Author: Ernst A. Heinz
Date: 16:24:35 01/06/99
Go up one level in this thread
On January 06, 1999 at 13:46:39, Jon Dart wrote: >I have been running a test suite from ECE (Encylopedia of Chess >Endings) for a while. > >The very first problem is this one: > >R7/8/8/8/3K4/8/3kp3/8 w - - bm Ra1; id "position 0001"; > >Crafty and Arasan both prefer Ra2+. Arasan's PV is >Ra2+ Kd1 Kd3 e1=N+ Kc3 Nf3 etc. In words, White forces >Black to under-promote and goes into a KR vs KN endgame, >which is almost certainly drawn. > >Ra1 draws immediately, after e1=Q Rxe1 Kxe1. > >It is pretty hard to say that Ra2+ is really worse, but >a human player would probably see that both approaches >draw and opt for the draw right away. Computers don't >do this because they see a material advantage in having >the Rook versus the Knight and this looks better than >the even score a draw gives (Arasan gives a reduced >material score in this case but it still favors having >the Knight). > >It looks to me like some of the other ECE problems have >a similar theme. The publicly available ECE test suites are notoriously buggy and incomplete. Many proposed "solutions" even contradict our omniscient endgame databases ... =Ernst=
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.