Author: Uri Blass
Date: 08:57:02 04/05/02
Go up one level in this thread
On April 05, 2002 at 11:20:01, Oliver Roese wrote: >Hi! > >This position is from >http://www.cs.ualberta.ca/~tony/RecentPapers/report.mac.pdf >(see also http://www.talkchess.com/forums/1/message.html?221364) > >[D]5k1b/7P/2B1KP1P/8/8/p7/P7/8 w - - 0 1 > >The author writes: >"After more than ten hours of play the position in [figure 9] was reached and >since >neither side was making progress the game was adjudicated after white's 111th >move of Bc6-d5. " >... >"Even today I doubt if many programs can do any better." > >I tested this somewhat with crafty and can confirm it . >Although this is a rather easy win for a human it is hard for an engine, >since the winning plan distracts several heuristics. In order to win, white has >to: >1) playing f7(Be8), thereby giving more room for the enemy pieces. >2) moving his king far away from the hotspot to the innocent looking pawn a3. >3) trading the "dangerous" pawn h7 against blacks pawn a3. > >[D]4Bk2/5P1P/7P/8/8/pK6/Pb6/8 w - - 0 1 > >In this position i abandonned the search after 30 minutes (although my system is >rather slow). > >Is there a program that could do better? >Would this position remain unsolved for many years? > >Oliver I am sure that there are programs that can find trading the h pawn for the a pawn. For example my program that it's only knowledge about pawn structure is that more than one pawn in the same file is bad and knows nothing about passed pawns can find it. I also believe that it is not hard to solve the problem by the right knowledge. 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.