Author: Shep
Date: 01:27:35 03/17/00
Go up one level in this thread
Very well said. Re: the tablebase analogy, this raises another question (which I haven't seen discussed before): How high is the percentage of positions where there is only one "perfect" move (i.e. "one move leads to a win, all others draw/lose" or "one move leads to a draw, all others lose")? I suspect it is much less than 0.1% in the 5-man case (someone with good knowledge of TBs could probably write a small program that can get this kind of info from the tablebases). I think it is not unreasonable to assume this also holds for any other N-man situation (5<N<=32). So there are many "perfect" moves in almost every position, making it quite likely a strong human can very often draw a perfect program. Re: increasing draw quota due to tablebases, I don't think that's the reason. Maybe in the Blitz games on ICC, but e.g. in my tournaments I see draws mostly because of improved evaluation and extended opening books. About 95% of the draws are reached before either program gets to probe its tablebases. --- Shep
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.