Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Rebel's anti-GM option

Author: Robert Hyatt

Date: 14:52:13 06/24/98

Go up one level in this thread


On June 24, 1998 at 15:02:30, Moritz Berger wrote:

>On June 24, 1998 at 13:43:13, Robert Hyatt wrote:
>
>
>>If anyone wants their program to sac like Tal, that's fine.  But to induce such
>>a strategy in a program, in order to prepare for playing GM's seems somewhat
>>risky, to say the least.  you want to *plug* holes, not *open* them up.  A good
>>GM is already capable of finding most of our weaknesses.  I don't see the point
>>in adding more.
>
>Crafty plays that move also, remember. It only needs slightly longer (3min) than
>the others, but on your quad-box it will certainly play it at blitz, also ;-)
>
>Moritz


I haven't tried it, but there are two points.  One, this is a position that
seems to be quite tricky to resolve with search.  I let crafty run a bit, and
it failed high, then low, so my current version won't play it since the fail
low makes it ignore the fail high.  That's probably just an extension in the
search that is misadjusted, as one or the other has to be wrong (high or low
fail, but not both).  But the other point is trying to play this thru a
speculative eval.. and that I don't want to do.  I certainly won't try the
Ne6 sac as my special imbalance code says "no sir"...



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.