This week, the Wall Street Journal‘s L. Gordon Crovitz wrote a strange column decrying the claim that the U.S. government created the internet as an “urban myth.” Crovitz was quickly debunked by myself and many others. The truth is even deeper. The U.S. high-technology industry, and the computer industry in particular, owe their existence to the government.
The computer was a war baby. The U.S. Army and Navy, needing fast ways to solve the differential equations they needed to aim long-range guns, funded major research projects. The Army effort, at the University of Pennsylvania, yielded ENIAC. The top engineers on the project, J. Presper Eckert and John Mauchly started their own company (soon sold to Sperry Rand) which built the first “private” commercial computer, UNIVAC.
In addition to being based on ENIAC, UNIVAC incorporated important advances from John von Neumann’s group at the Institute for Advanced Studies as well as work from Maurice Wilkes’ group at the University of Cambridge and Alan Turing’s cryptographic computing efforts at Bletchly Park. All of these efforts were funded by the U.S. and British governments. And the buyer of the first UNIVAC system: The U.S. Census Bureau.
IBM’s work with the Navy was less groundbreaking, but the company moved into electronic computers in a big way after the war. A declassified history of computing at the National Security Agency shows just how deeply the NSA was involved in the design and development of IBM’s 700-series computers. NSA also paid for development of the first supercomputer, the IBM 7030 Stretch, in the late 1950s and its successor, HARVEST. It really wasn’t until the 1960s that commercial demand for computers eclipsed government purchases and the design of those business systems, such as the IBM 1401, 7090, and System/360, was heavily influenced by the work that had been done on government contracts.
Government influence was pervasive elsewhere. Jay Forrester’s Whirlwind, the first real-time, distributed system, was originally a Navy project but evolved into the Air Force’s SAGE strategic defense control system. COBOL, the most widely used business programming language for decades, was developed in a private-public partnership led by the Commerce Dept.
It is doubtful that modern supercomputers would exist had government not financed their development down to the present. IBM now sells Blue Gene systems commercially, but the initial units were built for the Department of Energy’s national labs. The 10 most powerful supercomputers in the world are all at government-owned or government-sponsored labs; No. 1 is an IBM Blue Gene/Q system at the Lawrence Livermore National Lab. (And those are just the ones we know about; the NSA does not release any information on its advanced systems.)
None of this is to say that private companies and private investment didn’t play a huge role in the development of the industry. While the government continues to play an important part in funding some of the most advanced developments, its contribution is many orders of magnitude smaller than in the early days. But the ideologically-driven insistance that the government’s role in the creation of the industry was small or non-existent is simply denying a n important and well documented history.
Although computing was accelerated by the war effort, Professor Atanasoff of Iowa State University and his assistant Clifford Berry were already working on the Atanasoff-Berry Computer (ABC) in 1939 before they were called to serve the War Department. The ABC was designed to solve linear equations which it did successfully in 1942. It was distinguished from the ENIAC only through the lack of stored changeable program. It was the first computer to perform all calculations via electronics rather than gears, ratchets, or switches. It was also the first computer that separated computation and memory.
There is very little solid evidence that the ABC actually solved any linear equations. One secretary interviewed decades later thought that it did, but Atanasof testified at the ENIAC trial (much closer to the actual events) that it did not. The video demos of the modern machine based on the ABC design (but debugged) show that it took more time to solve a two-equation system than an alert high-school freshman could do with pencil and paper. Even worse, results from one stage of the computation had to be carried by hand from one part of the machine to another. Perhaps this is what is meant by the notion that it “separated computation and memory.” No gears or ratchets, maybe, but a hand transfer! ABC was designed to solve only one kind of problem, whereas ENIAC was general-purpose, automatic once it was programmed, faster than any previous machine (except perhaps for the top-secret Collossus developed for the British government). It was immediately put to work solving a frightfully complex problem–whether a hydogen bomb was feasible–that could not have been attempted by previously-known methods.
I don’t deny ENIAC was a qualitative advance. I’m saying the ABC happened without the action of the Federal Government beyond the original Land Grant to the University.
Was Research Corporation administered by the Smithsonian in the 1930s?
From the Smithsonian Archives:
“The Research Corporation was created under the laws of New York in 1912, having as its chief asset certain patents assigned to it by Frederic G. Cottrell. Cottrell, a chemist, had invented devices used in the electrostatic precipitation of pollutants. He wished to support the research of promising young scientists in various fields and approached the Smithsonian, asking it to act as administrator of his proposal. For various reasons the Smithsonian declined, and a non-profit corporation was created in its place. The Secretary of the Smithsonian has been an ex officio member of its board of directors from the corporation’s inception.”
The ABC NEVER did anything. Including work. I’m sorry. That’s not quite true. It did get used as a pawn to convince a judge with no knowledge of technology that a patent should be invalidated. I know, an absurd statement. After all, who could profit from a patent being invalidated. And what kind of a judge in that time period didn’t understand the technology he was ruling on.
It was distinguished in other ways, such as being partly mechanical and not working. Atanasoff invited Mauchly to look at it, and he told him the truth, that it wasn’t worth his time.
There is no evidence to support the strange statement that “UNIVAC incorporated important advances from John von Neumann’s group at the Institute for Advanced Studies as well as work from Maurice Wilkes’ group at the University of Cambridge and Alan Turing’s cryptographic computing efforts at Bletchly Park.” The IAS machine was a totally separate effort, although both UNIVAC and the IAS were based on the EDVAC machine which began to be conceived by the ENIAC group before von Neumann became a consultant with them. Wilkes based his machine on von Neumann’s “First Draft” report which in turn was based on the EDVAC work of the ENIAC inventors. The Bletchly Park work was secret until decades after UNIVAC. All of these projects received some degree of government support and the government deserves much credit for pioneering computer development, but the cross-pollination you assert is a fabrication.
Thanks for your clarification.
I wasn’t trying to write a definitive history of the computer, just highlight government’s role. ENIAC was not quite a true stored-program computer. That honor seems to belong to either EDVAC or EDSAC; no one seems quite sure. But the incorporation of true stored programming was an important UNIVAC improvement over ENIAC. And while the existence of Colossus remained classified until 1976, there sure was cross-fertilization of ideas between Bletchly Park and Wilke’s Cambridge lab, especially right after the war. There was an awful lot going on in a lot of places during and just after the war and I don’t think we’ll ever get the history completely untangled, even now that nearly all the records (I hope) have been declassified. I think the only guy who did important work in the area who didn’t get support from either the U.S. or British government was Konrad Zuse.
As for Atanasoff, not to detract from the several innovations which he could rightfully claim, but his machine was never capable of doing the job for which it was intended (solving large systems of simultaneous linear equations), and when he was hired by the Navy Ordnance Laboratory to build a computer, his project was cancelled–on von Neumann’s recommendation no less–for total absence of progress. See Calvin Mooers’ article on the subject for the Annals. In any case, Atanasoff, too, benefited from government grants and contracts, which supports the thesis of this article.
I think you could fairly credit the government for advances related to the integrated circuit and the miniaturization surrounding the space program. But the contribution of the government was in providing huge sacks of money via grants and defense contracts. They do deserve some credit for managing some of those early programs quite well.
Oh good. SO we can agree. Spending by the department of defense, and spending related to war produce results. Good. I’m glad we all see what happened here.