Author: Louis Fagliano
Date: 07:59:28 08/15/03
Go up one level in this thread
On August 15, 2003 at 10:28:54, Anthony Cozzie wrote: >nt I think the reason why Anthony Cozzie claims that computer shootouts are worthless for determining these things is that if the computer makes just one less than best for for either side all the remaining moves/anaylsis would be wrong. The best way to determine the truth for a position is to use Chessbase 8 and have your favorite engine running in the background. Make the move that the engine thinks is best by pressing the space bar and keep doing so every minute or so. Eventually one side or another's eval will start to increase or the eval will head for 0.00. Then most importantly, to find out if this is the correct adjudication of the original position, use the arrow keys to go backward in the variation and either use your own chess abilities to find better moves for the side who's eval started to get worse, or see if the engine itself can find a better move for the side that is starting to lose. The engine will keep hashtables of the variation as you go backward and know which side's eval started to get worse. Now seeing that position again when you went backward in the variation, it will "change it's mind" about was was the best for that side and will often find a better move for that side. Keep doing this for several iterations so that you have sub-variations and sub-sub-variations and you'll end up with a much better analysis of the truth of a certain position that just using the shootout feature.
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.