ABSTRACT
Mobilising all available scientific expertise in an attempt to objectify our social milieu.
(Bourdieu, 1988:5)
In the past two chapters we have explored two major ways in which graduate students learn to produce particular kinds of knowledge. The disciplines we have discussed in detail-social anthropology, human geography and biochemistry-are defined in large measure by those distinctive research orientations. Their defining characteristics include these major types of research practice: for the social anthropologist or human geographer, field research defines the paramount reality of the subject, while the biochemist relies implicitly on the work of laboratory experiment. For each, knowledge production is defined in terms of a method and a site-‘the field’ and ‘the bench’, respectively. In this chapter we turn to a third mode of organizing knowledge. We examine the work of computing, in two rather different disciplinary contexts-physical geography and artificial intelligence. The specific contents of these subjects are different, but their common features depend on the use of computer science to create models. (All scientists produce models and representations of the natural or social world, of course: here we are referring to the specific means and ends of the research process.) The bench scientist’s primary concern seems to be ‘Can I get my experiment to run?’ and the field researcher’s concern is ‘Can I survive and can I make sense of all this?’. The computer scientist’s interest is ‘Will this program run?’ and ‘Will this model yield the right predictions?’. While their programs and models do ultimately aim to reflect an independent reality, these scientists are strongly focused on the internal coherence of their computational work, and are often working at some remove from observations and data in the ‘real world’.
