Etienne Wenger's Communities of Practice (1998) deals with ‘a kind of community created over time by the sustained pursuit of a shared enterprise’ (1998:45). The shared enterprise to which he returns throughout the book is a case study of an insurance claims department, which may be new to most of us in its details but is familiar in its outlines to anyone who has ever had a routine job in an office: the forms, the questions, the boss, the coffee break. We can admire the skill, resourcefulness and collaboration of a team doing a complex and potentially boring task, dealing with the anomalies and changes of any such complex process involving many institutions. It is a useful insight to focus on the gradual process of learning through ‘legitimate peripheral participation’, informal consultation, community memory and trial and error. I thought of many parallel cases that Wenger's overview might illuminate: my work as admissions tutor in a university department, scientific research groups (Latour and Woolgar 1979, Knorr-Cetina 1981, Lynch 1985, Ochs and Jacoby 2000), Thomas Edison's workshop (Bazerman 1999), US political parties (Schudson 1998) or an advertising agency pitching for a new account (Rothenberg 1994). All these processes involve a community around a shared enterprise that goes beyond what one person could do and produces something that may be useful (even if it is just a particularly flashy ad for Subaru).

But what if the enterprise is producing not only wealth, but risks?