Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: A Blast from the past - DBII didn't win a single game!!

Author: Walter Faxon

Date: 20:09:13 05/01/05

Go up one level in this thread


On April 30, 2005 at 22:54:29, Robert Hyatt wrote:

>On April 30, 2005 at 17:20:57, Uri Blass wrote:
>
>>On April 30, 2005 at 15:34:28, Robert Hyatt wrote:
>>
>>>On April 29, 2005 at 16:32:16, Uri Blass wrote:
>>>
>>>>On April 29, 2005 at 06:50:46, Vasik Rajlich wrote:
>>>>
>>>>>On April 29, 2005 at 05:23:51, Uri Blass wrote:
>>>>>
>>>>>>On April 28, 2005 at 23:17:46, Robert Hyatt wrote:
>>>>>>
>>>>>>>On April 28, 2005 at 22:01:24, chandler yergin wrote:
>>>>>>>
>>>>>>>>On April 28, 2005 at 20:30:36, Robert Hyatt wrote:
>>>>>>>>
>>>>>>>>>On April 28, 2005 at 15:17:13, Rolf Tueschen wrote:
>>>>>>>>>
>>>>>>>>>>On April 28, 2005 at 14:12:33, Robert Hyatt wrote:
>>>>>>>>>>
>>>>>>>>>>>Hsu didn't win.  Deep Blue "won".
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>>No.
>>>>>>>>>>
>>>>>>>>>>1. DBII lost in game one. Very badly. Kasparov's superior chess!
>>>>>>>>>>
>>>>>>>>>>2. Then Kasparov gave up a drawn game. He was very confused.
>>>>>>>>>>
>>>>>>>>>>3. Then three draws.
>>>>>>>>>>
>>>>>>>>>>4. Then Kasparov threw another game. He was no longer in the match since the
>>>>>>>>>>second game...
>>>>>>>>>>
>>>>>>>>>>So, where DBII WON a single game?
>>>>>>>>>
>>>>>>>>>The _official_ history of the match has DB winning rounds 2 and 6, losing round
>>>>>>>>>1, and drawing the rest.  What part of that is confusing?  If you want to play
>>>>>>>>>word games and say that the 1-0 result in round two means black lost rather than
>>>>>>>>>that white won, fine.  In round 1 DB lost, Kasparov didn't win.
>>>>>>>>
>>>>>>>>Could not the same thing be said about Game 6?
>>>>>>>
>>>>>>>
>>>>>>>That was my point.  That was Rolf's claim in fact.  It works both ways.  But
>>>>>>>whether DB won a game, or Kasparov lost the game, the final result is _exactly_
>>>>>>>the same...
>>>>>>
>>>>>>The final result is not exactly the same in the meaning that if DB won games not
>>>>>>thanks to stupid errors of kasparov then the impression of the chess players
>>>>>>could be that DB is better than Kasparov and after the match the impression of
>>>>>>the chess players was not that DB is better than Kasparov.
>>>>>>
>>>>>>Both mistakes of kasparov in the games that he lost are mistakes that kasparov
>>>>>>does not do against humans and both were result of not understanding the
>>>>>>machine.
>>>>>>
>>>>>>In game 2 Kasparov believed that he has no chance for a perpetual check because
>>>>>>he believed that DB could see it after getting the impression that DB saw
>>>>>>another perpetual earlier in the game because it was the only way that kasparov
>>>>>>could explain DB's move to himself.
>>>>>>The result was that he even did not analyze this possibility.
>>>>>>
>>>>>>In game 6 Kasparov played for a line that he was not ready to play.
>>>>>>
>>>>>>DB did not get a losing position in game 1 because of a stupid error so the
>>>>>>story of game 1 is different.
>>>>>>
>>>>>>The impression of the chess players after the match was that DBII was not better
>>>>>>than Kasparov at 1997 inspite of the results.
>>>>>>
>>>>>>Uri
>>>>>
>>>>>If you want to say that DBII played just 6 games and therefore nothing was
>>>>>proved - that's fine. That was indeed a horrible waste of resources.
>>>>>
>>>>>But this crap about game 1 being a "quality win" while games 2 and 6 were junk -
>>>>>come on.
>>>>>
>>>>>The "cleanest" game was game 2. Yes, there was an extremely difficult perpetual
>>>>>at the end that was missed by both sides, but aside from that it was a nice
>>>>>"positional" game. You can just pretend that Kasparov resigned two or three
>>>>>moves earlier.
>>>>
>>>>No
>>>>
>>>>There was no reason to resign in game 2 earlier.
>>>>
>>>>You resign when you believe that you have no practical chances and the game
>>>>proves that kasparov had practical chances earlier.
>>>>
>>>>I believe that kasparov resigned only because of the fact that the opponent was
>>>>a computer and he believed that the program cannot miss a perpetual after it did
>>>>not miss a similiar idea of very deep perpetual earlier(at least this is what he
>>>>believed because he had no different explanation for earlier moves of DBII) so
>>>>he even did not check if he has a perpetual and assumed that he has no chances.
>>>>
>>>>Uri
>>>
>>>
>>>I don't buy that reasoning.  It assumes that somehow Kasparov "knew" that the
>>>computer thought it was winning before and after the move it played.  How would
>>>he know whether or not DB was making a move that led to a forced repetition, as
>>>its best alternative?  He couldn't see the display...
>>
>>He could not see the display but he could know that the position of deeper blue
>>is better based on positional knowledge so he also believed that the machine
>>knows it.
>>
>
>That is a circular definition.  He knows it because he knows it.
>
>I've seen many positions in both human and computer chess where one side appears
>to be better, but the opposite is true, or the position is a draw, even though
>one side appears to have a significant edge at first glance.
>
>
>
>
>>The machine had alternative moves that did not give Kasparov the chance to force
>>perpetual check and kasparov believed that the machine will choose one of them
>>in case that he has a perpetual check so he even did not analyze the line that
>>he sacrifice a piece and assume it is not drawn and analyzed only the line that
>>he get inferior endgame.
>
>Same problem.  It didn't play a move that didn't lead to a draw, because it was
>winning.  But what if in reality, those "non-drawing" moves were actually losing
>moves, or at least DB believed they were losing?
>
>I think it is completely foolish to play your game based on what you believe the
>machine might or might not see.  To do otherwise invites this very kind of
>problem.


I believe Kasparov "trusted" that DB would not have allowed a perpetual, so to
"save face" as well as effort, he resigned.  Not wise policy, but people often
do foolish things.

As an example of "trusting" your opponent, I have often read where, when a GM
takes a good long think then offers a sacrifice, his (or her) GM opponent will
decline immediately, on principle.  Why assume that you can out-think another GM
tactically?  Why assume that this particular sacrifice is unsound?  Instead,
just make good moves and make turn the time your opponent used into a waste --
which is always a clear gain.



>>I believe that kasparov never did the mistake of resigning in a drawn position
>>against humans so the fact that he resigned in a drawn position against deeper
>>blue needs some explanation.
>
>
>Yes, but a simple "I failed to notice the draw, I thought white had a way out
>but I had mis-calculated" would have been the better answer.
>
>
>>
>>Note that even if we get the wrong assumption that deeper blue cannot miss
>>perpetual check that humans can find then Kasparov could still hope that he may
>>force the machine to choose between letting him draw by perpetual check and
>>letting him another draw from inferior endgame by it's evaluation when the
>>machine does not allow him perpetual check so it is clear that not analyzing the
>>option of perpetual check was a mistake because even if Deeper blue can escape
>>from perpetual then the question how it can escape is important.
>>
>>Uri



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.