author = {Star, Susan Leigh}
title = {Infrastructure and ethnographic practice Working on the fringes}
keywords = {}


Infrastructure and ethnographic practice Infrastructure and ethnographic practice Working on the fringes Susan Leigh Star Department of Communication, University of California San Diego, lstar@ucsd.edu Abstract By bringing together science studies, information science and ethnographic fieldwork in interdisciplinary research the author argues for the relevance of ethnographic practices when studying information systems as infrastructures of communication. Ethnographic fieldwork focuses attention on fringes and materialities of infrastructures and renders the researcher able to read the invisible layers of control and access, to understand the changes in the social orderings that are brought about by information technology. Numerous examples and personal accounts of studies of infrastructures with ethnographic tools show how paying analytical attention to mundane aspects of information infrastructures helps to understand the consequences of the imbrication of infrastructure and human organization. Keywords Ethnographic fieldwork, information infrastructures, standards, fringes © Scandinavian Journal of Information Systems, 2002, 14(2): 107-122 107 Infrastructure and ethnographic practice Introduction Information science, especially the “social” side, is an emergent, interdisciplinary field. It compares historically with other fields such as educational research and nursing research. At first domain-driven, and closely linked with library science and information retrieval, it is now finding its own theoretical contributions. With the advent of networked computing, information science has also been pulled far beyond its original domain. It draws now on organization research, science and technology studies, cognitive science and artificial intelligence, anthropology and sociology, among others. As with theory, so also with methods. The traditional methods of information and library science – for example, transaction logging – have themselves been impacted both by networked computing and by the expansion of the field’s mandate. One of the notable borrowings in methods is fieldwork – organizational ethnography, expanded user studies (Bishop and Star, 1999), and the ethnographic study of the design and use of networked computing. Yet none of these face precisely the problems presented by information science as a domain, and its unique interpolation with infrastructure. In some sense, infrastructure is our domain, especially the infrastructures of communication. Using fieldwork to analyze it breaks new frontiers of methodology, for us and for other This paper addresses one aspect of ethnographic practice for information science: the importance of attending to infrastructure while conducting fieldwork. Traditional fieldwork in sociology has often passed rather lightly over questions of infrastructure (with some important exceptions, see e.g. Lindesmith, 1947 and Becker, 1982). It has been well-attuned to the nuances of and (this was my contribution) how language, membership, identity and learning. In nematologists2 use computers to keep track of anthropology, the emphasis in traditional their worm specimens. One must admit that fieldwork has included some infrastructural these topics are generally low profile (to put it issues, including tool use, symbolic artifacts, and the impact of technology on modernization processes. In organization studies, attention has been paid to the impact of computerization, communication with and about technology, and some emphasis on the built environment. 108 © Scandinavian Journal of Information Systems, 2002, 14(2): 107-122 Boring things Some five years ago, in Palo Alto, California, I joined with several colleagues, all ethnographers, to found a new professional society1. The idea for the society arose from a series of conversations we had about our somewhat unusual research topics - things that most people would find quite dull. We called it The Society of People Interested in Boring Things. All of us were, in some way, interested in a broad study of information technology, using ethnography. Among the boring topics presenters brought to the table were: the inscription of gender in unemployment forms used by the city government in Hamburg, Germany; the difficulties of measuring urine output in a post-surgical ward in the Netherlands, and how to design better cups for metrication; the company mascot and the slogans used by a large Midwestern insurance firm in its attempts to build “corporate culture”; mildly), and for most social scientists, adequately boring to qualify for membership in our new association. In addition, what they have in common is a concern with infrastructure, the invisible glue that binds disciplines together, within and across their boundaries. As I have noted, for historical reasons, infrastructure is usually singularly unexciting as a research object for ethnographers. The human, symbolic, interactive aspects of infrastructure are terribly difficult for ethnographers to “open up” in the way that we easily may open up conversations, rituals or gestures. Infrastructure socio-technical disciplines. This paper is a think-piece about some of those issues, and how they impact our emerging inter-discipline. The first barrier to using fieldwork is seeing infrastructure, and that means getting past the first take on information infrastructure – that it is boring, not as exciting as the traditional issues fieldwork is so good at “catching.” At first try, using fieldwork to stand and watch people punching keys and looking at screens is terribly difficult for trying to see social order. Or, in fact, to see much of anything. often appears simply as a list of numbers of surgery was developed. Again, an infrastructural technical specifications, or black boxes, wires learning device and form of communication and plugs, in the scientific/disciplinary between scientists was being developed – workplace. (Where is the human behavior side something that had been ignored in prior of that?) In my work as an ethnographer histories of British brain research, presumably studying life sciences and medicine, I have because it was too “boring.” found that infrastructure can also be messy and distasteful. For example, in studying museum representations, I found myself up close and personal with the history of taxidermy (Star, 1992). This research included tracking down the biological supply houses that had provided items such as standard-sized glass eyes for the different animals in the museum dioramas; home-made devices for shaving and softening animal skins, and other tools for preparing and preserving specimens and habitats. Figure One: This is a picture of part of the infrastructure of representation in building However, his department had constructed a list museum dioramas, an important communication of approved journals and ours (a young, technology for biologists trying to bring nature interdisciplinary one) was not on the list. The to the masses. The glass eyes are standardized purpose of the list? Not some imprimateur of wares sold to museums by biological supply scientific correctness. Rather, the department houses, after the craft of taxidermy came to be wanted articles to be published in journals that routine in this representational technology, were indexed, and thus counted, in the science during the late nineteenth century in the US (Star, 1992). In another study, of brain scientists, (Star, 1989), I learned about their difficulties in obtaining (often illegal) and preserving (technically quite a difficult undertaking) brains for study during the nineteenth century in Britain, where brain © Scandinavian Journal of Information Systems, 2002, 14(2): 107-122 109 Infrastructure and ethnographic practice These behind-the-scenes, messy or boring items form a crucial part of the materiality of how scholarly and scientific work is done. Lack of infrastructure directly impacts the flow of interdisciplinary knowledge. It is thus of particular importance in using fieldwork to understand the design and use of information systems. For instance, I recently invited a young Dutch colleague to submit an interesting article to a journal I help edit. He replied that he was interested in the journal, and agreed that it would be a good audience for his work. citation index. Then they could prove to their government funders, with hard numbers, that the research had impact according to the citation index. When I inquired about how one gets counted by the science citation index, I was told that one must contact the science citation publishers, and present them with letters and testimony from eminent scholars testifying that the journal is worthwhile. This sort of infrastructural barrier (or helpful facility, depending on one’s viewpoint) is pervasive in scholarly work as well as in all modern bureaucracies. As evidenced by this anecdote, it is biased against new, unorthodox, and interdisciplinary paths, knowledges or approaches that tend to appear first at the margins of disciplines, in social movements, small presses, or in independent media venues open to risk-taking. The barrier is, in this case, inherently conservative. Ethnography is very good at exposing these biases, when the right questions are asked. (For established fields, of course, tools like this are also vital aspects of communication and its quality.) Thus, in order to understand these sorts of communicative tools, we need to analyze their role in scientific Infrastructure and ethnographic practice and daily work – and play - and learn to read these invisible layers of control and access. phone book indicates a rural area; those that list only husband’s names for married couples indicate a heterosexually biased, sexist society. Unearthing the narratives behind boring aspects of infrastructure does, however, reveal, often in a very direct way, how knowledge is constrained, built and preserved. In addition, historical changes may provide clues. To In order to understand how this operates, however, it is necessary to "deconstruct" the Many aspects of infrastructure are more difficult boring, backstage parts of infrastructure, to to locate, for several reasons. First, people tend disembed the narratives it contains and the to discount this aspect of infrastructure as behind-the-scenes decisions (such as that extraneous to knowledge or to their tasks. They performed by the science citation index), as part therefore do not tend to include mention of them of material information science culture. This in official historical narratives (except in means overcoming the initial boredom and passing, see Clarke and Fujimura, 1992, for an reading the deeper social structures embedded excellent discussion of this problem). Second, in these tools. During the last several years, I details such as materials, standards and have been studying such tools, both formal and classification schemes do not always obviously informal. (Ethnography always examines the intersect those variables and processes familiar formal and informal, not taking either for to us in analyzing human interactions: gender, granted as the natural way to do things.) I have race, status, career, power, innovation investigated several scientific, medical, ordinary trajectories, and so forth. life, and political classification schemes, with an eye to understanding the values and work practices embedded in them. One of the infrastructures I studied was the International Classification of Diseases (ICD), a global information-collecting system continue with the phone book example, names administered by the World Health Organization and locations of services may change with (Bowker and Star, 1999). It is about one political currents and social movements. To hundred years old, and has been revised every quote again from Bowker and Star (1999): decade or so to reflect changes in medical and epidemiological knowledge. It takes the form of a long list of codes, numerals that stand for diseases and causes of death. The numbers are inscribed in medical and insurance software, death certificates, and other epidemiological and vital statistical tools. The volumes where these numbers reside are more than 2,000 pages long, basically a very large list with instructions for selecting the numerals. It is not the sort of book that usually compels dramatic reading: Reading the ICD is a lot like reading the telephone book. In fact, it is worse. The telephone book, especially the yellow pages, contains a more obvious degree of narrative structure. It tells how local businesses see themselves, how many restaurants of a given ethnicity there are in the locale, whether or not hot tubs or plastic surgeons are to be found there. (Yet most people don’t curl up with a good telephone book of a Saturday night.) (Bowker and Star, 1999, p. 56) In the Santa Cruz, California, phone book, Alcoholics Anonymous and Narcotics Anonymous are listed in emergency services; years ago they would have been listed under “rehabilitation” if at all. The changed status reflects the widespread recognition of the organizations’ reliability in crisis situations, as well as acceptance of their theory of addiction as a medical condition. Under the community events section in the beginning, next to the Garlic Festival and the celebration of the anniversary of the city’s founding, the Gay and Lesbian Pride Parade is listed as an annual event. Behind this simple telephone book listing lies decades of activism and conflict—for gays and lesbians, becoming part of the civic infrastructure in this way betokens a kind of public acceptance almost unthinkable 30 years ago… excursions into this aspect of information infrastructure can be stiflingly boring. Many classifications appear as nothing more than lists of numbers with labels attached, buried in software menus, users’ manuals, or other references.” (1999: 57) Other kinds of information in a telephone book can be read indirectly - for instance, a slender 110 © Scandinavian Journal of Information Systems, 2002, 14(2): 107-122 Infrastructure and ethnographic practice In the case of the International Classification of says, "graves and reminiscences can be neither Diseases, mentioned above, Western and conquered nor transferred" and that "to know a middle-class values and foci are inscribed in the language you must be able to pray in it, write list of mortality and morbidity labels. These love letters in it, and curse in it." (Schütz, 1944) labels are used, among other things, to fill out There is no such thing as a stand-alone word. death certificates and record epidemics around And Schütz was enamored of strangers as the globe. They are thus critical, often invisible windows into our thinking-as-usual, ways of resources for allocating aid and tracking disrupting what he called “thinking as usual.” international health concerns. As an example of There is a hopeful and often romantic aspect to Western, middle-class values, one can turn to this – strangers bring new perspectives, trouble the part of the classification scheme that our complacency (e.g. Stonequist, 1937). encodes accidents. According to the list, one may fall from a car or from a commode, but not from an elephant. One may be a heroin or an absinthe addict3, but addiction to sniffing gasoline is not represented. This latter is a serious problem among, for example, urban Aboriginal youth in Australia. Large-scale information infrastructures, such as the Web and digital libraries, are making strangers of all of us in this sense, both designers and users. We are constantly meeting up with the fringes of other languages, a space where neither keywords nor co-word analysis can supply us with graves and reminiscences. One way to understand these emphases, from an ethnographic point of view, is not to yell “ah ha, a bias! I knew it all along.” This is not only bad fieldwork, it becomes a silly sort of boredom after the first éclat that science indeed contains values and biases. What is of more concern to information sciences is how to use this ethnographic information to theorize about the information-communication aspects of social order, and to help us understand the changes wrought by information technology. I find the symbolic interactionist-phenomenological approach to the words themselves, the categories-in-use, to be a helpful source for making such theory. For example, both William James and Alfred Schütz usefully linked words/categories with questions of membership and belonging. Some of the fringes come from the necessarily interdisciplinary undertaking of building such large systems. Some come from the indexicality of the content within libraries and their texts where words mean one thing in one discipline, and another in another one, for example. This is an old problem, and one of the richest ones in information science, for builders of thesauri and designers of information retrieval systems. New faster, bigger databases and algorithms for disambiguation change some things about the problems - speed of processing, revising thesauri on the fly, brilliant insights into adjacency issues and modeling of problem spaces. In earlier times, changes to thesauri in print versions, could take years or decades, involve many committees and much heated, but invisible, discussion of revisions. These have by no means disappeared in the digital realm, but take a different form, some of it automated. However, while the smoke may have disappeared from the smoke-filled rooms of committees, the heat has not. To bring this back again to ethnographic practice, it is precisely the role of fieldwork to understand – through a kind of temporary membership – these fringes and nuances. The fact that there are clashes and differences in meaning is a commonplace of ethnography. What is new is the speed and complexity with which these fringes appear in our everyday lives, via information technologies. Methodologically, we need to learn to speak to Fringes The philosopher William James used to say that "words have fringes." He was quoted on this point by the sociologist Alfred Schütz in his classic essay, "The Stranger." Schütz spoke of the stranger as "one who comes and stays a while," not a mere passerby. A stranger often has trouble with the fringes of language, the nuances, the historical context, including its indexicality. Indexicality is that which cannot explicitly be put into a representation, but requires insider knowledge such as history, nuance, and context. To the extent that all representations are incomplete, indexicality fills in the necessary blanks. For instance, Schütz © Scandinavian Journal of Information Systems, 2002, 14(2): 107-122 111 Infrastructure and ethnographic practice them as a form of social ordering – not to sort chose instead to use Gopher and other simpler them out on behalf of others, but to take note of net utilities with less technical functionality the shape and nature of the clashes, their (later, of course, they turned to the Web). duration, and their consequences. We are not in Obviously, this crossed communication was a this sense social engineers, but always problem of some concern to us as system somewhat strangers, who analyze. Background of my own work I have worked since 1981 to build partnerships between computer and information scientists and social scientists. I first worked in the area of Artificial Intelligence modeling in the 1980s with Carl Hewitt at MIT, where my job more or less was to find things in nature whose properties could be translated into what was then called a highly parallel open system (Hewitt, 1985; 1986; Star, 1989). I was a purveyor of, in AI terms, "metaphors." In Greek, of course, "metaphor" means moving a thing from one place to another (moving vans are labeled "Metaphoros" for a literal point of view on this). In AI terms, this meant fetching good modelable data from phenomena "out there in the world". This was for me the beginning of finding the fringes between fields. I think of how long it took me to learn the meaning of "transparent" when I was a newbie stranger to the world of computer science, coming as I did from interpretive sociology. It really means opaque! In working on the Illinois Digital Library Project from 1994-1998 (Bishop, et al., 2000), and earlier on the Worm Community System based at the University of Arizona (1991-94) (Star and Ruhleder, 1996), I ran into an interesting set of fringes from both the design and use sides of the equation. On the Worm Community Project, my coinvestigator Karen Ruhleder and I found a world of clashing meanings between designers and users of the system. The project came just before the advent of the Web, and just as academe become fully saturated with email users (especially in the sciences), 1991-1994. We studied a scientific community and a custom-made system co-designed with the community. Most respondents said they liked the system, praising its ease of use and its understanding of the problem domain. On the other hand, most did not sign on; many then 112 © Scandinavian Journal of Information Systems, 2002, 14(2): 107-122 developers and evaluators. Despite good user prototype feedback and participation in the system development, there were unforeseen, complex challenges to usage involving infrastructural and organizational relationships. The system was neither widely adopted, nor did it have a sustained impact on the field as the resources and communication channels it proffered became available through other (often more accessible) means. It did provide important insights and models for continuing work on the technical side; it also provided insights for us as social scientists into the profound impact of the understanding of infrastructure on group interactions. In short, we found that we had underestimated the problems with local infrastructure. We had underestimated the impact of the colliding “fringes” between users and designers, and in general, we learned a lot about how people feel about and use infrastructure and changes to it, including such (to us then) unlikely things as feeling shame, guilt, fear, rage; lying (to the point of claiming to use the system and not using it) and sneaking around; and what is not at all now surprising, using one system to show the evaluators and then switching back to familiar technology in their routine work. On the Digital Library Project, and with the advent of the web, other fringes were to be found in the content of documents and web sites which are always, and interestingly, full of these meetings of strangers. There are many types: homonyms (again, an old, old problem in library science - much of the research from that field is ignored by computer scientists and systems builders, unfortunately). Another type comes from resistance and social movements that incorporate and re-appropriate language at lightning speed. In twenty years "queer" has gone from being a term of loathing and gender boundary patrolling of homosexuality to a positive term (in some circles) denoting radical and challenging approaches to gender roles and sexuality. Of course, on the street, it is usually still derogatory. The term "nigger" is halfway - it can be the ugliest of racial epithets or a Infrastructure and ethnographic practice positive re-appropriation by rap singers or fringes associated with standards, embedded African Americans speaking to each other in categories (as opposed to those visibly solidarity. I hardly know what to make of it, appearing in LCSH or specific cataloguing however, when I see my white middle class systems), sizes, and those now imprinted on surfer students greet each other on the UC San almost every object bought, observed, or every Diego campus with, "hey nigger, w'as up?" (Of process to which we as human beings are course we have words for this in sociology - subjected (medical tests, postgraduate cultural appropriation, the migration of standardized examinations (GREs), shopping, language forms across sub-cultures, and so on. traveling, eating, giving birth, becoming a This research is virtually unknown amongst citizen or getting a residency permit - and so on, system builders; even where known, the not to mention using the library). They have technical problems and the social research do some of the same characteristics as the others not match.) Fringes change with context, which described above; at the same time, they are is why they are fringes. There is another usually deeply invisible, as is the work involved opportunity here for ethnographic tools – in creating and using them. This ventures into observation, participation, interviewing – to the territory of the ethnography of everyday life, enter into and understand how these sorts of and how information science may be used to cultural nuances operate in how people use read aspects of daily life often neglected by information systems. fieldwork. The collisions and their politics, and the lack of Let me give a couple of examples of standards understanding in the technical community, is struggles. First, a mundane example taken from why I delight in the work of Sanford Berman, little maps. I recently bought a poppy seed fringe hero amongst librarians and pioneer into packet to plant poppies in my back yard. I found the ethics of categories and key words. His now- my attention drawn to the everyday information classic (or infamous, depending on who you talk embedded in the little package itself. In to) example of the information retrieval addition to the bar codes, which encode both problems associated with common objects, such price and agricultural information, there is a tiny as light bulbs, illustrates colliding fringes map at the bottom telling when to plant the between lay users and a professional elite. He poppies. My area indicates Sept-February. This explains, holding a light bulb over his head, map is of very coarse granularity, with four or someone trying to find out about light bulbs five degrees of differentiation, and completely could never do so using Library of Congress excluding Hawaii or Alaska. However, another Subject Headings (LCSH) categories. Instead, map, published by Sunset Magazine (a Western one would have to know to look under “electric US gardening magazine), and dedicated to the lamp, incandescent.” This is minor suffering, in microclimates of the American West, takes into a sense. More urgent are his politically charged account the coastal fog that extends inland about challenges to the Library of Congress, such as 4 miles in San Diego, and adjusts the planting attempting to remove “the Jewish Question”, times accordingly. The granularity is different “primitive,” and “Yellow Peril” as unquestioned because they are communicating to an audience categories (Berman, 1984; 1993). All of the of gardeners who need finer detail. (They also standard problems of social movements and do not use the US Department of Agriculture language, long a staple in fieldwork, appear here Climate Zones, a map used for commercial in an instant. This is an opportunity to link with agricultural purposes.) My “real” planting time sociological ethnographic work on social worlds is May-February. and social movements, and the appropriation of language. Many have argued that maps encode all sorts of arguments and targeted audiences, and embody just the sort of fringes and standards struggles found in textual documents. Granularity is political, and that is especially important in cartography. For example, the Peters Projection, a politically progressive, and not Standards as fringes I want to turn my attention to an entirely different aspect of fringes here, one that is not usually recognized as co-extensive with the same problems Berman addresses. These are the © Scandinavian Journal of Information Systems, 2002, 14(2): 107-122 113 Infrastructure and ethnographic practice cartographically very scientific attempted to remedy the bias toward countries of the North, as shown by the older Mercator projection. Becker gives us the example of maps that do not show elevation. He lives at the foot of Lombard St. in San Francisco, also know touristically as the “curviest street in world,” surrounded by some of the steepest urban streets in the US (1982). He often finds puzzled tourists on his street staring up 60 degrees and wondering how on earth they will make the climb. These two maps show different kinds of arguments and audiences, and different ways of dealing with the problem, or not dealing with it. There are now cognitive maps of every major city and region, many industries, and many political or diplomatic situations - all meant ironically, yet also seriously. All in some sense subvert, or make visible, the fringes embedded in standard representations. Again, they are also rich territory for using ethnography to explore information systems in everyday life. Another example shows both the cultural history and the seriousness of these processes: The U.S. Immigration and Naturalization (INS) form one must fill out in order to apply for citizenship, embeds another kind of example of categorical and standardization fringes. The application for a green card, or resident alien permit, includes questions such as, “are you mentally retarded,” “are you an alcoholic” or (perhaps my ironic favorite) “are you a cardcarrying anarchist.” I am married to an alien, who is also an academic, and when we came to one question of this form, “have you every sold your body for profit,” his first reply was, “of course – I’m an academic, aren’t I?” (Many of the questions about mental retardation and prostitution come from the eugenics movement of the early 20th century, which had a strong hand in building immigration laws. This raises the important point about the range and nature of what computer scientists would call "legacy systems" found in everyday life and in formal systems.) We have recently filled out the U.S. Citizenship form. The instructions come in the form of about one hundred pages of U.S History, from which citizenship questions are drawn in a quiz, where one is allowed short sentences and multiple choice. He also holds a PhD in History 114 © Scandinavian Journal of Information Systems, 2002, 14(2): 107-122 Intellectual background: science studies In the world of science, which has always has a close affiliation with information science, social science scholars began to study how laboratories work during the 1970s, work that was later to link with the concerns expressed above regarding infrastructure. In Europe and the US, notably with the 1979 publication of Latour and Woolgar's Laboratory Life (1979), people began to explore the laboratory as a kind of anthropological field, with scientists as the tribe. Laboratory Life was an ethnographic examination of the production of a scientific fact. It looked at the devices (called "inscription devices" by Latour and Woolgar) used by biologists to record and preserve data and at the gradual deletion of uncertainty and qualifications in the statements emerging from the laboratory. It explicitly tried to eschew the obvious categories that previous, more macroscale studies of science had produced - occupational stratification, the role of national cultures in science, and so forth. The idea was to approach science making afresh, to look empirically at knowledge production in a detailed, face-to-face context, much as an anthropologist would approach a new “tribe” (their metaphor). With the publication of Laboratory Life, a window was opened to a more qualitative, intensively observational set of studies of scientific work and practice. Many were produced over the next two decades, examining such interesting phenomena as talk in the laboratory, the acquisition of manual skills in performing tests, the ambiguity of scientific objects, the intersection of heterogeneous and Philosophy of Science. We came to the question, “what form of government does the United States have?” As a good historian, he began to answer, “Well, from post-colonial and globalization point of view, many argue that the form of government is now actually via multinational corporations and lobbyists, with a distinct media influence….” "No, no, no," I say. "The answer is bicameral representative democracy." "Oh," he says. Standards are standards, and they embody values, simplifications, and treaties. Another prime opportunity for ethnographic investigation. Infrastructure and ethnographic practice Disciplines and categories: viewpoints in making scientific theories, and, by the 1990s, the research community began disciplines are systematic studies of the design and use of information technologies (see e.g. Star, 1995). Against commonsense belief, scientific and This development towards the "technical turn" academic disciplines do not constitute a high in science studies, that is, the ethnographic degree of consensus. On the contrary, one might study of the design and use of advanced better define a knowledge discipline as a technologies, such as computers, had many commitment to engage in disagreements. research ramifications (Star, 1983). It used many Biologists do not agree on the nature of species; of the same techniques as earlier laboratory sociologists bicker about the nature of society; studies of science. However, it also directly literary critics diverge on notions of genre and engaged social scientists in studying style. What endures, however, are debates about communicating machines, the emergence of the PC and late the World Wide Web, and to observe attempts to model human behavior. In addition, in the early 1990s, several detailed studies of the materials aspects of scientific work began to appear, many of which began to pick up aspects of boring things and infrastructure (see e.g. Clarke, 1998). Recent science and technology studies have taken this combination of the technical turn and studies of materials deep into the investigation of infrastructure. The ethnographic eye that helped reveal the inner workings of science or technology research and development applies no less to the built scientific-technical broken in research on this topic - both in the environment, including information search for areas such as the "speech area" and in infrastructure. Conflicts about standardization, denunciations of the very idea. Participants selection and maintenance of tools, and the right came from physiology, surgery, anatomy, materials for the job of knowledge production psychology, hospital administration, and have very slowly come into center stage via this philosophy. In the end, their disagreements synthesis (Clarke and Fujimura, 1992). Along helped to form the basis for neurophysiology as with this has come a rediscovery of some of the a discipline. In biology, a similar arena emerged tools germane to cognate disciplines that had around the unit of analysis for species selection: previously analyzed material culture and the group or individual? Genes or environment? built environment. These have included, inter Biologists come in large part to self-define alia, fields such as architecture (where scholars around the stances they take on these issues. sometimes read the built environment as a kind of text); literary theory (especially those aspects of literary theory that help surface hidden stylistic assumptions and narrative structure), and social geography (where the values and biases inherent in such tools as maps are a lively topic of inquiry). My own work, on categories, boundary objects and standards as structuring knowledge owes much to these fields as well as to cognitive anthropology and linguistics, areas whose scholars have investigated the tool aspect and origin of various category systems. commitments to disagree the categories that constitute the core knowledge of the field. Insofar as these categories are inscribed in material objects, databases, and knowledge management tools such as thesauri and journal indexing terms, they themselves form a kind of glue that acts to keep the discipline communicating. The same is true of interdisciplinary communication. For example, in my earlier studies of neurophysiology and brain research, debates raged from the early nineteenth century to the present day about whether particular functions are localized in a particular part of the brain (Star, 1989). Dozens of careers were made and In none of these sorts of debates, however, are the basic terms of the debate questioned. Localizationists may have disagreed with diffusionists about the localized vs. distributed character of cognitive function, but almost none of them chose to look to the environment, whole body, or elsewhere for the seat of cognition, or to dismiss the question out of hand. Biologists all agree that speciation is a crucial phenomenon, whatever their causal allegiance. One important theoretical direction for information science is to develop a larger and deeper map of scientific debates, focusing on © Scandinavian Journal of Information Systems, 2002, 14(2): 107-122 115 Infrastructure and ethnographic practice the basic terms of the debate, linked with infrastructural constraints and historically Defining infrastructure is not as obvious as it inherited tools. Defining infrastructure may seem. I had a commonsense notion of What does emerge with some frequency are two infrastructure when I first started studying the kinds of structures within the debates. First, design of interdisciplinary computer systems – much of the infrastructural technology is used infrastructure as something that other things unquestioningly by everyone in the debate. In “run on,” things that are substrate to events and brain surgery, both localizationists and movements. Railroads, highways, plumbing, diffusionists used surgery, electricity, and more recently, the information electroencephalographs, and neurological superhighway. Good infrastructure is by testing to validate their claims, for instance. definition invisible, part of the background for Second, particular categories (rather than the other kinds of work. It is ready-to-hand. This classification scheme as a whole) become image holds up well enough for most targets for debate. Thus Kirk and Kutchins purposes—turn on the faucet for a drink of (1992) describe a fierce debate between gay water and you use a vast infrastructure of activists and psychiatrists about the plumbing and water regulation without usually medicalization of the category "homosexual" as thinking much about it. an illness in early versions of the Diagnostic and Statistical Manual (DSM), the psychiatric equivalent of the ICD. The DSM assigns categories to mental illnesses, and is widely used in psychiatric epidemiology as well as in such crucial infrastructural functions as thirdparty reimbursement for psychiatric care. At the same time, few on either side of the debate quarreled with the basic need for such a category system. Once the system was put into place, it acquired its own inertia and entanglements with the everyday bookkeeping and diagnostic practices of psychiatrists and other mental health professionals. However, in light of a deeper analysis of infrastructure, and especially to understand large-scale technical systems in the making, or to examine the situations of those who are not served by a particular infrastructure, this definition is both too shallow and too absolute. For a highway engineer, the tarmac is not infrastructure but topic of research and development. For the blind person, the graphics programming and standards for the World Wide Web are not helpful supporters of computer use, but barriers that must be worked around (Star, 1991). One person’s infrastructure is another’s brick wall, or in some cases, one person’s brick wall is another’s object of demolition. As Star and Ruhleder put it, infrastructure is a fundamentally relational concept, becoming real infrastructure in relation to organized practices (1996; see also Jewett and Kling, 1991). So, within a given cultural context, the teacher considers the blackboard as working The job of an ethnographer of scientific practice and the information contained within, therefore, is to raise these second- and third-order questions about the existence and nature of the whole classification scheme, the taken-forgranted tools used in intra- and interdisciplinary communication. One aspect of this work is to surface embedded biases in infrastructure integral to giving a lesson. For the representations of knowledge, both blatant (e.g. school architect, and for the janitor, it is a in advertisements) and subtle (e.g., categories in variable in a spatial planning process or a target databases). The critical study of cartography and for cleaning: “Analytically, infrastructure maps is especially import to this enterprise. appears only as a relational property, not as a Reading the invisible maps that border thing stripped of use.” (Star and Ruhleder, 1996: disciplines requires new sorts of metadata tools, 113) ones that can help us understand the traffic across disciplinary borders as well as the takenfor-granted questions to which disciplinary adherents are committed. A few ideas about how to approach this are sketched below. In my studies of the development of computer systems and scientific work, I have begun to see infrastructure as part of human organization, and as problematic as any other part. I’ve done a kind of Gestalt switch, what Bowker has called an “infrastructural inversion” – foregrounding 116 © Scandinavian Journal of Information Systems, 2002, 14(2): 107-122 Infrastructure and ethnographic practice the truly back stage elements of work practice, As well as the important studies of body the boring things (1994). Recent work in the snatching, identity tourism and trans-global history of science (Bowker, 1994; Hughes, knowledge networks, then, let us also attend 1983; 1989; Yates, 1989; Edwards, 1996; analytically to the plugs, settings, sizes, and Summerton, 1994) has begun to describe the other profoundly mundane aspects of history of large-scale systems in precisely this cyberspace, in some of the same ways we might way. In science as well as in culture more parse a telephone book. generally, we see and name things differently These studies are important, and in some ways, under different infrastructural regimes. natural to ethnographers, where the familiar Technological developments are processes and identity, membership, and learning issues take relations braided in with thought and work. In another, fascinating form. There are many fewer the study of nematologists mentioned above, studies on the effect of standardization or formal Ruhleder and I listed the properties of classification on group formation, the design of infrastructure as that which is embedded; networks and their import for various transparent; having reach or scope; is learned as communities, or on the fierce policy debates part of membership; has links with conventions about domain names, exchange protocols, or of practice; embodies standards; is built on an languages. installed base (and its inertia); becomes visible upon breakdown; and is fixed in modular increments, not centrally or from an overview. Perhaps this is not surprising, given the invisible and boring nature of many of these venues from the point of view of social science and humanities. The latter topics tend to occur in semi-private settings, or buried in inaccessible electronic code. Theirs is not the usual sort of anthropological or ethnographic strangeness. Rather, it is an embedded strangeness, a secondorder one, that of the forgotten, the background, the frozen in place. Struggles with infrastructure are built into the very fabric of technical work, and increasingly other domains of work and play (Neumann and Star, 1996). However, it is often easier to stay within the traditional purview of social scientific studies: talk, community, identity, and group processes as now mediated by information technology. There have been several good studies of MUDs and role-playing, The ecological effect of studying boring things distance-mediated identity, cyberspace (infrastructure, in this case) is in some ways community and status hierarchies. The similar. The ecology of the distributed high-tech challenges they present are non-trivial workplace, home or school is profoundly methodologically. How does one study action at impacted by the relatively unstudied a distance? How does one even observe the infrastructure that permeates all its functions. interaction of keyboard, embodied groups, and Study a city and neglect its sewers and power language? What are the ethics of studying supplies (as many have), and you miss essential people whose identity you may never know? aspects of distributional justice and planning When is an infrastructure finished, and how power (but see Latour and Hernant, 1998). would we know that? How do we understand Study an information system and neglect its the ecology of work as affected by standards, wires and settings, and you miss standardization and classification? What is equally essential aspects of aesthetics, justice, universal or local about standardized interfaces? and change. Your ethnography will be Perhaps most important of all, what values and incomplete. Perhaps, as ethnographers, if we ethical principles do we inscribe in the inner stopped thinking of computers as information depths of the built information environment? highways, and begin to think of them more (Hanseth and Monteiro, 1996; Goguen, 1994) modestly as symbol sewers, this realm would We need new methods to understand this open up a bit. imbrication of infrastructure and human organization. One promising direction is to apply the tools of ethnography to this imbrication. © Scandinavian Journal of Information Systems, 2002, 14(2): 107-122 117 Infrastructure and ethnographic practice From the history of ideas to interdisciplinary communication Most of the stories about communication across communities that have been told in the past have excluded analysis of shared infrastructure. Instead, they have emphasized the ideas and techniques that have migrated from one field to another, or, they have looked for structural similarities between disciplines’ knowledge structures that could explain affinities. Other theories have looked at the migration of individuals across boundaries, often graduate students moving from one lab to another, or senior researchers beginning a second career. All of these studies provide valuable insights into how science travels across disciplinary boundaries, or how new interdisciplinary fields are established. At the same time, the role of shared material objects and infrastructures is crucial for a full ethnographic picture. Shared objects One of the kinds of material/infrastructural arrangements that may occur across fields is the development of boundary objects (Star and Griesemer, 1989; Star, 1989). These are objects that dwell in more than one community of practice – a discipline, or a line of work, or a voluntary association. They have two important properties: they are loosely structured in common use, and become more tightly bound in particular locations. They are thus both ambiguous and clear, at different moments, for different purposes. I developed this notion with James Griesemer some years ago when we were conducting a study of the development of the Museum of Vertebrate Zoology (MVZ) at the University of California, Berkeley. Two individuals from different communities founded the MVZ early in the century: biologist Joseph Grinnell and philanthropist/amateur naturalist Annie M. Alexander. Grinnell was one of the founding figures of population ecology. His research questions and methods required the collection of large numbers of specimens from all over the state of California. To do this, he enlisted the aid of dozens of amateur collectors and naturalists, who were interested in the project for its value to the conservation movement, rather than for reasons of scientific accuracy. 118 © Scandinavian Journal of Information Systems, 2002, 14(2): 107-122 One of the interesting things about the development of the museum was the very different visions participants brought to the table. Most of the collectors were uninterested in Grinnell’s formal ecological theories, for example. At the same time, however, these heterogeneous groups were able to cooperate, even down to the level of collecting and painstakingly labeling specimens for Grinnell’s museum. How did this cooperation occur in the absence of intellectual consensus? ( I would now say: how did the fringes get institutionalized and how did people develop workarounds to account for the misunderstandings?) Specimens, maps, even the cases inside the museum grew into boundary objects shared across these different worlds. Thus, a local naturalist could come to the museum with a specimen, learn its name, and feel that he or she was contributing to conservation, via science. Professional biologists could use the collections in another way. The delicate diplomatic structures that composed this set of arrangements included several varieties of boundary object. They also relied on a certain degree of standardization across methods, for example in the collecting of specimens and the description of habitats. Boundary objects are everywhere, but the concept is especially important in ethnographic analysis of cooperation and issues of infrastructure. Often an infrastructural device such as a thesaurus or the science citation index discussed above becomes a boundary object (see e.g. Harvey and Chrisman, 1998). Even people who strongly disagree on theoretical matters come to refer to the tool in a similar fashion, and it provides a lingua franca for exchanges. Where these exchanges are stabilized, boundary objects develop facilitating heterogeneous cooperation. There is much work to be done to understand all of the ramifications of this approach to interdisciplinarity. We need to understand more, for example, about the behind-the-scenes decisions about things such as encoding and standardizing; tinkering and tailoring activities (see e.g. Gasser, 1986; Trigg and Bødker, 1994), and the observation and deconstruction of decisions carried into infrastructural forms Infrastructure and ethnographic practice classification systems, decisions to invest in one sort of system or another. (Bowker and Star, 1999, Chapter 3). We need to understand more about how boundary objects develop as well as how they fail to develop in various work settings. In the above-mentioned study of systems of classification, Bowker and I attempted to A deconstructive reading of infrastructure unearth the developmental aspects of quickly reveals the presence of what literary infrastructure creation and use. We discovered theorists would call a master narrative, that is, a many moments when the master narrative-insingle voice that does not problematize the-making became visible. For example, we diversity. This is the voice of the unconscious studied the creation of the system of race center, the pseudo-inclusive generic. An classification under apartheid in South Africa. example of this encoding into infrastructure From 1950 until the end of apartheid in the early would be a medical history form for women that 1990s, all South Africans were classified into encodes monogamous traditional four racial groups, European (white), Asian, heterosexuality as the only class of responses: “Bantu” (black African), and Coloured (mixed blanks for “maiden name” and “husband’s race). Of course, millions of people did not fit name,” blanks for “form of birth control” but into such oversimplified designations, which none for other sexual practices that may have conflated language groups, race and ethnicity, medical consequences, and no place at all for appearance and genetics, and many other partners other than husband to be called in a factors. This did not stop the government from medical emergency. Latour discusses the enforcing totalitarian control over the lives of narrative inscribed in the failed metro system, those so classified, including restrictions on Aramis, as encoding a particular size of car workplaces, residences, voting, and so on. based on the presumed nuclear family (1996). Band-Aids or mastectomy prostheses labeled “flesh colored” which are closest to the color of white people’s skin are another kind of example. In order to understand the cracks in the system and how it was enforced, we examined a number of cases of racial reclassification. These were legal cases where the person felt (or sometimes a government informant felt) that they had been wrongly classified. Common instances of this were among light-skinned people classified as Coloured, who felt that they should be classified White (a vastly more privileged category). In the reclassification process, the emergence of the master narrative, and how it fits in with information technology, is clear. There is no room for ambiguity on the form, whatever may be the ambiguities the As we learned long ago from Derrida and from feminist linguists such as Wendy Martyna, identifying and subverting these pseudo-generic voices means first identifying that that has been left silent. In Adrienne Rich’s words, this means listening first for “lies, secrets and silences.” Some of the literary devices that represent master narratives are: creating global actors, or turning a diverse set of activities and interests into one actor with a presumably monolithic agenda (“the United States stands for person lived with in everyday life. One could be democracy”); personification, or making a set of assigned only one category, eternal and actions into a single actor with volition ahistorical. From this would devolve (“science seeks a cure for cancer”); passive government statistics on racial groups, voice (“the data have revealed that”) and Parliamentary and police organizations, and deletion of modalities. This latter has been well even sports teams. Since the hearings on race described by sociologists of science—the reclassification were done in camera, the public process by which a scientific fact is gradually face of the master narrative was able to be stripped of the circumstances of its enforced in a vast system of bureaucracy, forms, development, and the attendant uncertainties, and layered “lies, secrets and silences.” and becomes an unvarnished truth. In terms of infrastructure, this may mean recovering the narrative before being able to analyze it. Again, this implies digging into the construction sites of infrastructures – standards setting, creating of In addition, much of the work that creates both boundary objects and master narratives becomes invisible once it is inscribed in infrastructure. In addition, many information systems represent and encode work processes, directly or © Scandinavian Journal of Information Systems, 2002, 14(2): 107-122 119 Infrastructure and ethnographic practice indirectly (payroll systems, time sheets, activity reports, and flow charts are among the many infrastructural tools that perform this function in the workplace). Such tools, like language itself, are always incomplete with reference to both the complexity and the indexicality of the processes represented. People are always adjusting, of discretion. working around the representations to get on with their jobs and their lives. Again, though, there is an opportunity for social archaeology for the analyst of infrastructure (Star and Strauss, 1999, discuss this in relation to the design of CSCW systems). In some instances, this means going backstage, in Erving Goffman’s terms, and recovering the mess obscured by the boring sameness of the are often at a disadvantage with promotion information represented. It is often in such back-stage work that important requirements are discovered. With any form of work, there are always people whose work goes unnoticed or not formally recognized (cleaners, janitors, maids, and often parents, for instance). Where the object of systems design is to support all work, leaving out what are locally perceived as “non-people” means that in fact the system does not work. Most computer systems designers arbitrarily cut off certain support personnel from the systems they are creating – sometimes secretaries (as with executive decision support systems, ignoring how many decisions are in fact made by secretaries for their bosses), usually janitors, cooks, and temporary personnel. The results are layers of silence built into the infrastructures that surround jobs. The solution to these silences and their negative consequences is not always, however, simply making things visible to all. So, for example, when Bowker and I were analyzing the attempts by a group of nurses to classify their work processes, we saw them walk a delicate line between visibility and invisibility. They wanted their work to be represented in order to be legitimated. At the same time, if they categorized all the tasks they did, and then built forms into hospital record keeping in order to track that work, they risked having the hospital accountants and HMO officials Taylorize their work and try to fob parts of it off on less expensive paraprofessionals. So leave the work tacit, and it fades into the wallpaper (in one 120 © Scandinavian Journal of Information Systems, 2002, 14(2): 107-122 1. Including, among others, anthropologists Charlotte Linde and Susan Anderson, historian Geoffrey Bowker, computer scientist David Levy, physician/philosopher Marc Berg. Conclusion We need to be able to theorize across the continuum of information infrastructures, from the old, historical, global to the everyday, simple and quintessentially invisible stuff of ordinariness. We need to see both layers of organizational complexity and demography to the minutiae of seed packets. The road in to both comes from many sources, through a myriad of exquisitely boring things. Information science offers a unique sort of lens of the world – how is it ordered, tagged, how do people find their ways through conflicting fringes, how is information retrieval changed by networked computing? Ethnographic methods offer the opportunity to go between the layers, to examine both the formal and information aspects of communication, to see how questions of membership, identity, learning and culture interleave with more traditional questions of surrogacy, retrieval, bibliometrics and cognitive style. We are in a new field with an old inheritance, and one that is at the center of vast social change. There is much to be done. Notes respondents’ words, “we are thrown in with the price of the room”). Make it explicit, and it will become a target for surveillance. The job of the nursing classifiers was to balance someone in the middle, making their work just visible enough for legitimation, but maintaining an area Much infrastructure is marked with this sort of invisible trouble. In academic departments, the question of what work should be visible and what should count for promotions and tenure often brings this to a head. Researchers who develop large information systems, performing and visual artists, those whose work may take a long time to come to fruition (such as architects) committees, who may not be able to evaluate or understand the invisible work that goes into research that does not culminate in a book or an article in a refereed journal. Infrastructure and ethnographic practice 2. Biologists who study worms, in this case those who were trying to sequence the genome of the nematode c. elegans. 3. The list began in Paris in the nineteenth century, thus the concern with what today, in most parts of the world, is a relatively minor addiction problem. References Becker, H. S. Art Worlds, Berkeley: University of California Press, 1982. Bowker, G. “Information Mythology and Infrastructure.” In Information Acumen: The Understanding and Use of Knowledge in Modern Business, L. Bud-Frierman (ed.) London: Routledge. 1994, pp. 231-247. Bowker, G. and Star, S. L. Sorting Things Out: Classification and Its Consequences, Cambridge, MA: MIT Press, 1999. Clarke, A. Disciplining Reproduction: Modernity, American Life Sciences and the ‘Problem of Sex,’ Berkeley: University of California Press, 1998. Clarke, A. E. and Fujimura, J. (eds). The Right Tools For The Job: At Work in Twentieth-Century Life Sciences, Princeton, NJ: Princeton University Press, 1992. Edwards, P. N. The Closed World: Computers and the Politics of Discourse in Cold War America, Cambridge, MA : MIT Press, 1996. Gasser, L. “The Integration of Computing and Routine Work,” ACM Transactions on Office Information Systems (4:3), 1986, pp. 205-225. Gerson, E. and Star, S. L. “Analyzing Due Process in the Workplace,” ACM Transactions on Office Information Systems (4:3), 1986, pp. 257- 270. Goguen, J. “Requirements Engineering as the Reconciliation of Technical and Social Issues,” in Requirements Engineering: Social and Technical Issues, M. Jirotka and J. Goguen (eds.) NY: Academic Press, 1994. © Scandinavian Journal of Information Systems, 2002, 14(2): 107-122 121 Hanseth, O. and Monteiro, E. “Inscribing Behavior in Information Infrastructure Standards,“ Accounting, Management & Information Technology (7:2), 1996, pp.183-211. Harvey, F. and Chrisman, N. “Boundary Objects and the Social Construction of GIS Technology, “ Environment and Planning A (30:41), 1998, pp. 1683-1694. Hewitt, C. “Offices are Open Systems,” ACM Transactions on Office Information Systems (4:3), 1986, pp. 271-287. Hughes, T. P. Networks of Power: Electrification in Western Society, 1880 –1930, Baltimore: Johns Hopkins University Press, 1983. Hughes, T. P. “The Evolution of Large Technological Systems,” in The Social Construction of Technological Systems, W. E. Bijker, T. P. Hughes, and T. Pinch (eds.), Cambridge: MIT Press, 1989, pp. 51-82. Jewett, Tom and Kling, Rob. “The Dynamics of Computerization in a Social Science Research Team: A Case Study of Infrastructure, Strategies, and Skills,” Social Science Computer Review (9:1), 1991, pp. 246-275. Kirk, S. A. and Kutchins, H. The Selling of the DSM: The Rhetoric of Science in Psychiatry, New York: Aldine de Gruyter, 1992. Latour, B. Aramis, or The Love of Technology. Cambridge, MA: Harvard University Press, 1996. Latour, B. and Hernant, E. Paris: Ville Invisible, Paris: La Decouverte, 1998. Latour, B. and Woolgar, S. Laboratory Life, Thousands Oaks, CA: SAGE, 1979. Lindesmith, A. Opiate addiction, Bloomington, Ind., Principia Press, 1947. Infrastructure and ethnographic practice Neumann, L. and Star, S. L. “Making Infrastructure: The Dream of a Common Language,” in Proceedings of PDC ‘96 (Participatory Design Conference), J. Blomberg, F. Kensing and E. Dykstra-Erickson (eds.), Palo Alto, CA: Computer Professionals for Social Responsibility, 1996, pp. 231-240. Robinson, M. “Double-level Languages and Co- Operative Working,” AI and Society (5:1), 1991, pp. 34-60. Star, S. L. Regions of the Mind: Brain Research and the Quest for Scientific Certainty, Stanford: Stanford University Press, 1989. Star, S.L. “Power, Technologies and the Phenomenology of Conventions: On Being Allergic to Onion,.” in A Sociology of Monsters: Essays on Power, Technology and Domination, J. Law (ed.), London: Routledge, 1991, pp. 26-56. Star, S. L. (ed.) Ecologies of Knowledge: Work and Politics in Science and Technology, Albany, NY: SUNY Press, 1995. Star, S. L. and Strauss, A. “Layers of Silence, Arenas of Voice: The Ecology of Visible and Invisible Work,” Computer-Supported Cooperative Work: The Journal of Collaborative Computing (8: 1), 1999, pp. 9-30. Summerton, J. (ed.) Changing Large Technical Systems, Boulder, CO: Westview Press, 1994. Trigg, R. and Bødker, S. “From Implementation to Design: Tailoring and the Emergence of Systematization in CSCW,” in Proceedings of ACM 1994 Conference on Computer-Supported Cooperative Work. New York: ACM Press, 1994, pp. 45-54. Yates, J. Control through Communication: The Rise of System in American Management, Baltimore, MD: Johns Hopkins University Press, 1989. 122 © Scandinavian Journal of Information Systems, 2002, 14(2): 107-122