Published online by Cambridge University Press: 13 December 2011
In 1969, when few commercial communications networks existed, a U.S. Defense Department research agency created an experimental system that would eventually become the Internet. Driven by both research and military considerations, the designers of the Internet created a complex, robust, and flexible system that differed in significant ways from contemporary commercial communications networks. In the 1970s and 1980s, computer manufacturers (mainly based in the United States) and telecommunications carriers (mainly operating outside the United States) vied to offer commercial network products and services, but no single company or technology was able to dominate the market, in part because computer users preferred the type of nonproprietary technical standards used in the Internet. In the 1980s, the National Science Foundation took over operation of the Internet, and in the 1990s the NSF turned over the network to privatesector operators. While the Internet has rapidly increased in scale under commercial ownership, the technology also continues to reflect the systems research origins.
1 For discussions of interchangeable parts, see Hounshell, David A., From the American System to Mass Production, 1800–1932 (Baltimore, 1984)Google Scholar; and Smith, Merritt Roe, Harpers Ferry Armory and the New Technology (Ithaca, N.Y., 1977Google Scholar). Studies of atomic weapons are numerous; see Rhodes, Richard, The Making of the Atomic Bomb (New York, 1986Google Scholar) for one detailed and readable account. The collection of essays in Smith, Merritt Roe, ed.. Military Enterprise and Technological Change (Cambridge, Mass., 1985Google Scholar) includes some of the major theorists on the military shaping of technology. Thomas P. Hughes discusses the more general concept of differences in “technological style” between organizations in Networks of Power (Baltimore, 1983Google Scholar).
2 See, for example, Norberg, Arthur L. and O'Neill, Judy E., Transforming Computer Technology: Information Processing for the Pentagon, 1962–1986 (Baltimore, 1996Google Scholar); Flamm, Kenneth, Creating the Computer: Government, Industry, and High Technology (Washington, D.C., 1988Google Scholar); MacKenzie, Donald, “The Influence of the Los Alamos and Livermore National Laboratories on the Development of Supercomputing,” Annals of the History of Computing 13 (1991Google Scholar). ARPA changed its name to DARPA (Defense Advanced Research Projects Agency) in 1972, and then changed it back to ARPA in 1993. I will use “ARPA” throughout to avoid confusion.
3 See David F. Noble, “Command Performance: A Perspective on the Social and Economic Consequences of Military Enterprise,” in Merritt Roe Smith, Military Enterprise and Technological Change, for a critical view of the military influence on technological design. See Hughes, Thomas P., Rescuing Prometheus (New York, 1998Google Scholar), for an analysis of nontraditional approaches to managing large technological systems, including the ARPANET.
6 The idea of packet switching was proposed by Paul Baran at RAND in 1964 and, independently, by Donald Davies at the British National Physical Laboratory in 1965.
7 An article on “The ARPA Network Design Decisions” by the project's main contractor, Bolt Beranek & Newman, acknowledged: “The costs of [reliability measures] are multiple: they add to the complexity of the system, and may degrade its performance, in addition to representing a dollars cost. In general, the communications subnetwork becomes more costly as these measures are improved.” See John M. McQuillan and David C. Walden, “The ARPA Network Design Decisions,” Computer Networks (1977).
8 Bolt Beranek & Newman subsequently built a series of networks for military intelligence agencies (Eric Elsam, telephone conversation with the author, 22 July 1997). Vint Cerf, one of main creators of the Internet, wrote frequently about the technology's military utility. See Cerf, Vinton G., “DARPA Activities in Packet Network Interconnection,” in Beauchamp, K. G., Interlinking of Computer Networks (Dordrecht, the Netherlands, 1979Google Scholar); Cerf, Vinton G. and Cain, Edward, “The DoD Internet Architecture Model,” Computer Networks 7 (1983Google Scholar); Cerf, Vinton G. and Lyons, Robert E., “Military Requirements for Packet-Switched Networks and Their Implications for Protocol Standardization,” Computer Networks 1 (1983).Google Scholar
9 Leonard Kleinrock, interview hy Judy O'Neill, 3 April 1990, Charles Babbage Institute archive; Schelonka, Edward P., “Resource Sharing with Arpanet,” in Abrams, Marshall D., Blanc, Robert P., and Cotton, Ira W., eds., Computer Networks (New York, 1976), 5–19.Google Scholar
10 Franklin F. Kuo, “Public Policy Issues Concerning ARPANET,” IEEE Fourth Data Communications Symposium 1975, 3–15.
11 Landweber, Lawrence H. and Solomon, Marvin H., “Use of Multiple Networks in CSNET,” IEEE Computing Society International Conference (Compcon), (New York, 1982), 401Google Scholar.
12 Davies, Donald W., “Proposal for a Digital Communication Network,” unpublished memorandum, National Physical Laboratory (NPL) 1966, 3Google Scholar; archived at the National Archive for the History of Computing (NAHC), Manchester, U.K.
13 Davies, Donald W., “Proposal for the Development of a National Communication Service for On-Line Data Processing,” unpublished memorandum, NPL, 1965Google Scholar; archived at NAHC.
14 Davies, Donald W., “A Computer Network for NPL,” unpublished memorandum, NPL, 1966, 2Google Scholar; archived at NAHC.
15 “NPL Communications System: Facilities Offered by Low and Medium Speed Terminal Hardware,” unpublished memorandum, NPL, 1967Google Scholar; archived at NAHC.
16 Tymshare's main computer, called the XDS 940, was actually based on an ARPA-sponsored experimental computer built at the University of California at Berkeley. This does not necessarily indicate a close connection between ARPA and Tymshare, but rather reflects ARPA's dominant role in funding computer science research during the 1960s.
17 J. Rinde, “TYMNET I: An Alternative to Packet Technology,” Third International Conference on Computer Communication, 1976, 272.
18 Gold, Michael M. and Selwyn, Lee L., “Real Time Computer Communications and the Public Interest,” AFIPS Fall Joint Computer Conference 33 (1968), 1474Google Scholar.
19 J. Rinde, “TYMNET I: An Alternative to Packet Technology.”
21 Paul L. Peck, “Effective Corporate Networking, Organization, and Standardization,” AFIPS Fall Joint Computer Conference (1971), 561–9.
22 Bell, C. Gordon and Newell, Allen, eds., Computer Structures: Readings and Examples (New York, 1971), 507Google Scholar.
23 Phil Hirsch, “SITA: Rating a Packet-Switched Network,” Datamation (March 1974).
24 Frisch, Ivan T. and Frank, Howard, “Computer Communications—How We Got Where We Are,” AFIPS National Computer Conference 44 (1975)Google Scholar.
25 Electronic mail, which already existed on time-sharing computers, was added to the ARPANET protocols in 1973. The World Wide Web was created by Tim Berners-Lee at CERN in 1990.
26 Lawrence G. Roberts, “Data by the Packet,” IEEE Spectrum (February 1974): 46. These cost savings were probably overstated, since ARPA would not necessarily have bought all of those extra computers for its contractors.
27 In fact, many contractors expressed a preference to have “their own machines” when the ARPANET idea was first presented by ARPA. See Abbate, Janet, Inventing the Internet (Cambridge, Mass., 1999), 50Google Scholar. A historical irony is that, since the price of computer power was rapidly falling, in a few years it might actually have been more cost-effective for ARPA to provide the additional computers than to build a network to share its existing machines. The window of opportunity between the time the ARPANET became technically feasible and when it ceased to make sense in purely economic terms may have been rather small. Of course, cost savings was only one goal of the ARPANET and probably meant more to ARPA's congressional sponsors than to the ARPANET community itself.
28 Lawrence G. Roberts, “Multiple Computer Networks and Intercomputer Communication,” ACM Symposium on Operating System Principles, 1967, 1.
29 A number of early nonprofit networks were built for this same reason: sharing computers without incurring excessive communications costs. A typical project was the 1966 Triangle Universities Computation Center Network, a cooperative project of Duke University, North Carolina State University, and University of North Carolina at Chapel Hill. This connected IBM 360 computers at the three computing centers, using leased lines that allowed users to transfer data between the three sites. See David J. Farber, “Networks: An Introduction,” Datamation (April 1972): 39. ARPA funded a similar network connecting IBM 360 computers at Carnegie-Mellon and Princeton. See Carr, C. Stephen, Crocker, Stephen D., and Cerf, Vinton G., “Host-Host Communication Protocol in the ARPA Network,” Proceedings AFIPS Spring Joint Computer Conference 36 (May 1970): 5–7, 77Google Scholar; Norberg and O'Neill, Transforming Computer Technology, 158. Dartmouth College had developed a network to connect student terminals to its time-sharing computers in 1964; with the help of an NSF grant, it expanded the network in 1967 to connect schools in Massachusetts and New Jersey. In 1965 General Electric's Information Systems division, which had provided equipment for Dartmouth, used a version of the Dartmouth system to build a network to support its own commercial time-sharing service, which served twenty-five U.S. cities as well as sites in Canada, Mexico, Britain, the Netherlands, and France. See Frisch, Ivan T. and Frank, Howard, “Computer Communications—How We Got Where We Are,” AFIPS National Computer Conference 44 (1975)Google Scholar.
30 Foy, Nancy, “Britain's Real-Time Club,” Annals of the History of Computing 8 (1986): 370Google Scholar.
32 Lawrence Roberts, “The Evolution of Packet Switching,” 1310. Telenet was acquired by GTE in 1979 and then by Sprint in 1986.
34 Henry C. Lucas Jr., Hugues Levecq, Robert Kraut, and Lynn Streeter, “France's Grass-Roots Data Net,” IEEE Spectrum (November 1995): 71–79.
35 Phil Hirsch, “Canada Network Won't Take SDLC Protocol,” Datamation (March 1975): 121–39.
36 See Janet Abbate, Inventing the Internet, 106–10, for an account of the rise of ARPANET e-mail.
37 MCI Mail was initiated by Bob Harchasik, who had come to MCI after serving as president of Tymnet, and the project was managed by Vint Cerf, who had led much of the Internet development at ARPA. Vint Cerf, interview by Judy O'Neill, 24 April 1990, Charles Babbage Institute archive.
38 Cantelon, Philip L., The History of MCI: The Early Years, 1968–1988 (Dallas, 1993), 366, 395Google Scholar.
40 L. David Passmore, “The Networking Standards Collision,” Datamation (1 Feb. 1985): 948.
42 L. David Passmore, “The Networking Standards Collision,” 105.
43 Matsushita's Japan Victor Company (JVC) brought out VHS in 1976; the competing Betamax system had been introduced by Sony in 1975.
44 The battle between public and private networking standards is described in more detail in Abbate, Janet, “The Internet Challenge: Conflict and Compromise in Computer Networking,” in Summerton, Jane, ed., Changing Large Technical Systems (Boulder, Colo., 1994Google Scholar); and in Abbate, Inventing the Internet, ch. 5.
45 Paul L. Peck, “Effective Corporate Networking, Organization, and Standardization,” 564.
48 See Band, Jonathan, “Competing Definitions of ‘Openness’ on the NII,” in Kahin, Brian and Abbate, Janet, eds., Standards Policy for Information Infrastructure (Cambridge, 1995Google Scholar), for a discussion of disagreements over the meaning of openness.
50 In contrast, later derivatives of UNIX have been in the public domain, most notably the Linux operating system. Linux was initially developed by Finnish programmer Linus Torvalds and was later adopted by the Free Software Foundation, based in Cambridge, Mass., and headed by former MIT computer scientist Richard Stallman.
51 ISO/TC 97/SC 16, “Provisional Model of Open-Systems Architecture,” ACM Computer Communication Review 8 (1978), 50Google Scholar.
52 Figallo, Cliff, “The WELL: A Regionally Based On-Line Community on the Internet,” in Kahin, Brian and Keller, James, eds., Public Access to the Internet (Cambridge, 1995), 51Google Scholar.
53 Quarterman, John S., The Matrix: Computer Networks and Conferencing Systems Worldwide (Burlington, Mass., 1990), 230–9, 243Google Scholar.
55 Heidi B. Heiden, telephone conversation with author, 30 July 1997.
56 Charles Hornig, “A Standard for the Transmission of IP Datagrams over Ethernet Net-works,” RFC 894, 1984.
57 John S. Quarterman, The Matrix, 301–38. The seven new networks were BARRNet (in the San Francisco Bay area); MIDNet (in the Midwest); NorthWestNet; NYSERNet (in the New York area); Sesquinet (in Texas); SURAnet (in the Southeast); and WESTNET (in the Rocky Mountain region).
58 Stephen Wolff, “Merit Retires NSFNET Backbone Service” (e-mail to com-priv and farnet members, 1991, available at nic.merit.edu/cise/pdp.txt).
59 Stephen Wolff, “Merit Retires NSFNET Backbone Service.” In fact, the granting of NSF contracts to IBM and MCI was somewhat controversial within the Internet community; NSF manager Stephen Wolff noted that there was “widespread skepticism” about this award, since neither of the companies had any experience with TCP/IP (ibid).
60 Merit, “NSFNET: A Partnership for High-Speed Networking” (http://www.merit.edu/merit/archive/nsfnet/final.report). Thanks to an anonymous reviewer for this reference.
61 The 1992 version of the NSF's Acceptable Use Policy is reprinted in Krol, Ed, The Whole Internet User's Guide & Catalog (Sebastopol, Calif, 1992), 353–4Google Scholar.
62 PSINet corporate web page, http://www.psi.net/profile/history.html.
63 PSINet corporate web page.
64 MERIT, “Merit Retires NSFNET Backbone Service,” nic.merit.edu/nsfnet/news. releases/nsfhet.retired, 1995; Stephen Wolff, “Merit Retires NSFNET Backbone Service,” e-mail to com-priv and farnet members, 1991, available at nic.merit.edu/cise/pdp.txt.
65 MERIT, “Merit Retires NSFNET Backbone Service.”