Saturday, October 21, 2006

Situated technologies, agency and underspecification

Last night around 80 people crowded into a small hot room at the Urban Center to listen to Mark Shepard, Omar Khan and Trebor Scholz introduce topics related to Architecture and Situated Technologies. I have notes scrawled over a programme that I'll sort out later, but a couple of things stuck in my head that I hope we get back to today.

First, there appeared to be some sort of assumption that 'things' are about to take on agency. This is a very strange claim to someone who studies material culture because agency is understood to always already include people and objects.

Second, there was quite a bit of focus on 'underspecification' - which solicited vigourous nodding amongst the audience but then seemed to lead conversation into a false dichotomy between underspecification and overspecialisation. (This also manifested in Trebor's call to stop "hiding behind professional language" as if there is some sort of evil intention - or weak character - behind specialist language.)

In any case, things get going in an hour or so this morning, so stay tuned!


Anonymous Larry Irons said...

Anne, what was meant by underspecification? I remember an interesting piece by Matthew Chalmers a couple of years ago that maintained any system necessarily involves underspecification of the situation of its use. It is an essay on the history of context.

Anonymous anne said...

Yes, I think they were talking about technological underspecification along the lines of hackability or modularity, or some sort of way to avoid the standard architectures of control. In one sense, this seemed rather obvious to me and something that I and others have argued for years now. In other ways it struck me as a bit utopian or reactionary, esepcially when it rubbed up against specialisation and not just specification. In any case, it didn't get unpacked the following day and I really don't know what to make of it now.

Anonymous Larry Irons said...

I think it is key to a concept of seamful design, as outlined by Chalmers, Dourish, and mentioned in Greenfield's Everyware. Andy Crabtree's work fits here as well, IMHO.

Anonymous anne said...

The concept of seamful design comes from Weiser, and yes, I've always agreed that is a good idea. I'm not quite sure how underspecification works in Crabtree's ethnomethodology though, can you please explain a bit how you see that?

Anonymous Larry Irons said...

Sure Anne, I’ll take a stab at it. But, first let me just respond to your point about Weiser because I left him out of my post on purpose.

I make the mistake at times of forgetting the complexity of what Weiser said and remembering the popularization of what he said, which emphasizes invisibility and seamlessness. Although, honestly, I don’t see much in what I have read that he wrote or presented that took the concept of seamful design as its topic, though he does of course mention it and is probably first in that regard. He looks more like a source of inspiration for those who do take it as a focus, rather than a resource for their actual work. I know that goes against the grain of what most people who knew him say, and I’ll gladly say I’m wrong if someone shows me something he wrote that actually uses the concept of seamful design to show how it is, or could be, done. Rather, we get the assertion that, “I think the value of invisibility is generally understood.” Weiser certainly did provide examples for invisible tools and, to answer your blog post from some time ago, I suspect the adjective “beautiful” was dropped from the term seamful because a seamful design by definition requires our attention, which Weiser does not appear to have considered a desirable property of an interface. As Chalmers has indicated, a seamful design allows people to reach in and tune the device to their preferences. The focus is very different than the ambient intelligence efforts to design systems, homes, automobiles, etc. that observe behavior and tune the device for us in the name of a seamless interface.

How will a ubiquitous home, car, phone, or other device respond to being sold, or traded?

But, to turn to your question about Crabtree. To an extent, Crabtree’s approach focuses on how to make seams visible in data about ubiquitous computing contexts rather than on using them in design per se.

Looking at Crabtree’s paper at DIS2004, “Design in the Absence of Practice: Breaching Experiments,” he uses GPS to conceptualize technological innovation as a breaching experiment. Whereas Garfinkel’s emphasis was on using breaching to make trouble in interaction so that properties of the social order are noticed, remarked upon, and accounted for. Crabtree uses breaching as a way to make practices of ubiquitous computing visible to the ethnographer, to “make visible the contingent ways in which the technology is made to work and the interactional practices providing for and organizing that work.” The most intriguing one of the experiments he reports is the Can You See Me Now? (CYSMN) game, where runners in the streets of a city chase and catch online players. The game is similar to the Seamful Game outlined by Chalmers in his “A Historical View of Context.” Crabtree offers the point that the GPS location service in the CYSMN game was unreliable in a technical sense, but did not hamper the runners in catching online players. In other words, he attempts to show the ways runners used the interactional order of their locational work to inform their understanding of how to apply the unreliability of GPS in the game to catch the online players. “In other words, the contingent formulation of game play strategies was directly informed by the runners’ working knowledge of GPS accuracy.” In this sense, the runners took advantage of errors and gaps in the media.

Crabtree also co-authored a paper at DIS2006, along with Chalmers and several other people, “Supporting Ethnographic Studies of Ubiquitous Computing in the Wild,” which focuses largely on how to capture, manage, and review various system logs and videos as ethnographic resources in the study of ubiquitous computing. The authors revisit the CYSMN game and several other projects to outline the challenges faced by ethnographers who want to use large sets of system logs and video recordings. Anyone who does conversation analysis, just dealing with audio, on hours of data will attest to the challenges involved in analyzing the turn taking sequences in the conversation alone. Yet, even though the Crabtree, et. al. attempt to outline what capabilities a toolset would need to meet the data management challenges for such ethnographic practice, they also point to the inexorable fact that “ethnographic data, like all social science data, is an active construct. Data is not simply contained in system recordings but produced through their manipulation.” They go on to talk about the difference between system time and interaction time, arguing that the latter needs more extensive support in ubiquitous computing. In other words, they are saying that developing tools for using large sets of system logs and video data is challenging, but needed since only by unpacking the interactional order can the ethnographer put into context the logged order of events, which they characterize as “a reified order.”

Anonymous anne said...

Larry - I agree that Weiser didn't flesh out these ideas very well, although you know I've been trained like a Pavlovian dog to cite my sources properly ;) And frankly, I've often assumed that the "beautiful" bit was dropped because most people working in the hard sciences tend not to know what to do with the softness of beauty. (It certainly has never gone well for me any time I've tried to inject explicit emotion into technoscientific conversations, despite a long tradition of feminist critiques of science.)

In any case, your description of Andy's work makes sense. For whatever reason I hadn't taken the connection between breaching and seams seriously. But something about those papers also strikes me as, quite simply, rather obvious.

I mean, it certainly seems clear to me that people's lived experience of place is not really amenable to being represented solely as GPS (lat/long) coordinates, and I understand what kind of challenge that presents to an ethnographer given a set of log-files of make sense of. It also seems clear to me that it is in actual interaction that we can see *how* people work through these things.

The novelty of their approach, it seems to me, is that they are presenting ways of thinking and doing that are a bit foreign to the contexts in which they are working. And that's really cool. But I'm not convinced it offers much to social and cultural theory and practice in return. But perhaps I'm missing something here...

Anonymous Larry Irons said...

Anne, the essential point I take from the concept of underspecification is the same point that Dreyfus made initially in "What Computers Can't Do" and Suchman elaborated and detailed in "Plans and Situated Action."

It is the mutual intelligibility of people acting together that provides the central point of analysis...people understand much more than what is said or written about devices because they experience their use. The devices don't experience their own use...the people do.


Post a Comment

<< Home

CC Copyright 2001-2009 by Anne Galloway. Some rights reserved. Powered by Blogger and hosted by Dreamhost.