Author: Uri Blass
Date: 01:50:38 11/26/02
Go up one level in this thread
On November 25, 2002 at 21:12:16, Dieter Buerssner wrote: >On November 25, 2002 at 11:19:50, Uri Blass wrote: > >>If you claim a win then you should beat programs with black and not chess tiger >>but programs that claim that it is a draw. >> >>Note that yace did not play 4.f4 in my analysis >> >>After 4.f4 Qc3 Yace feels bad with 4.f5 so finding Qc3 may be a good test >>position > >Uri, it is nice to see, that you use Yace. But I really think, in this sort of >endgame, one should not easily trust engines. Some arguments. Assume KQPKQ and >an engine without TBs. Also, ignore for the argument the 50 moves rule. There >are positions in KQPKQ, which need more than 50 moves, until a pawn will >advance. Earlier in this (or in a related) thread, you argued with tactics. When >50 moves are needed (or when 30 moves are needed) until a pawn advances, tactics >will be out of reach for (almost?) any chess engine. > >You critizised Tiger for the high eval. I did not look close enough, neither do >I have the knowledge. But in general it seems possible to me, that with rather >sophisticated knowledge in the engine, such a score is reasonable. You are theoretically right but practically engines know almost nothing and I am sure that they do not have sophisticated knowledge in queen endgame. Engine usually even do not know about position of KRPP vs KR that are drawn based on book so it is not logical to expect tiger to have that knowledge. A game of tiger against Crafty proved that tiger evaluation can drop to 0.00 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.