Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: X3D Fritz with the inferior hardware is much better than Deeper Blue

Author: Robert Hyatt

Date: 11:37:12 11/13/03

Go up one level in this thread


On November 13, 2003 at 11:17:24, Uri Blass wrote:

>On November 13, 2003 at 10:58:27, Robert Hyatt wrote:
>
>>On November 13, 2003 at 10:21:45, Uri Blass wrote:
>>
>>>On November 13, 2003 at 10:03:48, Robert Hyatt wrote:
>>>
>>>>On November 13, 2003 at 01:59:30, Les Fernandez wrote:
>>>>
>>>>>On November 12, 2003 at 18:21:27, Robert Hyatt wrote:
>>>>>
>>>>>>On November 12, 2003 at 18:04:18, Dan Sicerly wrote:
>>>>>>
>>>>>>>On November 12, 2003 at 13:24:32, Robert Hyatt wrote:
>>>>>>>
>>>>>>>>On November 12, 2003 at 03:56:09, Jorge Pichard wrote:
>>>>>>>>
>>>>>>>>>On November 12, 2003 at 03:21:46, Uri Blass wrote:
>>>>>>>>>
>>>>>>>>>>On November 12, 2003 at 03:16:44, Uri Blass wrote:
>>>>>>>>>>
>>>>>>>>>>>On November 12, 2003 at 02:11:04, Derek Paquette wrote:
>>>>>>>>>>>
>>>>>>>>>>>>I'm reading through all these posts and watched the games, and something
>>>>>>>>>>>>occurred to me, especially what the CEO of chessbase said, that kasparov was
>>>>>>>>>>>>sent the versions of x3d fritz as they were created...so...
>>>>>>>>>>>>
>>>>>>>>>>>>If chessbase were able to clone kasparov use him, study him, poke holes in his
>>>>>>>>>>>>game, analyse him for hundreds of hours, how well would they do?
>>>>>>>>>>>>
>>>>>>>>>>>>But thats a question we'll never be able to answer, but how about this ,
>>>>>>>>>>>>
>>>>>>>>>>>>If they sent x3d version fritz with the same hardware to a large super
>>>>>>>>>>>>grandmaster tournament, a round robin, where all the players not only had to
>>>>>>>>>>>>prepare for the computer, but other humans, how would it fair?
>>>>>>>>>>>>
>>>>>>>>>>>>I think we are seriously under estimating this beast...
>>>>>>>>>>>>just my opinion of course.
>>>>>>>>>>>>-Derek
>>>>>>>>>>>
>>>>>>>>>>>This is exactly what chessbase want you to believe.
>>>>>>>>>>>
>>>>>>>>>>>I think that all these matches are unimportant matches and it may be better if
>>>>>>>>>>>we ignore them.
>>>>>>>>>>>
>>>>>>>>>>>I do not trust the humans in these match to try to win the matches and I suspect
>>>>>>>>>>>that they believe that draw can give them more money.
>>>>>>>>>>>
>>>>>>>>>>>Uri
>>>>>>>>>>
>>>>>>>>>>I can add that the fact that kasparov got another match after drawing with Fritz
>>>>>>>>>>when Smirin never got another match after beating the machines increase my
>>>>>>>>>>impression(hardware was slower against smirin but time control was also faster
>>>>>>>>>>and the second factor is something that is supposed to help the machines so I
>>>>>>>>>>think that smirin's result is better than kasparov's result and kramnik's
>>>>>>>>>>result).
>>>>>>>>>>
>>>>>>>>>>Players may believe that it is better if they do not win because if they draw
>>>>>>>>>>they have better chances to get another match.
>>>>>>>>>>
>>>>>>>>>>Uri
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>Kasparov lost to IBM Deeper Blue and still has not gotten a rematch, and I
>>>>>>>>>believe that X3D Fritz with the inferior hardware is much better than Deeper
>>>>>>>>>Blue :-)
>>>>>>>>>
>>>>>>>>>Jorge
>>>>>>>>
>>>>>>>>Why don't you knock off the wild speculation, and just watch.  If fritz
>>>>>>>>wins, it might be as good as, or better than deep blue.  If it loses, it
>>>>>>>>isn't.
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>  You're not just being funny are you?  I have heard you say a million times
>>>>>>>that the numbers that deepblue calculates makes it impossible for any micro to
>>>>>>>beat it.
>>>>>>
>>>>>>I still believe that although the gap is closing.  IE I have seen a quad-cpu
>>>>>>machine run Crafty at 7M nodes per second.  Not to DB's speed, but definitely
>>>>>>closer than 3 years ago.
>>>>>>
>>>>>>But to make statements such as fritz is much better is simply ridiculous until
>>>>>>Fritz manages to beat Kasparov in the match.  Until it does, there is no
>>>>>>evidence to support this "superiority".  If Fritz doesn't win (ie it loses or
>>>>>>draws) there is also no evidence to support this since DB actually _won_.
>>>>>
>>>>>Hi Bob,
>>>>>
>>>>>I am not sure much can be deduced by this match if fritz loses since I strongly
>>>>>believe fatigue is a variable in man vs machine and this match is made up of
>>>>>only 4 games whereas the DB was 6.  To me it may be more conclusive if this was
>>>>>a similar 6 game format.  If I can find the results it might be interesting to
>>>>>see how Kaspy has done if we look at his win/loss record of the first half of
>>>>>all his games played on this level and compare it to the last half record.  Do
>>>>>you happen to have all the DB, Junior results? If so can you send them to me, I
>>>>>would be interested in them.
>>>>>
>>>>>Thx Bob,
>>>>
>>>>My only comment is that if Fritz wins, the debate can continue.  If Fritz
>>>>loses, then the discussion is moot for the time being.  Kasparov has, IMHO,
>>>>a significant disadvantage using the 3d stuff.  I have used that stuff
>>>>myself and I would not want to do it for four hours.  I notice he leaves
>>>>the "board" frequently, which probably helps a bit.  But this is a flawed
>>>>idea, IMHO, as he should be playing on a real board.
>>>>
>>>>If Fritz still manages to lose, then comparing it to DB is difficult, since
>>>>DB won 7 years ago, on hardware 7 years older than today's.  If Fritz wins,
>>>>then fritz and DB can be used in the same sentence with some sense.
>>>
>>>You forget that deeper blue had the advantage of not being a commercial program
>>>so Kasparov could not train at home against it to learn its weaknesses.
>>>
>>>Uri
>>
>>
>>I don't know that he can train against this version of DF either.  Remember
>>that he _had_ played the previous version in 1996.  This version that beat him
>>was put together right before the match so no practice would have been possible
>>had it been agreed to...
>
>It was possible by delaying the match.
>
>It is also clear that kasparov has at least the commercial fritz and it seems
>similiar to the real Fritz that is playing.
>
>Fritz8 likes Nxf2 exactly like the Fritz that was playing and people found that
>public Fritz8 can find the castling move that shredder does not like after
>enough time.
>
>Kasparov did not have deeper blue of few months before the match and he even did
>not have the version of 1996 to train against it.
>
>Uri


No, but one could make the case that the 1996 match _was_ training for
the 1997 match.  :)



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.