Ethnomethodology
In The You On AI Encyclopedia
Garfinkel's founding text, Studies in Ethnomethodology (1967), introduced a radically different way of doing sociology. Rather than explaining behavior through macro-structures (class, institutions, norms), ethnomethodology analyzed how participants themselves produced the recognizable features of social scenes through their interpretive work. The famous 'breaching experiments' — in which Garfinkel's students deliberately violated taken-for-granted expectations to make visible the work of maintaining them — demonstrated that apparently stable social order was a continuous achievement rather than a structural given.
The tradition's methodological commitments proved unusually portable to technology studies. Where computer science treated users as rational agents executing plans, ethnomethodology's close attention to practice revealed users as interpretive agents constantly improvising. Harvey Sacks's conversation analysis — which developed from ethnomethodology — provided tools for analyzing the sequential organization of interaction that Suchman adapted to human-machine encounters. The photocopier video studies that anchor Plans and Situated Actions are recognizably ethnomethodological in their attention to moment-by-moment interpretation.
Ethnomethodology's influence on technology studies runs through multiple generations. Suchman's work is the most visible, but the tradition also shaped the development of participatory design, computer-supported cooperative work (CSCW), and the broader strand of human-computer interaction that takes practice rather than cognition as its unit of analysis. Paul Dourish's Where the Action Is (2001) and the work of researchers at PARC, Lancaster, and elsewhere continue the tradition.
The tradition's relevance for AI is specific. Ethnomethodology insists that the sense-making work in any interaction is produced by participants through their interpretive practices. Applied to human-AI interaction, this insistence exposes the asymmetry Suchman has documented: the human does the sense-making; the machine does not. The tradition's refusal to take abstract models as adequate to practice is a permanent resource for the critique of AI systems that claim to model intelligent behavior.
Origin
Ethnomethodology emerged from Harold Garfinkel's work at UCLA in the 1950s and 1960s, drawing on Alfred Schutz's phenomenological sociology and Talcott Parsons's action theory. The 1967 Studies in Ethnomethodology consolidated the framework and launched the tradition. Harvey Sacks's development of conversation analysis in the same period extended the methodology into the close study of talk.
Suchman's training under Jack Whalen at Berkeley connected her to the ethnomethodological tradition through its West Coast center. The methodological commitments she brought to PARC were not universal in anthropology but were characteristic of ethnomethodologically-informed fieldwork.
Key Ideas
Order as achievement. Social order is not a structural given but an ongoing accomplishment produced through members' interpretive practices.
Breaching reveals maintenance. Deliberate violations of taken-for-granted expectations make visible the continuous work of maintaining them.
Close attention to practice. The tradition's method is sustained attention to what participants actually do, as opposed to what they or observers say they do.
Sense-making is participant work. Meaning is produced in the moment by participants' interpretive activity, not retrieved from stable structures.
Applied to AI. The tradition's insistence that sense-making is participant work exposes the asymmetry in human-machine interaction: the human makes sense; the machine does not.