Author: blass uri
Date: 04:19:19 10/22/99
Go up one level in this thread
On October 22, 1999 at 06:45:57, Georg v. Zimmermann wrote: >Hiarcs learning feature is interesting, but far from smart enought to be of real >help. If Bxb7 looses because of opening lines to the 0-0-0 white king, you can >show Hiarcs, it will not consider this move in the given position again. But it >will on the next move, and on the next move... and as black it will try to >protect b7 ... It is better than nothing and there are cases that losing by taking b7 is a mistake only for one move. >And: the guy who said that you get best results when making your own analyses >and using a comp-program (till it gets to maybe ply9,10,11) to look for >blunders was coorect. IMHO. >--Tec I do not think that using comp-programs only for blunders is a good idea because sometimes programs can find good positional ideas by search. I believe the best way to use computers is also to give it a long time for positions(when you sleep) and also to do a tree. I test Junior5.9 in correspondence games(3/4 final of Israel championship) because I am interested in the level of computers in correspondence games. Junior5.9 has now a clear advantage with black in one of my correspondence game(it evaluates it as 0.65 pawns and I evaluates it as a win for it or a draw) My opponent did not do tactical blunders(the evaluation never changed by more than 0.3 pawn). I know that my opponent told me that he does not use a computer and he does not use a lot of time for his moves but he was strong enough to get 5.5 out of 6 in the first stage(1/2 final). I talked with him by telephone for some days that he was not at home and he usually replied after 1 hour,2 hours or 10-20 minutes. 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.