27 Huxley, New Bottles for New Wine, 1318.
28 The computer scientist Bill Joy wrote a long piece in Wired warning of the outcomes of research conducted without ethical safeguards and the dangers of eugenics in the past, "Why the Future Doesn"t Need Us," Wired 8.4 [April 2000], (accessed 27 June 2005).
29 Vinge, "The Coming Technological Singularity."
30 Eugen Leitl, e-mail to Silk-list mailing list, 16 May 2000, Eugen Leitl, e-mail to Silk-list mailing list, 7 August 2000, Friedrich A. Hayek, Law, Legislation and Liberty, 1:20.
3. The Movement
1 For instance, Richard Stallman writes, "The Free Software movement and the Open Source movement are like two political camps within the free software community. Radical groups in the 1960s developed a reputation for factionalism: organizations split because of disagreements on details of strategy, and then treated each other as enemies. Or at least, such is the image people have of them, whether or not it was true. The relationship between the Free Software movement and the Open Source movement is just the opposite of that picture. We disagree on the basic principles, but agree more or less on the practical recommendations. So we can and do work together on many specific projects. We don"t think of the Open Source movement as an enemy. The enemy is proprietary software" ("Why "Free Software" Is Better than "Open Source,"" GNU"s Not Unix! [accessed 9 July 2006]). By contrast, the Open Source Initiative characterizes the relationship as follows: "How is "open source" related to "free software"? The Open Source Initiative is a marketing program for free software. It"s a pitch for "free software" because it works, not because it"s the only right thing to do. We"re selling freedom on its merits" ( [accessed 9 July 2006]). There are a large number of definitions of Free Software: canonical definitions include Richard Stallman"s writings on the Free Software Foundation"s Web site, www.fsf.org, including the "Free Software Definition" and "Confusing Words and Phrases that Are Worth Avoiding." From the Open Source side there is the "Open Source Definition" ( Unaffiliated definitions can be found at www.freedomdefined.org.
2 Moody, Rebel Code, 193.
3 Frank Hecker, quoted in Hamerly and Paquin, "Freeing the Source," 198.
4 See Moody, Rebel Code, chap. 11, for a more detailed version of the story.
5 Bruce Perens, "The Open Source Definition," 184.
6 Steven Weber, The Success of Open Source.
7 "Netscape Announces Plans to Make Next-Generation Communicator Source Code Available Free on the Net," Netscape press release, 22 January 1998, (accessed 25 Sept 2007).
8 On the history of software development methodologies, see Mahoney, "The Histories of Computing(s)" and "The Roots of Software Engineering."
9 Especially good descriptions of what this cycle is like can be found in Ullman, Close to the Machine and The Bug.
10 Jamie Zawinski, "resignation and postmortem," 31 March 1999, Ibid.
12 Ibid.
13 Ibid.
14 Ibid.
15 "Open Source Pioneers Meet in Historic Summit," press release, 14 April 1998, O"Reilly Press, See Hamerly and Paquin, "Freeing the Source." The story is elegantly related in Moody, Rebel Code, 182204. Raymond gives Christine Petersen of the Foresight Inst.i.tute credit for the term open source.
17 From Raymond, The Cathedral and the Bazaar. The changelog is available online only: ~esr/writings/cathedral-bazaar/cathedral-bazaar/.
18 Josh McHugh, "For the Love of Hacking," Forbes, 10 August 1998, 94100.
19 On social movements-the closest a.n.a.log, developed long ago-see Gerlach and Hine, People, Power, Change, and Freeman and Johnson, Waves of Protest. However, the Free Software and Open Source Movements do not have "causes" of the kind that conventional movements do, other than the perpetuation of Free and Open Source Software (see Coleman, "Political Agnosticism"; Chan, "Coding Free Software"). Similarly, there is no single development methodology that would cover only Open Source. Advocates of Open Source are all too willing to exclude those individuals or organizations who follow the same "development methodology" but do not use a Free Software license-such as Microsoft"s oft-mocked "shared-source" program. The list of licenses approved by both the Free Software Foundation and the Open Source Initiative is substantially the same. Further, the Debian Free Software Guidelines and the "Open Source Definition" are almost identical (compare with [both accessed 30 June 2006]).
20 It is, in the terms of Actor Network Theory, a process of "enrollment" in which partic.i.p.ants find ways to rhetorically align-and to disalign-their interests. It does not const.i.tute the substance of their interest, however. See Latour, Science in Action; Callon, "Some Elements of a Sociology of Translation."
21 Coleman, "Political Agnosticism."
22 See, respectively, Raymond, The Cathedral and the Bazaar, and Williams, Free as in Freedom.
23 For example, Castells, The Internet Galaxy, and Weber, The Success of Open Source both tell versions of the same story of origins and development.
4. Sharing Source Code
1 "Sharing" source code is not the only kind of sharing among geeks (e.g., informal sharing to communicate ideas), and UNIX is not the only shared software. Other examples that exhibit this kind of proliferation (e.g., the LISP programming language, the TeX text-formatting system) are as ubiquitous as UNIX today. The inverse of my argument here is that selling produces a different kind of order: many products that existed in much larger numbers than UNIX have since disappeared because they were never ported or forked; they are now part of dead-computer museums and collections, if they have survived at all.
2 The story of UNIX has not been told, and yet it has been told hundreds of thousands of times. Every hacker, programmer, computer scientist, and geek tells a version of UNIX history-a usable past. Thus, the sources for this chapter include these stories, heard and recorded throughout my fieldwork, but also easily accessible in academic work on Free Software, which enthusiastically partic.i.p.ates in this potted-history retailing. See, for example, Steven Weber, The Success of Open Source; Castells, The Internet Galaxy; Himanen, The Hacker Ethic; Benkler, The Wealth of Networks. To date there is but one detailed history of UNIX-A Quarter Century of UNIX, by Peter Salus-which I rely on extensively. Matt Ratto"s dissertation, "The Pressure of Openness," also contains an excellent a.n.a.lytic history of the events told in this chapter.
3 The intersection of UNIX and TCP/IP occurred around 1980 and led to the famous switch from the Network Control Protocol (NCP) to the Transmission Control Protocol/Internet Protocol that occurred on 1 January 1983 (see Salus, Casting the Net).
4 Light, "When Computers Were Women"; Grier, When Computers Were Human.
5 There is a large and growing scholarly history of software: Wexelblat, History of Programming Languages and Bergin and Gibson, History of Programming Languages 2 are collected papers by historians and partic.i.p.ants. Key works in history include Campbell-Kelly, From Airline Reservations to Sonic the Hedgehog; Akera and Nebeker, From 0 to 1; Has.h.a.gen, Keil-Slawik, and Norberg, History of Computing-Software Issues; Donald A. MacKenzie, Mechanizing Proof. Michael Mahoney has written by far the most about the early history of software; his relevant works include "The Roots of Software Engineering," "The Structures of Computation," "In Our Own Image," and "Finding a History for Software Engineering." On UNIX in particular, there is shockingly little historical work. Martin Campbell-Kelly and William Aspray devote a mere two pages in their general history Computer. As early as 1978, Ken Thompson and Dennis Ritchie were reflecting on the "history" of UNIX in "The UNIX Time-Sharing System: A Retrospective." Ritchie maintains a Web site that contains a valuable collection of early doc.u.ments and his own reminiscences ( Mahoney has also conducted interviews with the main partic.i.p.ants in the development of UNIX at Bell Labs. These interviews have not been published anywhere, but are drawn on as background in this chapter (interviews are in Mahoney"s personal files).
6 Turing, "On Computable Numbers." See also Davis, Engines of Logic, for a basic explanation.
7 Sharing programs makes sense in this period only in terms of user groups such as SHARE (IBM) and USE (DEC). These groups were indeed sharing source code and sharing programs they had written (see Akera, "Volunteerism and the Fruits of Collaboration"), but they were const.i.tuted around specific machines and manufacturers; brand loyalty and customization were familiar pursuits, but sharing source code across dissimilar computers was not.
8 See Waldrop, The Dream Machine, 14247.
9 A large number of editors were created in the 1970s; Richard Stallman"s EMACS and Bill Joy"s vi remain the most well known. Douglas Engelbart is somewhat too handsomely credited with the creation of the interactive computer, but the work of Butler Lampson and Peter Deutsch in Berkeley, as well as that of the Multics team, Ken Thompson, and others on early on-screen editors is surely more substantial in terms of the fundamental ideas and problems of manipulating text files on a screen. This story is largely undoc.u.mented, save for in the computer-science literature itself. On Engelbart, see Bardini, Bootstrapping.
10 See Campbell-Kelly, From Airline Reservations to Sonic the Hedgehog.
11 Ibid., 107.
12 Campbell-Kelly and Aspray, Computer, 2035.
13 Ultimately, the Department of Justice case against IBM used bundling as evidence of monopolistic behavior, in addition to claims about the creation of so-called Plug Compatible Machines, devices that were reverse-engineered by meticulously constructing both the mechanical interface and the software that would communicate with IBM mainframes. See Franklin M. Fischer, Folded, Spindled, and Mutilated; Brock, The Second Information Revolution.
14 The story of this project and the lessons Brooks learned are the subject of one of the most famous software-development handbooks, The Mythical Man-Month, by Frederick Brooks.
15 The computer industry has always relied heavily on trade secret, much less so on patent and copyright. Trade secret also produces its own form of order, access, and circulation, which was carried over into the early software industry as well. See Kidder, The Soul of a New Machine for a cla.s.sic account of secrecy and compet.i.tion in the computer industry.
16 On time sharing, see Lee et al., "Project MAC." Multics makes an appearance in nearly all histories of computing, the best resource by far being Tom van Vleck"s Web site Some widely admired technical innovations (many of which were borrowed from Multics) include: the hierarchical file system, the command sh.e.l.l for interacting with the system; the decision to treat everything, including external devices, as the same kind of ent.i.ty (a file), the "pipe" operator which allowed the output of one tool to be "piped" as input to another tool, facilitating the easy creation of complex tasks from simple tools.
18 Salus, A Quarter Century of UNIX, 3337.
19 Campbell-Kelly, From Airline Reservations to Sonic the Hedgehog, 143.
20 Ritchie"s Web site contains a copy of a 1974 license ( and a series of ads that exemplify the uneasy positioning of UNIX as a commercial product ( According to Don Libes and Sandy Ressler, "The original licenses were source licenses. . . . [C]ommercial inst.i.tutions paid fees on the order of $20,000. If you owned more than one machine, you had to buy binary licenses for every additional machine [i.e., you were not allowed to copy the source and install it] you wanted to install UNIX on. They were fairly pricey at $8000, considering you couldn"t resell them. On the other hand, educational inst.i.tutions could buy source licenses for several hundred dollars-just enough to cover Bell Labs" administrative overhead and the cost of the tapes" (Life with UNIX, 2021).
21 According to Salus, this licensing practice was also a direct result of Judge Thomas Meaney"s 1956 ant.i.trust consent decree which required AT&T to reveal and to license its patents for nominal fees (A Quarter Century of UNIX, 56); see also Brock, The Second Information Revolution, 11620.
22 Even in computer science, source code was rarely formally shared, and more likely presented in the form of theorems and proofs, or in various idealized higher-level languages such as Donald Knuth"s MIX language for presenting algorithms (Knuth, The Art of Computer Programming). Snippets of actual source code are much more likely to be found in printed form in handbooks, manuals, how-to guides, and other professional publications aimed at training programmers.
23 The simultaneous development of the operating system and the norms for creating, sharing, doc.u.menting, and extending it are often referred to as the "UNIX philosophy." It includes the central idea that one should build on the ideas (software) of others (see Gancarz, The Unix Philosophy and Linux and the UNIX Philosophy). See also Raymond, The Art of UNIX Programming.
24 Bell Labs threatened the nascent UNIX NEWS newsletter with trademark infringement, so "USENIX" was a concession that harkened back to the original USE users" group for DEC machines, but avoided explicitly using the name UNIX. Libes and Ressler, Life with UNIX, 9.
25 Salus, A Quarter Century of Unix, 138.
26 Ibid., emphasis added.
27 Ken Thompson and Dennis Ritchie, "The Unix Operating System," Bell Systems Technical Journal (1974).
28 Greg Rose, quoted in Lions, Commentary, n.p.
29 Lions, Commentary, n.p.
30 Ibid.
31 Tanenbaum"s two most famous textbooks are Operating Systems and Computer Networks, which have seen three and four editions respectively.
32 Tanenbaum was not the only person to follow this route. The other acknowledged giant in the computer-science textbook world, Douglas Comer, created Xinu and Xinu-PC (UNIX spelled backwards) in Operating Systems Design in 1984.
33 McKusick, "Twenty Years of Berkeley Unix," 32.
34 Libes and Ressler, Life with UNIX, 1617.
35 A recent court case between the Utah-based SCO-the current owner of the legal rights to the original UNIX source code-and IBM raised yet again the question of how much of the original UNIX source code exists in the BSD distribution. SCO alleges that IBM (and Linus Torvalds) inserted SCO-owned UNIX source code into the Linux kernel. However, the incredibly circuitous route of the "original" source code makes these claims hard to ferret out: it was developed at Bell Labs, licensed to multiple universities, used as a basis for BSD, sold to an earlier version of the company SCO (then known as the Santa Cruz Operation), which created a version called Xenix in cooperation with Microsoft. See the diagram by Eric Levenez at For more detail on this case, see www.groklaw.com.
36 See Vinton G. Cerf and Robert Kahn, "A Protocol for Packet Network Interconnection." For the history, see Abbate, Inventing the Internet; Norberg and O"Neill, A History of the Information Techniques Processing Office. Also see chapters 1 and 5 herein for more detail on the role of these protocols and the RFC process.
37 Waldrop, The Dream Machine, chaps. 5 and 6.
38 The exception being a not unimportant tool called Unix to Unix Copy Protocol, or uucp, which was widely used to transmit data by phone and formed the bases for the creation of the Usenet. See Hauben and Hauben, Netizens.
39 Salus, A Quarter Century of UNIX, 161.
40 TCP/IP Digest 1.6 (11 November 1981) contains Joy"s explanation of Berkeley"s intentions (Message-ID: anews.aucbvax.5236).
41 See Andrew Leonard, "BSD Unix: Power to the People, from the Code," Salon, 16 May 2000, Norberg and O"Neill, A History of the Information Techniques Processing Office, 18485. They cite Comer, Internetworking with TCP/IP, 6 for the figure.
5. Conceiving Open Systems
1 Quoted in Libes and Ressler, Life with UNIX, 67, and also in Critchley and Batty, Open Systems, 17. I first heard it in an interview with Sean Doyle in 1998.
2 Moral in this usage signals the "moral and social order" I explored through the concept of social imaginaries in chapter 1. Or, in the Scottish Enlightenment sense of Adam Smith, it points to the right organization and relations of exchange among humans.
3 There is, of course, a relatively robust discourse of open systems in biology, sociology, systems theory, and cybernetics; however, that meaning of open systems is more or less completely distinct from what openness and open systems came to mean in the computer industry in the period book-ended by the arrivals of the personal computer and the explosion of the Internet (ca. 198093). One relevant overlap between these two meanings can be found in the work of Carl Hewitt at the MIT Media Lab and in the interest in "agorics" taken by K. Eric Drexler, Bernardo Huberman, and Mark S. Miller. See Huberman, The Ecology of Computation.
4 Keves, "Open Systems Formal Evaluation Process," 87.
5 General Motors stirred strong interest in open systems by creating, in 1985, its Manufacturing Automation Protocol (MAP), which was built on UNIX. At the time, General Motors was the second-largest purchaser of computer equipment after the government. The Department of Defense and the U.S. Air Force also adopted and required POSIX-compliant UNIX systems early on.
6 Paul Fusco, "The Gospel According to Joy," New York Times, 27 March 1988, Sunday Magazine, 28.
7 "Dinosaur" entry, The Jargon File, Crichtley and Batty, Open Systems, 10.
9 An excellent counterpoint here is Paul Edwards"s The Closed World, which clearly demonstrates the appeal of a thoroughly and hierarchically controlled system such as the Semi-Automated Ground Environment (SAGE) of the Department of Defense against the emergence of more "green world" models of openness.
10 Crichtley and Batty, Open Systems, 13.
11 McKenna, Who"s Afraid of Big Blue? 178, emphasis added. McKenna goes on to suggest that computer companies can differentiate themselves by adding services, better interfaces, or higher reliability-ironically similar to arguments that the Open Source Initiative would make ten years later.
12 Richard Stallman, echoing the image of medieval manacled wretches, characterized the blind spot thus: "Unix does not give the user any more legal freedom than Windows does. What they mean by "open systems" is that you can mix and match components, so you can decide to have, say, a Sun chain on your right leg and some other company"s chain on your left leg, and maybe some third company"s chain on your right arm, and this is supposed to be better than having to choose to have Sun chains on all your limbs, or Microsoft chains on all your limbs. You know, I don"t care whose chains are on each limb. What I want is not to be chained by anyone" ("Richard Stallman: High School Misfit, Symbol of Free Software, MacArthur-certified Genius," interview by Michael Gross, Cambridge, Ma.s.s., 1999, 5, A similar story can be told about the emergence, in the late 1960s and early 1970s, of manufacturers of "plug-compatible" devices, peripherals that plugged into IBM machines (see Takahashi, "The Rise and Fall of the Plug Compatible Manufacturers"). Similarly, in the 1990s the story of browser compatibility and the World Wide Web Consortium (W3C) standards is another recapitulation.
14 McKenna, Who"s Afraid of Big Blue? 178.