Reconfiguring a Classic
In 1987, the personal computer revolution was accelerating into the consumer market, and a boom in artificial intelligence (AI) research promised a further revolution for business applications of computing. Xerox Palo Alto Research Center (PARC) seemed the ideal institution to bring these two opportunities together. PARC had developed a vision of high-powered personal computer workstations during the 1970s, building groundbreaking products that inaugurated the graphical user interfaces later popularized by Macintosh and Windows. The PARC Systems Science Laboratory was also a leading center of commercial AI research. Furthermore, PARC had developed a tradition of interdisciplinary innovation, including pioneering uses of philosophy, linguistics, and cognitive science as tools for designing the human–computer interface (Blackwell Reference Blackwell2006).
It was into this environment that Lucy Suchman arrived to conduct the research leading to Plans and Situated Actions, the first edition of the book that is now republished in expanded form twenty years later. In a period during which advanced Xerox workstations were being deployed in offices far removed from computer science departments (even the White House), it became clear that controlled laboratory studies were no longer sufficient to understand the ways that computers, let alone the ubiquitous Xerox photocopiers, would be used. Instead, PARC recruited anthropologists, including Suchman, to conduct ethnographic studies of technology use in the workplace. However, anthropologists are adept at questioning not only the people they study but also themselves and the social context from which they launch their investigations. Suchman's analysis of PARC's requirements, or rather its implicit sociotechnical presumptions, caused Plans and Situated Actions to have huge impact within the fields of cognitive science and human–computer interaction.
In this ambitiously extended second edition, which in addition to much new text carries the revised title Human–Machine Reconfigurations, Suchman grasps the opportunity to reflect on the context in which the original book had so much impact, the intellectual developments that have arisen from its publication, and its consequences for her own subsequent research.
When she arrived at PARC, initially as a Ph.D. student, Suchman brought not only the conventional fieldwork approaches of the anthropologist but also the ethnomethodological analysis techniques of Conversation Analysis. It was this analytic perspective that led to the most provocative aspect of her original work and still forms one of the central elements of her research philosophy. By analyzing interaction between people and machines in terms that had previously been applied to interpersonal interaction, she emphasized the ways in which AI approaches (where planning and goal analysis precedes action) were unable to account for the normal ways that people help each other to accomplish everyday work (where plans are only an element of action rather than its primary foundation). The results had substantial impact at Xerox, where even the operation of photocopiers was incorporating more complex interaction than the users wished. But her critique of the conventional view of human–machine interaction (as a dialog in which machine intelligence might assist) had even more impact outside Xerox, both in human–computer interaction and in Cognitive Science (1993).
This second edition of her influential book offers a considered evaluation and response to the various controversies in which the original text has participated. In particular, it offers a chance for Suchman to respond to the more extreme readings in which both critics and popularizers accused her of denying the reality of plans as an element of human cognition. The original text is therefore reproduced at the heart of the book and is offered for new consideration in the light of many footnotes that clarify and elaborate on that subsequent commentary. I found a particularly welcome element, in these footnotes, to be the many citations of more recent works that would form an excellent reading list for the new or returning researcher concerned with these questions.
In addition to that original text (and an evaluative synopsis of the central argument and its impact by way of introduction), the new edition also includes five new chapters that address the theme of human–machine configuration from richer theoretical grounds in the social study and philosophy of science. In these new chapters, Suchman revisits the essential concerns of human–machine interaction and communication from the perspective of contemporary social studies of science and technology, drawing additionally on anthropology and feminist philosophy. The result not only brings her original work up to date with regard to subsequent intellectual development but also adds broader intellectual foundations to those earlier insights that were more specifically rooted in the methods and perspectives of ethnomethodology and Conversation Analysis.
Her overall concern remains the question of how humans are described and imagined, in the light of machines that imitate humans and in the work of researchers who design such machines. Many contemporary research trends bring with them assumptions of agency in machines, for example in those descriptions of ubiquitous and wearable computing that imagine a perfect invisible infrastructure, fronted by a Jeeves-like servant sufficiently like his master to anticipate not only material needs but also implicitly social and moral actions. Although research projects motivated by such visions come nowhere near the practical reality to test such imagined futures, agency is problematic even in the laboratory, where artificial life can be made to stand in place of those who create it, producing an artifact that, while in one sense a communicative actor, can also be made to assume the status of an objective and neutral scientific instrument when convenient.
These issues become even more critical in the case of machines that are designed precisely to be humanlike, by giving them bodies, emotional awareness through affective computing, or imitation of developmental social skills. However, rather than repeat her earlier critique of technical aspirations, by simply observing the ways in which machines are deficient in enacting social relations, Suchman now aims for new analytic perspectives that allow new kinds of relationship to be described. One such opportunity is that of new media art, in the light of which laboratory encounters with research prototypes can be viewed as a particular kind of performance, exploring the possibility of new kinds of relation by acting them out. Rather than require agency as a condition of a machine that can be brought into relation with a person, this draws on anthropological perspectives in which many objects can be assigned personhood as a component of their participation in social networks. Her advice is to play with the difference between people and machines, rather than explain one through imitation by the other.
This final portion of the book provides a valuable introduction to much contemporary writing in the social study of science that is relevant to the design of human–computer interfaces. It is written not only for an audience of other social scientists (although it probably demands a reader with some sympathy to their vocabulary and aims) but also for those technology designers who take a serious critical attitude to their work and who welcome perspectives that allow them to inspect the tacit assumptions of their own everyday research communities. The final chapter extends into material that is likely to be far outside the ‘comfort zone’ for most readers of this journal and perhaps might try the patience of even sympathetic readers, but it is rich in intellectual challenges for those who seek them. This book is far more than a simple update of the first edition, in some ways approaching a sequel. For those who have benefited, or would have benefited, from the original book, this edition will be an invaluable replacement.
