Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Deep Junior is better even if it is slower than Deeper Blue !

Author: Robert Hyatt

Date: 17:36:08 01/28/03

Go up one level in this thread


On January 28, 2003 at 18:00:32, Rolf Tueschen wrote:

>On January 27, 2003 at 21:15:52, Robert Hyatt wrote:
>
>>On January 27, 2003 at 15:38:14, Rolf Tueschen wrote:
>>
>>>On January 27, 2003 at 15:01:00, Robert Hyatt wrote:
>>>
>>>>On January 27, 2003 at 11:36:28, Sune Fischer wrote:
>>>>
>>>>>On January 27, 2003 at 10:57:26, Matthew Hull wrote:
>>>>>
>>>>>>[snip]
>>>>>>
>>>>>>
>>>>>>>When did Junior ever "remember" the table bases?
>>>>>>
>>>>>>When did Frenzee ever remember the Ruy Lopez?  Same issue, yes?  But I don't see
>>>>>>you arguing that opening books are unfair, a la Rolf Tueschen.
>>>>>
>>>>>That doesn't mean I don't agree with Rolf :)
>>>>>
>>>>>I just think it is possible to put a bit more of the engine personality into the
>>>>>book, for instance with learning. If you win against a higher rated player,
>>>>>remember this position as good, if you lose to a lower rated player don't play
>>>>>this again.
>>>>>
>>>>>However, you can see clearly from the discussions going on here that the book is
>>>>>not perceived as being part of Junior, statements like: "Junior was lost out of
>>>>>book" clearly distinguishes the engine "Junior" from the book _used_ by Junior.
>>>>>It is as though Junior didn't lose, it was the book that lost the game on behalf
>>>>>of Junior! This is also the natural way to think, it really is two very
>>>>>different pieces of software.
>>>>
>>>>I don't agree.  I have heard _many_ humans say "I got into a lost position from
>>>>the book line I played."  Did they play the line, did they remember it, or did
>>>>they do _both_???
>>>
>>>
>>>Surely you didn't hear GM talk this way becauase it would be a contradiction in
>>>itself because GM don't have books. They have analyses. But these are not books
>>>yet. Of course GM remember their analyses but at least it's their stuff.
>>>
>>
>>I have heard the following:
>>
>>1.  I did not remember the sequence of moves correctly, bungled the order, and
>>played into a lost position.  (Kasparov, round 6, 1997 vs deep blue).  Implies
>>he played _moves_ from memory.  Not something he had "learned" by playing the
>>games himself.
>>
>>2.  I overlooked that reply in my home-analysis I had prepared, and ended up
>>in a lost position.  This has been stated by _many_ GM players over the years,
>>after they were "out-booked" by their opponent's pre-tournament or pre-match
>>preparation.
>>
>>3.  My opponent had done his homework and hit me with a TN I had not seen
>>before, leaving me in a terrible position.
>>
>>Etc...
>
>Heard or read? Anyway I don't buy that. That is a chapter of research that
>hasn't been written yet. They don't talk too much abot this.
>
>


"heard".






>>
>>
>>>
>>>>
>>>>I don't think the line is that sharp between book and engine, particularly when
>>>>there
>>>>is "learning" involved.
>>>>
>>>>
>>>>>
>>>>>The same applies to the egtbs, Junior wouldn't be winning, the egtbs would be.
>>>>
>>>>Suppose Junior had the code to _build_ the egtbs built in?  What then?
>>>>
>>>>Suppose the EGTBs could be built on the fly, on demand, while the program was
>>>>running.  What then?
>>>
>>>
>>>Suppose the engine would create a new engine on the fly - say Junior would give
>>>birth to Fritz on the fly. Then Kasparov could really take the train back to
>>>Moscow. BTW I announce copyright for that idea in CC! Why should that be
>>>impossible? Could a Grandmaster NOT change his personality in the middle of a
>>>game??? Aha! Bingo. QED. 500000$$$ please.
>>>
>>>
>>
>>Sure they can.  Kramnik did.  It backfired.  Kasparov did.  It backfired.
>
>No, also here I doubt that they tried it. Playing a "new" opening does't mean
>that change of personality we know from CC.
>
>Since I read that someone said that Kasparov played a new opening on Sunday let
>me say that Kasparov played the opening with both colours.
>
>Rolf Tueschen
>
>
>
>>
>>>:)
>>>
>>>
>>>Rolf Tueschen
>>>
>>>
>>>
>>>
>>>>
>>>>
>>>>
>>>>
>>>>>
>>>>>However I must agree (I guess) that "the machine" is all of the above, so
>>>>>depends on what you want from the match. I think Kasparov wanted to play Junior
>>>>>and not the egtbs, so for that reason I think the rules are fine.
>>>>>
>>>>>-S.
>>>>>
>>>>>
>>>>>>:)
>>>>>>Matt



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.