>First of all, you are assuming a frequentist interpretation of probabilities
>when you talk about drawing outcomes from a distribution. But the whole point
>of the Bayesian approach is that the laws of probability theory apply equally
>well to degrees of confidence/plausibility/belief -- you don't have to tie them
>to limiting frequencies. Used in this way, a probability distribution encodes
>one's state of information, and says nothing (directly) about frequencies.
However you interpret Bayes rule, you make the assumption that the
evidence upon which you are conditioning is germane to the proposition
in question. If you apply Bayes rule to induction, you are assuming
that the past is germane to the question of whether induction will hold
in the future. This is circular.
The frequentist/subjectivist issue is a red herring.
--
Ron Parr email: [EMAIL PROTECTED]
--------------------------------------------------------------------------
Home Page: http://robotics.stanford.edu/~parr
