Computer Chess Club Archives


Search

Terms

Messages

Subject: About exact and safe control of the machine's output in computerchess

Author: Hans Gerber

Date: 12:04:19 06/23/00

Go up one level in this thread


On June 23, 2000 at 10:11:12, Eugene Nalimov wrote:

>On June 23, 2000 at 06:50:16, Hans Gerber wrote:

>>Must I repeat [to R. Hyatt] that it was and still is new and informative that you declared
>>something wrong for several weeks and that I gave proof for that. It is about
>>the mean accusation against Kasparov that he had accused deliberately the DB
>>team of cheating. What I achieved is the proof that in the given documents
>>Kasparov did never do this.
>
>That's not true. Yes, he refused to name it "cheating". Instead he said "there
>are serious questions. It maid the move no one computer can make, and than just
>never saw a 8-10 ply draw".
>
>For me that is equivalent to saying "DB team probably was cheating".
>
>Eugene
>

Perhaps you did not have the time to follow former exchanges and the recent one
between "Pete.R." and me.

Of course you are right in a sense, but then you missed the point for the
mentioned reasons.

All IMO of course.



I try to discuss the questions of how scientists in the DB team should have
handled the problems with their test person = Kasparov. Simply because I believe
that it is a great advantage to be a scientist and therefore being able to rely
on certain universally accepted principles. Kasparov for the opposite might be a
genial chessplayer but he has all the rights in the world to behave on his
emotions and personality.

I continued that even if Kasparov had behaved worse, the scientists should have
dealt with his agitation in a calm, open and friendly way. All that with the
goal in mind to get good games from Kasparov against their creation.


Now to the detail about what was the meaning of what Kasparov had said in the
quoting you gave. For a very simple reason your conclusion is not forced here.
Simply because there could also be cheating from the outside of the DB team.

I was also surprised when I read the in-detail explanation R. Hyatt gave here.

You might find it in the collected articles of CCC.

Let me summarize:
___________________

- From the early days in computerchess on there were attemptions to exclude
cheating.

- Inspite of several discussions a solution could not be found.

- Not only could it be excluded that cheaters came from the outside by all kind
of wave transmissions but also internally there is no way to get a 100% safe
documentation of the thought process of the machine. This could be done in the
machine itself, at the output and then moments or days afterwards.

- Especially with the multi-processor entities it is impossible to get a kind of
always the same chosen move in a given position. So no determinism.

Here I insisted and asked him if he would not think that the scientists should
reflect that problem in advance and make 'sure' that they got as output what the
machine really had decided. Problem of authenticity.

He answered, no, this would not be possible but it had been debated...


Since I am not a computerchess scientist I must take these answers as state of
the art for the computerchess of the last three decades.

From a science understanding it is unbelievable that there could be no exact and
safe documenting of the thought process of a computer. BTW I am sure that
methods of control could be found. Perhaps by the introduction  of a processor
just to collect the data flow and then to freeze it step by step in sort of
photo frames. So that you could examin the turning points where new moves, new
evaluations had been found.

My argumentation is this, in times of academic spare time acticities on weekends
it was enough to document the legal move itself, cheating was completely out of
any imagination, if however you want to participate in professional sports you
should guarantee a control of the machine's output. It might be more expensive
but it should be do-able. Otherwise you won't get accepted in the world of
chess.


Is all this not worth a serious debate among computerchess experts?


In interdisciplinary sciences it is always accepted that you could give your
input of ideas and observations. What I want is to come over the simple level of
defaming sometimes this side and then the opposite.

There is a serious problem in computerchess and Kasparov with his instinct and
understanding of chess smelt it. We do not have exact and safe controls about
the thought process of our machines. Strange in view of the millions of dollar
investments...



Hans



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.