Vannevar Bush

Vannevar Bush (March 11, 1890 – June 28, 1974) was an American engineer, inventor, and science administrator.

Career

Bush received BS and MS degrees from Tufts University in 1913 and a doctorate in engineering jointly from MIT and Harvard University in 1916. In 1919 he joined the Department of Electrical Engineering at Massachusetts Institute of Technology (MIT) and in 1922 he founded the company that became the Raytheon Company. Bush became vice president of MIT and dean of the MIT School of Engineering in 1932, and in 1938 president of the Carnegie Institution of Washington.

During World War II Bush headed the U.S. Office of Scientific Research and Development (OSRD), through which almost all wartime military R&D was carried out, including important developments in radar and the initiation and early administration of the Manhattan Project. He emphasized the importance of scientific research to national security and economic well-being, and he was chiefly responsible for the movement that led to the creation of the National Science Foundation.

During his career, Bush patented numerous inventions. He is known particularly for his engineering work on analog computers, for leading the science effort during the Second World War, for proposing the National Science Foundation, and for the imaginary Memex in his widely cited essay “As we many think,” which made him folk hero in information science.

This article is concerned with Bush’s work related to documentation and information science. For the rest of his life and work see the Wikipedia [[1]] and the biography by G. Pascal Zachary.

Work Related to Documentation

In the 1930s Bush became interested in the potential of combining two technologies that were then really important: microfilm and photoelectric sensing. This led to many years of work on rapid selector machines and a popular essay, “As we may think,” in which he imagines development of that technology as an imaginary system he called a “memex.” based on that technology.

The Microfilm Rapid Selector

Microfilm was by far the leading mass storage medium. “Rapid selector” refers to a class of search engines designed to identify documents copied on to rolls of microfilm with indexing codes expressed as opaque or translucent dots adjacent to each document. Searches would usepattern recognition to detect desired index codes as the film passed between a set of thin beams of light arranged to express a coded search query and one or more photoelectric cells. A “hit” when the pattern of dots on the film exactly blocked all the light beams. Each identified frame could then be projected or copied. Unlike card files, random access was not feasible. Every search had to be a serial search of one or more rolls of film.

In 1938 Kodak and National Cash Register agreed to fund Bush to design and build a selection machine that would search 35 mm film rapidly. Bush employed students including Russell Cleven Coile and, briefly, Claude E. Shannon. The design was comparable to contemporary designs using sensors to search decks of punch cards, but potentially faster.

Bush and his staff were apparently unaware that microfilm rapid selector technology had already been developed a decade earlier by Emanuel Goldberg, head of Zeiss Ikon the leading German photographic equipment manufacturing company. Goldberg had published descriptions in English and German, demonstrated two working prototypes, and already held a patents for it in the US and other countries (Buckland 2006).

There were significant differences: First, Bush’s design based on light reaching just the right combination of photoelectric sensors, which required multiple sensors and complex circuitry, whereas Goldberg’s design used the extinction of light to a single sensor, a much simpler method. Second, the content of Bush’s records was limited to a newly-keyed six-line abstracts, whereas Goldberg’s design used images of anything that could be microfilmed. Third, Bush’s design had the film flowing continuously (and rapidly!) over the sensors whereas Goldberg ran the film through a modified movie gate, one frame at a time but at an accelerated rate, which provided greater control and stability. It stopped to project or to automatically print a copy of each document found. Bush intended to use a synchronized stroboscopic flash to make an instant copy of each found document without stopping. Bush’s design would have been more rapid – if it had worked, but the technology used was unequal to the task.

Bush persisted and after the war, a new prototype was developed by Engineering Research Associates in collaboration with Ralph Robert Shaw, head of the National Agriculture Library, with widespread publicity but this this prototype also failed. Shaw claimed that two patent searches had been done and was embarrassed when a visitor who asked to see the machine claimed to have patented the design. It was Emanuel Goldberg.

Bush’s visions were not technically practical or affordable. After a great deal of expenditure over more than twenty years, a single prototype was eventually put to practical use in the US Navy’s Bureau of Ships in 1961. By then others had abandoned rolls of film for chips of microfilm in a few very expensive machines and electronic digital technology was emerging as a better choice. Bush’s decades-long quest to make a functioning rapid selector is chronicled in detail in Colin Burke’s Information and Secrecy: Vannevar Bush, Ultra, and the Other Memex (Metuchen, NJ: Scarecrow Press, 1994). ISBN 0 8108 2783 2.

The Comparator

Before and during the Second World War Bush was involved in efforts to get the US Navy codebreaking unit to make greater use of machine-based methods. Specifically, he wanted the Navy to fund a rapid selector machine for cryptanalysis. Traditional encryption used codes with a stable equivalence between original text and encoded text, but new cipher techniques, notably the German Enigma machines, encrypt each character in a new way each time in complex ways. Because the possibilities were very large and random-seeming, traditional code-breaking techniques were decreasingly effective. As with searching indexes, machinery was needed to help detect patterns.

The US naval cryptanalysts were making progress using familiar, reliable IBM tabulators. Bush proposed use of rapid selector machines for faster analysis and he was secretly commissioned to provide technical reports detailed major components for coincidence counting whereby two texts are compared of the occurrence of the same sequence of letters. Such a comparator could be versatile, useful in principle for analysis of any encryption system. A requirement was that one text could be shifted one step relative to the other for another comparison until all positions in one text were compared with all in the other. As enciphering machines became more complex, longer and longer messages were required. This could be done by hand comparing long strips of paper, letter by letter. Holes in punch cards representing characters could be used in tabulator machines but all methods were very slow and labor intensive.

Unfortunately, in 1937 existing technology was unsuited for high-speed input, sensing, counting and recording. Meanwhile, Bush was heavily over-committed to other important activities, including important committees and becoming president of the Carnegie Institution of Washington in 1938. Suitable staff were very hard to find, especially for top-secret work. Eventually a machine designed to use holes punched across 70 mm wide rolls of opaque paper was delivered, but key parts were missing and it did not work.

In 1940 there was a renewed effort funded by the National Defense Research Committee. The paper-tape machine was modified and found some use helping to crack the Japanese cipher machine, but he was using untested technologies not made by any manufacturer and it still malfunctioned. (For a detailed accounts see Burke (1994)).

"As We May Think"

In 1939 when his first microfilm rapid selector was being developed, Bush drafted a lively essay noting both the difficulty of finding relevant documents and advances in photography, circuitry, and the use of photoelectric sensors. He speculated on what could be done with a microfilm reader and a personal collection of microfilmed documents, with a search capability and display screens built into a desk. It was published in 1945, attracted attention, and was repeatedly reprinted.

Few of the ideas were new and his proposed “trails” for knowledge organization was conventional indexing under a new name. Nevertheless, it was new and exciting for readers new to the field and it inspired several, notably Douglas C. Engelbart.

The essay quickly became a popular symbol for what could be done and for the next fifty years citing it uncritically was an expectation for writings on information retrieval or library automation even though it was not concerned with digital computers or networked services. Linda Smith has shown that authors citing it often had not read, or had not understood, the article (Linda Smith 1981 & 1991).

In a rare knowledgeable critique, Robert Fairthorne commented in 1951,

Bush’s paper was timely, even though few of his suggestions were original [… ] A large public had its imagination stirred and its eyes and purses opened. Perhaps fortunately at this stage few people noticed that the paper, and the state of mind it typified, had considered what machines could do, rather than what they should do.

Publication

”As we may think.” Atlantic Monthly 176 (1945): 101-108. [[2]] Frequently reprinted.