Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: More about Knowledge Vs Fast Searchers

Author: Thorsten Czub

Date: 14:53:47 05/22/98

Go up one level in this thread


On May 22, 1998 at 17:18:37, Don Dailey wrote:
>Thorsten,
>
>I see you constantly post about knowledge and search speed as if
>it is completely separate,  there is some reason to believe you
>do indeed partition this stuff into black and white.  Generally
>you seem to view speed contempously and knowledge with warmth.

1 from 4 times. Maybe. I can live with 1/4.


>But this is only my general impression, forgive me if I am wrong
>here.

:-))

I have forgiven you. :-))

>I like to view a chess program as a "black box" that spits out
>moves.  We don't care what happens inside the box, and whatever
>happens there is hidden from view.   A good question for all of
>us is could we really tell the difference if we could not see
>node counts, iteration numbers and claims of speed.  What if all
>we saw as output was a move?   Could you identify the fast
>programs?
>
>- Don

Maybe not. I would not have identified Tiger 11.2 !!
I would have identified Nimzo by watching the game:
Paderborn 1998, CTiger-Nimzo98

There Nimzo made many "null-moves" that would have made it easy to
identify it.

Of course I can make mistakes by doing the turing-test.
But studying the main-lines, evals and the MOMENT (=time) correlation
WHEN which program SAW the point, gives me exactly THE NEEDED
information to identify about chess-strength.

If I would not see this information, and only have black box, working
with programmers would not very good possible.
My friend Bernd has 2430 ELO. He is not able to tell you by looking at
the main-lines/evals which chess program is stronger.
But he can tell you which program played stronger.
BUT : he still is unable to say from ONE game which program is stronger.
I can do so from watching ONE game.
HE needs many games, and THAN he is able to follow it.
I have tried this with him in an experiment 1993 in munuch,
championship.
At the beginning of the tournament I told him hiarcs would win.
I showed him the program Mark and I tested before munich.
He was unable to FEEL it.
I said to him, wait until we are in munich, i will teach you to
understand - to watch the right way on the main-lines.
After a few games of hiarcs he was able.
We replayed all munich games in the train on the journey back from
munich to Dortmund and I saw by his comments that he WAS able to
understand what I felt before.

So even a strong chess player is no guaranty to diagnosis about
chess-strength. Strong chess players often do not know WHAT is
complicate for programs and what not. They project from their own
problems in chess and think the machines are the same.
But Bernd WAS able to FEEL what I felt after some days in munich.
In the end he was as enthusiastic about hiarcs as I was, but he had of
course the much deeper understanding of chess. He told/advised me about
HIS insights, and i told him what to watch out when you want to value
chess programs.

A black box would handicap me. But not him.

I NEED the information from the programs.
Sorry. I am not a good chess player but I have seen strong chess players
telling bullshit during the games (Ossi Weiner often bluffs and also
Rainer Menningen : both Mephisto team behave like POPES , especially
when they are wrong. During playoff-game Genius-Hiarcs MARK and BERND
and I had seen what Menningen and Weiner did not saw although they had
the same data in front).
Also Ossi showed no insights in the game Genius-CSTal in Paderborn 1995.

In Paderborn 1998 i have seen the same behaviour in the beginning of the
game
Clever + smart vs. ChessTiger 11.2.
Ingo and Stefan THOUGHT and expressed their ideas about the game.
I never subscribed their ideas. And when the fail lows appeared, i was
right.

It was the same in the game comet-Tiger and Tiger-Gandalf. I saw very
early that Tiger-Gandalf will be lost for Tiger. I knew Gandalf and
Tiger. No matter what CHESSSSSSS is on board. No matter that Vincent has
more knowledge about computerchess and chess (he is a programmer and a
strong chess player). All these FACTS are not important. I knew that
Tiger would ugly lose.
This was my main pain. I sat there, saw that a draw would be enough for
winning the tournament. But I also knew that Tiger would be lost in this
kind of masked/hidden king attack.
When the attack came into SEARCH area, it was too late.
Chess Tiger should have seen the problem WHEN IT WAS NOT LOST, BEFORE
the king attack happened. But - there WAS NOTHING to see (with search).
Gandalf SAW this because of static evaluations. He moved the right
moves, without SEEING something concrete.

If you want we can analyse the games here, if this helps.
I am not good in chess. Not in programming. But I know (=feel) when
something goes wrong. And than I speak it out. Call it ignorance. When
my RESULTS in prophecy are as good as the ones who can play better chess
or program , than why criticize me for beeing b/w ? I am not. I get all
my info out of the way the programs play. out of the game, the main-line
and the evals. Search depth and correlation of fail-lows and other
things.
If you make me a black-box (like in the dedicated machine times) I need
more time than with information.







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.