The model proposed above raises a quite different set of questions from those explicitly tackled by Kuhn and Popper, and shifts the focus of the debate. How should the balance between functions be struck? That is to say, for any given group of scientists, how many should be fulfilling puzzle solving functions, rather than critical functions? And of those performing critical functions, how many should, say, be performing evaluative functions? These questions,and others like them, may not have ‘hard and fast’ contextually invariant answers. Instead, the proper distribution of activity may depend on the skill base available—e.g. perhaps despite their
best efforts, some people cannot feign being dogmatic when they are not, in so far as they cannot really push themselves to defend a theory come what may—and also the state of science at the time. If T4 were evaluated as suffering from severe defects but there was nothing else available to put in its place, for instance, then perhaps more imaginative effort would be required. Similarly, if T5 remained
untested and unchallenged, then perhaps more offensive and defensive interplay concerning T5 would be merited. (So note also that the wisdom of occasional episodes resembling revolutions, but not quite so extreme and wide-ranging as Kuhn’s model demands, may be accounted for.)
I should emphasise that I have not denied that there is a fact of the matter about what an individual scientist might best do (or best be directed to do) in a particular context of inquiry. Rather, I have suggested that determining what this is requires reference not only to the state of science understood as a body of propositions (or as knowledge) but also to what other scientists are doing and the
capacities of the individual scientist. Consider a new professional scientist, going into his first postdoctoral research project; and let his capacity for good work be fixed by his interests, desires, and experience. Assume he could work just as well in group B as in group A. It might be preferable for him to join A because its line of inquiry is
more promising than that of B, on current indicators, although it has fewer members.21 So in short, I take there to be measures—even if they are rough measures, such as Popper’s corroboration function22— of how theories (and/or research programmes, modelling procedures, etc.) are faring. And these, given the resources at our disposal, determine how we should respond. A simple analogy may help. Imagine you, the chess player, are managing science. The pieces are the scientists under your command, and their capacities vary in accordance with their type (e.g. pawn or rook). The position on the board—nature is playing the opposing side—reflects the status quo. And now imagine you are told that, against the rules of normal chess, you are allowed to introduce a new pawn (which you can place on any unoccupied square).23 (This is akin to the introduction of a new scientist; pieces
working in combination on your side can be thought of as research groups, and so on.) Some moves will be better than others, given your aim of winning the game, and in some circumstances it will be clear that one available move is best. So my own view is that considering social structure neither precludes employing insights from what might be called the ‘logical’ tradition in the philosophy of science—formal apparatus, such as confirmation or corroboration functions, for instance—nor requires
acceptance of the view that studies in scientific method always require reference to the history of science. Social structure is relevant to questions of scientific method; but it is hardly as if when we discuss groups, rather than individuals, we suddenly find ourselves in territory where the ‘logical’ tradition has nothing to offer.24 The picture presented in this paper is complex, and the questions enumerated in this concluding section are daunting. It may prove to be the case that they are beyond our power to answer satisfactorily except in highly idealized contexts. Nevertheless, it appears that complexity is necessary if we are to truly get to grips with the question of how science should work. At the very least, the model here considered, e.g. as presented in Fig. 6, provides a basic framework with which to tackle practical questions when considering the research activity of a group (or groups). And even
if that model is rejected, to consider functions at the level of the group is arguably to make an important conceptual breakthrough in understanding (and therefore shaping) science. If there is one message to take away, it is that ideal science may be realizable in more than one way.