Technological Diffusion in the Development of the Mainframe Computer and Early Semiconductors Lav Varshney Inventing an Information Society ENGRG/ECE 298 and S&TS/HIST 292 Third Essay Assignment Question 3 April 21, 2003
Technological diffusion is the spread of technology from one place to another. Technological diffusion within an industry refers to the spread of technology among different universities and companies. There are many modes by which technology can be transferred among different institutions. One method of diffusion is open publication and demonstration by the inventing body. This can take the form of journal articles, symposia, lecture series, and granting of permission to observe manufacturing processes. A second way that technological diffusion occurs is through the transfer of scientists and engineers from the inventing institution to other institutions. Other institutions acquire the scientific and technical knowledge that an employee has gained and embedded into his human capital. Reverse engineering and industrial espionage are other ways that technology can diffuse through an industry. In addition to these direct modes of technological diffusion, there are indirect modes as well. Gaining the knowledge that a technology has been created, but not knowledge of any specific technical details, can inspire researchers to create a technology with similar functionality. Similarly, patents can inspire others to develop technology with similar functionality, but different technical bases, through a process known as inventing around a patent. The technological diffusion from the center of invention to the rest of the industry is often critical in the development of a new technology. This process was equally important for the development of the mainframe computer and for the development of early semiconductors up to 1970. Even if indirect modes of technological diffusion are ignored, it can be said that almost all development in both industries either occurred at the center of invention or was a direct result of technological diffusion from the center. The Electronic Numerical Integrator and Computer (ENIAC), developed by researchers at the Moore School at the University of Pennsylvania during World War II, is considered to be the first mainframe computer. 1 The ENIAC was John Presper Eckert and John Mauchly s project, with later contributions by John von Neumann. 2 The Electronic Discrete Variable Automatic Computer (EDVAC), the 1
successor to the ENIAC, was conceived by Eckert, Mauchly, von Neumann and others as the ENIAC was being built. 3 In 1945, von Neumann wrote up the plans for the EDVAC in a report entitled A First Draft of a Report on the EDVAC, which became the, seminal document describing the stored-program computer. 4 The report described the complete logical design, but not the engineering design, of the EDVAC. Even though the report was intended only for internal circulation, computer builders across the globe came into possession of copies. 5 The report ultimately was the technological basis for the worldwide computer industry. 6 Even though the report contained no engineering details and no procedure for actually implementing the computer, the stored-program architecture was used in all subsequent mainframe computers, demonstrating the great influence of the EDVAC on future developments. Technological diffusion of the mainframe computer from the Moore School to other institutions did not stop with circulation of the EDVAC Report. There was a secret conference at MIT entirely without publicity and attendance by invitation only at which von Neumann and his Moore School colleagues disclosed the details of ENIAC and EDVAC to the nascent American computing community. 7 Later, various British and American university, government, and industrial laboratories requested permission for their researchers to spend some time working at the Moore School, so that they could learn and bring back the computer technology. 8 Due to the large demand, the Moore School organized a summer school for about forty invitation-only participants, representatives from the major research institutions involved with computing. 9 Among the most prominent attendees was Cambridge University s Maurice Wilkes, who later developed the Electronic Delay Storage Automatic Calculator (EDSAC), the first functioning stored-program computer. He had used precisely the logical architecture he learned at the Moore School, but with a different physical implementation. 10 Project Whirlwind was an aircraft simulation project that required real-time control, but even the computers developed by Project 2
Whirlwind used the Pennsylvania Technique of electronic digital calculation used in the ENIAC and in its successor, the EDVAC. 11 The influence of the Moore School was so extensive that it is possible to trace links between the Moore School and virtually all the government, university, and industrial laboratories that established computer projects in America and Britain in the late 1940s. 12 Technological diffusion from the Moore School was also very important to the development of later mainframe computers. Eckert, Mauchly, and von Neumann all left the Moore School for other pursuits, bringing with them the technical expertise that they had acquired from the ENIAC and EDVAC. Eckert and Mauchly established their Electronic Control Company, and proceeded to design and build the Universal Automatic Computer (UNIVAC). 13 The UNIVAC was the first commercially sold American electronic data processing computer, and influenced many later mainframes. Von Neumann returned to the Institute for Advanced Study (IAS), after his work on the ENIAC and EDVAC, and developed the IAS computer. 14 As an academician, von Neumann made the design freely available. Consequently, the design of this computer formed the basis for many other computers, as its technology diffused across the country. The ILLIAC at the University of Illinois and the MANIAC at Los Alamos National Laboratory were directly modeled after the IAS computer. 15 IBM s Defense Calculator, the IBM 701, was also based on the IAS computer. 16 In fact the whole line of computers that sprang from the IBM 701, including the IBM 704, were ultimately the result of technological diffusion from von Neumann and the Moore School. As seen, the entirety of mainframe computer development can be traced to technological diffusion from the Moore School. The process of invention and development for the transistor took a much more complicated path than the invention and development of the computer. Despite this fact, all major developments of the transistor and the whole semiconductor industry can be traced to technological diffusion from Bell 3
Laboratories, where the transistor was invented. John Bardeen, Walter Brattain, and William Shockley, researchers at Bell Labs, are considered to be the inventors of the transistor. Bell Labs kept the invention of the transistor a secret until the appropriate patent applications had been filed, completely stifling technological diffusion. 17 After that, however, they began an aggressive campaign of promotion and technological diffusion that spread transistor technology throughout industry, spurring major advances elsewhere. The first revelation of the transistor was made in a demonstration given to representatives of the Army, Navy, and Air Force. 18 At the same time, Shockley sent three short papers about the transistor for publication in the Physical Review. 19 The next week, a large demonstration was made to the press about the transistor. The next step in Bell Labs plans for going public was another, more technical demonstration A week after the press conference, hundreds of letters went out to scientists, engineers, and radio manufacturers, inviting them to a presentation at Murray Hill. 20 Bardeen, Brattain, and Shockley also gave numerous talks before scientists and engineers. 21 Later, a five-day symposium about the transistor was held by Bell Labs for more than three hundred scientists, engineers, military officers, and government bureaucrats from across the United States. 22 The junction transistor was also made available directly from Bell Labs for experimental purposes. 23 The transistor symposium did not mention fabrication techniques, an omission similar to the lack of engineering details in the EDVAC report. However, in the case of the transistor, the fabrication techniques were vital. Bell Labs held a second symposium, the Transistor Technology Symposium and this time around the labs was much more open about manufacturing art. It made a concerted effort to reveal everything it knew about making transistors both point-contact and junction. 24 The attendees included representatives from 26 American and 14 foreign companies, both large and small. It was at this second symposium that a great deal of technological diffusion occurred. The 4
proceedings of the symposium were published, and the book became fondly known as Mother Bell s Cookbook by members of the burgeoning industry. 25 The symposium and proceedings allowed many companies to start making transistors, and in the process advance the art of semiconductor manufacturing. Totsuko, a Japanese company got into the transistor business, and created one of the most well known applications of the transistor, the transistor radio, as well as the technique of making transistors with phosphorus doped germanium. Totsuko s engineers had gleaned the proceedings of the symposium, but there was little information about manufacturing processes or equipment. The head of the Totsuko transistor development team, Kazuo Iwama visited Western Electric s Allentown facility, and sent back details of what he had observed. 26 The process of technological diffusion from Bell to Totsuko resulted in a new kind of transistor and a major application of transistors. A few years after the invention of the transistor, members of the transistor development team began leaving Bell Labs for other opportunities. Gordon Teal left Bell Labs and joined Texas Instruments as their research director. Teal had been the main researcher into crystal growing techniques, and brought with him the advanced silicon crystal growing techniques that he had been developing. Under his direction, Texas Instruments developed the first grown-junction silicon transistor. 27 Technological diffusion from Bell Labs to Texas Instruments was key to the development of silicon transistors. William Shockley also left Bell Labs and started Shockley Semiconductor. Although the company was a complete failure, it did a masterful job of diffusing technologies developed at Bell Labs, and sparking later developments. Shockley still had influence within Bell Labs, and, was often on the phone, wheedling further details from them. 28 Occasionally Bell Labs scientists would consult for Shockley Semiconductor as well. A large group of Shockley s employees left his company, further diffusing the technology of Bell Labs, and stimulating developments in the field of semiconductors. 29 In particular, Robert Noyce 5
exemplified the development that results through technological diffusion. He left Shockley Semiconductor and joined Fairchild, and then invented the Integrated Circuit. After he left Fairchild to go to Intel, he spurred the development of the microprocessor. 30 The original technological diffusion from Bell Labs through Shockley s migration to Silicon Valley resulted in the development of the newest semiconductor advances. All important semiconductor developments can be attributed to technological diffusion from Bell Labs to the rest of the industry. The diffusion of the stored-program computer architecture idea, originated at the Moore School, led to advances in different physical implementations and also improvements to the general stored-program idea for mainframe computers. The diffusion of the transistor from Bell Labs to other companies led to advances in transistors, transistor materials, transistor manufacturing techniques, and new semiconductor devices such as the integrated circuit and the microprocessor. Technological diffusion from the center of invention to the rest of the industry was therefore equally important to both the mainframe computer and early semiconductors because all important developments in both fields were the direct result of the diffusion. 1 Ronald Kline, Electronic Computers in World War II, March 12, 2003. 2 Martin Campbell-Kelly and William Aspray, Computer, New York, pp. 85-95. 3 Martin Campbell-Kelly and William Aspray, Computer, New York, p. 92. 4 Martin Campbell-Kelly and William Aspray, Computer, New York, p. 94. 5 Martin Campbell-Kelly and William Aspray, Computer, New York, p. 95. 6 Martin Campbell-Kelly and William Aspray, Computer, New York, p. 94. 7 Martin Campbell-Kelly and William Aspray, Computer, New York, pp. 95-96. 8 Martin Campbell-Kelly and William Aspray, Computer, New York, p. 98. 9 Martin Campbell-Kelly and William Aspray, Computer, New York, p. 98. 10 Martin Campbell-Kelly and William Aspray, Computer, New York, pp. 100-104. 11 Martin Campbell-Kelly and William Aspray, Computer, New York, p. 160. 12 Martin Campbell-Kelly and William Aspray, Computer, New York, p. 99. 13 Martin Campbell-Kelly and William Aspray, Computer, New York, pp. 107-109. 14 Ronald Kline, Early Development of the Mainframe Computer in the Cold War, March 14, 2003. 15 Ronald Kline, Early Development of the Mainframe Computer in the Cold War, March 14, 2003. 16 Ronald Kline, The Rise of IBM, March 28, 2003. 17 Michael Riordan and Lillian Hoddeson, Crystal Fire, New York 18 Michael Riordan and Lillian Hoddeson, Crystal Fire, New York, pp. 161-162 19 Michael Riordan and Lillian Hoddeson, Crystal Fire, New York, p. 163 20 Michael Riordan and Lillian Hoddeson, Crystal Fire, New York, p. 165 21 Michael Riordan and Lillian Hoddeson, Crystal Fire, New York, p. 176 6
22 Michael Riordan and Lillian Hoddeson, Crystal Fire, New York, p. 195 23 Michael Riordan and Lillian Hoddeson, Crystal Fire, New York, p. 195 24 Michael Riordan and Lillian Hoddeson, Crystal Fire, New York, p. 197 25 Michael Riordan and Lillian Hoddeson, Crystal Fire, New York, p. 197 26 Michael Riordan and Lillian Hoddeson, Crystal Fire, New York, p. 215 27 Michael Riordan and Lillian Hoddeson, Crystal Fire, New York, pp. 206-208 28 Michael Riordan and Lillian Hoddeson, Crystal Fire, New York, p. 240 29 Ronald Kline, Development of the Transistor and Invention of the Integrated Circuit, April 9, 2003. 30 Ronald Kline, Development of the Transistor and Invention of the Integrated Circuit, April 9, 2003. 7