Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: DB and bias

Author: Dave Gomboc

Date: 02:20:02 07/22/00

Go up one level in this thread


On July 21, 2000 at 13:43:46, Robert Hyatt wrote:

>On July 21, 2000 at 13:12:33, Dave Gomboc wrote:
>
>>On July 21, 2000 at 13:10:28, Dave Gomboc wrote:
>>
>>>On July 21, 2000 at 12:14:44, Graham Laight wrote:
>>>
>>>>On July 21, 2000 at 11:21:57, Robert Hyatt wrote:
>>>>
>>>>>Not at all.  Hsu was hired by IBM to build a new chess machine that would beat
>>>>>Kasparov.  The machine was first assembled in late 1995 or early 1996, not too
>>>>>much before the actual match happened.  Prior to that, all they had was the
>>>>>deep thought hardware that was still located at Carnegie-Melon university
>>>>>according to Hsu's book.
>>>>
>>>>I've just checked Amazon - there doesn't seem to be anything related to chess
>>>>available by a "Hsu".
>>>>
>>>>Any idea when we'll be able to buy this book, please?
>>>>
>>>>-g
>>>
>>>Tough to say.  I know of a first cut, and a rewrite.  Perhaps there will be
>>>major additional changes, perhaps not.  I'm not that familiar with the
>>>book-writing business.
>>>
>>>Dave
>>
>>For a more frank account of the match, I'd suggest lobbying Joel Benjamin to
>>publish his book.
>>
>>Dave
>
>
>The problem with Joel is that he doesn't have the technical insight into the
>background of the machine.  He knows what it could do playing chess, and he
>knows the many suggested evaluation features he requested to make it play
>better.  But he doesn't know much about the underlying architecture, the history
>of the thing (dating back to chiptest) and so forth.
>
>I think that a book by each would be nice complementary works and would likely
>give a more rounded view of the thing.
>
>Of course, there will be cries of 'biased reporting' and there we go again,
>since Joel did work for IBM on the project...

There's already 'biased reporting', because nobody from the Deep Blue team has
put down on paper what must be said.

For instance, GK got mauled in game 2 in 1997.  How often does one see GK have
nothing: no material, no initiative, no activity... just a bad position.  He
suffered for the _entire_ game.  You could see it on his face, you could see it
in his body language.  Despite the position being closed -- computers are
supposed to suck at closed positions -- Deep Blue rolled him up like a wet paper
bag.  You'd probably have to go back to Karpov-Kasparov I to find another game
where GK had been so completely outplayed.

But Deep Blue, as good as it is, isn't perfect, and it played Kf1 instead of
Kh1, most likely because of some king centralization term for the endgame that
wasn't revelant in the specific position.  So what's the party line?  "GK
resigned in a drawn position." -- a summation that hardly does justice to DB's
accomplishment.

Hsu is _extremely_ kind to GK in his book; you must know that.

Dave



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.