You haven’t convinced me that Occam’s prior is false.
I would claim that example 1 is an invalid criticism because Occam’s prior only claims to work on sets of mutually exclusive hypotheses, example 2 is an invalid criticism because K-complexity is uncomputable, so the quoted hypothesis is outlawed (which may be unsatisfactory, I don’t know the solution to that), and example 3 is an invalid criticism because in the case of P(Hypothesis|Prior Information) Occam’s prior should only care about the stuff to the left of the “|”.
The other problem with disregarding Occam’s prior is that it has a proof. The proof is at least good enough to refute claims that there’s no connection with complexity at all.
Re-reading your post I think that your definition of hypothesis is wider than mine, and when I say hypothesis I mean what you call a “predictor”. If you apply Occam’s prior to predictors (instead of the “wide”/”narrow” thing) do our positions become the same?
The fourth one is the important one, and matches what Vladimir_Nesov was saying. In my second paragraph I’m explaining why each of your examples aren’t things that you should apply Occam’s prior to, even though they might be called “hypotheses”. I can’t really see how my paragraphs 1 and 3 could be unparsable.
You haven’t convinced me that Occam’s prior is false.
I would claim that example 1 is an invalid criticism because Occam’s prior only claims to work on sets of mutually exclusive hypotheses, example 2 is an invalid criticism because K-complexity is uncomputable, so the quoted hypothesis is outlawed (which may be unsatisfactory, I don’t know the solution to that), and example 3 is an invalid criticism because in the case of P(Hypothesis|Prior Information) Occam’s prior should only care about the stuff to the left of the “|”.
The other problem with disregarding Occam’s prior is that it has a proof. The proof is at least good enough to refute claims that there’s no connection with complexity at all.
Re-reading your post I think that your definition of hypothesis is wider than mine, and when I say hypothesis I mean what you call a “predictor”. If you apply Occam’s prior to predictors (instead of the “wide”/”narrow” thing) do our positions become the same?
Cannot parse your first three paragraphs. Completely agree with the fourth.
The fourth one is the important one, and matches what Vladimir_Nesov was saying. In my second paragraph I’m explaining why each of your examples aren’t things that you should apply Occam’s prior to, even though they might be called “hypotheses”. I can’t really see how my paragraphs 1 and 3 could be unparsable.