The Birth of Computer Science as a Discipline

Abstract

Computer science did not emerge as an academic discipline; it was dragged into existence by a combination of Cold War anxiety, military funding, and the embarrassing fact that the machines being built in the 1940s and 1950s had no intellectual home. Mathematics departments considered them engineering problems. Engineering departments considered them mathematical abstractions. The result was a two-decade argument about what computer science actually was, conducted by some of the most brilliant people of the twentieth century, in the shadow of Sputnik and the nuclear arms race, and ultimately resolved not by consensus but by money.

Before There Were Departments

The first computers were built by physicists, mathematicians, and engineers working on specific wartime problems. ENIAC (1945) at the University of Pennsylvania was funded by the Army Ballistics Research Laboratory to calculate artillery firing tables. Manchester’s Baby (1948) was a test bed for a memory technology, not a research program in computation. The Cambridge EDSAC (1949) was built by Maurice Wilkes to serve as a tool for numerical calculation across the university. None of these machines were built to instantiate a discipline. Their builders were solving engineering problems.

The people who thought seriously about the theoretical foundations of what these machines were doing (Turing, Church, Gödel) worked in mathematics departments and thought in terms of formal logic, not computing machinery. The separation between theoretical and practical computation was sharp and, in the early 1950s, seemed permanent. Building computers was engineering work. Proving theorems about what was computable was mathematics. The idea that these activities belonged to a single academic field had few advocates.

The first serious academic program in computing in the United States emerged at the University of Pennsylvania in 1946, not as computer science but as the Moore School of Electrical Engineering’s intensive eight-week course on the theory and techniques of electronic digital computers, a course designed to spread knowledge of ENIAC-style machines to other institutions. It trained the people who would build the second generation of American computers. It was an engineering workshop, not a curriculum.

The conference circuit came first. MIT held a meeting on computing in October 1945; the first in Europe, the Conference on High Speed Automatic Calculating-Machines, ran in Cambridge from 22 to 25 June 1949 around the newly completed EDSAC, with about 150 participants and a talk by Alan Turing on checking programs. Harvard’s symposium that September drew more than 700. The largest was UNESCO’s International Conference on Information Processing in Paris, 15 to 20 June 1959, with some 2,000 participants from 37 countries, Howard Aiken as president and Sergei Lebedev, John Mauchly, Maurice Wilkes and Konrad Zuse among the vice-presidents; it led to the founding of IFIP, the International Federation for Information Processing.

The Sputnik Shock and the Funding Flood

On October 4, 1957, the Soviet Union launched Sputnik, the first artificial satellite, and the American scientific establishment entered a state of institutional panic. The question was not merely technical; it was existential: had American science and engineering fallen behind? The political response was immediate. The National Defense Education Act (1958) poured federal money into science, mathematics, and engineering education at American universities. The Advanced Research Projects Agency (ARPA, later DARPA) was created in 1958 to ensure American technological supremacy.

For computing, the effects were transformational. ARPA needed to fund research in computing, not to build specific machines, but to develop the theoretical and practical foundations of the field. This required researchers, and researchers required academic homes. The practical consequence was that American universities began creating centers, institutes, and eventually departments specifically for computing research.

MIT’s Project MAC (Mathematics and Computation) was established in 1963 with $2 million in ARPA funding, an extraordinary sum. It was designed to explore time-sharing operating systems (the idea that multiple users could share a single computer simultaneously) and artificial intelligence. Project MAC started out on the Compatible Time-Sharing System (CTSS), which Fernando Corbató’s group at the MIT Computation Center had first demonstrated in 1961, and went on to produce Multics (whose ideas Unix borrowed) and fundamental work in programming languages and AI.

Info

ARPA’s funding model was unusual and important: it provided large grants with minimal direction, trusting researchers to pursue interesting problems rather than requiring specific deliverables. This “program manager” model (in which individual ARPA program managers made large bets on researchers they believed in) funded nearly every foundational advance in American computer science from 1960 through 1990. The internet (ARPANET), time-sharing, AI, personal computing, and the mouse all trace directly to ARPA funding.

The First Departments

The question of which university established the first computer science department in the United States is contested but consequential. The answer depends on what you mean by “computer science department”, a naming and organization question that reflects the deeper argument about what the field was.

Purdue University established a Department of Computer Science in 1962, making it the first institution in the United States to create a free-standing academic department with “computer science” in its name. The Purdue department emerged from the Department of Mathematics, reflecting the view (held strongly by mathematicians at the time) that computing was applied mathematics.

Stanford University created its Computer Science Department in 1965, separating from the Mathematics department under the leadership of George Forsythe, who had argued since the late 1950s that computer science required its own disciplinary identity. Forsythe’s Stanford program attracted John McCarthy (who coined the term “artificial intelligence” in the 1955 proposal for the 1956 Dartmouth Conference and was computing’s early evangelist for the view that intelligence itself was a computable phenomenon) and Donald Knuth, whose multi-volume The Art of Computer Programming (first volume 1968) became the field’s foundational textbook and articulated a vision of computer science as the rigorous study of algorithms.

Carnegie Mellon University (then Carnegie Institute of Technology) created its Computer Science Department in 1965. CMU was distinctive in its approach: Herbert Simon and Allen Newell, who shared the 1975 Turing Award, argued that computing was not primarily about machines or mathematics but about information processing, a view that connected computer science to psychology, economics, and the emerging field of cognitive science. Their work on the General Problem Solver (1957) and production systems shaped both AI research and the broader theory of computation as symbol manipulation.

University of California, Berkeley built its program through the 1960s, creating a computer science department at the end of the decade and folding it into EECS in 1973. Berkeley’s particular contribution was systems software: the Berkeley Software Distribution (BSD) of Unix became the vehicle through which TCP/IP spread to the academic world, and researchers like William Kahan (whose work on floating-point arithmetic earned the 1989 Turing Award) and later David Patterson (RISC architecture, 1980s, in parallel with John Hennessy at Stanford) shaped the hardware-software interface.

By the start of the 1970s there were dozens of such departments; by the end of that decade they were standard equipment at American universities. The degree numbers show the scale: 1,588 master’s degrees in computer and information sciences in 1970–71 and 8,070 in 1985–86, a fivefold rise (an increase of 408 percent).

The Discipline’s Identity Crisis

The rapid institutionalization of computer science did not resolve the question of what it was. Three competing visions dominated the debates of the 1960s and 1970s:

Computer science as mathematics. Proponents (many trained in formal logic and discrete mathematics) argued that the core of the discipline was the study of algorithms and computational complexity. The proper questions were: what can be computed? How efficiently? What problems are fundamentally intractable? This tradition produced complexity theory (Cook, Karp, 1971–1972, establishing the P vs NP framework), formal language theory, and the theoretical foundations of programming language semantics. Its strongest expression was in the work of European computer scientists, particularly Dijkstra and Hoare (see European Computer Science Academia).

Computer science as engineering. Proponents argued that the central problems were practical: building systems that worked, at scale, reliably. The interesting questions were not mathematical proofs but design principles: how do you organize a large software system? How do you design a processor? How do you build a network? This tradition dominated at Berkeley and produced the systems research that built the internet’s infrastructure.

Computer science as science. The ACM published curriculum recommendations in 1968 (the “Curriculum 68” report) that tried to resolve the debate by defining computer science as a discipline with theoretical foundations, experimental methods, and engineering applications, analogous to physics, not mathematics or engineering. This framing gave departments a political justification for independence but did not resolve the intellectual tension.

The tension was never fully resolved. It persists in contemporary computer science departments as an argument between theorists and systems researchers, between those who value mathematical proof and those who value working systems. What it produced, however, was a field with unusual breadth, one that could claim Knuth’s algorithmic analysis and Thompson’s Unix implementation as equally central.

Warning

The “Is computer science a science?” question was not merely philosophical. It determined which college within a university the department reported to (and therefore its budget), whether computer science PhDs could count toward national science statistics, and how federal research grants were categorized. The line usually quoted here (“Computer science is no more about computers than astronomy is about telescopes”) was a position in an institutional argument as much as a philosophical observation: it claimed territory in the mathematics/science quadrant and away from engineering.

Myth: Dijkstra said “computer science is no more about computers than astronomy is about telescopes”

No source in Dijkstra’s writings contains it. The earliest version of the argument belongs to Alan Perlis, who in 1968 objected that there is no science of an instrument, and that computer science is no more the science of computers than astronomy is of telescopes. The saying in its familiar wording surfaces in print only in the mid-1980s and was popularized as a slogan by Michael Fellows and Ian Parberry around 1991. See Myths and Misconceptions.

Naming the Field

The field’s names arrived late and differed by country. German newspapers spoke of electronic brains and “thinking machines” (Fritz Mauthner’s philosophical dictionary had used the word in 1910 for the logic machines of William Jevons, John Venn and Allan Marquand); Der Spiegel used the English “Computer” in only a handful of articles before 1965, when it appeared in 34 issues. France got ordinateur from the classicist Jacques Perret, whom IBM France had asked for a name; he proposed it in a letter of 16 April 1955. The engineer Philippe Dreyfus coined informatique in 1962. In Germany the word Informatik first appeared in 1956 in the name of Standard Elektrik Lorenz’s “Informatikwerk” in Stuttgart, where Karl Steinbuch built the Informatik-System Quelle; it entered public use in July 1968, when the federal research ministry circulated recommendations for a new degree course in “Informatik” modelled on American computer science and Research Minister Gerhard Stoltenberg used the word on 22 July, opening the conference “Der Computer in der Universität” at the Technical University of Berlin.

Curriculum and the ACM

The Association for Computing Machinery (ACM), founded in 1947, played a critical role in institutionalizing computer science by providing a professional organization that could define the field’s boundaries and publish curriculum recommendations. The ACM’s 1968 Curriculum report defined a standard undergraduate CS curriculum: data structures, algorithms, numerical methods, programming languages, computer organization, and systems programming. The report was not binding, but it gave new departments a template.

More influential were the ACM Turing Award lectures, established in 1966 (by 2017 the award had gone to 65 people, only three of them women: Frances Allen, Shafi Goldwasser and Barbara Liskov; the youngest winner was Donald Knuth, aged 36 in 1974) and named for Alan Turing. The Turing Award became the field’s most prestigious recognition (its Nobel Prize equivalent) and the lectures associated with it became canonical statements of what the recipients believed computer science was. Alan Perlis (first recipient, 1966), who had chaired the committee that designed the ALGOL programming language, used his lecture to argue for programming as a fundamental intellectual activity. Edsger Dijkstra (1972) argued for mathematical rigor. Donald Knuth (1974), in “Computer Programming as an Art,” argued that programming was an art as well as a science. The Turing Awards traced the field’s intellectual genealogy in a way no curriculum document could.

The Graduate Research Machine

American computer science’s most significant institutional contribution was not the undergraduate curriculum but the PhD program. American research universities built graduate programs in CS that trained researchers from around the world, and many of those researchers returned to their home countries carrying American research norms, American problem definitions, and connections to American funding agencies.

The model was simple: faculty competed for DARPA, NSF, and later industry grants; grant funding paid for graduate students; graduate students did research that generated publications, which established the faculty member’s reputation, which secured more grants. The loop compounded: strong programs attracted strong applicants, produced influential alumni, and generated the publications that defined the field’s research agenda.

In 1988–89 the 129 PhD-granting computer science departments in the United States and Canada produced 625 doctorates, 40 percent of them to foreign students, a small number compared to today (1,651 in 2014). A significant fraction of these PhDs went to non-American students who returned home: the American graduate school system was simultaneously training the world’s computer scientists and establishing American research priorities as the field’s default agenda.

Industry and the Research Triangle

American computer science developed in unusual proximity to industry. Bell Laboratories, IBM Research, Xerox PARC, and later Microsoft Research and Google Brain were not universities, but they produced research of academic quality and employed people with academic training. The boundaries were porous: researchers moved between universities and industrial labs, and the most important industrial research labs (Bell in the 1970s, PARC in the same period) operated like academic departments with unlimited budgets and no teaching obligations.

This proximity shaped what problems academic computer science worked on. Bell Labs produced Unix and C; universities adopted both and built curricula around them. IBM’s systems architecture defined the computing environment that students learned to work in. When Xerox PARC invented the graphical user interface, personal computing, and ethernet, academic CS departments incorporated these into their curricula within years.

The relationship was not always comfortable. Some academics worried that proximity to industry distorted research priorities toward commercially relevant problems and away from fundamental theory. Others argued that the field’s practical orientation was its strength: that computer science was more useful than physics precisely because it had never fully separated itself from engineering.

The Field at Scale

By 2000, computer science was one of the most popular undergraduate majors in the United States. The economic pull was direct: software engineering offered entry-level salaries that few other fields matched. The intellectual pull was also real: computing had become the infrastructure of every other discipline, and understanding it was increasingly a prerequisite for working in biology, economics, physics, and the social sciences.

The field that had no departments in 1955 had hundreds by 2000, built on Cold War funding and industry demand.

The field also learned to write its own history. Henry Prevost Babbage collected his father’s papers in 1889; Vivian Bowden’s anthology Faster Than Thought appeared in 1953, and Brian Randell’s The Origins of Digital Computers (1973) gathered the primary sources from Babbage to EDSAC. Uta Merzbach, born in Berlin in 1933 and a survivor of Theresienstadt, joined the Smithsonian in 1964 as associate curator for calculating machines and computers (full curator from 1971) and recorded early oral histories, among them a 1968 interview with Konrad Zuse. Exhibitions followed: IBM’s A Computer Perspective by Charles and Ray Eames in New York (1971–75), Heinz Zemanek’s data-processing department at the Technical Museum Vienna, opened on 2 May 1974, Mens en Computer in The Hague from 21 September 1977, which drew 300,000 visitors with working machines, and the Digital Computer Museum that DEC opened in Marlborough, Massachusetts, on 24 September 1979. As The Computer Museum it moved to a Boston harbour warehouse in 1984 and closed in 1999; its collection became the core of the Computer History Museum. The Heinz Nixdorf MuseumsForum in Paderborn opened on 24 October 1996.

What it produced (in addition to the software and systems that run the modern world) was a set of intellectual frameworks that spread into every other field: the algorithm as a unit of analysis, the database as a model for structured knowledge, the network as a model for social and biological systems. Computer science did not merely build the tools the world uses. It provided the conceptual vocabulary through which the world increasingly describes itself.