Top banner


A response from Harry Collins and Robert Evans to Michael Lynch's review of Rethinking Expertise

July 15, 2008

Readers of American Scientist who are not party to the internecine battles in the field of the sociology of scientific knowledge may be puzzled about the tone and content of Michael Lynch's review of our book Rethinking Expertise. On the one hand, the book is criticised for reporting ("strange") experiments and being too empirical, yet it is also criticised for being too philosophical. How can this be?

Lynch adheres to a very small school called "ethnomethodologists."In the 1950s the talented sociologist Harold Garfinkel asked the perceptive question: "How do members of a jury, encountering a new social situation, know how to act as proper jury-persons?" He observed juries and studied how they worked out how to act. The methods that ordinary people use to organise their world, "ethnomethodology," became the topic of the new school. Lynch was one of Garfinkel's first graduate students. Unfortunately, things soon took a bad turn and ethnomethodologists began to see themselves as an elite group opposed to mainstream sociology and, in particular, the kind of sociology that developed theories or looked too scientific. For the core group, sociology, if it was to continue to exist, had to become nothing but the minutely detailed study of ordinary peoples' (actors') methods which eschewed the creation of new scientific ideas (ideas based on"analysts' categories"). That's what Lynch means when he says our work"does not touch the ground," the current review merely reiterating oft-repeated arguments.

Those who know the history of philosophy of science will recognise in ethnomethodology a throwback to the ambitions of logical positivism.Logical positivism tried to limit sound knowledge to the immediate apprehension of the senses. But every scientific observation must involve theory or you would not know what to look for; logical positivism and science were incompatible. Likewise, no sociological generalisation can manage without analysts' categories as well as actors' categories; social science and a purist ethnomethodology are incompatible.

The ideology, then, requires ethnomethodologists to cauterise their minds against any kind of abstraction or invention of new categories which are not immediately and intimately related to how ordinary people act from moment to moment. The cauterisation procedure explains Lynch's inability to comprehend the purpose of what he calls the "strange" experiments: They involve looking at one thing in order to find out about another through analytic inference. Analytic inference is beyond the conceptual purview of thoroughgoing ethnomethodology because it involves connections made by the analyst that might not be immediately obvious to the actors described. Thus, though Lynch, like any good ethnomethodologist, has taught himself not to be able to infer over any conceptual distance, scientists to whom we have explained the experiments find nothing at all strange about them and grasp the point immediately; an account of them was published in a high-ranking peer reviewed journal and they were subsequently written up as a news item in Nature. Not only cannot ethnomethodologists infer, they also cannot make sense of the idea that analysts might invent new categories and develop new theories that only later become part of ordinary peoples' worlds – something that scientists do all the time and that is the aim of the book.

Lynch says, correctly, that our work has been the subject of much debate over the past six years. The paper to which he elliptically refers is, in fact, the second most cited paper in the history of the journal Lynch edits. You would not know this from the review and,whilst it would be wrong for us to claim all these citations as support for our position, it is surely just as wrong to imply that the "extensive debate" has been entirely negative and that we have simply chosen to ignore it.

Finally, near the start of the review, Lynch complains that the concept of interactional expertise does not tell one "how to decide in any given case which substantive fields of expertise are relevant,which experts or bodies of experts can be trusted, or just how much interactional expertise might be adequate." But as we state in the book, the "Periodic Table of Expertises" is not designed to answer that question but to provide a conceptual framework within which analysts,and with luck, actors too in the longer term, can think about it. On the other hand, Chapter 5 of the book, which is entitled "Demarcation Criteria," does deal with the distinguishing between different kinds of expert and it surprising that Lynch does not mention it. This does little to help readers of his review get a sense of the book's content or arguments.

In sum, therefore, reading the review tells one quite a bit about the concerns of ethnomethodology but very little about the book. For that it would be better to read the book or look at our website: The theory and a selection of the associated empirical work are available for download at the site.

Harry Collins and Robert Evans

comments powered by Disqus


Bottom Banner