Date: Sat, 4 Nov 1995 05:49:43 -0800 (PST) From: Phil Agre To: rre@weber.ucsd.edu Subject: history of the internet [This is a transcript of Geof Bowker's remarks as discussant on three very interesting papers on the history of the Internet at the recent conference of the Society for Social Studies of Science at the University of Virginia. It may seem quite out of context without the papers it is commenting on. But I'm forwarding it to the list anyway because it includes some very useful comments about what it would mean to tell the history of the Internet in a complex way that takes various factors into account. I have annotated Geof's remarks with contact information for the authors whose papers he discusses.] Date: Wed, 1 Nov 1995 15:35:37 -0600 (CST) From: Geof Bowker To: pagre@weber.ucsd.edu Subject: Thoughts on the history of the internet [...] Comments on session on the history of the Internet at the Society for Social Studies of Science and Society for the History of Technology Annual General Meetings, Charlottesville, Virginia, 18-22 October, 1995 Session Organizer: Jon Guice Juan Rogers , Characterizations of Internet History: toward a multistranded account Janet Abbate , Open Systems' as a Socio-Technical Model for the Internet Judy O'Neill , Developing the Internet: ARPA's view Comments by Geoffrey C. Bowker These are an excellent set of papers; together they provide a set of tools for conceptualizing the history of the Internet and thereby drawing some lessons for the future. Before commenting on the papers in more detail, I will make three general points about the history of the Internet. 1. In the beginning, there is the problem of origins. Judy O'Neill writes that: "while many people and institutions contributed to the development of the Internet, its roots are firmly planted in the DOD". Juan Rogers for his part highlights: "the peculiar institutional dynamic of science and the circles of science and technology policy-making during the 1980s that generated a national networking agenda and hosted its implementation". Janet Abbate describes the dynamic of the relationship between international telecommunications companies (the national PTTs) creating the OSI and ARPA. I shall come back to each of these in turn, but let me begin by some remarks about the problems of origins in historical writing. Michel Serres, in his work on the origins of geometry, speaks of the search for an origin as the act of working one's way upstream, from a delta to a mountain lake or spring which is the source' of a river. He notes that once you get to the source you have not arrived at an origin: there are a myriad of trickles of water arriving through capillary action into the spring that feeds the river - coming from all directions at once. What is different about the source is that from here on in all the water will flow in a single direction. I find this metaphor very useful when thinking about the history of the Internet. When any large socio-technical system is in place, it defines its own set of past trajectories: so that the series of choices that were made come to seem inevitable - a progressive story of technical optimization. However, when we look back into the history, we find that a series of choices are made which later on (downstream) become inevitable and optimal because the world has changed - but which at the time are quite contingent. Donald MacKenzie writes beautifully about this in his book Inventing Accuracy about missile guidance systems: once the choice has been made to go from inertial guidance (initially thought to be impossible) then radio guidance gets left behind as what appears to be a technically poorer and less secure alternative. Yet if the same amount of resources had been poured into the latter, MacKenzie argues, then inertial guidance could equally well have languished, and could have gone down in history as technically unfeasible. Similarly with the history of the Internet, we need to be sensitive to the ways in which contingent choices become inevitable and entrenched. 2. The Internet itself has changed the nature of the academic and business enterprise. The world is not the same before and after its introduction. This is a difficult problem historiographically: generally the historian can assume that her story unfolds against an unchanging backdrop. When the backdrop changes, the writing becomes more difficult - there are new actors in the world now (as Latour reminds us when describing Pasteur's microbes). The story of the Internet is, as Rogers reminds us, also the history of the development of high energy physics, the carving out of an institutional niche for academic computing, and the story of the intertwining of these two strands. 3. There is no natural hierarchy of layers in an emerging information infrastructure. Janet Abbate has written about this beautifully in her work. When we look at the relationship between TCP/IP and OSI, the question of what sits on what is not easily answered, she notes. Thus we can with her ask what does open' mean in OSI. A position of complete monopoly provides exactly the same open' transparency as does a system of well articulated competing systems. Openness is a relative, organizationally grounded concept. These general points made, let me turn to some more specific remarks. One thing common to all three papers - and indeed to much writing about the history of computing - is that they do not deal with any wider literature. (It is, of course, somewhat churlish of me to mention this in the context of such short papers). However, I cannot accept that the Internet is de novo in this sense, even though I think that it is one of the most important creations of recent history. We should ask ourselves constantly the question: what kind of thing is the Internet, what is it like? One history into which all three of these papers could sink is the history of standardization. This has been an exciting area in the past 10 to 20 years in the history of science and technology: for example, Thomas Hughes (in Networks of Power), Joseph O'Connell (on electrical standards - in Social Studies of Science), Simon Schaffer, Gabrielle Dupuy on the networked city, Alain Gras on the development of air traffic control, Paul David on the economics of networks. Why mention these? Well you could *expect* continuities here from one network form to the next. Certainly if you look at the development of techniques in network technologies you can find a transfer of skills, personnel, ideas going from (starting in the late eighteenth century) water to gas to electric to telephone. Surely the Internet is not so new and strange - and is not out of this sequence. Let me note here that often people are unaware themselves of such connections when they describe their own work. In a stunning work on the development of classification systems in the eighteenth and nineteenth centuries, Patrick Tort traces links from etymological to geological and mineralogical to biological to physiognomic classifications to classifications of work practice and disease. Each history before then had been written as if the problem of classification was new every time: and genetic classifications (classifications based on the origin of the classes) were created anew in each field. However Tort traces the links and shows that there is a pattern of influence that directly hooked one field into the next: thus he shows that genetic classifications are not natural' features of scientific disciplines, but that they emerged and spread contingently. A second history into which all three could sink, especially that of Rogers, is that of the mutual construction of organizations and the technologies that are used in them. JoAnne Yates has done some wonderful work here, looking at the development of office technology (the copy book, carbon paper, the filing cabinet) in the nineteenth century. She develops Gidden's notion of structuration to deal with this mutual causation. Could our authors have drawn on these resources? A third approach they might have drawn on is that of Bruno Latour - notably in his forthcoming book on the Aramis socio-technical system (an attempt in Paris to provide point to point mass transit, by creating virtual trains' with semi-autonomous carriages coupling and decoupling freely with others to form longer trains for part of the journey and then shunting off to particular stations). Latour argues that sociotechnical objects have a variable ontology': they become more and less real over time. Rogers points to this when he argues: "I think I have shown that the only way to offer a plausible account of the evolution of internetworking is to follow several strands, identify their points of interconnection, the times and reasons when they become intertwined and even merge and so on". Do you need to draw on this kind of theoretical resources - viz one that poses basic ontological and epistemological questions - in order to understand the history of the Internet? I suspect so: precisely because we are dealing with an information infrastructure. And now some specific comments on the papers. Juan Rogers draws our attention to the stories that get told in the popular press about the Internet - either as revolutionary democratic force or as arm of the military industrial complex. He says that his history is not setting the record straight'; but he does not tell us clearly what these popular stories that he deconstructs so well actually do. I would suggest to him that these stories (and compare here the ubiquitous and frequently inaccurate encapsulated histories of science in scientific textbooks): 1. Draw boundaries between inside and outside. They tell us who is of the community and who is not; what is part of the Internet and what is part of the background. And so... 2. They tell you what you can forget/ignore. You don't need to have lived the history any more. You just need to understand the capsule story and read the current literature. The story provides a context for you to work forward from - and to do this, it does not much matter if the context' is accurate or not. 3. In the process, these stories encapsulate a politics of the present. They define a trajectory that motivates current actors and drives them into future actions. They show us not the way it was, but which way the stream is flowing now. Janet Abbate has as her final sentence: "The enduring influences of the open system concept illustrate the power of a model to identify aspects of a technological system as sites for political intervention". I agree with her identification of politics here, but what does it mean to speak of the enduring influence of open systems'. Has she not also shown that this is a vacuous concept - for example the four definitions of open' that she draws from Band? Political interventions abound in the history of the Internet: their clustering around an empty concept is an interesting occurrence; but one we can only deal with if we are very careful with our attributions of intention and causality. Judy O'Neill's paper is very different from the other two. She highlights Cerf and Kahn as links with the military, and has nothing on the ISO actors of Abbate or the situated computing in different institutional sections of Rogers (the relationship of academic computing and the high energy physics community, the role of university administrators in BITNET etc). I think that her account is accurate, as is Rogers' and Abbate's. But in order to articulate the accounts, we must abandon statements like O'Neill's: while many people and institutions contributed to the development of the Internet, its roots are firmly planted in the DOD". We need alternatives to roots and trees. Deleuze and Guattari in their work (notable Mille Plateaux) offer us the metaphor of the rhizome. Where the tree of knowledge has a single root, the rhizome is protean and has many nodes. We need to play with such metaphors if we are to describe and understand the Internet. To conclude. Neither Abbate nor Rogers tells us what their papers are really *about*. Abbate describes a model which is powerful but empty. Rogers describes narratives which are powerful but inaccurate. What is happening with this emptiness and inaccuracy? The interesting thing in Internet history is that we are still at a juncture where model and narrative are genuinely contingent. These papers honestly and beautifully reflect this; and provide some initial tools for thinking about it.