CHI update #3: Paul Dourish smacks the CHI community upside the head
Paul Dourish gave a fabulous talk about the real roles ethnography should play in human-computer interaction, and the problems that currently exist with the assumptions of ethnography. I highly encourage anyone in human-computer interaction to read the paper - it's very well-written. Here's a summary of the talk, as much as I could get down.
Ethnography is a long-term, immersive, participant observation; it is a text; and it is inherently interpretive and analytic. There are a lot of misunderstandings of ethnography in HCI, though. Ethnographers have received many negative reviews for the lack of a (strong) "implications for design" section - these are a few Dourish collected from others:
"If you were to write a proposal of new technologies and use your data to support the design, the result would likely be a quite strong paper."
"How does this inform design in new ways?"
"The more technically inclined audience could be forgiven to think that CHI research on the topic is all talk."
"Does not provide an actionable framework ..."
... and more (including one from a paper I helped write ... *cough cough*)
In particular, these four issues exist between ethnography and HCI/design:
Marginalization of theory: ethnography was seen as a "toolbox" of field techniques, the ethnographer as a "tape recorder." (Diana Forsythe famously said that "an ethnographer is not a tape recorder" -- an ethnographer must take an analytic stance, choose an interpretive practice.) Objectivity and subjectivity: Paul notices that many scientific/engineering students are uncomfortable with ethnography because they see interpretation as subjective, but ethnographic data isn't just "collected," it's generated from the encounter of the ethnographer and the field settings, and proceeds from a consciousness of what is or may be subjective. Interpretation and analysis are central. Doesn't mean that everything is "hopelessly" and "problematically" subjective, though.
Disciplinary power relations: what is implied by the insistence for implications for design? Instead, perhaps we should be asking what are the implications for theory. Why is design a natural end-point? Why does theory/analysis seem like an unreasonable end-point? Design is a privileged activity in HCI -- this shows the asymmetry between the disciplines and how much they're valued.
Relationship between technology and practice: the common view is that ethnography will uncover problems that design can fix. This assumes that the world is problematic and can be fixed by (technological) design. A better approach would to have a broader view of practice, including how technology is put to use (and adopted, adapted, repurposed, and appropriated), how people create new circumstances and consequences of technology use, and how technologies take on social meaning. To formulate practice as "deficient" or "needing to be fixed" presupposes a lot, and also puts design outside of the domain of the ethnographer.
Representation: over the last two centuries, anthropological ethnography has grown from "objective, instrumental, actionable" accounts to situated encounters. The former is now what is requested in technology and product design, though.
Finally, it's not that implications for design are bad -- they can be productive, and part of conversations between ethnographers and others. Ethnography also isn't just abstract and academic. It's just that the absence of implications for design shouldn't disqualify an ethnography -- they're a poor metric for evaluating ethnographic work. The narrow focus on design can underplay ethnography's contribution. Interdisciplinary engagement (ethnographers, designers, etc.) is also important, but we have to be conscious of the power relations that are going on between disciplines. (For example, CHI has out-priced a lot of non-technical disciplines!)
... Speaking of fees, Paul isn't the first to grumble about the fee hike this year. Jofish blogged it a while ago, and a group of students held a "CHI Student Fee Bake Sale" in protest. Hopefully ACM gets the message.