Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Garry still singing the same Deep Blue blues...

Author: Ed Schröder

Date: 13:16:07 05/07/00

Go up one level in this thread


On May 07, 2000 at 14:36:42, Robert Hyatt wrote:

>On May 07, 2000 at 10:34:26, Ed Schröder wrote:
>
>>On May 06, 2000 at 23:16:37, Robert Hyatt wrote:
>>
>>>On May 06, 2000 at 17:44:53, Hans Gerber wrote:
>>>
>>>>On May 06, 2000 at 03:09:53, Pete Galati wrote:
>>>>
>>>>>On May 05, 2000 at 21:17:14, Robert Hyatt wrote:
>>>>
>>>>>>
>>>>>>I see no obligation for anything, other than to push pieces of wood around
>>>>>>on the table, and then say "I resign" or "thank you when your opponent resigns."
>>>>>>I have played a lot of chess.  I have never seen an event that would _require_
>>>>>>that my opponent give me a post-mortem session after I lost a game to him.  I
>>>>>>might ask, but if he said "sorry, but no" I would hardly feel offended, and
>>>>>>I certainly wouldn't accuse him of "Aha!  you have something to hide.  Prove
>>>>>>that you didn't have a receiver in your ear and an accomplice outside the room
>>>>>>operating a computer."
>>>>
>>>>>>
>>>>>>
>>>>>>Not to most of us.  They accomplished something we have dreamed about for many
>>>>>>years.  Kasparov lost the 'image'.. not Hsu...
>>>>>>
>>>>>>
>>>>>>
>>>>>>There will _never_ be any "proof".  It is _impossible_ to prove they didn't
>>>>>>cheat.  _impossible_ <period>.  If they had had the display on the wall in the
>>>>>>analysis hall, they could _still_ have cheated had they wanted...
>>>>>>
>>>>>>this is a useless argument point.  Whether they provided the logs during the
>>>>>>game, after the game makes no difference.  If they had provided the logs
>>>>>>_before_ the game, there might have been reason for suspicion, of course. :)
>>>>>>
>>>>>>
>>>>>
>>>>>The thing about the cheating issue is just like in a court of law, who's the
>>>>>burden of proof with?  IBM doesn't realistically _need_ to prove that they're
>>>>>not guilty of cheating, because _nobody_ has proven that they _did_ cheat.
>>>>>
>>>>>Kasparov & his people speculating speculating or claiming that IBM has cheated
>>>>>means absolutely nothing at all because they don't have any proof.
>>>>>
>>>>>Pete
>>>>>
>>>>
>>>>We have interesting points of view. Both of you reject my giving priority one to
>>>>_science_.
>>>>
>>>>Of course your are right that a) in chess between chessplayers there is no place
>>>>for the obligation to inform the opponent about one's preparation and tricks,
>>>>and b) as a matter of justice, a cheating has not been proven yet.
>>>>
>>>>However let's take a look into science. Many times we see a lot of tricky
>>>>experimental settings. Most of the time the real goal of a research is hidden,
>>>>in social sciences. Here in our case we have scientists with their machine, and
>>>>the machine should play chess against a very strong human opponent. The first
>>>>question that should be examined is if the machine really does play the game
>>>>without influences from the outside. The second if the machine should be able to
>>>>play _a whole match_ on his own. Depending on the goal of the research, this
>>>>should be clearly defined. Likewise the output of the machine should be
>>>>objectively controlled. If this control is impossible to attend, then the whole
>>>>attempt is condemned to end without reliable data. Since scientists were the
>>>>actors, not just chessplayers, we can hold them responsible for such a bad
>>>>outcome. The proof of "cheating" however has not been achieved. In science you
>>>>loose your reputation a bit earlier than in a courtroom trial. If you have
>>>>missed to follow trivial rules in science, you are responsible for a bad
>>>>outcome, for any kind of possible irritation about the "results".
>>>>
>>>>If I understand Kasparov right he seems to be dissapointed by the behavior of
>>>>scientists who treated him like a friend during the whole time before the match
>>>>and the first match. To always repeat that Kasparov had his fingers on all the
>>>>rules of the match is at least misleading. Why should he have spent a minute on
>>>>the reflection that these scientists could try to cheat him or to treat him
>>>>badly? So, this argumental line is a dead one.
>>>>
>>>
>>>
>>>It isn't dead at all.  This was about playing chess.  After game 2 he suddenly
>>>decided that he needed to see inside the DB machine, in a way that he could not
>>>hope to do if DB were a human instead.  His request was bogus.  As were his
>>>outrageous claims/statements about the game...
>>>
>>>
>>>>Why is it so difficult to understand that the suspicion (due to the faults of
>>>>the scientists) of a cheat fires back onto these scientists, who had neglected
>>>>the iron rules of scientifical methodology?
>>>
>>>
>>>They didn't neglect a thing.  The way they played Kasparov was the way _all_ of
>>>computer chess has been playing games, for years.  We change programs between
>>>rounds.  We all show up at ACM tournaments, or WMCCC tournaments, or WCCC
>>>tournaments, with code that is new and has not been seen by others before.
>>>
>>>They did nothing 'different'.  They didn't do 'bad science'.  Believe what you
>>>want, but if you want to make statements, at _least_ make them based on fact.
>>>And the facts of the match are pretty much public knowledge.
>>>
>>>
>>>
>>>
>>>>
>>>>So, we have the complicated situation, that the applauded "result" of the match
>>>>is invalid.
>>>
>>>
>>>The match was _not_ invalid.
>>
>>Right, the match was not invalid. However the point Hans is making is
>>that the match was presented (by IBM) as being "science" which is to
>>laugh about. If it was about science the request of Kasparov to see
>>the logfile should have been granted 10 minutes after the request and
>>not 2 years later.
>>
>>Ed
>
>
>Why?  What would be the justification for letting Kasparov see how deep it
>was searching, what its evaluation was at various points... _during_ the match?
>
>Humans don't do that.  Why does the machine have to?
>
>For Hsu/Campbell this _was_ about science.  It _always_ was.  For IBM, it
>was about marketing and revenue.  It _always_ was.  One doesn't preclude the
>other by any stretch.  Otherwise "Intel Inside" would mean the pentiums are
>not about 'science'...

There are 2 kinds of science, the one about money and the orginal one
which is about complete openness. Guess about which one I was talking :)

Ed



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.