Nowdays every serious attempt to justify scientific realism is obliged to somehow manage van Fraassen’s employment of evolutionary epistemology against the important realistic «no-miracles argument». Thus a systematic translation of arguments pro et contra sufficiency of natural selection for evolutionary progress into epistemological language is needed. The main thesis of this paper is that it is difficult and maybe even impossible to reconcile scientific realism with classic evolutionary epistemology based on selectionist models.
Selectionist evolutionary epistemology allows for both realistic (Popper) and anti-realistic (Toulmin) interpretation while catastrophist/saltationist evolutionary epistemology (Kuhn) leads to anti-realism.
Classical, selectionist (adaptationist) evolutionary epistemology of science draws an analogy between development of science and natural selection. But natural selection immediately increases only the relative fitness of organisms with regard to specific and changing environment. Therefore evolutionary epistemology of science is exploited (by van Fraassen in particular) against scientific realism which presumes existence of absolute scientific progress as an approach to truth. In modern biology in order to explain absolute evolutionary progress nonadaptationist, nonselectionist models based on a passive trend mechanism (a random walk limited by walls) were worked out, the ratchet model in particular. This paper suggests nonadaptationist extension of evolutionary epistemology of science, namely the ratchet model of scientific progress and illustrates it by history of thermodynamics. This model enables to combine realistic concept of scientific progress as an approach to truth with antirealistic concept of scientific development as an improvement of problem-solving ability: the former is ascribed to scientific dynamics on a global scale and the latter on a local.