How and Why to Localise the Scientific Realism Debate: Making Historical Arguments in the Scientific Realism Debate Compatible with Methodological Pluralism

I argue for a new way of localising historical arguments in the scientific realism debate. We should see historical arguments as attempts to empirically assess scientific methods. For such arguments to be good, they need to be about a single method. Therefore, only if there is a unified method of science can historical inductions on science license general conclusions about the epistemic status of current science. However the consensus seems to be that there is no such unified scientific method. Various versions of methodological pluralism seem to undermine any attempt to assess scientific methods through historical means, as they make it hard to see methods as persisting through theory change or as applying beyond a very specific field. In particular, views of scientific methods that see them as highly context specific seem to undermine any kind of historical realism debate. I attempt to outline a way in which we can individuate scientific methods in order to empirically test them. I also argue that the impact of context can be accounted for in a way that still leaves room for historical assessments of methods if we categorise contexts according to types of difficulty. The view of the historical scientific realism debate we end up with is one in which various methodological resources are argued to be either unreliable or reliable for a given type of difficulty, based on evidence from the history of science. These conclusions about the reliability of methods may be relevant to the epistemic status of a given theory, but establishing which methods and difficulties are present in an actual scientific context requires detailed engagement with local evidence. I compare my position with other localist views and explain how it offers more role for historical inductions on science than some other localist writers.