The Community for Technology Leaders

From the Editor's Desk

Jeffrey R. Yost, Editor in Chief

Perhaps no computer has been more aptly named than IBM's Stretch—a system designed for Los Alamos Scientific Laboratory in the second half of the 1950s to substantially extend and expand existing computing technology.

In the early computer industry, government contracts were an important source of funding to facilitate R&D. In the mid-1950s, IBM was successful in receiving the primary computer contract to design, develop, and supply AN/FSQ-7 vacuum-tube computers for the Semi-Automatic Ground Environment (computer and radar air defense system). This contract was extremely lucrative, bringing hundreds of millions of dollars to Big Blue, and it quickly eclipsed any perceived first-mover advantage Remington Rand (Sperry Rand after 1955) might have achieved by being the first to market digital computers. (Univac I was delivered in early 1951.) In reality, IBM had been investing substantially in electronics and computing research in the late 1940s and early 1950s and, in terms of capabilities, had not fallen behind.

In the mid-1950s, IBM and Sperry Rand's Sperry Univac computer division competed for the contract to supply an advanced computer for the Atomic Energy Commission's Livermore Research Laboratory. Sperry Univac won this contract—resulting in LARC—leaving IBM all the more determined to secure the next major government scientific computer contract, this time for Los Alamos lab. Project Stretch was awarded to IBM in 1956.

Stretch differed markedly from the SAGE computer development project. The contract called for a single computer of unparalleled capability, and unlike SAGE AN/FSQ-7 machines, it was a next-generation transistor-based computer. Although Stretch was the most powerful computer in the world at its completion in 1960, it did not meet original performance specifications and initially disappointed IBM's president, Thomas Watson, Jr. From a narrow project accounting sense, Stretch lost millions of dollars for the firm. However, no case is better than Stretch at emphasizing how different short-term accounting on a project can be from the longer term, less tangible, advancement of capabilities. In this latter sense, Stretch was a tremendous success.

While Stretch has hardly gone unnoticed, it remains understudied compared to System/360 and some of the other IBM systems it influenced. In the opening article of this Annals issue, leading computer scientist and author of The Mythical Man-Month, Frederick Brooks provides important insights into the fundamental technological achievements with Stretch and the legacy of these developments for IBM.

Also in this issue

Janet Abbate provides an important analysis detailing the fundamental role of the National Science Foundation in privatizing the Internet. Contrary to the relatively smooth transition emphasized at a recent NSFnet celebration, she demonstrates the different visions, conflicts, and politics involved and the resulting technical and social transformations of the Internet.

A substantial portion of the existing literature on the British history of computing focuses on key early centers (Manchester, Cambridge, and the National Physical Laboratory) or on shortcomings of government policy. Alternatively, Martyn Clark's useful study details the work of the University Grants Committee and the Advisory Committee on High-Speed Calculating Machines in extending computing technology to different universities in the UK.

Silvio Hénin uncovers the lost history of two Italian keyboard calculators—one by Milanese Luigi Torchi (1834) and another by Florentine Tito Gonnella (1858). He discusses how Torchi's calculator might have been the first to include a direct-multiplication mechanism.

Arthe Van Laer explores the European Economic Community (EEC) Commission's policies and attempts to promote and advance Europe's position within the international computer industry—unified efforts that were largely unsuccessful given various countries' unwillingness to relinquish control in this industry.

Finally, David Laws provides a rich examination of the talent and technologies of Fairchild Semiconductor and the deep legacy this firm had, technologically and organizationally, on the future of semiconductor devices.


Tim Bergin and Thomas Haigh requested the following correction/clarification regarding their article in the previous issue of the Annals (October–December 2009). Table 2 in their article incorrectly gives March 1971 as the first IMS installation. This date was taken from the Datapro Report (full citation in article) for 1974 but actually refers to IMS-2. As pointed out in Robert Patrick's article, the first installation of IMS version 1 was in 1968 as a Type II IBM program (i.e. not officially supported). In the "unbundling" announcement of 23 June 1969, IMS was classified as a program product available for a monthly charge of $600. The authors thank Bob Patrick for pointing this out. According to Burton Grad (one of the guest editors of the History of Database Management Systems special issue), IBM did not begin charging license fees until January 1970.

63 ms
(Ver 3.x)