[par1970]

Yeah, maybe you are right. But is doing math and reasoning about Turing machines a priori? If so, then it seems plausible to me that reasoning about a codebase (without running it) is also ‘a priori’.