The below are my notes on "Ways of Knowing in HCI".

Author: Paul Dourish

Paul Dourish offers the following simple definition:

Ethnography is an approach to understanding cultural life that is founded not on witnessing but on participation, with the goal of understanding not simply what people are doing, but how they experience what they do.

The idea is to, rather than studying a group from the outside looking in, to embed oneself in the group, and take note of not only the activities of the group (as one could have done by looking from the outside), but also the way that the group's activities personally affect the researcher, which is an aspect that would be impossible to understand from outside observation. In this way the ethnographer can gain insight which would otherwise be impossible to gather by traditional means of observation.

The obvious next question is, “doesn't the ethnographer alter things by being there?” The answer is, of course, yes, but the argument goes that every participant alters the scene by being there, and so the thing that the ethnographer studies is not so much the group in a vacuum, but the group in the presence of another (which is the state of the group anyway, by its definition).

Another way I think about this question is that the ethnographer is not granted “special privilege.” They happen to be taking note of the scene, maybe with an intent of brining the cultural information they gather to their own, perhaps different cultural group, but every participant also takes note of the interactions in the group, and has their own motivations. So, the role of the ethnographer might not be so different from an “ordinary” member of the group after all.

In the following discussion about the history of ethnographic practice, the notion of “multi-sited ethnography” seems especially relevant to the study of computer science. The idea is that the objects of ethnographic inquiry are not confined to a single site. Rather, they have far reaching influence, and so must be studied with the understanding that they might affect and have different meaning and purpose to different groups.

Ethnography is a practice which is concerned with details. That is, it tends to resist generalization. As Dourish says:

Ethnographic work at the conceptual level may work best not by proving answers but by raising questions, challenging perceived understandings, giving silenced perspectives voice, and creating new conceptual understandings. That is, it may be destabilizing rather than instrumental

In the context of design, ethnography seems well suited to understanding design requirements, and as a means of engaging with communities who might engage with the designed artifact. Ethnographic methods might give insight into these artifacts as places where culture is generated and interpreted. That is, it might lend insight into how people perceive the technology that they use (e.g., social media), giving a high-level view where other methods might be too concerned with minutia.

The chapter ends with some questions we might ask about ethnographic research:

• What are the work's empirical claims?
  • In other words, what has been observed? What conversations/observations have been generated by the ethnographer?
• What are the work's conceptual claims?
  • What has the work attempted to explain by generalization?
  • It is important to differentiate these claims from the empirical ones.
• How was the work produced, and in what ways?
  • It is here that we should expect to see some evidence that the ethnographer was a participant in generating their observations. Omitting the ethnographer's own role isn't correct, because the ethnographer's presence is inseparable from the observations.
  • We would also expect some understanding of the context in which the observations were made.
• How does the work contribute to the corpus?

And some questions which don't really make sense in the context of ethnographic research:

• “Is this a representative sample?”
  • As described above, ethnography is concerned with reporting detailed, singular events. It is also especially adept at targeting underrepresented groups. Statements made by participants can't be interpreted as anything more than evidence of the exact occurrences themselves.
• “How can you tell if what people told you is right?”
  • Dourish says ethnography is “not a question of dividing the world into true statements and false ones; all statements and all actions at all times are produced to meet the immediate circumstances of the moment.” A lie is just as informative as the truth; what's important is the response of the participant.
• “What should I build now that I know this?”
  • Dourish says “the implications for design … lie not within the ethnographic text itself, but rather in the way in which it reframes the contexts and questions of design.” In some sense, the ethnographic text is “low-level.” It is used, as mentioned above, as a destabilizing force, a way to understand different perspectives, a way to understand unintended effects. Design occurs at a higher level than the text, since it necessitates generalization.