Author: Robert Hyatt
Date: 06:45:25 10/13/98
Go up one level in this thread
On October 13, 1998 at 02:18:58, blass uri wrote: > >On October 13, 1998 at 01:53:42, Jouni Uski wrote: > >>On October 12, 1998 at 22:01:43, Robert Hyatt wrote: >> >>>On October 11, 1998 at 13:41:16, John Coffey wrote: >>> >>>>Can a program like crafty win KQ vs. KR without tablebases? (Assuming a >>>>minute or so per move?) >>>> >>>>John Coffey >>> >>> >>>Crafty can't... I doubt any program can. Most GM's can't until they >>>study it... I was at the ACM event where Browne tried to beat Belle and >>>failed miserably several times. He studied it, then at the rematch >>>he won easily. But it is non-trivial the first time, and I have had >>>an occasional GM get into this ending on ICC and offer a draw immediately. >> >>In SSDF big game archive I found about 40 KQ-KR games and for my BIG surprise >>all except one were won by Q side and this without any tablebases!! >How do you know that it is without any tablebases? >I know that fritz5 has tablebases for KQ-KR and maybe other programs also use >tablebases for this ending. > >It is not surprising that the Q won without tablebases because the defender may >do mistakes. >I think that the results between humans are similiar because it is more >difficult to find the right defence than to find the right winning moves. > >Uri >> I think your explanation is dead right. Normal "mate-finder" evaluations would keep the pieces close to the kings. But in KQ vs KR, this is often wrong and the rook has to go to the other side of the board to cover one key square. Without a tablebase, finding these moves may well be impossible, because not finding the move still doesn't result in losing the rook or getting mated within a normal search horizon... >>Jouni
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.