Note: I wrote a better article about this; you should read that instead of this stub: The Problem of the Criterion.
I've previously argued for the existence of what I've called a "free variable" in epistemology that forces a choice between ways of knowing because no one way of knowing (system of epistemology or simply an epistemology) can be both complete and consistent. In the process of working on a current project and not wanting to have to rederive anything someone else has already argued for in academic literature, I discovered this feature already has a name and has been written about: epistemic circularity.
I find it sort of surprising we've not addressed this more within the LW community, although it's perhaps less surprising than might otherwise be expected given LW's positivist leanings. I don't have much to say on epistemic circularity at the moment, although I do consider it critical to my worldview and a crux of my thinking about philosophical conservatism for alignment research, but I did at least want to bring some wider attention to a concept that, to my recollection, we've ignored as a community.