Author: Robin Smith
Date: 15:15:01 02/26/03
Go up one level in this thread
On February 26, 2003 at 10:57:50, Stephen Ham wrote: >Dear Jose, > >Thanks for noticing. I'm not the computer chess expert that you folks are, so I >wrote it through the eyes of an average person. > >There were a couple small errors that I wish I had caught. My review was >completed in mid-January, well in advance of the February rating list. So I made >a couple of quick revisions to my review when the latest list came out, but now >see that I missed a spot. I also found a typo where I wrote "speach" rather than >"speed." I hope that didn't ruin the review for you. > >As a non-expert regarding chess computers, please allow me to ask a question >that's probably very stupid. It's clear that I found Shredder 7 to be a very >strong chess engine, in fact, one of the strongest to date. Yet I also found >that its ability to evaluate positions was not equally good. In fact, I think >Fritz 6a and Fritz 7 have more accurate evaluation functions. Is there not a >direct coorelation between a chess engine's ability to pick the right move and >its ability to corectly evaluate the position? > >Thanks in advance. > >Stephen The simple answer to your question is no. For example if you were to arbitrarily multiply every evaluation of a program by a factor of 10, the program would play identically. Shredder has a more "optimistic" eval than Fritz. I often see Shredder thinking one side is winning, while Fritz will think things are about =. Rarely is it the other way around, unless Fritz sees something tactical that Shredder is missing. In some positions Fritz is objectively more accurate, but in others Shredder will be. Programs often simply don't agree. I routinely see evaluation deltas of a pawn or more. Sometimes far more. And not just in artificial or unusual positions. Robin Smith
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.