Implications for Design - Dourish 2006

CHI 2006: I missed attending one of the more provoking presentations of the conference -- Paul Dourish's Implications for Design. So, I was forced to read the paper instead.

What did this paper say?

CHI papers that report on ethnography have a common closing section: "Implications for Design". CHI reviewers are evaluating the quality of ethnographies based on this closing section. Dourish critiques this measure of ethnography. He proposes that ethnography has more to offer the field of human-computer interaction (HCI) than a list of design ideas. (Worse, ethnographies are not tools that are meant to find "gaps" or unfulfilled technological needs. This means ethnography does not lend towards design solutions.)

What is ethnography good for? One, Dourish does not dismiss design implications outright. What he does is to ask researchers to articulate why they thought of a particular design implication: "the analytic and conceptual work that lies behind ... is often where the substantive intellectual achievement is to be found." This is because ethnography is not just a fact-report, it involves interpretation and perspective. Explicitly writing out one's interpretations and perspectives is a lionshare of what an ethnography is. One ought to include this analysis; and, if one draws design implications from this ethnography, one should detail the rationale that connects the analysis to the design implications. Design implications are weaker if they lack a strong connection to the observations.

Second, Dourish points out that ethnographies are very valuable in themselves. An ethnography is an ethno-graphy. That is "people/race/culture"-"writing". It is a written perspective of a culture. Because of this, implications of design do not simply fall out of an ethnography. This leads to the first point above (requiring researchers to be explicit about their steps of analysis). This also leads to the point that there are other benefits of describing and analyzing an ethnographic site. For example, Dourish suggests that you can have implications for non-design ("we've discovered that you shouldn't use IM here"), implications for policy (rather than technology design solutions), and implications for the nature of design itself (revisiting the traditional idea of designer versus user). Notice that all the benefits I list here ultimately result in design-related decisions.

Response

Despite Dourish's critique of "implications for design", I still see this paper in accord with the basic goal of the CHI/HCI community, which is to engage in design. ("CSCW is basically a design oriented research area." - Schmidt and Bannon 1992) I do not believe this is a paper that says, "Design implications are evil". Nor do I believe it is a paper that says, "We should not care about design; we should care only ethnographies for their own sake." This paper still accepts the HCI desire for change (i.e. 'design'). Although Dourish has noted that sometimes we should just leave the people we study alone; and, that design implications we learn from studying them don't have to apply to them - maybe to other people instead.

Also, I am interested in seeing how ethnography can really change our understanding of design. Perhaps there are more complicated categories than "designer" and "user". (I guess its time to read about participatory design and etc...)

Overall, it ought to be interesting to see the direction of HCI as the roles of "designer", "builder", and "user" get redefined.

Schmidt, K., & Bannon, L. (1992). Taking CSCW seriously: Supporting articulation work. Computer Supported Cooperative Work, 1, 1-2, 7-40. PDF: http://www.itu.dk/people/schmidt/papers/cscw_seriously.pdf

Related posts:
Intuity, pasta&vinegar1,pasta&vinegar2,chimerically,m.blog (posse sticker),technotaste (anthro pov)

//Methodology //CommentsorTrackbacks: 12 //permanent link

Currently Reading: The Discovery of Grounded Theory by Glaser and Strauss
The Discovery of Grounded Theory

Just a note from Chapter one so far:

In contrasting grounded theory with logico-deductive theory and discussing and assessing their relative merits in ability to fit and work (predict, explain, and be relevant), we have the position that the adequacy of a theory for sociology today cannot be divorced from the process by which it is generated. Thus one canon for judging the usefulness of a theory is how it was generated--and we suggest that it is likely to be a better theory to the degree that it has been inductively developed from social research.

"likely to be a better theory". I will dwell on this for today. Is it reasonable to pick and choose among methodologies for theory-generation? What if I wrote a purely conceptual paper, say on patterns of CMC communication or comics-information behavior--whatever, and stated clearly what is my method of theory-generation (say, logical deduction or some kind of conceptual exercise). Would that be acceptable? I think that the answer is, "If your audience agrees with you, then so be it."

If so, then what if I considered myself a very flexible theory-generator? Today, I will generate a theory using the principles of grounded-theory. Tomorrow, I will use a purely logical argument. Thursday afternoon, something else. Thursday evening, grounded-theory again. Is that possible? Why or why not?

As a footnote, what are other methods of theory-generation?

//Methodology //CommentsorTrackbacks: 0 //permanent link