Author: Albert Silver
Date: 08:01:11 06/15/00
Go up one level in this thread
On June 15, 2000 at 08:09:00, Hans Gerber wrote:
>On June 14, 2000 at 23:37:14, Albert Silver wrote:
>
>>On June 14, 2000 at 09:14:25, Hans Gerber wrote:
>>
>>>Topic: Certain details of the match between DB and Kasparov
>>>
>>>On June 13, 2000 at 22:57:09, Albert Silver wrote:
>>>
>>>>On June 13, 2000 at 18:08:16, Hans Gerber wrote:
>>>>
>>>The isolated solution of positions is not the same as a game of chess. Kasparov
>>>thought that certain decisions by the machine simply contradicted the typically
>>>machine-like style of playing chess.
>>
>>He thought so based on specific moves, moves that he presented. Why anyone would
>>expect a multi-million dollar computer chess project to play the same as Fritz
>>thinking all night is beyond me. Also, I'd like to know how Kasparov qualified
>>as an expert on what Deep Blue was capable of seeing in chess. If anyone knew,
>>it was Deep blue's creators.
>
>
>Yes, and the meaning of that? That they should have sat together...
>
Meaning his accusation isn't based on an expert opinion but based on his
over-active imagination and lack of information. He isn't saying, "Based on my
expert knowledge of what computers (supercomputers) are capable of doing, and
the limitations of what programming and technology can bring, these moves could
not have been made by a machine." Instead he is saying, "Nice moves, cheating
must be involved." Based on nothing.
>
>>
>>>
>>>Unless you do not want to accuse Kasparov of cheap tries to excuse his own weak
>>>playing you must take his questions for serious.
>>>
>>>(Please keep in mind this point for the other debate about the actual situation
>>>between computerchess and human chessplayers.)
>>>
>>>
>>>
>>>>The Deep Blue team is
>>>>confounded. Why does he want the printouts? He never asked for them in the first
>>>>match, nor did he request them either before or after the first game, so why
>>>>now?
>>>
>>>
>>>Let me ask you in return: why should the Deep Blue team ask such questions when
>>>it is well known that they do not have the same understanding of chess like
>>>Kasparov? Why did'nt they accept his questions as quite normal coming from a
>>>human champion? Do you see my point?
>>
>>No, as I explained above.
>>
>
>
>Are you interested in science?
Yes, what does that have to do with it?
>
>
>
>>> You agreed that they should have treated
>>>him with all courtesy but then why they should have doubted any of his actions
>>>concerning chess and the whole event? Did you see a situation before where a
>>>computer rejected to win material? No please, not again the examples where also
>>>Crafty rejected such pills.
>>
>>What kind of examples do you want then?
>
>
>Examples for a machine's rejection of a clear material advantage...
>
>In connection with a whole game, not just the results of a special tuning for a
>concrete position!
It happens all the time. Neglecting a material advantage in order to avoid a
positional disadvantage or worsening of the position is not uncommon, plus I
have seen many instances where the program would forego winning material to
increase the pressure. Deliberately sacrificing material to open lines or gain
another positional advantage is also more and more common.
>
>
>
>>
>>>
>>>
>>>
>>>> Furthermore, the printouts will certainly reveal _much_ to GK just by
>>>>allowing him to see what the computer thinks of key positions and the analysis
>>>>that led it to those evaluations. That's quite a lot.
>>>
>>>
>>>
>>>It is interesting that experts like R. Hyatt stated that Kasparov might have
>>>gained _nothing_ from the viewing of the prints! If this is true then it is not
>>>to understand why they refused to show the logs.
>>
>>That statement doesn't constitute proof and if he stated that, he could also be
>>wrong. Kasparov is an intelligent man, and if he could discern the lines of
>>thinking and the evaluations the computer attributed to the positions, he could
>>learn much. I have no doubt about it.
>>
>
>
>Let me try to understand. Do you think it could be a disadvantage if Kasparov
>could _learn_? Are you believing in that concept of secrecy that allowed DEEP
>BLUE to survive against Kasparov in the three games with the two colors?
>
>Do you believe in the strength of the machine or do you prefer to rely on all
>sort of tricky strategies from the team of humans around the machine?
What tricky strategies? Here is the computer: want to play? Kasparov never had
any problems playing computers in the past. He played Genius, he played Fritz,
he played Deep Thought, and a first match against Deep Blue, and NEVER did he
say, "show me your printouts". So why now, when the computer did no better than
draw against him (the fact that he resigned is his own problem), should
everything be different?
>
>I was always thinking that we had a match between a chessplayer and a machine...
>
>
>
>
>>>
>>>
>>>> Naturally they ask him. GK
>>>>may beat around the bush a little, but eventually he will have to tell them. I
>>>>don't believe they would have swallowed a cute little speech about scientific
>>>>truth by him. Not there and then, in a match with a million-dollar prizefund.
>>>>This may be coffee change for IBM, but it isn't for GK, and there is no reason
>>>>to just hand the match to him. So he tells them. He doesn't even have to be
>>>>rude. Let us imagine he is as tactful as possible under the circumstances: he
>>>>can't believe a computer could play some of the moves it did. Only a human could
>>>>play such moves. And lest you say that GK would never say such a thing, I'd
>>>>suggest reading the interview he gave to Playboy magazine some years back.
>>>
>>>
>>>Of course! He thought that and still thinks that. And note, he still thinks that
>>>although the "prints" have been published. What does that mean? When R. Hyatt
>>>explained that the publication of such prints meant nothing to the question
>>>wheter there was a cheat or not? Why do you think the published prints would
>>>change reality? The reality Kasparov saw in 1997? Why don't you think about the
>>>possibility that the complete DB team might be innocent but others might have
>>>cheated?
>>
>>How?
>
>
>Please let us wait until R. Hyatt gives us further explanations. However that
>could be contra-productiv to give potential cheaters too much information. All I
>can say is that it was stated that in principle such cheating was possible.
Without any of the DB team knowing? I doubt that very much unless you concoct a
scenario akin to a Mission Impossible episode.
>
>
>
>>
>>> We only have one fact and that is the astonishment Kasparov showed in
>>>view of certain moves.
>>>
>>>Let me confirm you that personally I would have behaved different if I had been
>>>the head of the DB team. I would have been highly interested in Kasparov's
>>>opinions and I would have analysed with him and the machine as long as needed.
>>
>>I don't know why you believe he was proposing to analyze anything with them. He
>>demanded to see the printouts, that's all.
>
>
>
>I am not a clairvoyant. I tried to imagine what they could have done. Very
>easily as scientists.
>
They DID all they could do: "Jim, go ask the IBM marketing crew if we can give
the printouts of Deep Blue's thoughts on these moves."
>
>>
>>
>>>Since when could I have a better aid to understand what my baby was able to
>>>achieve? Remember that I am a scientist. I am not a gambler!
>>>
>>>
>>>
>>>>His
>>>>opinions on what women could or could not do were straight from the Stone Age.
>>>>In any case, it may have been eloquent but it is still an accusation of
>>>>cheating. If you have a different theory on why he asked or what he would say to
>>>>justify it, please say, but remember that he stood by his accusations in public.
>>>>I never heard any denials from him.
>>>
>>>
>>>
>>>Yes, and I hope he will not let it go for cheap resons or some extra money!
>>>
>>>Why do we have to make it so difficult? Only non-scientists can take offense by
>>>Kasparov accusations. Are you sure that you must interprete his questions as
>>>accusations? Please read again what R. Hyatt wrote about the possibilities to
>>>cheat in such events. For me as Hsu it is the simplest thing to sit together
>>>with Kasparov and to take his ideas for serious. Since I myself I don't want to
>>>win a match by cheating! I ask myself for how long I must repeat that a question
>>>is nothing negative. A question is always something positive and helpful. I know
>>>that nobody of my team cheated, so the more I would be interested what Kasparov
>>>might think about certain moves!
>>>
>>>
>>>
>>>>
>>>>So there it is. The world champion wants to see the printouts to make sure no
>>>>cheating is going on. How would you have reacted? Do you believe that courtesy
>>>>should also go that far?
>>>
>>>
>>>Yes, of course! Since I have read the explanations of R. Hyatt!! Of course I
>>>would be highly interested in all kind of possible suspicions of unfair
>>>intervention in my own scientifical experiment!
>>>
>>>
>>>> So the request is heard by the DB marketing team from
>>>>IBM (I doubt Hsu was empowered to make that sort of decision). They basically
>>>>are being asked to give up vital information in the beginning of the match in
>>>>order to satisfy GK that they aren't cheaters.
>>>
>>>
>>>Now after all my arguments I hope that you see the mistake in your reasoning
>>>here. Why do you argue that Kasparov has accused the team itself? Where does
>>>your conclusion come from?
>>
>>I do not believe there was some secret plot involving cheating. I also don't
>>believe cheating could have occured without the DB team's knowledge. I don't
>>think the DB team would listen to this and think there was a mysterious plot
>>involved. I think that Kasparov's accusation had no substance whatsoever other
>>than what his imagination was able to provide.
>
>
>I must accept that as your thinking. All I can say is that I am astonished that
>Kasparov's thoughts on chess seem to count so little.
>
Not chess, computer chess. He is the foremost expert on chess, not computer
chess. There are many people in this forum who are better qualified to say what
a _computer_ can or can't do in terms of chess.
>>>As I said before, Hsu will come down to normal the moment he will see what he
>>>has lost when he treated Kasparov like an unwanted guest when in real Kasparov
>>>was the very special favorite who brought computerchess into the daily
>>>newspapers... The denial of the logfiles was the most stupid and shortsighted
>>>action the leader of a scientifical project (that consisted of a close
>>>cooperation between a single human being and a team of scientists) could
>>>produce.
>>
>>There was no cooperation. It was a match. A publicity stunt for all involved.
>>That is all.
>>
>>>
>>>How Hsu could have behaved better?
>>
>>
>>I doubt Hsu had anything to do with the decisions involved in the printouts
>>episode.
>>
>> Albert Silver
>
>
>This is a very delicate situation. Perhaps it would go too far if I continued to
>put my finger into that wound of bad behavior in science. As I wrote weeks ago
>the leading scientist of some project can not be satisfied if his setting is
>destroyed and suddenly a completely different motivation is dominating. Earlier
>it was the machine's strength that should be tested in a little match and now it
>was more a typical psycho war to weaken the strengths of the human player.
I don't believe in any Psycho War. Any problems caused by this episode were
instigated by Kasparov alone. The conditions of this match compared to prevous
ones were either identical or better, so his sudden antics were possibly a part
of his own psycho war tactics, not the DB team's. I think he was just preparing
(unconsciously) his long list of excuses as he was getting very nervous and
realized the machine was stronger than he had anticipated. At least he was
honest enough in the immediate press conference after game 6, when he admitted
to having been afraid. He never again repeated that though, and commentary went
along the lines of cheating, plots to take him down, etc... If we play a game,
and you make a few moves that impress me deeply, should you have to prove you
aren't guilty of cheating? That is all I see here factually, nothing more.
Albert Silver
>
>Do you really think that the quality of the machine could be proven this way?
>
>
>
>Hans Gerber
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.