Two summers ago I was flattered and fortunate to be an invited guest to a research workshop on “Knowledge Infrastructures” at the University of Michigan School of Information. Sponsored by both the National Science Foundation and the Sloan Foundation, I found this event to be a highly interdisciplinary, refreshingly speculative, and entirely energizing exercise. (Plus, I met a lot of really smart and very kind people.) Over two dozen of us hailing from a bunch of different disciplines — history and sociology, computer science and information science, technology studies and digital humanities — spent several days together in a state-of-the-art workshop space to discuss and debate how “knowledge infrastructures” might best be understood, studied, and evaluated.
This week I received the formal result of that conference: a freely-downloadable 32-page report entitled Knowledge Infrastructures: Intellectual Frameworks and Research Challenges, written by the eight principal scholars behind the project: Paul N. Edwards, Steven J. Jackson, Melissa K. Chalmers, Geoffrey C. Bowker, Christine L. Borgman, David Ribes, Matt Burton, and Scout Calvert. In the words of the authors (p. iii):
This document reports the outcomes, organized around three central questions: How are knowledge infrastructures changing? How do changes in knowledge infrastructures reinforce or redistribute authority, influence, and power? And how can we best study, know, and imagine knowledge infrastructures moving forward?
(You can see a candid snapshot from the workshop session below … and although I appear to be scowling in the middle of the photo, I was really having a great time!)
Although the conference organizers specifically tried to defer pinning down the definition of “knowledge infrastructures” too tightly, to avoid deterring interesting lines of imaginative contribution from the participants, Edwards usefully defined knowledge infrastructures as (p. 5):
[R]obust networks of people, artifacts, and institutions that generate, share, and maintain specific knowledge about the human and natural worlds.
I like this definition of a knowledge infrastructure because it is flexible enough to encompass phenomena as diverse as the US public school system, the interlocking joint financing, production, and distribution arrangements among major global entertainment media firms, or the hardware, software, data and marketing activities of Google.
I suspect my own work on the ways that human labor is mobilized, made visible or invisible, and valued or devalued within knowledge infrastructures was the reason I was asked to participate in this workshop, and I was pleased to see that labor concerns were key to this report, right from the start (p. 2):
This challenge is of more than intellectual concern. The institutions in which most knowledge workers live and labor have not kept pace, or have done so piecemeal, without a long-term vision or a strategy. For example, the widespread excitement about crowdsourced knowledge, assembled by unpaid individuals who volunteer their time out of personal interest, ignores the fact that most knowledge workers’ salaries are still paid by bricks-and-mortar organizations with hierarchical structures, established institutional cultures, systems of credit and compensation, and other “sticky” processes and routines.
For example, in discussing the current frenzy around “big data” — whether distributed data, user-generated data, process-generated data, sensor-derived data, or object-oriented data — the authors pointed out that a key resource to mobilize such data with value across space and time was missing (p. 10):
[W]e have not yet developed a cadre of metadata workers who could effectively address the issues, and we have not yet fully faced the implications of the basic infrastructural problem of maintenance. We do know that it takes enormous work to shift a database from one medium to another, let alone to adjust its outputs and algorithms so that it can remain useful both to its home discipline and to neighboring ones. Thus three results of today’s scramble to post every available scrap of data online are, first, a plethora of “dirty” data, whose quality may be impossible for other investigators to evaluate; second, weak or nonexistent guarantees of long-term persistence for many data sources; and finally, inconsistent metadata practices that may render reuse of data impossible — despite their intent to do the opposite.
After defining the need for human intervention in what is often termed (rather narrowly) the “curation” of data, the authors point to some real and stark consequences of the way this labor ends up being structured and restructured as the techhologies, applications, and ownership of data change over time (p. 13):
[K]nowledge infrastructures often carry significant distributional consequences, advancing the interests of some and actively damaging the prospects of others. For example, moves toward expanded data sharing may simultaneously devalue or commodify certain kinds of data production […] may undermine the practice and recognition of long-standing craft traditions […] [and] more efficient forms of information exchange have redistributed labor and eliminated whole categories of workers […]
Our point is not that these are always bad things, nor that we should abandon the language of sharing, access, or any of the other real and positive potentials of new knowledge infrastructure development. Rather, we argue that the consequences of change are rarely socially, culturally, or economically neutral.
Coincidentally, I’m going to be back at the University of Michigan next week, to talk about many of these same issues. I was kindly invited by one of the report authors, Melissa Chalmers, to speak at their Science, Technology, Medicine, and Society (STeMS) faculty/graduate colloquium series. (I was enrolled and trained in a similar program as a graduate student, even though today I work in a journalism school and an information school.) The topic of my talk will be another collaborative project in which I was lucky to be involved, a book titled Media Technologies which will soon be released by MIT Press. Edited by STS and communication scholars Tarleton Gillespie, Pablo J. Boczkowski and Kirsten A. Foot, this volume looks at many of the same issues as the Knowledge Infrastructures report, claiming that sociotechnical information and communication systems can best be explored by an approach that combines scholarship in science and technology studies with scholarship in communication studies.
My contribution to this volume, which I’ll be talking about at Michigan on Monday, was titled “Making Media Work: Time, Space, Identity, and Labor in the Analysis of Information and Communication Infrastructures.” One of the best parts of writing the chapter was the collaborative and interactive nature of the process — the authors of the Media Technologies volume engaged in the same kind of workshop discussions as the Knowledge Infrastructures group (you can see my scribbled concept map from one of our brainstorming sessions above). In the remaining space of this blog post, I thought I’d excerpt the first part of that chapter, both to echo and expand upon the ideas in the Knowledge Infrastructures report and to provide a nice preview for my colleagues in Ann Arbor:
Making Media Technology Work
I work with media all day, every day, as both a researcher and a teacher in a public university, wrestling with questions about our relationship to information and communication technology, both past and present. As a researcher, I proudly identify as a historian and geographer of technology, and my own most detailed case studies so far—in telegraphy, librarianship, and stenography—all have roots in the nineteenth century. But each of those topics also extends tendrils into the twenty-first century: urban bicycle messengers using smartphones recall the telegraph messenger boys of decades past; library catalogers retool their standards and practices to produce metadata for digital libraries accessible over the web; and live closed captioning on cable television is often created through computer-aided stenography. As a teacher, the pattern is reversed; my specialty is addressing the “new media” concerns of the current day, with syllabi on The Information Society, Digital Divides and Differences, and Media Fluency for the Digital Age littering my website, my Facebook page, and my Twitter feed. But each of these new media courses holds at its core a set of historical examples and arguments drawn from old media. This chapter presents my humble attempt to distill and defend the main insight that I’ve drawn from these productive contradictions over the last fifteen years or so: that a wide range of human “information labor,” enabling and constraining the constant circulation of information across a wide range of technological and social contexts, remains crucial to making media technologies work.
That idea of “work” lurks within our relationship to present-day media technologies in a variety of ways. Ask my current undergraduates on the first day of the semester about the “work” they do to find information today, and they will reply that finding information is no work at all. The answer to any brief question of fact is just a Google search away, as likely (in their minds) to lead to an amateur blog post as to a professional piece of journalism. The explanation for any named but unfamiliar event or idea is as close as the next Wikipedia page, a source they know was shunned by their former high school teachers, but which they suspect is secretly employed by their time-pressed college professors. Questions of a vaguer nature might be posed to their vast social network of friends, relations, and acquaintances through Facebook, as a public “wall” or “status” post inviting the crowd to reply. And any informational product which eludes these three strategies, somehow not available to them instantly as a web-based link to a downloadable digital file, can certainly be delivered in physical form—they still remember “books”—in twenty-four hours or less, if one is willing and able to pay, from the mega-retailer Amazon.com. Often sounding a bit too much like advertising copy, my students regularly inform me that these new tools “free” them from the pesky work of having to travel to the library, having to read through long and turgid books, and having to remember facts and definitions that are only a click away.
Press them further, of course, and they will agree that there is much “work” still to be done once any given bit of information has been supplied by the network. There are exams to study for, papers to write, presentations to compose. This kind of creative labor is easy for them to see, and easy for them to see themselves performing. It is written into the university curriculum as “complex communication” and “critical thinking” (Bok 2007; Booth, Colomb, and Williams 2008). It is the kind of high-status, high-value labor that they are paying to practice and master with their college tuition dollars in the first place. Such labor experiences will provide them with entry into those elite areas of the “space of flows” of the information society (Castells 1996)—what decades of scholars have called the postindustrial bourgeoisie, the symbolic analysts, or the creative class (Bell 1976; Reich 1991; Florida 2002, respectively). With the raw materials of information, gathered from their vast and always-on data, content, and knowledge networks, these students trust that they will end up on the correct side of the digital divide (Eubanks 2011; Norris 2001; Warschauer 2003).
In this trust, many of my students display a narrow understanding of history and geography, which underpins their narrow understanding of their own position and privilege. Media history for them is a textbook teleology of technological advances (from print culture to radio culture to television culture to digital culture) and market redefinitions (from elite audience to mass audience to individual audience), which result in their own ultimate emergence at the top of an information food chain as both target market and content originator (Baughman 1992; Downey 2011; John 2010; Starr 2004). Structures of media consolidation and content personalization have made it all too easy for them to live within a self-reinforcing informational geography of safe and satisfying answers, an “echo chamber” whether on the political right, the political left, or the political apathetic (Bagdikian 2004; Kovach and Rosenstiel 2010; Jamieson and Cappella 2008; McChesney 2008; Pariser 2011). And their own amateur media activity—whether uploading photos to their social network profile or downloading the latest cultural content outside of intellectual property paywalls—reinforces the fiction that information circulation is driven simply by “play” and that information content is simply available for “free” (Gillespie 2007; Jenkins 2006; Kline, Dyer-Witheford, and de Peuter 2003; Lessig 2010; Vaidhyanathan 2001). No wonder they are unable to see much of the actual work that underpins this media.
How might we as teachers break through this narrow, instrumental, and rather triumphalist understanding of new media infrastructures? The standard strategy of “media literacy” is to demonstrate to students that none of these admittedly extraordinary technologies, Google or Wikipedia or Facebook or Amazon, are able to deliver an information experience that entirely frees the user from further work, especially when one approaches these services with anything more than a trivial question (Fallows 2011; Gillmor 2008; Jenkins 2006; Levinson 2009; Lievrouw and Livingstone 2006; Martens 2010; Manovich 2001). It is easy to Google the name of a well-known corporation, access the Wikipedia biography of a well-known historical figure, discover a Facebook friend of similar interests, or find plenty of competing purchase options on Amazon for a mainstream, bestselling book. But pose a more complicated question on Google, and the search transforms from an “I feel lucky!” first hit success to an information overload of pages and pages of dubious result candidates, computed not simply from the original “PageRank” algorithm whereby sites with lots of links, from sites with lots of links, recursively float higher in the search rankings, but from an increasingly complicated set of contextual variables including the searcher’s own geographic location, query history, and “psychographic” marketing profile (Battelle 2005; Kink and Hess 2008; Morozov 2011). Seek a more contextual summary from Wikipedia of a broad time period in which a historical figure lived, and the “encyclopedia that anyone can edit” reaches the limits of its “no original research” restriction and its lack of professional historian contributors (Hansen, Berente, and Lyytinen 2009; Mangu-Ward 2007; Pentzold 2010; Poe 2006; Rosenzweig 2006). Attempt to define yourself on Facebook using markers other than “Likes” of purchase choices and pop-culture affiliations, and the social network application programming interface (API) is unable to parse your descriptors (Papacharissi 2009; Watkins 2009). And ask for Amazon user suggestions about a more obscure, out of print text and you may very well fall prey to the “review spam” of paid advertising disguised as customer satisfaction, or the petty squabbling of zealots who give zero stars to any work that suggests a difference of opinion with their immovable, ideological worldview (Auletta 2010; Robinson 2010; Roychoudhuri 2010). Media literacy exercises can be an eye-opening demonstration of the limits of automation in many of these seemingly laborless systems, revealing both the need for users to learn and apply sophisticated query strategies, and the influence of layers of algorithms which combine to produce complex and sometimes contradictory results (Gillespie, chapter 9, this volume).
However, teaching media literacy skills, no matter how effective, still keeps the focus of work on the students themselves. The greater challenge is to convince them that even when tools like Google and Wikipedia and Facebook and Amazon work as intended at the moment they are invoked, behind the scenes and before the fact there actually occurred great amounts of design, organization, production, reproduction, and “repair” labor on the part of many, many others besides themselves (Jackson, chapter 11, this volume). A cursory understanding of the battles between Google engineers and the outside “search engine optimization” (SEO) vendors reveals an ongoing arms race through which the search algorithm is constantly repaired from the inside and then reverse-engineered from the outside, in an environment where dropping off of the first page of a Google search can mean significant and unexpected revenue loss for online retailers (Basen 2011). Similarly, the briefest exploration into Wikipedia article production reveals the power-law division of labor represented by the small number of users who actually write substantive original articles for the site, versus the larger number who merely tweak and reorganize and spell-check and, yes, sometimes vandalize those articles—not to mention the work of algorithms known as “bots” to flag and queue articles for quality and revision (Niederer and van Dijck 2010). For the first time, journalists and scholars are beginning to reveal that social networking sites like Facebook demand constant human content moderation and censorship of photos, videos, and even text speech that violate legal terms of service, zones of personal privacy, and community norms of propriety. And once one peeks behind the virtual facade of Amazon, it is easy to see the material realities of logistics and fulfillment and customer service, with (high-paid) technology workers keeping server farms running in one region, (low-paid) warehouse workers packing product in another region, and an Internetwork of both public and private delivery services shuttling boxes back and forth from suppliers and customers in between. The best outcome for me as a teacher is when students realize that the media literacy skills that they employ in order to effectively use and critically evaluate such web tools are useless without an understanding of the deeper context of how those tools deliver what they promise. In other words, the sporadic information labor of my students as Google, Wikipedia, Facebook, and Amazon users is intimately connected to the ongoing information labor of the many, many behind-the-scenes designers, builders, operators, and maintainers of Google, Wikipedia, Face- book, and Amazon themselves.
As these present-day examples suggest, what I am loosely calling “information labor” here represents a broad diversity. Certainly some information labor reflects the same expectations of my students for their own successful futures: expensive, individual, high-status, high-value labor (or “knowledge work”), as predicted by the post-industrialists, produced by the elite universities, and circulated among the leading transnational corporations (Deuze 2007; Levy and Murnane 2004). Other information labor is collectively organized work outside of a formal organization, aggregated over a network into the so-called “wisdom of crowds” (Kreiss, Finn, and Turner 2011; Shirky 2008; van Dijck and Nieborg 2009). When such labor occurs outside of a formal wage or salary relation, it goes by various names: some have called such labor “gift exchange,” such as in the case of advice provided within online communities; others have termed it “produsage,” if it comes as a consequence of a formal user or customer relationship; or it may be termed “playbor,” if it is considered to have both entertainment value and exchange value (Bermejo 2009; Elk 2011; Kollock 1999). Still more information labor is only a little more expensive than free: contingent labor assembled by the temporary agencies and independent contracting arrangements of digital distributed work online, or emplaced in the sprawling factories of free trade zones, for a wage hopefully considered livable in its local context, but likely considered subminimum in its employing context (Rogers 2000; Benner 2002; Christensen and Barker 1998). And a growing portion of information labor is almost entirely abstracted from human minds and hands, existing as automated, algorithmic labor forever capturing some previous human expertise, judgment, pattern, or intention as replicable and executable code (Gillespie, chapter 9, this volume).
All of these forms of information labor share a crucial aspect, however: users tend not to see it. For one thing, this labor is obscured by the perpetual marketing claims of both the technologies that surround it and the content that flows through it—after all, customers are motivated to buy iPhones and apps, not the aggregated and morselized labor power of factory workers and developers. Whether through user-friendly interfaces, supply-chain intermediaries, cultural myths of smart technology, or plain old “commodity fetishism” (where a single-minded focus on the price of a good or service distracts us from considering the conditions of production for that good or service), information laborers of all sorts are likely to be hidden, out of sight and out of mind, from those who encounter their products and processes on a daily basis (Downey 2001, 2004b). The clickstream engineers of Google, the volunteer editors of Wikipedia, the outsourced moderators for Facebook, and the logistics army behind Amazon—all must be revealed, situated, and explored in order for us to reveal, situate, and explore our own daily labor with these systems.
Conceptualizing and Exploring Information Labor
What kinds of research interventions, from science and technology studies on one hand, and communication and media studies on the other, can help us with this task of “uncovering information labor” in the classroom? It is helpful to start by putting the concepts of “information” and “technology” in context. After all, the very purpose of information and communication technology is to make information—whether conceptualized as data, content, or knowledge—accessible across space and across time, from one context to another, from one community of practice to the next. But all information and communication technologies also depend, both for their daily functioning and for their overall meaning, on different forms of human labor, each with its own temporal and spatial characteristics as well.
Tools from both history and geography can be brought to bear on the question. All of the contexts and all of the communities in which we might look for information, technology, and labor are necessarily situated geographically and temporally, a condition we can analyze in terms of place, space, and scale. Individual places support or constrain certain kinds of informational activities, which are structured by their users and inhabitants, their natural and built environments, and the social meanings ascribed to them. Places connect through relationships of all sorts—technological, social, political, and economic—into broader conceptual spaces for action, be it the state space of government and military control, the market space linking raw material extraction to component assembly to consumer retail, the cyberspace interface of bodies and technologies exchanging encoded electronic communication, or the imagined space of a cultural or diasporic or aspirational community fragmented across other national, economic, and technological boundaries. And finally, these spaces are assembled, reassembled and, sometimes, disassembled at both small and large scales simultaneously, with complicated arrangements of power and uneven possibilities for making change (Downey 2007b, 2009).
Information and communication technologies, and the larger media infrastructures within which they are situated, developed, used, and understood, by their very nature exist to transcend history and geography, storing ideas across time and moving ideas across space in an organized and productive manner (Edwards 2003; Star and Bowker 2006; Wright 2007). This work of making information accessible is really about (a) making information useful (or what we might call “realizing its use value”) and (b) bringing that information into motion (or what we might call “putting information into circulation”). Especially in our current, overwhelmingly capitalist, global political economy, these two issues—how a society values information, and how information circulates through a society—are not just connected. In what we might call a dialectical relationship, each concept helps to define the other: to be useful, information must circulate through many minds (and eventually through yours); and to circulate, many minds must judge some piece of information to be (at least potentially) useful. All the agency that we bring to information along the way—whether producing information as a part of work or play, appropriating information as private property, commodifying information for market exchange, offering information up in a creative commons, or claiming the right to information in the public interest—must be understood within this basic structure of value and circulation (Dyer-Witheford 1999; Harvey 2001, 2010; Schiller 1999).
Setting up the parameters of structure and agency in this way gives us a framework for understanding media infrastructures as sites for the performance of information labor, but it doesn’t give us any clues as to what to look for when investigating the laborers themselves. Fortunately, recent scholars of technology have established quite clearly that spatial, temporal, and technological circumstances are inevitably part and parcel of social relations and cultural meanings (Nakamura 2002; Smith and Kollock 1999; Turner 2009). Within information infrastructures, for example, the evolving and overlapping categories of computer engineers, scientists, entrepreneurs, and enthusiasts over the last several decades have been revealed to involve profound meanings in terms of lots of “identity” categories—age, gender, class, race/ethnicity, political philosophy, and nationality, for example—especially as the labor practice of computer programming has been professionalized in the capitalist workplace, institutionalized in the college curriculum, implicated in interdisciplinary science, and globalized in the network economy (Aneesh 2006; Ensmenger 2010; Light 1999; Nelson, Tu, and Hines 2001; Turner 2005).
Thus what starts out as a simple classroom question about “how do you find out what you need to know?” turns out to be a rich and complicated set of related research questions about one’s place in a whole set of extended relationships of information circulation—in other words, a question about “who does what kind of information work, when and where and why?” To explain to our students what is necessary to “make media work,” I believe we must study both information and labor, in both spatial and temporal context, with attention to social relations: (1) how human labor applied to information always takes place in, and depends on, a particular spatial/temporal and political-economic context; (2) how that human labor, and the social relations and cultural meanings attached to it, both enable and constrain the ability of information itself to move from one context to another; and (3) how that circulation of information from one context to another comes full circle to affect the subsequent spatial/temporal patterns, political-economic conditions, social relations, and cultural meanings for further labor.
That can be a lot to juggle in a single research project. But attention to this basic dialectical relationship of change—where labor of a particular sort is mobilized to circulate information, and the circulation of that information helps to alter the parameters of that labor—brings a useful insight. In order to productively categorize, historicize, analyze, and, yes, teach about any “new” media infrastructure (be it the “lightning lines” of the telegraph in the 1840s, the “electronic brain” of the digital computer in the 1940s, the “electronic hearth” of the television in the 1970s, or the “information superhighway” represented by the World Wide Web today) we must continue to pay attention to the space, time, and social relations of the human laborers who are bound up in that infrastructure as well.
If that excerpt of my chapter was at all of interest, look for the full book when it comes out soon — the other chapters by my colleagues are even more intriguing, believe me — and I’ll see you in Ann Arbor!