Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: How much time does your program need to see the draw?

Author: Robert Henry Durrett

Date: 09:11:57 06/21/02

Go up one level in this thread


On June 21, 2002 at 10:24:11, Uri Blass wrote:

>On June 21, 2002 at 09:24:04, Richard Pijl wrote:
>
>>On June 21, 2002 at 08:53:00, Robert Henry Durrett wrote:
>>
>>>On June 21, 2002 at 06:28:27, Kurt Utzinger wrote:
>>>
>>>>On June 21, 2002 at 04:38:04, Uri Blass wrote:
>>>>
>>>>>[D]8/8/8/5P2/1p1b4/pNk5/K7/8 b - - 0 103
>>>>>
>>>>
>>>>There is no chance for today's computer programs to see that this is a draw.
>>>>Even without the white pawn, none of the 20 programs tested by me have an idea
>>>>of draw.
>>>>
>>>>Kurt
>>>
>>>Why, specifically, is this true?  [I do not challenge the statement, but ask
>>>about the software.]
>>>
>>>What, specifically, would have to be done to a chess engine to make it see these
>>>things?
>>>
>>It doesn't seem very difficult to add this knowledge, just a matter of doing it.
>>The problem is that white has sufficient moves to avoid zugzwang, the squares
>>before the pawns are controlled by white and the bishop is of the wrong color,
>>not able to support the advance of the pawns.
>>Richard.
>
>Yes 6 squares is practically enough to avoid zunzwang when the danger of threat
>is only from a king and a bishop
>
>The bishop can control 4 and the king can control 2 but they cannot control all
>the relevant 6  different squares.
>
>king at e2 and bishop at e5 can control 6 different squares that the knight at
>e4 controls but the knight at e4 has 8 legal moves and when the knight has only
>6 legal moves it is not possible.
>
>Uri

To me, the real issue is whether or not the given position is representative of
a class of positions which present the same sort of problem, as opposed to being
an isolated case.

Introducing knowledge specific to the given position would surely be a waste of
time.  The "fix" needs to be for an entire class or type of position.

Is the given position representative of a large number of other positions for
which modern chess engines all have the same problem?  If so, what would all of
these positions have in common?  Is there a "fix" which would make everybody
happy?

Bob D.



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.