Author: Albert Silver
Date: 14:33:58 06/04/00
Go up one level in this thread
On June 03, 2000 at 19:24:08, Robert Hyatt wrote:
>On June 03, 2000 at 10:41:58, blass uri wrote:
>
>>My observation is that programmers did not try to teach program some rules
>>about endgames(for example about the fact that KRPPP vs KRPP is almost always a
>>draw when all pawns are on the same side).
>>
>>
>>I know no program that knows that this position is probably a draw(evaluation of
>>less than 0.5 pawns advantage for white):
>>
>>[D]5rk1/6pp/8/8/8/8/5PPP/5RK1 w - - 0 1
>>
>>I am interested to know if there is a programmer who try to teach his(her)
>>program as much as possible about endgames by teaching it rules from chess books
>>without caring about nps and about being better in games.
>>
>>Uri
>
>
>I do this all the time. I just haven't gotten to this case yet. Nor the case
>of Q+P's is _very_ difficult to win with a 1 P advantage.
Yes, it has all sorts of curiousities. Take a simple K+Q+P vs. K+Q. The best
strategy for the defending side usually is to paradoxically move the king _away_
from the pawn. The reason is simply that although the pawn will probably advance
very far (7th), in the end the attacking side can easily be caught in a
perpetual and chances of it finding a square where it is safe becomes
impossible. If the defending king is too close there are all sorts of spots
where the queen can't check due to a defending counter check that trades queens
or gives the pawn a tempo to queen. There are also the numerous problematic
situations where the rook is as strong as the queen, such as fortresses or a
rook behind an advanced passed pawn.
Albert Silver
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.