Author: Bob Durrett
Date: 10:09:42 11/17/02
Go up one level in this thread
On November 17, 2002 at 12:25:19, Sune Fischer wrote: >On November 17, 2002 at 12:15:47, Bob Durrett wrote: > >I think you should read here what Bruce says about that, the solution is to >extend with a small search to quiet out the position, it's call a quiescent >search: > >http://www.seanet.com/~brucemo/topics/quiescent.htm > >Really you should read his whole site! > >-S. I will! I've got it on my priority list, right after the thesis I'm looking at. Bob D. > > >>The best way for a human to evaluate a tactical position is, presumably, to look >>at lines (beginning with forced move sequences) emanating from that position. >> >>But I wonder . . . >> >>Is it possible to just look for "indicators" in the position which would >>indicate the likelihood of a combination being present? I have read some >>writings of GMs in hardcopy chess books about this. They seem to suggest that >>positions containing combinations "smell" like a combination. In other words, >>the GM is alerted to the possibility of a combination even though he has not yet >>found it. >> >>Assume, for the sake of discussion, that this is true. i.e. that it is POSSIBLE >>to detect the likely presence of a combination, without looking at any move >>sequences. It would be sufficient if it worked most of the time. >> >>If true, then it might be possible to use this to improve evaluation of >>positions, especially leaf node positions. >> >>Maybe it would take a GM to specify what the chess engine's position evaluator >>would have to look for. But is it doable? >> >>Better yet, do the present-day top chess engines already do this during position >>evaluation? At least to some extent? >> >>Just a thought. >> >>Bob D.
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.