Karl Popper’s ‘constitution of science’ (2)

By the end of the first part of this series I listed the most important norms included in what Ian Jarvie ¹ calls Karl Popper’s ‘Proto-constitution of science’. Now I will start discussing several important problems of this Popperian rules. One first problem we can identify is whether Popper proposes these norms as just a suggestion, and “invitation to play science” according to these rules instead of according to a different set of them, as a kind of normative mandate, or if he thinks that it is the “definition” or the “essence” of science to be played in exactly this way, so that those that are not following these norms are just not doing “science” properly speaking. Another way of putting this question would be to ask if, for Popper, someone not following these norms would be a bad scientist, or just no scientist at all. Of course, having here a set of norms (rather than a single one), and taking into account that many of them could in principle be disobeyed to a higher or lower extent (something to which I will come back later), it is conceivable that the norms were followed more or less, so that one could be a “better” or “worse” practitioner of Popperian science.

The Logic of Scientific Discovery, ² with its insistence on the problem of demarcation, is written in such a way that a categorical reading of the methodological norms seems more natural with Popper’s intentions: i.e., if you don’t strictly follow the falsificationist rules, you are simply not doing science, but ‘metaphysics’ or something like that. But this seems too strict: couldn’t one prefer some times a theory with a little lower degree of testability, say, and still be a scientist, even a falsificationist scientist, as one can be a professional athlete and skip some few training days occasionally? Perhaps it is better, and more coherent with Popper’s thought, to interpret this “methodological constitution” as a (Weberian) ideal type, or as a kind of (Kantian) regulative idea. This, however, does not totally solve the problem of whether these norms constitute a definition of science, or have just to be taken as a suggestion to carry out research in a certain way. As we have seen above, that methodological norms are conventions (and not ‘truths’) invites to take more seriously the latter option. But this option would obviously lead to a more fundamental question: why should I follow those norms, instead of other possible ones (perhaps slightly different, or perhaps more disparate)? Popper’s immediate answer is, of course, that you should follow these norms in order to attain the goal of keeping your theories (or ‘systems’) as testable as possible. But of course this leads to ask the following questions:

A) would following Popper’s rules, such as they are formulated, actually lead us to maximise the degree of testability of our theories? I.e., wouldn’t it be possible to attain that goal through some more or less different means?

B) why should be testability the most important goal and perhaps the only goal of scientific practice? After all, ‘testability’ sounds like something that is most naturally a means to attain other valuable things, something that has an instrumental value, not something that is valuable ‘in itself’. Popper often speaks in The Logic of Scientific Discovery of “the progress of science”, “the progress of knowledge”, or just “progress”, as the prize that following his recipes would make us to attain. Hence, why is the maximisation of testability (and just testability) the only means to further the progress of science. What if the goals of science (and of scientists) are more diverse, more complex, and if there are more labyrinthine ways of approaching them than through an obtuse fixation with testability?

I shall devote the rest of this entry to discuss the first of these two questions. A simple way in which Popperian rules could be not efficient enough is if the maximisation of testability within a particular segment of the complex network of statements involved in any realistic research process could sometimes entail a significant reduction of testability in other segments. Perhaps in order to keep a theory highly testable we should some times accept rather dogmatically other theoretical hypotheses, either more general or profound (having to do with general presuppositions of the field), or more technical ones (like those about the use of measurement devices or statistical assumptions). I doubt that the Popperian system could offer something like a global measure of testability (if developing a workable local measure were not utopian enough) such that in cases like this, a clear recommendation would emerge. Perhaps Popper would insist that his norms are recommendations for the individual scientist, which has the obligation of maximising the testability of her portion of research. But even in this case, it is an open question whether a high number of individual researchers proceeding in this way would lead science to globally maximise its degree of testability (and ‘progress’), of if there could exist some complex, negative feedback mechanisms, making that the maximisation of global progress would better be served by at least some individual scientists behaving less ‘honestly’.

Just to be a little bit more specific: imagine that an individual scientists could choose between developing their preferred theories following in the strictest possible way the falsificationist rules, or doing it in a more ‘verificationist’ way, i.e., ‘protecting’ those theories from falsification in a high proportion of cases. Is there a guarantee that the first strategy would necessarily lead to a higher degree of global progress than the second one (which, by the way, is reminiscent of the Lakatosian ‘methodology of research programmes’³)? Perhaps in the case of science private vices also transform at times into public virtues in a Mandevillian way ⁴.