In "As We May Think" Vannevar Bush Envisions Mechanized Information Retrieval and the Concept of Hypertext

In July 1945 American Engineer Vannevar Bush published a popular description of the aims of his Rapid Selector information retrieval machine in his article, As We May Think, that appeared in the Atlantic Monthly, Vol. 176, No. 1 (1945) 641-49. This visionary paper described the Memex,” an electromechanical microfilm machine, which Bush began developing conceptually in 1938. As conceived, the Memex was capable of making permanent associative links in information. Features of the hypothetical Memex foreshadowed aspects of the personal computer and hyperlinks on the Internet. Bush was unable to patent his Rapid Selector because of its similarity to aspects of prior work on electronic document retrieval previously patented by Emanuel Goldberg.

On September 10, 1945 Bush published a condensed, illustrated version of "As We May Think" in Life magazine, 19, No. 11 (1945) 112-114, 116, 121, 123-24. Life's editors added the following subtitle: "A Top U.S. Scientist Foresees a Possible Future World in Which Man-Made Machines Will Start to Think." They also replaced the Atlantic Monthly's numbered sections with headings, and added illustrations of the "cyclops camera," the "supersecretary" and the "Memex" microfilm machine in the form of a desk. This was the first published illustration of what the Memex might look like. In From Memex to Hypertext: Vannever Bush and the Mind's Machine (1991) James Nyce and Paul Kahn published a version of "As We May Think" that shows the differences between the two different versions of Bush's essay published in 1945. Nyce and Kahn also developed a brief animated film showing how the Memex might have operated. Bush, himself, never seems to have developed a working version of the machine, though his group worked on a prototype.

In August 1947 Ralph R. Shaw, Director of Libraries for the U.S. Department of Agriculture, in collaboration with Engineering Research Associates of St. Paul, Minnesota, using funds provided by the Office of Technical Services of the Department of Commerce, began the development of the Rapid Selector machine for the electronic searching of information recorded in reels of microfilm. Shaw's device incorporated technology developed by Emanuel Goldberg in 1928-1931, and by Bush starting in 1938. Shaw's Rapid Selector was an attempt to realize goals described in Bush's 1945 publication, As We May Think. Shaw's machine

"was based on the earlier prototype developed from 1938 to 1940 by a team at MIT under Bush's direction. The project manager for the Bush prototype was John H. Howard and the research assistants were Russell C. Coile, John Coombs, Claude Shannon, and Lawrence Steinhardt. Eastman Kodak and National Cash Register each provided $10,000 funding. The project's objective was to develop, within two years, a prototype machine capable of selecting microfilmed business records from microfilm rapidly: A microfilm rapid selector. Bush's selector was indeed rapid because it took advantage of two new developments: Improved photoelectric cell technology; and the stroboscopic lamp pioneered by his colleague Harold E. Edgerton. By creating a bright flash of light lasting only one-millionth of a second, the stroboscopic lamp made it possible to copy a selected microfilm image "on the fly," without stopping the film (and the search) to make a copy. The Bush microfilm selector was never used operationally, except that it seems to have been used for cryptanalysis: It was, after all, designed to be effective at identifying (selecting) every occurrence of a specified code" (http://people.ischool.berkeley.edu/~buckland/goldbush.html, accessed 02-20-2012).

Until December 2013 I was never able to find any truly detailed information on the version of the Rapid Selector built after World War II. I did learn that in 1951 physicist Louis N. Ridenour, librarian, inventor and publisher Ralph R. Shaw, and physicist Albert G. Hill published a thin volume entitled Bibliography in an Age of Science. This book included three lectures delivered at the University of Illinois the previous year, one of which described the Rapid Selector which had been built under Shaw's supervision, asserting that it did operate. This work I came across several years after publishing Origins of Cyberspace and From Gutenberg to the Internet. Shaw's chapter included illustrations on pp. 60-61 of the Rapid Selector prototype which was in operation at this time. This machine stored 72,000 frames of information on a 2,000 foot reel of film. The prototype could search through data at the rate of 78,000 "codes per minute." "Improvement of this searching speed to 120,000 codes per minute is now in sight."

However, further information about Shaw's Rapid Selector in use eluded me for several more years, and I wondered whether it really operated like Shaw claimed. In December 2011 I acquired a copy of Roberto Busa's Varia specimina concordantiarum (Milano, 1951). This bi-lingual work with texts in English and Italian was subtitled, "A First Example of Word Index Automatically Compiled and Printed by IBM Punched Card Machines." Before deciding to employ IBM electric punched card tabulators to produce his concordance Father Busa took the opportunity to see the Rapid Selector in operation at the Department of Agriculture in Washington, D.C. He wrote that he was able to see it operating in November 1949, and that:

"Its principal feature is the whirlwiind speed with which it explores the reels of microfilm— 10,000 photograms per minute— and instantaneously rephotographs on another microfilm strip all and only those photograms which bear a determined item.

"I shall not give a detailed description because I thought not suitable to apply this system to the composition of concordances; I will only say that, besides not allowing automatic printing of the concordances, such as can be done with the system hereunder, the rapid selector necessitates on the one hand that all the cards, to be made from the sorted microfilm, be of photosensitive paper, and on the other hand all the different words and forms of each word be previously coded, for the entire text must be translated into numerical symbols by hand" (Busa, op cit, 22.)

Then in December 2013 I discovered that the Hathi Trust had digitized and made available Report for the Microfilm Rapid Selector. Contract Cac-47-24. 20 June 1949 published by Engineering Research Associates. This 29-page report with 11 illustrations provided all the detail that one might desire concerning the design and characteristics of the machine, without providing information concerning its efficiency or utility. From the Foreword I quote:

"The incentive for this development arose form a basic need for a more efficient mechanism for organization and dissemination of scientific information. The facilities of the Department of Agriculture Library and the specialized experience of its Librarian and staff fitted the requirement for a testing agency equipped to handle varied categories of technical data in large volumes. Hence, the project developed cooperatively between the Department of Commerce and Agriculture and Engineering Research Associates, Inc.. Specifications for a system meeting the requirements were drawn up by ERA in August, 1947, under the title "General Description and Proposed Technical Specifications for Microfilm Selector'. In general, the machine developed meets the goals set up in that document.

"In brief, the system provides for microfilm storage of abstracts and corresponding code areas by which each abstract may be associated with six different fields of interest. The Microfilm Selector scans the film at the rate of more that 10,000 frames per minute which may correspond to as many as 60,000 subjects per minute. It selects all abstracts which are associated with an interest category specified by the operator, and recopies the selected items on a separate roll of 35mm film by the use of high-speed photoflash techniques. (p. ii)

"Report-Microfilm Rapid Selector

 "This machine is similar in basic concept to a prior experimental development known as the Bush Rapid Selector which was announced by the Massachusetts Institute of Technology in 1940. Several members of the ERA staff were engaged in this earlier development, and it was possible to utilize their experience as a starting point in the present project.

"The intent of the original contract (which was to have been concluded by 30 June 1948) was toward the construction of a pilot machine to demonstrate the principles involved. In recognition of the immediate needs of the Department of Commerce Library, however, and as evidence of ERA's special interest in the development, it was decided to continue the work beyond the term and scope of the original contract, at the Contractor's own expense. Thus it was possible to complete a practical working machine which would fully demonstrate the possibilities of the system. The resulting Microfilm Selector (completed 25 January 1949) goes well beyond the requirements of an experimental model; it is, in fact, a close approach to an engineering model.

"At the present time, the Microfilm Selector has not yet been subjected to thorough performance tests. On the basis of preliminary tests, howover, it is considered that all of the important components have been proved fundamentally sound. It would be very surprising if the intitial period of use did not reveal some weaknesses in design and construction, but there is every reason to believe that such faults will be minor in character, and capable of correction without extensive rebuilding or further development." (p. iii).

Hook & Norman, Origins of Cyberspace (2002) no. 244, and other entries.