It Began with Babbage (28 page)

So, strictly speaking, the baby Mark I was the first stored-program computer to
become operational
.
⁸⁸
However, when the baby first ran, it had no provision for automatic input and output. It was not a complete—that is, it was not a
fully
operational—automatic stored-program computer.

The Manchester designers acknowledged that the baby was only a first step—a pilot study. It was a small-scale experimental machine intended to test the viability of the storage principle and to gain some experience of working with this kind of a machine.
⁸⁹
Yet, the baby was a universal machine in the sense that it could be used to solve any problem that could be reduced to a program of elementary instructions.
⁹⁰

In the case of the EDSAC, by February 1949, the input (paper tape reader) and output (teleprinter) mechanisms had been attached to the computer. On May 6, a program to
compute a table of squares and print the results was read into the machine, and all of a sudden the results were printed out. Wheeler then wrote a program to compute a table of prime numbers that was soon after executed correctly.
⁹¹

In the light of what happened in the future in the realm of computers and computing—and what in present-centered language is called
information technology
—this event of May 6, 1949, must count as a small step for a small group of people working obscurely in a Cambridge laboratory that would entail a giant leap for humankind.

Wilkes, no doubt, had a flair for the dramatic. A month and a half later, he hosted a conference on automatic computing in Cambridge, which began with a demonstration of the EDSAC in action.
⁹²
The program, written by Wilkes for computing and printing the squares of numbers and their first difference, and Wheeler's program to compute and print a table of primes were run.

As for the Manchester group, a fully operational machine, the Mark I, with input and output facilities was completed in October 1949.
⁹³
Hence, the place of EDSAC as the world's first fully operational automatic stored-program computer.

An architectural concept: the stored-program computer principle. For both Wilkes in Cambridge and Williams in Manchester, this shared paradigmatic core became the foundation of their respective mental schemas for an automatic digital computer (see Section V, this chapter). Thereafter, their respective schemas were refined and elaborated in two different ways. Each researcher instantiated his schema differently, with the outcome being two distinct schemas with a common core. Each postulated a particular design for a particular machine that conformed to the stored-program computer architectural concept. Each design became the theory for a particular computer such that each of them, Wilkes and Williams, believed that if a computer was built according to their respective designs it would “best” realize the stored-program computer architectural principle as laid out in the EDVAC report.

The implementations of the two machines were two distinct experiments, each a test of a distinct design-as-theory. However, the separation in time between design and implementation, theory and experiment, was not “clean.” Rather, the process of implementation went hand in hand with the process of design; theory and experiment overlapped.

Of course, neither Wilkes nor Williams was on his own. Each built a small team; each communicated his respective schema (inner vision, as it were) of what a stored-program computer
should be like
to members of his team. Their schemas were externalized and shared with others. If there was an unfolding paradigm for which the stored-computer concept was the core architectural principle, then we find here the nucleation of two
subparadigms
, each expanding and refining the core principle in a different way.

Something else is noteworthy here. If we think of the EDSAC and the Manchester Mark I designs as representing subparadigms within the overall emerging paradigm,
they did not become alternative, competing, mutually exclusive theories of how a stored-program computer should look. Rather, each project was a complementary experiment as a whole that tested the validity of the stored-program computer concept. At the same time, we must keep in mind that the latter concept was an abstract entity. The two projects, in Cambridge and Manchester, represent different pathways to how the abstract stored-program computer concept could give rise to a material computational artifact. Each project was in the business of producing an individual computer with its own design-as-theory. The emerging paradigm of this particular science of the artificial was to be populated, at this early stage in its life, with at least two distinct designs-as-theories, two distinct implementations—using a biological analogy, two distinct species of computational artifacts. Like biological species, these two cultural species (for artifacts are aspects of culture) might survive or might become extinct in the future, or they may give rise to evolutionary descendants. There was nothing to suggest that these two species (and others that might arise in the future either as their descendants or from other principles) embodied alternate, mutually exclusive scientific theories as, for example, Darwinian and Lamarckian evolutionary theories were in 19th-century biology, or deterministic and probabilistic theories were in 20th-century subatomic physics. The subparadigms representing the EDSAC and the Manchester Mark I designs could coexist peacefully, perhaps in friendly competition, as part of the larger computational paradigm.

But what of the EDVAC itself, with a theory and design that had given birth to the stored-program computer paradigm?

As it happened, the ENIAC team dispersed soon after the ENIAC was commissioned. Eckert and Mauchly left to form their own company, and Goldstine and Burks joined von Neumann at the Institute of Advanced Study at Princeton to plan, design, and build a machine along the EDVAC principles using an electrostatic storage tube as the basis for memory.
⁹⁴
This device was to be developed at the nearby RCA Laboratory by electrical engineer and inventor Jan A. Rajchman (1911–1989); the device was called the
Selectron
.

The detailed principles of the machine (the IAS computer, as it would be called)—its logical design—was delineated in a report issued in June 1946 titled
Preliminary Discussion of the Logical Design of an Electronic Computing Instrument
.
⁹⁵
Like the EDSAC and the Manchester Mark I, the IAS computer was designed as a single-address computer. The main memory was to be 4096 40-binary digits words; physically, the memory was comprised of 40 Selectron tubes each of 4096 binary digit capacity. A single word of memory would comprise identical locations on all 40 Selectron tubes.

Like the EDVAC report, this report by Burks and colleagues was highly influential in consolidating the stored-program paradigm. However, the IAS computer was not finished until 1952.
⁹⁶
The main contribution of the paper by Burks and colleagues was, thus,
to refine further the logical, conceptual, and theoretical aspects of the paradigm along a separate pathway from those of the Cambridge and Manchester projects.

As for the EDVAC, even though the original conceivers of this machine had left, work on building this machine continued at the Moore School until 1949, when it was transferred to the BRL in Aberdeen, Maryland, where it was completed and became operational in 1951. The EDVAC was discontinued in 1962.
⁹⁷
With the transfer of the EDVAC, the Moore School's place in the history of computing came to an end.

A paradigm in science is created by a process that is part social, part cultural, and part intellectual. That a paradigm entails acceptance by a community of practitioners of a system of ideas and beliefs is undoubtedly a social process. Sometimes, it is even a political process. And it is cultural insofar that it is consistent with the belief systems, the manners and mores of the practitioners. But, it is never an irrational or unintellectual process. There is reason and logic at the core of paradigm formation.

However, science also entails
communication
. Scientists need to let others
know
about their work—others outside their own laboratories. In a natural science such as physics, scientists (or natural philosophers, as they were once called) would correspond with one another. This also enabled scientists to examine, criticize, and test one another's ideas. When Sir Isaac Newton (1642–1727) wrote to fellow natural philosopher Robert Hooke (1635–1703) that “If I have seen further it is by standing on the shoulders of giants,” he was surely referring as much to his living contemporaries (such as Hooke) as to his dead predecessors.

During the 17th century, the means of communicating scientific results was greatly enriched, indeed altered, by two related events: the formation of
scientific societies
such as the Royal Society of London, founded in 1660, and the Académie des Sciences, founded in Paris in 1666; and the establishment of
scientific periodicals
, of which the oldest (and still preeminent) was the
Philosophical Transactions of the Royal Society
(1665). Scientific societies formalized the social nature of the scientific enterprise by enabling its members to meet in a common space and for a common cause. Scientific periodicals facilitated communication of results in more permanent and public form than letters between scientists.

Implicit in the founding of both societies and periodicals is the presumption that there is a shared field of interest and common agreement about the broad principles underpinning the field. They contribute to the social fabric of paradigm formation.

The 1940s witnessed the emergence of this communicative element among the small but growing community of people interested in computers. On January 7–10, 1945, Aiken organized, at Harvard University, a Symposium on Large-Scale Digital Calculating Machines, sponsored jointly by the university and the U.S. Navy's Bureau of Ordinance. The meeting was synchronized with the formal opening of Harvard's new Computation
Laboratory (later renamed the Aiken Computation Laboratory). The program included a demonstration of the Harvard Mark I, and the formal sessions included papers on such topics as the Mark I, the Bell Telephone Laboratory relay computers, delay line memory, electrostatic storage tubes, computational methods for the solution of mathematical problems, and the preparation of problems for automatic computation. The official list of registrants for the symposium numbered more than 325 people from academia, industry, and government, mostly from within the United States (but including Turing).
⁹⁸

And (as we have noted) Wilkes organized, in 1949, the Conference on High-Speed Automatic Calculating-Machines in Cambridge, England, sponsored by the University Mathematical Laboratory and the Ministry of Supply of the U.K. government.
⁹⁹
This was attended by more than 140 participants from Britain, France, Sweden, Holland, and Germany (but none from the United States). In addition to the EDSAC demonstration, there were sessions on relay computers, electrostatic storage tubes, methods of preparing problems for automatic computation, different kinds of storage technologies, the Manchester Mark I, the NPL's ACE, and circuit design.

As for societies, in 1947, some people in the United States began floating the idea of starting an association of those interested in computing machines.
¹⁰⁰
Thus began a “mimeographed campaign” for founding the world's first computer society. Originally named Eastern Association for Computing Machinery and formed at a meeting on September 15, 1947 (attended by 78 people), the organization's name was changed to Association for Computing Machinery (ACM) in January 1948.
¹⁰¹
And so, even before the successful completion of the first stored-program computers, a society was in place. Mauchly was the vice-president in the first year and president in the second.

As for periodicals devoted exclusively to computing, the process was rather more tardy. The first such periodical was
Mathematical Tables and Other Aids to Computation
(
MTAC
), a quarterly founded in 1943 and published by the National Research Council (United States).
¹⁰²
People like Wilkes and Renwick (on the EDSAC); Womersley, head of the NPL's mathematics division; Herman and Adele Goldstine (on the ENIAC); Wallace Eckert (on IBM's plugboard relay machines), Leslie Comrie (on scientific computing); and Franz Alt (on the Bell Laboratory computers) published papers in the
MTAC
. The ACM was surprisingly sluggish; its first journal,
Journal of the ACM
, a quarterly, first appeared in 1954.
¹⁰³
Clearly, the social aspect of the new paradigm—in terms of the formation of societies and periodicals—was very sparse during the 1940s, perhaps evidence that a paradigm, although born, was still very much in its earliest infancy.

  
1
. J. L. Casti. (2003).
The one true platonic heaven
(p. xii). Washington, DC: Joseph Henry Press.

  
2
. H. H. Goldstine. (1972).
The computer from Pascal to von Neumann
(p. 245). Princeton, NJ: Princeton University Press.

  
3
. Ibid., p. 179.

  
4
. Ibid., p. 182.

  
5
. Ibid., p. 183.

  
6
. Ibid., p. 182.

  
7
. Ibid., p. 186.

  
8
. S. M. Ulam. (1980). Von Neumann: The interaction of mathematics and computing. In N. Metropolis, J. Howlett & G.- C. Rota (Eds.),
A history of computing in the twentieth century
(pp. 93–99). New York: Academic Press (see especially p. 94).

  
9
. N. Stern. (1980). John von Neumann's influence on electronic digital computing, 1944–1946.
Annals of the History of Computing, 2
, 349–362.