Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: IBM hired the wrong people because it won?

Author: Robert Hyatt

Date: 06:23:06 01/12/99

Go up one level in this thread


On January 11, 1999 at 21:15:47, Graham Laight wrote:

>On January 11, 1999 at 09:33:03, Robert Hyatt wrote:
>
>>>In the last game kasparov simply replaced the order of moves.
>>>He played a line that he recommended not to play in his book.
>>>This is not something that he does all the time.
>>>
>>>Uri
>>
>>Or, depending on who you believe, Kasparov tried that sequence of moves out
>>against Fritz over and over and won every game, even if fritz did sacrifice
>>the piece on e6.  To date no computer has been able to win the white side of
>>this game against a strong human (or computer) opponent.  So even this game
>>shows that DB is fearsome...
>
>I have seen a lot of great wisdom and straight talking common sense from Bob
>posting on the newsgroups over the years. Mostly, I agree with him - even when
>I'm on the other side of the discussion.
>
>But on this occasion - come on Bob, who are you trying to kid?
>
>At the risk of breaching copyright, I quote directly from the book "Kasparov V
>Deeper Blue" by Daniel King.
>
>*** Start Of Quoted Text ***
>
>Although this has been seen a few times at grandmaster level, it is known to be
>a mistake because of...
>
>8 Nxe6
>
>Garry shook his head in disbelief.
>
>He had made a finger-slip allowing a known sacrifice which, of course, was all
>in Deep Blue's database.
>
>*** End Of Quoted Text ***
>
>This was written not only by a chess GM, but also a man who was watching Garry
>on the video monitor.
>
>I think that most of us would conclude that Garry made a mistake. Daniel's
>description is incongruent with the concept of a man with a secret plan who is
>playing a computer that has no idea about the opponent's emotional reactions.
>
>Graham.

I am basing my comment only on comments by the Kasparov team before/after the
match.  I don't claim that my conjecture is true.  I only claim that it is
possible.  IE at one of the ACM events, I sat across the board from Ken Thompson
helping him analyze a position.  He was playing "nuchess" in a decisive game and
needed a win to win the tournament.  He found a line that looked good and wanted
other opinions.  It was based on the idea that a computer will often give up a
piece to win a rook on h1, but it doesn't realize that the knight that captured
on a1 is trapped and gets lost eventually.  We tried the position on Cray Blitz
and it fell for it.  So we assumed that Nuchess would also since they weren't
as fast.  They did, Belle won.

It is certainly possible that as Kasparov played game after game against Fritz
when preparing for the event, that he noticed that this position could arise and
that as black, he could thump fritz every time.  And feeling desparate enough
in game 6, he could certainly have tried to unleash his "trap" only to have it
backfire.

I make no claim this is true.  Or that it is even probably.  I don't know.  But
I do know it is definitely 'possible'.  I have seen humans do this many times
against computers.  One statement Kasparov made was _really_ revealing however.
Before the match, he made a statement that he thought that DB would be able to
beat fritz 80-90% of the time.  From results reported by Hsu, this was more than
just a little "low".  And it could most definitely have led to trying something
unsound.

And let me point out once again, after playing Nxe6 as white, to date no
computer has been able to win with the white pieces, against another computer
or against strong human opposition.  Which means that it is certainly possible
that Kasparov had noticed this and hoped for a swindle.

That's all I said.  With no idea of how probable it is that this actually
happened.  But in light of the above, you'd have to agree that it is certainly
not an impossible scenario?

I've done this against computers.  At USM we used to have monthly chess club
tournaments, and when we had an odd number, we would slip in a computer, like
the Novag supercon or whatever.  And I generally won these games because I'd
do just such planning.  :)



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.