For a number of reasons, systems designers have recently shown considerable interest in ethnography. Principally this has been as a method for the specification of end-user requirements for systems. This paper argues that most of this interest is predicated in a misunderstanding of ethnography's role in social science. Instead of fixing upon its analytic aspects, designers have defined it as a form of data collection. They have done this for very good, design-relevant reasons. But designers do

more »

... ot need ethnography to do what they wish to do. In the central part of the paper, an approach to analytic ethnography in HCI is set out and illustrated. The latter sections take this approach and show how it opens up "the play of possibilities" for design. These are illustrated by counterposing a summary logic of organisational structure such as that associated with the calculus of efficiency and productivity with the "local logics" of daily organisational life.