratories that engaged in largely secret defense related work. Additionally Congress created administrative entities in support of national defense objectives. It authorized funding for the Office of Naval Research (ONR) (see Sapolsky Reference Sapolsky1990),Footnote ¹³ the Atomic Energy Commission, the National Advisory Committee for Aeronautics (later NASA), the Defense Applied Research Projects Agency (DARPA), and the Air Force's Research and Development center in Santa Monica, an initially private corporation to be known by its acronym, RAND. These research laboratories enjoyed freedoms unknown in academic institutions that had to represent epistemic virtues in public.

During Marschak's time as director, but particularly after Koopmans’ research directorship beginning the summer of 1948, the Cowles Commission would find a new place in-between these two sites of the post-1945 production of knowledge. This new position of Cowles became apparent following Koopmans’ newly established relationship with RAND responding to the withdrawal of its major funding sources, the Social Science Research Council and the Rockefeller Foundation (see Mirowski Reference Mirowski2002, 262).Footnote ¹⁴ And so the Cowles Commission became a hybrid institution somewhere between these two sites of science, between a university department and a national laboratory. The activity analysis conference can be understood historically as the successful attempt to negotiate the tension that existed between these two sites, between newer and older images of knowledge. In fact this negotiation required distance from both, from military purposes on the one hand, and the economics departments on the other.

Distancing from RAND

RAND stands for the continuity of pre- and postwar research. Having been part of the military industry as “Project Rand,” it was founded as an autonomous, military financed research center in 1948 in order “to insure the continuance of teamwork among the military, other government agencies, industry, and the universities.”Footnote ¹⁵ Closed off from the public, and without commitments to a disciplinary order, RAND sponsored highly eclectic research ranging across weapons systems engineering, abstract mathematical inquiry, studies in logic, computational technology, operations research, and game theory all under the label of “systems analysis.” RAND was a socially committed trans-disciplinary environment without immediate relation to any of the traditional concerns of economic theory. If one can speak of a shared vision at all, it was to seek to apply the principles of rational agency to politics and warfare.

Some historians have emphasized the role of RAND for the development of Cowles’ research program (see Mirowski Reference Mirowski2002; and Leonard Reference Leonard2010). Mirowski goes as far as arguing that “RAND was the primary intellectual influence upon the Cowles Commission in the 1950s, which is tantamount to saying RAND was the inspiration for much of the advanced mathematical formalization of the neoclassical orthodoxy in the immediate postwar period” (Mirowski Reference Mirowski2002, 208). It is vital however to recognize, as Mirowski does not, that the RAND community had no disciplinary commitments. If RAND had really been the dominant force behind Cowles’ impact on economics it would not have produced a transformation of economic theory, but rather a diffusion of tools and methods of wartime research into an amalgam of research areas and proto-disciplines that were amenable to technical formulation. RAND did not provide closure to disciplines, but embedded them in a trans-disciplinary meta-science (as individuals like Anatol Rapoport and Herbert Simon had hoped for economics). For RAND, the shape of economics as a discipline was simply irrelevant. Indeed, Mirowski repeatedly points out that Cowles’ commitment to economic theory, specifically traditional equilibrium analysis, served, in his terms, to “ward off cyborgs” for decades (ibid., 220, 255, 270). If Cowles indeed held off the cyborg's debut in economics, it must have been the case that the arcane spirit of mathematical Walrasian economic theory was the villain (or hero). The inspiration for the integration of economics through the agency of equilibrium analysis thus did not come from the military.Footnote ¹⁶

While RAND was not limited by disciplinary concerns, academia was. In the years to come universities, with institutions like departmental instructional units and undergraduate curricula, fostered the differentiation of scientific fields. The end of the Second World War was the beginning of a re-negotiation of the disciplinary arrangements of the late nineteenth century. The rhetoric of difference between science and engineering, between pure and applied, between literary and scientific work took hold. C. P. Snow's Two Cultures emerged in this period (Ortolano Reference Ortolano2009). As a result, mathematics departments gained an autonomy they never had before as they became independent from the physical sciences, as we discuss in the next section. At that time the “social sciences” also grew apart as differentiated disciplines of business economics (Augier and March Reference Augier and March2011), sociology, cultural anthropology, political science, and economics (Backhouse and Fontaine Reference Backhouse and Fontaine2010). This is one of the oddities of early Cold War Science: alongside the differentiation of disciplines, ongoing today, there was a shared commitment to certain kinds of techniques among the elites of these disciplines. As methods easily cross disciplinary boundaries when being detached from a system of belief in their early stages, they can later come to define these boundaries when “thickened” with doxic content regarding the deeper strictures of reality (Chandler Reference Chandler2009).

In other words, RAND might have provided resources, but it was of no help to Cowlesmen in confronting the issue of legitimacy they faced within academia. This pressing issue is apparent in Koopmans’ “Introduction” to the conference proceedings when he referred to the nature of military funding:

Apart from this rhetorical strategy of distancing from RAND, one should also note that there were a number of scholars at Cowles who had no connection with RAND or its activities. Roy Radner, Stanley Reiter, Leo Hurwicz, and Gerard Debreu engaged in mathematical modeling that was more Platonist and less computational. Between Cowles and RAND stood, as we are going to see, the purity of Bourbaki mathematics. Those attracted to purity and rigor remained rather ignorant about what their colleagues did at RAND. As Debreu for example asserted later, somewhat ambiguously:

Some went, Debreu knew. But he preferred not to ask who went. Who knows what they do there?

Distancing from Politics

Legitimacy shaped the conflict between Cowles and its host, the University of Chicago Economics Department. One of Schultz's and later Hotelling's PhD students, Milton Friedman, was outspoken in his distaste for the theoretical approach to data-mining – a distaste that culminated in his confusing methodological essay of 1953. This conflict might have appeared to be technical in nature, but it was nourished by a climate of political suspicion. Chicago had grown into a “school” known, among American economists, as staunchly opposed to the Roosevelt administration and New Deal policies (van Horn et al. Reference Van Horn, Mirowski and Stapleford2011). Paul Douglas had been a liberal outsider among conservative economists like Henry Simons, Jacob Viner, and Frank Knight, and their students Gregg Lewis and George Stigler. The Cowles people in contrast were a collection of European socialists and social democrats, and homegrown left-liberals. “We members of the Cowles Commission,” Lawrence Klein witnessed, “were seeking an objective that would permit state intervention and guidance for economic policy, and this approach was eschewed by both the National Bureau and the Chicago School” (quoted in Mirowski Reference Mirowski2002, 243). Indeed Klein joined the Communist Party while at Cowles. Koopmans’ very left-wing views at the time, not so uncommon among American scholars, found a sympathetic audience among the European social democrats at Cowles. Indeed Marschak, in his Menshevik days, had been the minister of labor in the Soviet Republic of Terek in 1918.

It was not only the economics department that questioned the legitimacy of Cowles, but the university's administrators and government officialdom at large. In that postwar period in which interest in socialism was tantamount to sedition, past Communist Party membership was grounds for employment termination, and interests in economic planning were best left unremarked, the need for de-politicization was evident to Koopmans. For example Domar and Kaplan spoke at the conference on applying linear models to the Soviet economy, but Koopmans would not include them in the proceedings (see Backhouse Reference Backhouse2012).

The fact that economists interested in advancing technical tools in economic theory, specifically Europeans, were rather “left” should not have surprised anyone at Chicago. Having fled Nazi and fascist Europe, many Cowlesmen had been active earlier in what was known as the “socialist calculation debate.” Oskar Lange, who left Chicago on a leave of absence to go to New York to work for the Polish government in exile from 1943–1945, resigned his professorship at Chicago in 1945 in order to help plan and build the postwar economy in Poland. It had been Lange who brought forward the Walrasian model as a planning device.Footnote ¹⁷ Arrow recalled:

Since activity analysis was concerned with organizational questions of production based on systems of equations and programming, it had a clear connection with market socialism. Thus, underneath the surface interest in activity analysis was an older set of arguments about the possibility that some kind of socialist planning model could produce the same efficient outcomes that a competitive market economy might produce. Discussing a theory of production in the second half of the 1940s in Chicago unavoidably evoked Lange's theory of planning.

It was this historical burden of competitive analysis that increased concerns about the legitimacy of the Cowles research program. Koopmans could not avoid mentioning Ludwig von Mises and Lange on the first pages of the Proceedings’ introduction in the following terms: “Particular use is made of those discussions in welfare economics (opened by a challenge of L. von Mises) that dealt with the possibility of economics calculation in a socialist society. The notion of prices as constituting the information that should circulate between centers of decision to make consistent allocation possible emerged from the discussion by Lange, Lerner and others” (Koopmans Reference Koopmans1951a, 3). When Koopmans circulated the introduction to various participants prior to its publication, it was Paul Samuelson who suggested that he simply skip reference to Lange and von Mises altogether (Paul A. Samuelson Papers, Box 45, “Tjalling Koopmans”). But for Koopmans, who as the spokesman of the conference sought a conceptual framework that would mediate conflicting demands, this question had to be settled rather than silenced. For him, planning was no longer a political option but an organizational necessity.

Economies “where competitive markets do not penetrate” in 1949 immediately evoked the war economy which necessarily had been a centrally planned economy. The U.S. did not achieve victory in the war by letting free or competitive markets decide what armaments to produce and how scientists should be allocated to particular military tasks. Koopmans clearly walked a tightrope in a politically heated world. He urged the use of a theory of planning in ways compatible with a democratic U.S. society. Koopmans’ attempt to de-politicize the theory of production and welfare economics through activity analysis was an attempt to deal with the tension under which the Cowles Commission was working: utilizing the planning tools developed during war time research and reshaping their previous meanings to the new environment. We have evidence of this tightrope-walk as he tried to frame a central notion of activity analysis: prices! On March 1, 1950, Koopmans wrote to Samuelson:

Koopmans’ insecurity as late as 1950 about such a basic question as whether to speak of “prices” or of “values” can only be understood against the background of the political poison that “planning” had become in the McCarthy period.

This confusion regarding terminology would later be at the heart of a debate about scientific priority: Leonid Kantorovich independently had discovered the principles of linear programming in the Soviet Union (see Bockman and Bernstein Reference Bockman and Bernstein2008). Whatever might have been the reason for giving Kantorovich and Koopmans the Nobel Prize jointly, it produced a debate that echoed the terms of the calculation debate. Kantorovich, of course, did not use the term “prices,” which became a problem once his article was to be translated. Without using this term, could his formulation of linear programming be considered the same discovery? If a theory of socialist planning and a market theory are formally equivalent, this difference of framing would not translate to a difference of credit. Yet Abraham Charnes and W.W. Cooper argued against the equivalent achievement of Kantorovich on the basis of an essentialist notion of prices, assuming that a Stalinist state could not bring forth work worthy of a Nobel Prize. Of course formal equivalence versus irreducibility of an economic theory of socialism and markets had been the core of the calculation debate between Lange and von Mises.

In sum, the spirit of the conference, responding to pressures of legitimacy, was supposed to be a-political. Applying new tools to economic theory, Koopmans’ research program stands for a de-politicization of a kind of economic theory previously used in contexts of socialist or war planning. The tacit agreement among the participants was that the significance of technical tools derived in some way or the other from the possibility of social engineering – whatever might be the political or moral justification for doing so. This de-politicization of the theory, as we see in the next section, was not so much achieved by Koopmans’ rhetorical moves and re-interpretations. Instead it was an unintended consequence of the theory's “moving” to the mathematics department.

Mathematics at Chicago and Princeton

The self-protective a-political turn was apparent not just to the economists at the Cowles Commission but equally to mathematicians who had been active in scientific war engineering, especially at Princeton and Chicago (where most mathematicians present at the conference were located). Both departments had been involved in the recruitment of mathematicians during WWII, and both were to become central to the postwar institutions of science. Without understanding the influence of Princeton and Chicago mathematicians on Cowles, we cannot understand the transformation of economic theory that was birthed there.

After WWII, mathematics in the United States needed to balance newly emerging fields of application that threatened its traditional internal organization (e.g. computation, partial differential equations, probability, and statistics). Mathematics could not rise to primus inter pares among the sciences by simply providing tools for the special sciences. It also had to strengthen its disciplinary autarky. American mathematics could best flourish in the postwar period if it stood apart from those applied disciplines whose political commitments produced continual scrutiny in the McCarthy period. If all mathematics were applied mathematics, and applications bore political weight in the cold war era, mathematicians would live under the same security regime as did nuclear physicists. This is the background of the renewed valorized distinction between “pure” and “applied” mathematics: the latter was potentially contaminated by the applications which existed outside mathematics, while the former entailed no commitment to any ideas which were not mathematical ideas.Footnote ²⁰ This relief from responsibility associated with mathematical purism was openly assumed by Jean Dieudonné, writing on behalf of Bourbaki:

This Bourbakian aloofness was vital for its success in the United States. Bourbaki mathematics became metonymous for pure mathematics, particularly at Chicago.

The story of the reconstruction of the Chicago mathematics department begins with Marshall Stone. Having done classified research for the U.S. Navy and the Department of War until 1945 while on leave from Harvard, he was, on the advice of von Neumann, asked by Chicago's President Robert Hutchins to re-build Chicago's mathematics department in 1946 (after the war had left five senior positions vacant). Chicago had hosted the Manhattan project and in order to hold on to scientists such as Enrico Fermi, James Franck, and Harold Urey, the mathematics department needed strengthening (see Mac Lane Reference Mac Lane and Duren1989a, 146). Stone was a magnificent administrator who knew that to pilot mathematics into new postwar waters he should take advantage of the opportunity to recruit first-rate émigré mathematicians, above all the single most important hire, André Weil – one of Bourbaki's leaders who managed to escape Europe in 1941.Footnote ²¹

As German mathematics had self-destructed with the purges of Jewish scholars, and with American mathematics still backward compared to European scholarship, Bourbaki represented a new and exciting integrative vision of the discipline of mathematics, one which took hold internationally in the period from the end of the 1940s through the late 1960s (Corry Reference Corry1992; Weintraub and Mirowski Reference Weintraub and Mirowski1994). The Theory of Sets volume had appeared in 1939, but the war's dislocations meant that subsequent volumes only began appearing in 1947. Those new volumes took up topology, algebra, functions of one real variable, and integration. The younger Bourbaki group members who were either members of, or passed through, the departments in Chicago and Princeton were (besides André Weil) Samuel Eilenberg, Armand Borel, Serge Lang, Claude Chevalley, and John Tate. They in turn had close contacts with other members like Pierre Samuel, Jean-Pierre Serre, and Jean-Louis Koszul (see Weil Reference Weil1991, 178).

The connections between the Chicago mathematics department and the Cowles group were manifest in the work of several mathematicians. Israel N. Herstein, who was to write the important textbook Topics in Algebra, was connected to the Cowles group after his 1952 appointment as Assistant Professor at Chicago in Mathematics and Economics. In mathematics he worked closely with Albert, while he also wrote several Cowles Commission Discussion Papers on optimization concepts and efficiency. That same year Herstein would write a paper jointly with the Princeton and RAND mathematician (and future Field Medalist) John Milnor, which must have been a blueprint for the style the future “Neo-Walrasians” wished to develop (Herstein and Milnor Reference Herstein and Milnor1953).Footnote ²² We have already mentioned that Murray Gerstenhaber, a Ph.D student of Albert, wrote the exceptional paper on convex polyhedral cones for the Activity Analysis Conference and volume.

Morton L. Slater was another mathematician affiliated with Cowles as a “Research Consultant.” He was to serve as the referee of most of the papers at the 1948 activity analysis conference. Slater was a University of Wisconsin graduate who did his graduate work in mathematics at Harvard. In that period he also had the title of Senior Mathematician, (Navy) Ordinance Research, Chicago, Illinois. In Cowles’ Annual Report for the year 1950–51, Slater is described as a Research Associate who “provided mathematical advice and criticism with regard to the work of many staff members and provided expository presentations of mathematical results to economists” (Cowles Commission 1951).

Not only did the mathematics department influence Cowles; the reverse was also true. Marshall Stone had certainly strengthened pure mathematics at Chicago but he was also concerned about the place of applied mathematics in his department. In seeking to broaden the Chicago mathematics community, Stone was to turn to the Cowles group for help.

Thus the presence of the Cowles Commission diminished pressure on the mathematics department to integrate applied mathematics into its curriculum. Put simply, Cowles and the Chicago mathematics department were in a mutually stabilizing relationship: Cowles allowed the mathematics department to maintain the disciplinary hierarchies and thus its disciplinary autarky while connection to the mathematics department rather than the economics department allowed Cowles to avoid the charge of political advocacy. Mathematical economics functioned as a buffer that allowed the “image conflict” within mathematics to become visible but not fraught.Footnote ²³ Given this shared dependence it is obvious that the ties of the Cowles Commission to the mathematics department were tighter than to the economics department, and helps to explain why few at Cowles were deeply involved in the controversies energizing the larger economics profession.

Unlike Chicago, Princeton's mathematics department hardly needed strengthening. Einstein, von Neumann, Gödel, and Weyl brought Göttingen and Vienna to the Institute for Advanced Study in the early 1930s, and with Oswald Veblen and James Alexander moving to the Institute from the university, it became the center of the mathematical universe. Veblen, who had been Princeton's department chair, helped leadership pass to Solomon Lefschetz who was soon to recruit a remarkable faculty by replacing senior retirees with young stars like Alonzo Church, William Feller, A. W. Tucker, S. S. Wilks, and Emil Artin.Footnote ²⁴

In contrast to the unified mathematical spirit developed by Marshall Stone and André Weil at Chicago, the mathematical hothouse of Fine Hall, home to the Princeton University mathematics department, hosted an assemblage of subgroups in which a hierarchy emerged from topology at the top – represented by Salomon Lefschetz – to game theory at the bottom – represented by Albert William Tucker.Footnote ²⁵ Tucker, a former student of Lefschetz’s, supervised virtually all the top game theory students: David Gale (PhD 1949), John Nash (PhD 1950), Lloyd Shapley (PhD 1954), and Marvin Minsky (PhD 1954). Though supervised by Fox, Harold Kuhn belonged to this group too, as well did Leon Henkin. Gale, Kuhn, and Tucker ran a weekly seminar on game theory and its computational uses.Footnote ²⁶

While the Cowles Commission had some productive contact with the Chicago economics department especially in teaching statistics and mathematical economics, Tucker's group was entirely uninvolved with the Princeton economics department. Oskar Morgenstern was the lone representative of recent developments in economic theory though his students rather shunned economic theory. Shubik recalled:

None of the participants at the Activity Analysis conference, however, would yawn. Von Neumann's ([1936] Reference von Neumann1945) earlier work in growth theory influenced nearly all conferees. That paper examined the possibility of existence of equilibrium of an economy whose various interconnected parts exhibited uniform rates of growth. It was the first paper related to economics which used a topological fixed point argument to achieve an equilibrium solution. The fact that general equilibrium analysis was reinvigorated at Cowles after 1949 was a direct result of the rediscovery of this paper and the important role it played by introducing topological methods.

John von Neumann

Von Neumann, hosted at the Institute for Advanced Study, was the shining mathematical star of war-time engineering, operations research, and game theory. As André Weil, the Chicago representative of Bourbaki, personified the new intellectual purity in economic theory, at Princeton John von Neumann's authority fused pure mathematics with the eclectic spirit of applied research.Footnote ²⁷ Both as a mathematician and wartime consultant on numerous secret projects, von Neumann was akin to the old-fashioned long-distance telephone operator of the 1930s: he coordinated all conversations.

There were two ironies in the resurrection of von Neumann's growth model. Von Neumann, just as his co-author and conference participant Oskar Morgenstern, was hostile to competitive economic analysis. The central theorem of the growth equilibrium paper was connected to his 1928 proof of the minimax theorem for two-person zero-sum games. While in that earlier paper (von Neumann Reference von Neumann1928) he had provided no reference to any economic tradition, the growth model paper of 1936 referred to a “typical economic equation system” (ibid., 1). While the former modeled strategic behavior, the latter modeled competition. In his and Morgenstern's 1944 Theory of Games and Economic Behavior, strategic behavior and competitive behavior were expressly opposed to one another ([Reference von Neumann and Morgenstern1944] 1953, 15) while the two papers’ central theorems were “oddly connected,” being reducible to one another via a saddle point ([1936] 1945, 1). Von Neumann and Morgenstern wondered whether “there may be some deeper formal connections here . . . . The subject should be clarified further” Reference von Neumann and Morgenstern([1944] 1953, 154). Without anyone at the time clarifying this further, the equivalence made possible the steady crowding out of game theory as an alternative “paradigm” to competition. The different economic intuitions that had separately evolved in game theory and in perfect competition analysis merged since they were both amenable to the same topological methods. Von Neumann himself did no further work in competitive equilibrium analysis, or any economics, nor did he return to work on topological fixed point theory.

There is a second irony to von Neumann's influence. By the late 1940s, von Neumann began pursuing algorithmics and the computational use of mathematics eschewing the Bourbakist approach to mathematics. His 1947 paper “The Mathematician” stated clearly his view that difficult problems in science generated the most important mathematical discoveries, and that a mathematics dependent only upon itself would soon become uninteresting, even to mathematicians. This view was entirely anti-Bourbaki. His ideas were represented at the conference by his colleague A. W. Tucker. Kuhn and Tucker developed von Neumann's minimax theorem for computational uses thus following von Neumann's anti-Bourbaki lead in mathematical engineering. It is for this reason that the developments in linear programming, game theory, and operations research began to merge. The notion of strategic behavior was lost to economics as it moved into the foreground of organizational questions. The Theory of Games and Economic Behavior was thus pushed away from the mainstream culture in economics.Footnote ²⁸ As Shubik witnessed,

While Shubik contrasted von Neumann's game theory with programming, Dantzig, who had discovered linear programming, could equally credit von Neumann as his source. The link is revealed in Robert Dorfman's (Reference Dorfman1984) particularly lucid account of Dantzig's “Eureka moment” of discovering linear programming two years before the conference. Dorfman quoted from a then unpublished 1976 memoirFootnote ²⁹ by George Dantzig as follows:

There followed a series of memoranda back and forth between von Neumann and Dantzig in one of which there is a footnote crediting Koopmans for “making suggestions on which the procedure for moving from one simplex to a better one is based” (Dorfman Reference Dorfman1984, 292; see also Backhouse Reference Backhouse2012, 6). Dantzig visited Cowles in June 1947, met Koopmans, and discovered the same interests. In June 1948, Dantzig visited Princeton to work with Kuhn and Gale on a solution of the duality problem, from which the Kuhn-Tucker theorem emerged. And so Koopmans, Dantzig, and Kuhn became the central figures in the conference for activity analysis in June 1949.

The Conference as Origin

With the preceding as multiple contexts for the 1949 conference, we are now able to interpret in what ways the conference was an “origin” of the changes that were to occur in economic theory. The conference produced the historical conditions for the integration of economic theory first by taking mathematical values of rigor and axiomatization for granted, and second by disengaging with the intellectual values that had shaped economics departments. Both the discipline of economics and the profession's postwar institutions were simply out of focus for the conferees. The conference appears to have been a gathering of scholars pulling at the same string instead of the public emergence of a sub-community with a shared purpose of converting the rest of the profession – not even the Econometric Society let alone the AEA – to their vision of the future of economic research. The conference could be an “originating event” only because the participants had little quarrel regarding the possible consequences of their work. In the no-man’s-land between university and laboratory as well as between the traditional disciplines, the tensions that war-time research created for both the mathematics and economics department could be ignored at the conference. Different institutional backgrounds, different theoretical threads, different epistemological justification for, and pragmatic relevance of, their work merged without being confronted one with another. The participants, instead, enjoyed living their skills while only implicitly agreeing on their meaning.

Certainly some participants might have believed that their work was going to represent new technical standards in the growing Econometric Society. But they would also have to know that, in 1949, at least half of the Econometric Society, and certainly most members of the American Economic Association, were in no position to appreciate their work. If the late 1940s graduate textbooks and syllabi taught no appreciation of Samuelson's modernization of Keynes and Hicks, how could the conferees believe they could convince economists of the power of the new techniques that went far beyond those of Samuelson and Hicks? Who of the participants would have been willing to proclaim the emergence of the new mathematical economics at an AEA roundtable on mathematics and economics that had been organized at the 1951 meetings with panelists Samuelson, Knight and Machlup! (Machlup Reference Machlup1952). What was topology to Knight or Machlup?

There certainly were some participants who were willing to confront others’ views with their vision of social science and its responsibilities. We have already noted that Tjalling Koopmans acted as the spokesman of the conference, framing the possible interpretations and uses of activity analysis while contextualizing its underlying traditions. It was Koopmans who was willing to present his vision in front of a joint meeting of the Econometric Society and the American Economic Association half a year after the conference (Koopmans Reference Koopmans1951a). It was he who was active in linking the various emergent theories with the theoretical tradition of welfare economics. It was also Koopmans who in 1957 would write the only piece that defended the theoretical turn at Cowles in methodological terms, in his famous Three Essays on the State of Economic Science (1957).

In contrast there was only one conferee willing to challenge the emergent Cowles perspective, namely Oskar Morgenstern (discussing method, as he already did and was known for ever since the 1930s) (see Boumans Reference Boumans, Maas and Morgan2012). His three-page paper warned readers of the lack of accuracy of the “observations” underlying the data used in linear programming. There is neither evidence for the belief, nor does the previous contextualization suggest, that other participants would react with strong emotions.Footnote ³⁰ One related point of disagreement might have been Leontief's input-output model and its apparent empirical applications. But given Leontief's absence, one might easily imagine that his model was not discussed as a challenge to theory, but as an interesting addition to method.Footnote ³¹

Since these challenges were hardly public, the activity analysis conference was little noted in the community of economists at the time. It was not seen at the time as a signal event. Unless economists were regular readers of Econometrica they would not even have been aware of these new ideas in mathematical theory, and even statistically interested readers of that journal would have had little connection to the conference's papers. In retrospect, the conference became a marker event because it was both exclusive and informal. The conferees had the freedom to not worry about the historical weight of their work, a liberty that had its source in the presentist problem-directed culture of mathematics departments.

The conference showcased the shared and interrogated methodological commitment free from any doxic baggage (Chandler Reference Chandler2009). This communal character characterizes a shift from economics writing as a literary production to economics as an exercise in scientific authorship.Footnote ³² This shift was heralded in Koopmans’ letter to the authors of the articles in the conference volume on July 8, 1949: “Since the subject of the volume has been furthered through conversations and conferences as much as through publications, it is suggested that authors feel free to refer to individuals as the source of ideas whenever there are no publications to which to refer” (Paul A. Samuelson Papers, Box 45, “Tjalling Koopmans”). Issues of scientific credit have a particular character in such an intellectual environment. If science in-the-making is a communal effort, a matter of conversation rather than contemplation, the intimate connection between authors and ideas becomes problematic (Düppe and Weintraub Reference Düppe and Weintraub2014). Koopmans’ letter alluded to this problem.

Given his own history of taking matters of scholarship personally, it is hardly surprising that Nicholas Georgescu-Roegen was the first to raise issues of credit and priority with respect to the conference. He claimed, as Koopmans quoted him in a letter to Samuelson on February 3, 1950, that his results were “first presented at March 22, 1949, at a meeting of the staff of Harvard Economics Research Project” (Paul A. Samuelson Papers, Box 45, “Tjalling Koopmans”). That is, Georgescu-Roegen claimed equal credit for Samuelson's theorem on the possibility of substitution in Leontief models. Samuelson replied on February 5: “I see no reason to question Georgescu-Roegen claim to independent discovery of the theorem concerning non-substitutability in the Leontief system. I only wonder that it was not discovered sooner by someone” (ibid). This short exchange documents the confrontation of two scientific attitudes. In one there is reference to a world in which individuals have agency in conceiving truth. In the other truth is the outcome of shared projects. In the former individuals claim credit, in the latter it is virtuous to give maximum credit to the larger community.

One cannot undervalue the Activities Analysis Conference. It announced, in Karen Knorr-Cetina's phrase, the birth of a new epistemic community. The conference showcased a new intellectual culture in economics as a communal effort based on shared and un-interrogated standards of evidence, rigor, and techniques embracing different and even conflicting visions of the nature of economic science. The new economics was to be defined by skills and techniques rather than by ideas and canonical texts created by solitary geniuses and founders of “schools of thought.” The conference performed this new culture in which establishing truth requires the cooperation of specialized intellects with a variety of skills. The conferees did not quarrel about their different backgrounds, their different intellectual socializations, or the diverse possible consequences of their work. Instead they collectively (modulo Georgescu-Roegen) enjoyed living their skills and sharing them, one with another. The conference thus pointed a way forward in which the sense of shared purpose that characterized a great deal of wartime science could carry over to peacetime economic analysis.