Saturday, November 11, 2006

Reading :: Acting with Technology

Originally posted: Sat, 11 Nov 2006 02:04:06

Acting with Technology: Activity Theory and Interaction Design

by Victor Kaptelinin and Bonnie A. Nardi

Activity theory is complex, intricate, and often difficult to understand, and until now, no full-length text has managed to provide a good solid understanding of what it entails. That's especially problematic since AT has been adopted by people in so many fields and used to do so many things, leading to some conceptual drift. When I taught a graduate course on activity theory and human-computer interaction a couple of years ago, I ended up using Nardi's Context and Consciousness and excerpts from several other works, but students rightly objected that the threads of AT had not been brought together in a coherent way.

If I were to teach that course again, I'd use this book. Kaptelinin and Nardi have worked hard to develop a single text that lays out AT for those without much background in it, and in general they do a really fine job. In particular, the chapters that summarize AT's principles, development, and treatment of objects are all quite strong. The case studies, which are largely based on published articles, are also valuable but are sometimes not well integrated with the overview and theory chapters. But that's okay -- the authors do a great job of referring to external AT studies and resources, so readers can easily find additional, relevant articles and cases.

So what does the book say about AT? The authors emphasize AT's coherence and its emphasis on human intentionality, themes that contrast with the characteristics the authors see in rival frameworks such as distributed cognition and actor-network theory (see, e.g., p.10). They also discuss how AT's new interest in Bakhtinian dialogism and multivocality (p.23) help to position it to study recent changes in work, since "work is more distributed, more contingent, less stable" (p.26). In fact, in their chapter on objects, they do a really nice job of discussing "polymotivation" (p.138), a key move in addressing the issue of networked work organization. And their chapter discussing AT's development and discontinuities is tremendously helpful for anyone who has wondered how AT relates to Vygotsky and Leont'ev and Bakhtin, or whether Engestrom's triangles are an essential part of an AT analysis, or why they're called "objects."

The authors also discuss AT in relationship to other postcognitivist theories in Chapters 9 and 10, particularly actor-network theory and distributed cognition. Bonnie was kind enough to let me review draft copies of these two chapters, and also kind enough to cite my work in them. My reaction, then and now, is that these chapters do important work -- but they tend to take AT's goals as the gold standard for comparison and criticize the other approaches for falling short of those standards. Since ANT and DC have very different objects (to use AT's terms), of course they fall short. I should say here that I consider myself an activity theorist, and I have a strong interest in individual human activity, development, and ingenuity (as the authors themselves point out by kindly citing my work). But I do think that other projects can be pursued and that examining their separate objects could have produced a more amicable comparison.

Despite that wrinkle, the comparison is tremendously valuable, especially in Chapter 10, where the authors compare treatments of agency. When activity theorists criticize how symmetrical approaches handle agency, the obvious counterclaim is that in emphasizing individuals so strongly, activity theorists actually ignore the incredibly important concept of mediation developed by Vygotsky. Isn't one obvious implication of mediation that cultural activity is mediated activity, i.e., activity that only exists because heterogeneous elements such as workers, practices, artifacts, and material objects form assemblages that mutually change each other? That is, doesn't AT point to a form of symmetry?

Kaptelinin and Nardi, anticipating that counterclaim, furnish what is perhaps the most important new development in the book. They produce a typology of agents (p.244), breaking down the concept of agency into several subtypes and demonstrating how these are operationalized differently through different agents. For instance,

  • Natural things have conditional agency -- they produce effects -- but they don't demonstrate delegated agency or needs-based agency.
  • Cultural things have conditional and delegated agency, but not needs-based agency.
  • nonhuman, domesticated living beings have conditional, delegated, and partial needs-based agency (they act according to their own biological needs, but not their cultural needs).
  • only human beings (such as "Spinuzzi's traffic engineers") exhibit all these kinds of agency in full.

This really is quite brilliant, as it nails down some of the slippages in the term "agency" that have allowed actor-network theorists to claim agency for things. For instance, Latour argues elegantly that speed bumps have agency just as much as people do, since these "sleeping policemen" are delegated the responsibilities that human policemen have, and often discharge those responsibilities more faithfully. But no, this chart says, Latour is confusing two very different types of agency. We all intuitively know this, the argument goes, but now we can point to exactly where the confusion is and we can avoid that mistake. Or so the argument goes; I'm still processing it.

Well. As you can tell, I'm really quite impressed with this smart and timely book. Although I've spent a bit of time in this review questioning assumptions in a couple of the chapters, partly because I've been so engaged with them lately, the book overall is a huge contribution to the AT literature. I'll certainly refer to it, cite it over and over, and think hard about its implications.

Blogged with Flock


Alan said...

Clay, Well, I've read the Intro and Chs. 9 and 10 in Kaptelinin and Nardi's book. There are many things to like about the parts of the test I've read. There are others that leave me a bit uncertain. From the intro they set up an arena that goes society-subjects-tools-world, but are really interested in subjects-tools relations... in some ways the old Marxist terrain of work- or labor-process studies. Interestingly, in their graph of agency in Ch.10, we get things (natural & cultural), non-human living beings (natural and cultural), human beings (who are said to have natural and cultural needs in Ch.9) and social entities. In general I like this, though I think it misses a good bit of historicity in the production of the naturalness and culturalness of things and living beings... Where I get pushed out of the text is in the apparent de-situated reification of subjectivity -- the repeated assertions of the social contextualization of subjectivity, intentionality and creativity notwithstanding. It strikes me that there is a great deal of theorizing of the relation between subjects and objects and effectively none, except to disparage overly structuralist approaches, of the relations between subjects and social contexts. In many ways it comes off as social interactionism where objects are activated but the social theory critical and feminist interactionists have fought to add over the last thirty years remains at arms length.

Similarly, the emphasis on creativity in the approach to activity theory presented in the chapters I've read appear vulnerable to Latour's critique of science scholars who focus on controversies... creativity and controversies are of great interest but -- and it is rather surprising to me that I'm writing this -- the vast majority of activities with objects are mundane or "normal scientific" these days. In short, the emphasis on technological empowerment in activity theory (as presented here) misses or skips over the widespread prevalence of technological disempowerment. In terms of normal science and disempowerment, I was repeatedly reminded me of Steven Lukes' critique of pluralist theories of power in Power: A Radical View during my reading.

I also think that the critiques of ANT are a bit unfair. Latour, in particular, stepped away from the Macchiavellian approach to human actants and did so without moving to the network level of analysis K&N insist upon tarring him/ANT with. I agree the refusal to acknowledge the difference w/r/t intentionality and beyond between active subjects and active objects in ANT is a HUGE problem, but my sense is that ANT simply glosses over the difference in its case studies rather than actually ignoring it... which is pretty damning in itself, and is a point K&N make.

Two more things. On the one hand, I really do sense that K&N make a mistake when they strongly emphasize the intentional roots of human culture and cultural change. A central point of the foundational work by Marx, Weber, Durkheim, Mead, Freud and beyond was that the problem with modernity is that the production of markets, bureaucracies, solidarities, selves and psyches is overly structured, staggeringly chaotic and deeply undemocratic. On the other hand, there is an emphasis on subject-level resistance to intended uses of technological objects but that is the extent of the range of subjective resistance allowed in activity theory... This, in fact, has its opposite/counterpart (I couldn't decide) in the celebration of the creativity of disembodied virtual game players -- where the positive connections and social relations are far more like pen pals, fan clubs, sports team affiliations and being transported by fiction than the authors acknowledge. Are there really no jerks playing virtual games as implied in the text?

Nearing the end... they cite your traffic safety study. Do they accurately represent your account? It seems to me that, as with their misunderstanding of ANT, they reify the creativity of the data crunchers, the manager and/or the freshman computer programmer (you?). Which is not to say that these folks were not creative but is to say that someone committed to ANT would want to unpack the "specialized domain knowledge," "mathematical tools", and "statistical analysis" black boxes, as well as the sustained and surely collective efforts to transform traffic safety work and traffic safety in Iowa.

It strikes me that AT is focused on the "microgenesis of change" (226), which is a great thing to focus on, but that K&N over-reach in terms of their conclusions about phenomenology, distributed cognition and ANT because many of the critiques of these latter traditions both fail to acknowledge the utility gained by the limited focus of these traditions and fall short in terms of acknowledging the efficacy and limits AT has because of its own concerns. In all four traditions, as I have come to see in my own work, the strengths and enablements of my tradition in political ecology come from its limited perspective and constraints.

This is what I like so much about Haraways emphasis on situated knowledges... I see it as a perspective that fosters the advance of individual and collective subjectivities and politics because of its insistence on co-constructing a shared, rather than individual, project and politics. Having written this, it might just be that K&N and I are very very close to agreeing and could work it all out pretty easily.

Lastly, my sense of the limits of the graph in Ch.10 is that the argument is made that the Mars probe can "produce an effect according to an intention" even though that intention is not the probes. As I see it, this will often undermine the natural-cultural split in things and nonhuman living beings in terms of natural things and beings being able to realize the intentions of (other) human beings.

Thanks for encouraging me to look this over.

Clay Spinuzzi said...

And thanks for following up on this! I think your comment is almost as long as the review -- and I think that's appropriate, since there are many unplumbed depths here.

Activity theory is grounded pretty strongly in a particular milieu, influenced heavily in the US by strands of anthropology, sociology, and especially educational psychology. The resulting body of theory and research tends to be fairly modernist and oriented toward observable, verifiable phenomena -- something that turns out to be fairly useful when applied to human-computer interaction and other traditional domains of cognitive psychology. In fact, activity theorists are used to being perceived as being "soft" scientists and have oriented their arguments to demonstrate how their "soft" approach measures up well under the "hard" criteria of cognitive psychology and informatics. To give an example, they mention my study of traffic engineers (in _Tracing Genres through Organizations_), which strongly emphasized individual creativity in order to contradict a prevalent HCI view that users are relatively passive parts of an informatic system.

But -- and I address this in my _Net Work_ manuscript, currently under review -- AT's expansion into different domains is bumping into the expansion of ANT and other "soft" approaches in those same domains, and ANT is founded on a very different set of assumptions, warrants, and argumentative strategies. At their contact points, the two orientations tend to talk past each other.

My sense from reading this book and from other discussions with Bonnie is that she still wants to evaluate ANT and similar approaches through the criteria that she has been using to justify AT in HCI circles. By those criteria, ANT is too "soft" and is self-evidently claptrap. (I hasten to add that from my perspective in rhetoric, ANT is not claptrap.) For an HCI audience, K&N's taxonomy of subjects is a strong angle to pursue, since it decomposes the notion of subject into categories that can be individually tested. For an audience grounded in ANT, STS, and related fields, it does not have as big a payoff for the reasons you cite. Which is to say that to some extent the two sides are talking past each other.

My pet theory about Latour's Machiavellianism, which you are free to dismiss -- Bonnie is skeptical about this too -- is that Latour remains Machiavellian, but in the original sense rather than in the pejorative might-makes-right sense that people typically apply to Machiavelli's thought. He's distanced himself from the name "Machiavelli" but not from the principles of analysis that his compatriots extracted from Machiavelli. I hope this is the case, because I think that Machiavelli still has a lot to offer our analysis of political-rhetorical interactions. For more on this, see my reviews of Machiavelli's work and of _Pandora's Hope_. Or look for _Net Work_ once I've published it!