Author: Christophe Theron
Date: 21:18:26 04/20/99
Go up one level in this thread
On April 20, 1999 at 19:35:24, Will Singleton wrote:
>
>On April 20, 1999 at 15:34:34, Christophe Theron wrote:
>
>>On April 20, 1999 at 14:40:08, Will Singleton wrote:
>>
>>>
>>>On April 20, 1999 at 14:10:20, blass uri wrote:
>>>
>>>>
>>>>Some programs knows that it is a draw but instead of stopping to play and claim
>>>>that it is a draw they start to do stupid mistakes and lose the game.
>>>>
>>>
>>>Yes, my program used to do that. Once past 50 (either actual or in the search),
>>>it would think all moves were equivalent, and would pick whatever happened to be
>>>first. Now I take into account potential loss of material.
>>>
>>>There is a similar bug some progs have, that when a repetition is found, it
>>>doesn't consider whose move it is. So, it might move into a repeat position and
>>>score it a draw, except that when the position occurred before it was the progs
>>>turn to move, and now it's going to lose its piece.
>>>
>>>Will
>>
>>Yes! The 16 bits version of Tiger had that bug one time, and lost against a
>>human player in an important tournament game! But it was easy to fix once I
>>realized what it was...
>>
>>
>> Christophe
>
>
>Right, server play helps to find a lot of these "gotchas."
No no, it was a real tournament game, against a strong player!
>Another one for me
>is the idiotic Bxa2 b3 ooops problem. I'm currently testing a "fix" for this,
>so it doesn't horizon it off. Needs work, though.
>
>Will
I almost never see this problem. I have a small penalty for a bishop on these
squares (based on some trigered conditions) and it looks like it's good enough.
The penalty is 0.20 pawn IIRC. I tried high values (up to 1.5 pawns), but it
didn't work at all. Tiger would let the opponent take the a2 pawn and try to
trap the bishop, but very often the opponent gets away with murder (especially
if it is a computer)...
Christophe
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.