Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Wrong Colored Bishop Endings

Author: Robert Hyatt

Date: 07:13:58 11/26/97

Go up one level in this thread


On November 26, 1997 at 01:15:55, Jouni Uski wrote:

>Eval is -4,xx, but does it matter? If correct move is found by wrong
>reason, it
>still is found...

possibly. But in this position, *any* move but Na5 loses.  Na5 draws,
but
if a program doesn't know this, this position doesn't test its knowledge
of
B+RP endings at all, just its tactical speed.  As I mentioned, Cray
Blitz
finds this in a couple of secnds, but *never* says DRAW, even though it
has B+RP code.  This is convincing that search alone won't find this,
because
I doubt there is anyprogram (other than DB) that can come close to the
tactical ability of CB (don't forget, this program solves *all* 300 WAC
positions in 1 second or less.  297 in under cond...

One of the other positions in the 5 position test are good knowledge
tests,
#4.  the other three are generally tactical as they are tests to find a
win
tactically, although knowledge does play into a couple of them as it is
necessary to avoid obvious draws...


But if the goal is to test knowledge, #5 doesn't do it, as a program
with *no*
knowledge about B+RP endings will still discover that Na5 is forced...



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.