The philosophy of science has produced numerous accounts of how scientific facts are generated, from very specific facilitators of belief, such as neo-Kantian constitutive principles, to global frameworks, such as Kuhnian paradigms. I consider a recent addition to this canon: van Fraassen’s notion of an epistemic stance—a collection of attitudes and policies governing the generation of factual beliefs—and his commitment to voluntarism in this context: the idea that contrary stances and sets of beliefs are rationally permissible. I argue that while (...) scientific inquiry inevitably favours a high degree of consensus in our choices of stance, there is no parallel constraint in the case of philosophical inquiry, such as that concerned with how scientific knowledge should be interpreted. This leads, in the latter case, to a fundamental and apparently irresolvable mystery at the heart of stance voluntarism, regarding the grounds for choosing basic epistemic stances. (shrink)