Author: Uri Blass
Date: 16:00:19 12/10/04
Go up one level in this thread
On December 10, 2004 at 17:48:31, Albert Silver wrote: >On December 10, 2004 at 13:20:58, Uri Blass wrote: > >>On December 10, 2004 at 11:23:22, Dana Turnmire wrote: >> >>>Wouldn't it test the evaluation function? Just a thought. >> >>I doubt if it is possible to tell programs to search to the same depth with >>exactly the same extensions and the same pruning and even if it is possible then >>better evaluation function does not mean a better program. >> >>The problem is that programs may never evaluate some types of unclear >>positions in normal games thank to extensions when they are forced to evaluate >>them in fixed depth Brute force search. >> >>Uri > >Another problem is that it will favor the program that depends entirely on the >evalaution function to produce quality moves. There is more than one approach to >chess programming as is completely obvious, so imposing this sort of limit is to >deliberately handicap the fast searchers that compensate their lack of >"knowledge" by searching deeper. It's not unreasonable either if you remember >the adage: "Chess is 99% tactics". > > Albert I do not think that fast searchers are always programs with faster evaluation. I also think that there is some knowledge that is impossible to replace always by going deeper(a piece may be trapped when many plies may be needed to capture it and in part of the cases the capture of the piece can happen many plies after the losing moves and programs without the right knowledge are unable to see it even with a long search). Uri
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.