Is It Reasonable to Assume that Fluctuations Are Probabilistically Distributed?

Abstract: This paper asks the question whether it is reasonable to suppose that fluctuations or perturbations affecting the course of some physical process can be represented probabilistically, as is often assumed in the construction of scientific models of biological and social phenomena. I argue that, in the absence of knowledge about the kind of system being modeled, there is no good reason to make the assumption, except perhaps that it has worked in the past. Assuming that the assumption has worked in the past, this should be regarded as a surprising truth, rather than something we ought to have taken for granted, and it stands in need of an explanation.