Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: How people could detect if a game was cooked? A Summary

Author: Rolf Tueschen

Date: 06:16:18 09/16/02

Go up one level in this thread


On September 15, 2002 at 18:00:59, Ed Schröder wrote:

>About pre-arranging games, anybody can do it, play xxx games, delete yyy lost
>games from the PGN output and post the remaining. The SSDF can do, all the
>super-chess-freaks playing their valuable tournaments can do it, and and and. In
>the end it all comes down if you trust the poster.
>
>Science is nice, but how does one scientifically proof someone is lying?


After all I have to comment here because people here seem willing to digest
everything from great experts. As if everything could be possible with the
exception of a proof with 'science'. The truth is that we need a little bit of
logic, not Astrophysics or stuff like that, to know what is going on.

At first I quote a retired collegue of Ed, let's call the "Not-To-Mentionable"
e.g. AB. Well, AB wrote me that XY (a famous tester) always understood his
testings as a WorldChampionship with XY as operator. So testing for him is
always testing program&operator, not the program alone.

Then I have talked about the roughly 30 games in total. In 5, and these were
important wins of the new Macheide program version of Rebel, I could present
data, food for thought. In three (!) of the examples the same "cooked" opening
was played where Shredder was without defense. All that by chance as the testing
design normally required? I seriously doubt that. But it's clear that with such
a "chosen" bias the whole testing is worthless. So we need no science but only a
bit thinking to come to that conclusion.

Because it's not true what Ed said.

Ed is thinking that such cheating, leaving out the losses and only posting the
wins, would be not to discover. This is wrong. Simply because many are testing
and they know about the probable proceeding in tests. Too many wins, or too few
losses simply make them suspicious. So, it is clear that a specific environment
leads to comparable results. With the usual variance. Extreme results were
possible but not probable. If however a certain tester _always_ gets supergood
results for the program which is his favorite at the time, then it's clear that
this must have reasons in the bias input by the tester himself. I wouldn't call
it cheating, it's better defined as 'operator influence'. Of course then this
has nothing to do with sound testing.

Jonas made a good argument.

Jonas makes a difference between normal testing with all the science involved
and a human approach of an individual (I called him operator after the
information of "AB") who doesn't want to do classical testing but who wants to
prove that he's able to work on a program version with all his creativity to
make it a better program. The exact detils of his method remain a secret. I
would say that this is completely ok.

If we forget about the announced claim of presenting data as a proof of the
secret. Because the presented games have nothing to do with the secret. BTW that
was the information "XY" wrote here into CCC after he had recovered from a first
confusion when he insulted me as the one who outed the secret of the secret. In
fact I did nothing else but showing that the games could never prove the better
strength of the Macheide Rebel. Neither statistically nor content-orientedly.

But was the whole attempt worth nothing at all?

With Jonas I think that "XY" did an interesting job. The problem is that we know
that he can't prove a thing, that the presented "proof" is a delusion full of
logical errors. So the job of the experts was to analyse the conditions of the
whole event. And although they came to the conclusion that nothing could be
proven, the fact, that a special style existed called Macheide, after the famous
Worldchampion LASKER's book, that is a real hype. So beyond all the necessary
reservation of scientists we have to realize that our modern times live from
magic and pretentiousness. The American motion picture Wag the Dog is the
artistic masterpiece where our modern media society got its memorial. As we know
from chess the threat is stronger -as idea- than being moved and having reality.
So the idea is stonger than what really happens. In truth we know that charisma
is 99% of the performance.

Likewise we in CC have our own spin doctors and charismatic figures. Not only
the programmers can create programs but also we creative testers. Not what is
really happening is important but the creation of ideas of a potential
happening. As in chess the potential is richer than reality. Isn't that a good
side effect of such a hobby? Isn't reality much brighter with such dreams?

So, in nuce I'm proud that I could contribute something. The SSDF published many
rankings, so why it shouldn't be allowed to create a Macheide version of Rebel?
Wasn't it the secret of Lasker, that he always won although he never (excuse me
but I want to present my case without unnecessary complexity) had better
positions? I think it's a great event when Lasker's strength could be utilized
in our modern chessplayers, the programs. We have many unsolved problems outside
classical science. Morphy (!) fields could be the next creative improvement of
chess programs. I know that Sheldrake is also a big outsider of science... ;)

Rolf Tueschen




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.