The problem of shared understanding, or mutual intelligibility. has defined the field of social studies for the past hundred years. On the one hand, interpreting the actions of others has been the social scientist's task; to come up with accounts of the significance of human actions is, after all, the principal charge of ethnographic anthropology. On the other hand, to understand the mutual intelligibility of action as a mundane, practical accomplishment of members of the society is, in large measure, the social scientist's problem or subject matter. An account of that accomplishment would constitute an account of the foundation of social order. Although studies of mutual intelligibility have been concerned exclusively with human action, we now have a technology that has brought with it the idea that rather than just using machines, we interact with them as well. Already, the notion of “human–machine interaction” pervades both technical and popular discussion of computers, whether about their design or their use. In the debate over specific problems in the design and use of interactive machines, however, no question is raised about the bases for the very idea of human–machine interaction as such.