Eric T. Meyer and Ralph Schroeder
Knowledge Machines: Digital Transformations of the Sciences and Humanities
Cambridge: The MIT Press, 2015, 271 pp.
Eric T. Meyer and Ralph Schroeder’s Knowledge Machines defines e-research, the book’s topic, as “the use of digital tools and data for the distributed and collaborative production of knowledge.” Noting that the Internet and related networked technologies have changed the ways academic researchers produce and consume knowledge, they ask, broadly, “What difference have digital technologies made to research, and how have they changed the direction and the practice of research?” In answering these questions, the book draws on a series of case studies focused on diverse research communities: Galaxy Zoo, a project that uses citizen scientists to classify galaxies from the Sloan Digital Sky Survey; the Genetic Association Information Network (GAIN), a project that encourages data sharing among medical researchers; SwissBioGrid, a collaboration between private companies and academic institutions that uses distributed computing to further life science research; the Virtual Observatory for the Study of Online Networks (VOSON), a web-based tool designed for social scientists; and Pynchon Wiki, which utilizes crowdsourcing to annotate novels.
In the book’s first chapter, “Conceptualizing e-Research,” Meyer and Schroeder set out their theoretical framework, arguing that understanding e-research requires both internal perspectives (that see the production of scientific knowledge as autonomous) and external perspectives (that focus more on social forces). In reviewing internal perspectives on scientific knowledge, the authors focus on Ian Hacking’s six “styles” of science, which consist of postulation, experimentation, modeling, taxonomic, probability and historiogenetic. The strength of these styles, they argue, is that they are applicable across disciplinary boundaries; in later chapters, for example, they describe the Galaxy Zoo project as experimental and taxonomic, the Pynchon Wiki as only taxonomic, and VOSON as related to modeling and probability. While the reliance on Hacking’s styles at times seems to simplify social science and humanities research in order to fit it into a framework developed in relation to the natural sciences, Meyer and Schroeder justify this decision by arguing for the increasing “scientization” of disciplines that have traditionally had distinct ways of producing knowledge.
The external perspectives introduced are drawn from social informatics, a research area that, as Meyer and Schroeder note, “tries to understand sociotechnical configurations in the world by focusing on the relationship between the social, or ‘socio,’ and the ‘technical,’ without a priori privileging either side.” Drawing heavily on the work of Rob Kling, Meyer and Schroeder conceptualize e-research as a process that involves “bringing reluctant users in, pushing for widespread adoption of new research technologies, and learning how these technologies strengthen communities and organizations.” Basing their perspective on social informatics, the authors explicitly set aside the social constructivist positions of actor-network theory or social construction of technology (which tend to focus on non-comparative studies that elucidate the uniqueness of each case), pointing again to the need for a theoretical perspective that can bridge the disciplines impacted by e-research.
The third chapter, “The Rise of Digital Research,” considers e-research funding and publication output, drawing for the latter primarily on data from online databases such as Scopus and JSTOR. The chapter concludes that e-research has grown over the past ten years, largely based in the United States and the United Kingdom. Computer science, engineering and mathematics journals publish more work drawing on computational approaches than other disciplines—however, the authors also note considerable activity in social science, business, information science and communications venues.
The remainder of the book considers various aspects of e-research, with chapters focusing on collaboration (e.g., distinctions between human collaborative complexity and computational complexity), data (e.g., tensions between flexible use and standards), limitations of e-research (e.g., lack of funding and lack of use of produced tools and infrastructures), and disciplinary concerns related to the sciences, social sciences and humanities. One of the strengths of the book in these middle chapters is the application of theoretical concepts such as mutual dependence and task certainty (the degree to which researchers are dependent on the actions of their peers and the degree to which the steps to be taken in research are understood and agreed upon). For example, while I raise concerns below related to the authors’ choice of cases from different disciplines, they usefully characterize the humanities as exhibiting low mutual dependence and high task uncertainty and, consequently, as being slower than the sciences to adopt networked technology. While the authors draw such distinctions between disciplines (claiming, for example, that social scientists are more enthusiastic about e-research than humanities scholars), they ultimately conclude that e-research has broad consequences that span disciplines, such as the increased scale and speed of research and the increased visibility of research (through blogs and easily available citation information). Ultimately, Meyer and Schroeder argue that, through the development of e-research projects, knowledge production becomes increasingly automated and mediated, performed through sociotechnical “knowledge machines” and subject to the processes of software tools and workflows (such as those that link the various institutions involved in sharing genomic data or the distributed individuals who work together to annotate a novel).
While such arguments do provide ways to consider the uptake of digitally enabled research in a variety of fields, the book’s treatment of humanities research reveals some of the challenges of such a far-reaching study. Primarily, despite the large number of cases included, it becomes clear that the burden of comparing large-scale physics research with social science and humanities projects greatly restricts the choices of the latter. Regarding social science research, Meyer and Schroeder concede that they had trouble finding “cases … where the tools of e-research were adopted in the same ways they have been in the sciences” (my emphasis). While they note that social science projects are often small-scale and tend to avoid building infrastructure, the social science projects used as case studies are large-scale and do build infrastructure. Similarly, Meyer and Schroeder focus on humanities research that involves digitizing, organizing and providing access to material, as projects involved in these tasks tend to be comparable to science projects and can more easily be described in relation to Hacking’s styles of science (although not always persuasively; it is difficult to consider the work of annotation apart from interpretation, which is absent from Hacking’s typology). In focusing on projects from the social sciences and the humanities that resemble projects in the sciences, Meyer and Schroeder fail to capture unique aspects of these disciplines such as the ways individual researchers apply computational tools in ways that differ from their use in the sciences.
Ultimately, Knowledge Machines is most persuasive when working at a scale that falls between the individual researcher, project or discipline and the grand scale of all e-research. Between these two, concepts such as mutual dependence and task certainty, drawn from the sociology of science, turn out to be very useful in distinguishing between different disciplines and the ways they might adopt new tools. While Meyer and Schroeder ask large, important questions, they rightly note that the transformations they seek to theorize are ongoing. Knowledge Machines suggests a few of the ways these transformations might play out in the future, but it’s probably most useful for suggesting mid-range theories that can be applied in the present.