Re: Optimal Prediction

[H J Ruhl]

Dear Hal Finney:

That is interesting.  But then I model things as: no predictor whether it 
has worked before or not can be trusted to perform the same re any future 
states.

Hal Ruhl