Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Is Deep Blue still considered better than Deep Junior ?

Author: Robert Hyatt

Date: 19:55:28 08/19/02

Go up one level in this thread


On August 19, 2002 at 17:44:36, Chris Carson wrote:

>On August 19, 2002 at 17:12:05, Robert Hyatt wrote:
>
>>On August 19, 2002 at 13:36:49, Chris Carson wrote:
>>
>>>On August 19, 2002 at 11:27:15, Robert Hyatt wrote:
>>>
>>>>On August 19, 2002 at 09:12:31, Chris Carson wrote:
>>>>
>>>>>On August 18, 2002 at 21:51:35, Robert Hyatt wrote:
>>>>>
>>>>>>On August 18, 2002 at 12:48:02, Chris Carson wrote:
>>>>>>
>>>>>>>On August 18, 2002 at 10:38:19, Robert Hyatt wrote:
>>>>>>>
>>>>>>>
>>>>>>>>
>>>>>>>>How would you figure that?  Deep Thought produced a 2650 rating over 25
>>>>>>>>consecutive 40/2hr games against GM players.  Anybody else done that yet?
>>>>>>>>Not that I have seen.
>>>>>>>>
>>>>>>>
>>>>>>>Deep Thought was a great piece of HW/SW for it's time.  Just a few observations
>>>>>>>when comparing to today's program:
>>>>>>>
>>>>>>>1.  DT did that aginst GM's with average ELO of 2475 (not very strong for GM's).
>>>>>>>
>>>>>>>2.  DT played against GM's that had far less experience (none) using
>>>>>>>anti-computer tatics.
>>>>>>>
>>>>>>>3.  There are a lot of tatical/strategic holes in DT games that modern
>>>>>>>commercial programs do not make.
>>>>>>>
>>>>>>>4.  Today's programs would crush DT.
>>>>>>>
>>>>>>>5.  The GM's that DJ, DF, CT, Rebel, Hiarcs, The King have played recently would
>>>>>>>crush DT.
>>>>>>>
>>>>>>>6.  DT would have no chance in a match with a 2700 GM (today) that had 100 games
>>>>>>>of preparation, Rebel played this GM even for 4 games.
>>>>>>>
>>>>>>>7.  I am pretty sure I could crush DT today.
>>>>>>>
>>>>>>>Chris
>>>>>>
>>>>>>
>>>>>>I can't prove anything about deep thought, but I'll offer you a chance to
>>>>>>put up or shut up on point 7.  Deep Thought easily beat Cray Blitz every time
>>>>>>we played (except for the first, where we could have won had we not run into
>>>>>>a serious bug).  Cray Blitz waxed Crafty in a 10 game match last year.
>>>>>>That means _you_ should be able to wax crafty on my quad xeon easily if you
>>>>>>could crush deep thought.
>>>>>>
>>>>>>Let me know when you are ready to "take the test".
>>>>>>
>>>>>>I don't believe you can do it for a minute...
>>>>>>
>>>>>>But the ball is in your court to accept the challenge...
>>>>>
>>>>>Here are all the PGN's for DT and DB.  Cherry pick any you like.  I do not think
>>>>>DT/DB or HSU needs the help, their accomplishments were very good, but no need
>>>>>to make them out to be better than today's programs or mis-represent them, you
>>>>>do more harm than help.
>>>>>
>>>>>Here are the ELO's against FIDE rated players for DT/DB and Commercials, the
>>>>>commercials have passed DT and DB96 ELO performance, DB97 will fall when it
>>>>>falls:
>>>>>
>>>>>PGN's:  http://home.interact.se/~w100107/welcome.htm
>>>>>
>>>>>Ratings HvC: http://home.interact.se/~w100107/manmachine.htm
>>>>>
>>>>>I am done discussing this with you unless you can "show with PGN's" something
>>>>>different.
>>>>>
>>>>>Chris
>>>>
>>>>
>>>>I never collected DT pgns.  However, I _know_ what happened to win the
>>>>Fredkin prize.  The opponents and results _might_ be in an older ISSUE of
>>>>the JICCA.  I will check when I have time...
>>>>
>>>>I haven't seen _any_ results where a micro produced a 2650 over 25 consecutive
>>>>games.  Yes, you can cherry-pick a tournament here and there.  But 25 games
>>>>is at least 3 events in a row....
>>>
>>>You ignore two points, or at least discount them:
>>>
>>>1.  Commercial programs can be played anytime.  A GM played 100 games against
>>>Rebel, I will ask Ed what the score was, then played a 4 game match.  DT/DB did
>>>not allow people to play prep games, you have held it against the commercials in
>>>the past, but not DT/DB.  DT did play a lot of games however, but noone
>>>individual had a lot of games against it.
>>
>>A few did, but that is the way it goes, of course.
>>
>>
>>>
>>>2.  Anti-computer tactics.  This would reduce DT rating by 100 or more points
>>>and a second hundred point reduction if the 25 games were match play.  Most of
>>>the commercials games are match play, in tournament they do a lot better (about
>>>100 points or more)
>>
>>I don't believe this is as true as you think.  First, most GM players are _not_
>>using anti-computer yet.  A few do.  But then again, a few did back in
>>the 1980's and 1990's as well.  So that hasn't changed much at all.  It _may_
>>change in the future as the micros continue to get stronger, however.  But
>>today, I don't believe most GM-level players even know what anti-computer
>>means, other than "they hate 'em and don't want 'em in the tournaments."
>>
>>>
>>>Again, DT/DB was great, HSU deserves a lot of credit.  The commercials have
>>>exceded it thought, DB97 would be crushed by Kramnik or Kasparov with
>>>preparation games.  Just my opinion, but based on DT/DB and commercial
>>>performance.
>>
>>I don't believe _any_ GM would be able to "crush" deep blue 1997.  Not now.
>>Not _ever_.  Perhaps beat it 3 of every 4 games?  Maybe, with some luck.  But
>>not "crushing"...
>
>Winning 3 out of 4 points is a "crushing win" in my opinion.  That would be a
>300 point ELO difference.  That is a pretty big difference.


First, three out of four is 200, not 300.  Second, notice I said "3 out of 4
with luck".  Not 3 out of 4 consistently.  I would bet that in a match between
DB2 and any top GM of today, at _least_ 1/2 of the games would end in draws,
and maybe more.  I don't believe the GM would win all of the others by any
stretch...

I don't believe a GM will really _crush_ any computer now playing, personally.
I believe they can win matches handily, but that is as far as I would go in
predicting any outcome...



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.