Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: The superior Rybka chess knowledge (With Corrections)

Author: Rolf Tueschen

Date: 13:15:12 01/18/06

Go up one level in this thread


The whole position, the who standpoint, the whole opinion of Chrilly is wrong,
but still his message is a highlight for CCC because it's thought-provoking and
most interesting. But still it is totally wrong. Typically nobody called Chrilly
a troll. I would call him similar to Socrates. At least he makes the readers
think.

Why is all that in his message wrong?

Because Chrilly has never understood human chess and it seems as if he hasnt
understood CC either. Therefore here the explanations.

BTW who is Lutz? Yes, he's a fine GM, but he has NOT the overall education of
one Huebner.

But now to the details.

The History of Human Chess is bound to the process of concretely playing chess.
It's a centuries-long attempt to understand the optimal chess strategies. It's
also the history of the Beauty of Chess as an Art. So a chessplayer is on one
side satisfied by the sports in chess and on the other side by the art of it.

An important detail of optimal human chess is the factor time. Nobody sane
enough would refuse to win a Q while taking a little P only to have the future
endgame of hard work with the plus of a single P. A Q win could and would end it
at once. Now what would happen if some chessplayer would begin to play
sub-optimal moves with the satisfaction to win P after P? Such a player would
badly lose almost all his games. Because in human chess we have a second factor,
and this is called "tiredness" after hard work. There are too many possibilities
to make little mistakes and then lose the advantage of a P. There are too many
positions in chess where you cant even win with the plus of a P. So, every
chessplayer is conditioned to play optimal moves to "win" time for avoiding
tiredness.

Now comes Chrilly with the advantage of 9 moves overall played at optimal
quality and several other moves more speculative, and deeper if some
unneccessary knowledge could be avoided in the programming process. Let me
define that detail. You can make a chess machine stronger - if you once have
reached an error-less base of 9 moves - by just making it look deeper by
throwing classical knowledge from human chess into the bin.

I'm not a strong GM, but still I claim that this concept is BS. Here the
explanation:

For a super GM with the incentive of enough money an errorless play of 9 moves
deep is piece of cake. And then he relies on his chess experience and artistry
while the machine is digging into fog, no, in the last region into the dark of
Nowhere (where a machine is completely blind until it can start its table
bases). A human chessplayer with a good training can make a chess machine look
like nuts. Of course for 99,9% of the rest the chess machine plays a too
errorless "chess". Against the FIDE rules, but we leave out this aspect here for
the sake of the argument.

You know what a chess machine is in terms of mountain climbing? It's a climber
who makes thousands of steps of 10 centimeters. One after the other. No matter
if the sun is shining or the moon. You know what will happen? A huge avalanche
of snow will roll downhill and take the little machine head down into the dark.
Period.

The avalanche of snow in mountain climbing is the superior knowledge of a
chessmaster who knows thousands of positions who look like draw but still are
won. That knowledge lies far beyond what a chess machine could know nowadays.
Fine that Rybka seems to play after the 10 centimeters paradigm, this gives a
human player extra chances to bully. Of course in mere CC (= engine vs engine
chess) the bigger depth is a plus extra against the nonsense of the mobility
code for the Q.

Sorry to all, because here I didnt play Socrates but played the spoon-feeder of
the "Nuernberger Trichter", a centuries-old German expression, which became the
role-model for generations of stupid teachers. Next time I'm writing again from
Greece... ;)



This page took 0.01 seconds to execute

Last modified: Thu, 07 Jul 11 08:48:38 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.