I think we’re splitting unnecessary hairs here; obviously we shouldn’t update our belief to something more specific than we can justify. At the same time, we want to formulate hypotheses in advance of the tests, and test whether these hypotheses are mistaken or worthy of promotion to belief, which to me seems a perfectly reasonable interpretation of what shokwave wrote.
More accurately, we should test more specific things, then become more specific. First make the test, then update the beliefs.
I think we’re splitting unnecessary hairs here; obviously we shouldn’t update our belief to something more specific than we can justify. At the same time, we want to formulate hypotheses in advance of the tests, and test whether these hypotheses are mistaken or worthy of promotion to belief, which to me seems a perfectly reasonable interpretation of what shokwave wrote.