Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: evaluation function - Secrecies? What is reality in CC?

Author: Rolf Tueschen

Date: 15:17:08 08/30/02

Go up one level in this thread


On August 30, 2002 at 17:13:38, Matthew Hull wrote:

>On August 30, 2002 at 16:48:10, Rolf Tueschen wrote:
>
>>On August 30, 2002 at 16:38:39, Matthew Hull wrote:
>>
>>>On August 30, 2002 at 16:24:41, Rolf Tueschen wrote:
>>>
>>>>On August 28, 2002 at 12:54:35, Robert Hyatt wrote:
>>>>
>>>>>On August 28, 2002 at 12:13:11, Vincent Diepeveen wrote:
>>>>>
>>>>>>On August 28, 2002 at 11:01:52, Jayakumar Ramanathan wrote:
>>>>>>
>>>>>>the truth is that all you will find out is a very primitif information
>>>>>>about it in programs like gnuchess and crafty which are open source.
>>>>>>
>>>>>>the real good evaluations are top secret, both from most amateurs as well
>>>>>>as from commercials. so please what you see in these programs gives
>>>>>>a good idea how it works. it doesn't represent quantity or quality of
>>>>>>the real ones.
>>>>>>
>>>>>
>>>>>
>>>>>
>>>>>The inverse is also true.  Just because you _can't_ see it doesn't mean
>>>>>that the quantity/quality is any better than what you _can_ see...
>>>>
>>>>In military sciences as in highly sensitive State sciences the exact details are
>>>>never published. Please do elaborate how secrecies entered into CC! What is the
>>>>tradition?
>>>
>>>You see, there's this thing called a competition.  It's a contest.  Even
>>>sometimes their are egos involved.  Sometimes people like to keep winning, so
>>>they don't tell everyone how they do it.
>>>
>>>Then there's some people who want to make merchandise of their program.  In
>>>order to sell the most copies, they don't tell anyone their winning secrets so
>>>as to keep others from duplicating and diluting their profits.
>>>
>>>Hope that helps.
>>
>>_No_, this doesn't help at all. Matthew, I didn't ask for these secrets out of
>>competition, you know. I was asking about the secrecies about the identity of
>>the machine's processes and the displayed reality. Know what I mean? I mean in
>>sports we have the control against doping. We are interested in the identity of
>>the competitive performance. We don't just take the apparent for the real
>>reality. Know what I mean? As it seems in CC a great part of competition is
>>about the fake of reality? Here I wanted to have some further details from the
>>past. As to your point of business, such a split reality is even more annoying,
>>don't you think so? I mean, we chessplayers want to know what the real eval is
>>for the particular position.
>>
>>Rolf Tueschen
>
>My apologies for the mis-understanding.

Not accepted. You didn't make anything wrong! :)
This is the good thing in online communication, you can always specify if
needed. Even with your actual answer you only touch slightly what I see as the
real problem. Let's see if we can find some unterstanding.

>
>As for your clarified question, I was not aware of the wide-spread
>disinformation or "fake of reality" in CC competitions.  A few frauds have been
>exposed in the past, but for the most part, it does not seem to be a problem.

I here is a misunderstanding. If the displayed reality is not a good mirror of
the real processes in the machine sports events aren't directly hurt. The human
opponent or other progs don't see the display anyway. But think of the testing.
In tests and autoplayer matches the orgnizers usually depend on the displayed
information. Also if a chessplayer wants to analyse chess positions with a
program or if he wants to play the machine and control his own performance. Here
the split reality is a real mess. I asked Dr. Hyatt because he's deeper in the
CC technology and also in the history of CC. And he's doing some science. I
think we are touching a crucial point here.


>It's pretty hard to pull off a real "Turq" masquerade in a WCCC or other
>important event.  It's less hard to pass off a clone program as an original
>work.  But, I don't get the impression that it's a huge problem.
>
>On the Internet, there are problems, but most everyone knows who the good guys
>are and if some annonymous interloper "program" starts winning everything,
>people get suspicious.  But that's just life in a free society.
>
>Just about any engine you want to download or purchase gives quite good
>information on what the engine is doing.  Although, to the un-initiated, it
>looks like prefect gibberish.
>
>It sounds like you are suggesting some "doping controls" or other measures as
>safeguards against "faking", and to satisfy the skeptical mind that everthing is
>scientifically in order.  There probably are some things that could be done, but
>that might take much of the fun out of it.  After all, it's only a game, and a
>pretty esoteric corner of it at that.

Is this true? For me CC came out of computer sciences and was one specific field
to do some practical research. Would you say that other fields have the same
uncontrolled myst? Say the whole field of AI with the robots. Is this full of
such fraud too? I never heard of before. I don't understand what you mean with
"fun". Could you specify where is some fun and through what this fun could be
lost? I don't understand you. I am convinced that we must accept reality as it
is and we want to research not to have myst and fun but we want to deepen our
knowledge of reality. I think that the "fun" comes always out of the doing it
'comme il faut'. No matter what. Therefore I had so many difficulties to
understand why experts in CC have so much fun in doing almost the same the
others do. Because for me it would be more important to do research for what
should be done and what would be successful. I could never understand why people
are happy with the excluding of chess out of CC. Nowadays certain nullmove
gambling is the symbol for the concentrating on pure performance without too
much chess knowledge in the progs. It seems as if most experts are in zeitnot,
so that they want instant success. -- Against human GM this is hopeless. But
then "we" still have the twisting during a match. Once we gamble, then we do
brute force only. In a short match we could confuse humans. Fantastic fun! But
not for me. This way you can't become a good chessplayer!
But this is another chapter.


>
>It's not a giant scientific endeavor, just an interesting hobby to squander away
>the precious few hours of your life on.

Although this becomes OT, we should try to realize that philosophically our
individual life and that of the human race has no real meaning at all. But
therefore it is so important to find the fun - if you want - in everything you
are doing. Therefore sciences are so interesting. You make good documentations
and even if you can't find the brilliant solutions some next generations might
be motivated by your data. So, the meaning of what we're doing lies in the
perfection how we're doing it. No matter in what field or on what level. If you
read a bit Montaigne or Confucius, I think this could help. The way is always
more important than the result.

Back to CC. :)

I wished we could establish a journalism where real analyses are more important
than just reporting who came in first or second. I wished to read something
competent about chess in modern CC. But for such nonsense ChessBase does no
sponsoring. Critical questions are being censored. So we have no investigative
journalism at all. Then you are right. It's just about quickly evaporating fun.
And hoping for the next generation of processors. Archeologists in future will
have some real work to do...

Rolf Tueschen


>
>Regards,
>Matt
>>
>>
>>>
>>>
>>>> Or did it start with DEEP BLUE? And then, how do you feel when you
>>>>one time have to defend the factual reality of the invisible (matches of DB2
>>>>against commercial progs) and then here you remind us of the existence of the
>>>>claimed real in opposition to the invisible. Do you see any need to reform
>>>>"classical" CC?
>>>>
>>>>Rolf Tueschen
>>>>
>>>>
>>>>
>>>>
>>>>>
>>>>>
>>>>>>>What is the best place to read about how chess evaluation functions are
>>>>>>>constructed? I am not an expert in programming but would like to know the
>>>>>>>details of how numerical values are assigned to various positional concepts
>>>>>>>(material, open files, etc.) and how they are weighted in the final value that
>>>>>>>the routine outputs.
>>>>>>>
>>>>>>>Thanks in advance,
>>>>>>>diomedesX



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.