Computer Chess Club Archives


Search

Terms

Messages

Subject: A Little Nonsense In Computerchess as such

Author: Rolf Tueschen

Date: 05:47:24 01/09/06

Go up one level in this thread


You give a nice insight into shrewed details of computerchess but you conclude
into false directions.

Just a couple of thoughts:

1) If Rybka is still winning little matches despite these weaknesses it's bad
for the opponents.

2) It proves that machines themselves have no idea of their own situation. You
must always tell them if that - then that. But this that must be as specified as
hell. There is no chance for genralisations.

3) What you discovered is not relevant for machine vs machine chess. Because the
probability is very low for such situations with underpromotion etc. Of course
humans could seek it and could program a machine that it seeks it but to what
avail?

4) your examples show very well how low the ovrall intelligence of such machine
monsters really is. They are dumber like most stupid humans. But the in the rest
of the game they are strong as hell. We call it idiot savants in psychiatry.

5) you are unfair because Vasik always informed us that he's not yet at the end
of his programming. He didnt claim anything.

6) What annoys me is the fact that Vasik could have created a search monster
that outplays by outsearching in very deep while all other programs intend to
present surround efforts. If you leave out all such sides you win space for your
search. At least I would think so as a lay. So I have no idea where Rybka will
land if it's overall a good machine.


Perhaps these are already ideas enough for further debates.

You know, we had the same debate years ago about CSTal. Tal played incredible
attacks but then was outplayed because of its more sloppy search mode. The
attacks just didnt succeed in a longer match. Here with Rybka we seem to have a
search monster WITH good knowledge but still with a lot of dark spots. But what
will happen if the opponents begin to tune against it? Czub's Hiarcs attempts
point into such a direction.

Perhaps in the end this heat will cool down to normal and we will already have a
new favorite in a couple of weeks...



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.