Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: A problem with tablebases

Author: José Antônio Fabiano Mendes

Date: 10:15:53 04/03/00

Go up one level in this thread


On April 03, 2000 at 12:47:43, James T. Walker wrote:

>On April 03, 2000 at 12:11:53, KarinsDad wrote:
>
>>On April 03, 2000 at 11:58:30, Dave Gomboc wrote:
>>
>>>On April 03, 2000 at 11:53:34, KarinsDad wrote:
>>>
>>>>On April 03, 2000 at 11:44:44, blass uri wrote:
>>>>
>>>>>On April 03, 2000 at 10:54:21, KarinsDad wrote:
>>>>>
>>>>>>On April 03, 2000 at 06:10:21, Jeremiah Penery wrote:
>>>>>>
>>>>>>>There is a problem with tablebases.
>>>>>>>
>>>>>>>No, they don't produce wrong answers.  And no, they don't make engines play
>>>>>>>weaker.  The problem is that engines would rather be -9 and not in TBs than see
>>>>>>>mate against itself and be in TBs.  I have one good example, from a game my
>>>>>>>modified Crafty played on ICC, and I may try to find it, if possible.  But I do
>>>>>>>remember what happened:
>>>>>>>
>>>>>>>I had something like a knight and a pawn, and my opponent had 2 knights and a
>>>>>>>pawn, about to promote.  I had the choice to let him promote, and have a score
>>>>>>>of about -9.  I would easily lose.  My other choice was to sacrifice my knight
>>>>>>>for his pawn, and enter a lost KNNKP endgame, somewhere around Mate in 75.  I
>>>>>>>would not easily lose this, unless my opponent had the TB, in which case it
>>>>>>>could be a draw by 50-moves.
>>>>>>
>>>>>>
>>>>>>Excuse me, but what am I missing?
>>>>>>
>>>>>>Isn't a KNNKP ending almost always a draw (assuming nothing special) if the side
>>>>>>with the pawn wants it to be? Cannot the side with the pawn force the draw
>>>>>>(regardless of 50 move rule) since KNNK is a draw?
>>>>>
>>>>>No
>>>>>The side with the two knight can sometimes win against a pawn because the side
>>>>>with the 2 knights can avoid stalemate when there is a pawn because the pawn can
>>>>>move and cannot avoid stalemate if he use the same plan when there is no pawn.
>>>>>
>>>>>Uri
>>>>
>>>>Huh?
>
>That's pretty close.  I have already seen _o_n_e_ game of Crafty vs Junior where
>Crafty won the KNNKP ending.  It was a fairly short mate though because Junior
>thought it was safely in a drawn endgame and traded right into it. (Junior 5)
>Jim Walker
>Jim Walker
>>>>
>>>>The side with the pawn pushes the pawn. If the other side captures, it is KNNK.
>>>>If the other side blocks it with a knight, then it is KNK (effectively). The
>>>>side with the pawn sits his king in the middle of the board and jumps back and
>>>>forth. If the side with the knights comes after the king, then the side with the
>>>>king pushes the pawn again.
>>>>
>>>>This still seems like it can be a forced draw by the side with the pawn.
>>>>
>>>>Now, of course, if you have some special position where the two knights can mate
>>>>the opposing king because it is trapped against the edge due to his own pawn,
>>>>then I can understand that it COULD happen. But that does not seem to be the
>>>>case in the example sited since he is talking about a forced mate in 75.
>>>>
>>>>KarinsDad :)
>>>
>>>There are a large percentage of wins in KNNKP.  The pawn must be blockaded
>>>before it gets too close to the promotion square.  The specific distance depends
>>>on the file it is on.  IIRC, this was extremely well-analyzed by a human long
>>>before a tablebase was generated for it.
Trotzky,1934. Source:"Test Tube Chess",1972,page 200 JAFM[ the longest
mate,found by computers, takes some 115 moves].
>>>
>>>Dave
>>
>>Ok, so what you are saying is that if the pawn is sufficiently advanced, then
>>the side with the pawn can force the draw. Otherwise, the side without the pawn
>>can first blockade the pawn and later put the opposing king in check often
>>enough so that even though the pawn can push, the king is forced into the corner
>>and checkmated before the pawn can queen.
>>
>>Is this correct?
>>
>>KarinsDad :)



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.