Author: Robin Smith
Date: 11:09:18 11/09/01
Go up one level in this thread
On November 09, 2001 at 11:01:00, William Penn wrote: >On November 09, 2001 at 10:47:25, Gian-Carlo Pascutto wrote: > >>On November 09, 2001 at 10:44:53, William Penn wrote: >> >>>Maybe, maybe not, I >>>have no idea. But my main interest is accurate analysis, so it is a serious >>>fault to me. >> >>If you need _accurate_ analysis, you should never use anything >>but the first move from the PV anyway. >> >>If you let it search to depth n, the second move is based on a >>n-1 depth search. The third one on an n-2 depth search. The moves >>at the end are quite often outright blunders. >> >>-- >>GCP > >No! That is not correct! > >Only the Fritz 7 (with MMX) engine produces outright analysis blunders, based on >my long experience of many years, running many different engines then >scrutinizing the analysis carefully. > >True, analysis is less accurate the deeper you go, but there should be no >outright blunders. Even at the end of the line, although the moves may be weak, >they should not be stupid blunders such as giving away one's Queen. > >Again, I have not seen such serious analysis faults (genuine, outright >blunders!) with any other engine in the last few years. None of them have ever >given away their Queen, or overlooked taking opponent's Queen, in my experience >to date -- except the new Fritz 7 (with MMX) engine!? >WP I have seen Shredder 5 show PV's that look like giveaway chess, throwing away piece after piece for no good reason. As I understand it, some programs have trouble reconstructing a PV from hash tables. The PV's and position eval end up having nothing in common, with only the eval and 1st move of the PV being valid. This could be what you are seeing with Fritz7. -rs
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.