Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Why "positional" test positions are physically impossible! WM Test

Author: Bob Durrett

Date: 17:12:29 12/04/02

Go up one level in this thread


Rolf:  Yours is a very long bulletin and I do tend to lose the forest for the
trees, but that is MY problem.  It sounds like you are talking about something
I'm interested in, so here is my attempt to communicate my thoughts about what I
think you are saying.  Forgive me in advance if I miss the mark occasionally.  :
)

On December 03, 2002 at 12:54:24, Rolf Tueschen wrote:

<snip>
>says that these tests can't show the strength of play or as it was
>claimed for this test, the "ability to analyse".

My impression is that people here are using test positions for a different
purpose entirely.  In other words, NOT to "show the strength of play" nor to
demonstrate a chess engine's "ability to analyze."

Instead, they seem preoccupied in finding "holes" resulting from errors in the
design, configuration and coding of the algorithms they are using for "search"
and "position evaluation."  When they find the holes, they modify their programs
to "plug" the holes.  Of particular concern to these guys seems to be to
identify relative strengths and weaknesses in various algorithms.

Your question to CCC members seems to be whether or not anyone thinks suitable
test positions COULD be found and used to "show the strength of play" &/or to
demonstrate a chess engine's "ability to analyze."  Do I have your question
right? [Incidentally, the best definition of the word "analysis" might be
different depending on whether the analyzer is human or a chess engine.]

<snip>

>testing with even these top class positions is nonsense. And why it has nothing
>to do, well, almost nothing, with _real_ strength.

Well, that may be true if the purpose of testing is to determine "strength."
But that does not mean that testing is a waste of time for any other purpose.

<snip>

>Look, if you have a positional game of chess, where do you choose the point for
>a test? Of course, in this WM-Test of Gurevich et al you take the position when
>exactly a certain by the experts well commented and mostly beautiful move has
>been made. Because there the commentators said: only with this move he could
>conservate the slight advantage.

If the testing were for the purpose of finding and plugging holes, then the
position in which the "key move" had to be made would be the right position to
test.  [If the engine failed to pass the test, then the programmer might make up
additional related positions to test to help the programmer better understand
the "hole" and to then debug ["de-hole"] the program.]

>But the truth is that often the engines see - in
>their actually possible realm - two solutions very closely together. And in
>general it could be said that for positional positions without tactics the evals
>are not very impressive at all.

That does not seem to follow logically.  The evaluation may depend on how the
evaluation coding works.  One engine may do position evaluation differently from
another.  Some may do a credible job of evaluating positional factors.

<snip>

>that "nice move" (that caught th attention of the commentators) is
>by no means the most important moment for the decision making.

I do not see the logic of this assertion.  True for some cases and not true
others would be what I would expect.

As an example, consider the buildup leading to a king attack.  Early in the
process, setting up the preconditions for a successful attack may preoccupy the
human attacking chessplayer. If chess engine position evaluation software were
designed to look for such preconditions, then that evaluation software would be
doing "positional assessments."  But I don't see that setting up of
preconditions is necessarily any more important to success of the attack than
finding the concrete "key move(s)."

Bob D.

<snip>

>
>Rolf Tueschen

<snip>



This page took 0.01 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.