Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Behind deep Blue

Author: Robert Hyatt

Date: 11:25:13 10/23/02

Go up one level in this thread


On October 23, 2002 at 13:19:41, Vincent Diepeveen wrote:

>On October 23, 2002 at 11:26:38, Robert Hyatt wrote:
>
>>On October 23, 2002 at 05:08:11, Vincent Diepeveen wrote:
>>
>>>On October 22, 2002 at 17:29:53, martin fierz wrote:
>>>
>>>let's be clear. the kramnik guy was happy to receive
>>>a million dollar in advance. Without much effort he played
>>>a few moves and it was 3-1. Then everyone started complaining
>>>that the match got no publicity and got no excitement.
>>>
>>>He then gives away a piece in a clear drawn position with
>>>a 1b trick (1 check in between). That's bullet blunder level.
>>>
>>>In fact i don't make such mistakes that much at bullet and
>>>last time i made such a mistake at slow level was a year or
>>>10 ago. Kramnik had plenty of time.
>>>
>>>0% chance he didn't deliberately blunder there.
>>
>>I think that is a totally stupid statement to make.  I can point out GM blunders
>>in _every_ tournament I have watched online.  I have seen them overlook a mate
>>in 2.  A hanging queen.  You-name-it.  Human GMs _do_ make mistakes.  Not as
>>often as non-GM players, but also far more often than "never".
>
>Ha the 'expert' is speaking here.
>
>In important matches the only few blunders i remember are also from
>a year or 10 ago, and most definitely not from Kramnik, and always
>in *big* time trouble.

Not in every game I have seen.  I saw a GM think for 20 minutes, with plenty of
time left, and he made a move that forced him to give up his queen to avoid a
mate
the very next move.  It happens, regardless of your hand-waving.


>
>Kramnik had 15 minutes left here...
>
>So forget your online toying. Online games aren't earning money.



Vincent, I was talking about Las Palmas a few years ago.  GM vs GM.  Not online
vs a computer.  Please read first, and then respond, rather than writing stuff
that is
unrelated to my point in any way...





>
>You know that, and i know it.

I know that GM players make ugly blunders in real games vs other GM players,
even
after thinking 20 minutes.  Kramnik dropped a piece after a long think against
Fritz.
It happens.




>
>What i remember is something from the 80s. A match Kasparov-Karpov,
>where Karpov makes a mistake in the endgame. If i show you the position
>you will most likely call it 'every day blunder stuff'.
>
>You should realize how well these guys are playing always.

Never said otherwise.  But they _do_ make basic blunders on occasion, and not
only when in dire time trouble either...



>
>They are not people who make a blunder in every game they play.

So?  Kramnik didn't make a blunder in every game he played vs Fritz either...



>
>They have for years nightmares when they blunder in a game without
>major time pressure.
>
>Kramnik had no pressure here.
>
>Apart from that, what's the last game you played 40 in 2 online?
>
>I can't remember any GM playing 40 in 2 online. Do you?

I wasn't talking about GM vs Computer, as I said.  Crafty was simply giving
analysis
as ICC relayed live games from events like Los Palmas (los or las, I don't
remember)...

>
>
>>
>>>
>>>He just didn't care. He wasn't shocked after the blunder.
>>>
>>>It was 3-2.
>>>
>>>then some beginner game at round 6. he has white. No way
>>>with his default play to even draw it. He would win it
>>>without problems. So kramnik decides to put some fire in
>>>the game and plays open position. A line he has never played
>>>before.
>>>
>>>To play Bxf7 objective criteria are not there. There is just
>>>one criteria: "make the match exciting".
>>>
>>>Of course that's very hard if you play your entire life 1.Nf3,
>>>so for a change he played some other d4 c4 line and opened the
>>>position. Then give away a piece for a few checks and the
>>>crowd has something
>>>
>>>3-3
>>>
>>>Exciting enough for the sponsor?
>>>
>>>Yes. So 2 quick draws are enough. Kramnik could have won game 8
>>>if he wanted to. He didn't want to. he gave it draw.
>>>
>>>Of course that openings line he normally doesn't play.
>>>
>>>So i was unhappy, i had expected kramnik to just make default moves
>>>and win at least with 4.5 - 3.5
>>>
>>>Anyway, this match drew no attention at all. No newspaper and no TV
>>>station and no one saw it. I can't speak for german newspapers, but
>>>at ARD,ZDF and WDR i didn't see anything broadcasted yet.
>>>
>>>30 seconds at CNN, and even those 30 seconds i missed. That's all that
>>>is broadcasted AFAIK.
>>>
>>>This match was not important. This match is not important. This match
>>>will never be important except for sales for chessbase and 4-4 on their
>>>box not showing a picture of kramnik, becuase no one knows the guy.
>>>
>>>I find kramnik stupid to not play 1 good game at the end.
>>>
>>>Whole match has been dominated again by games with stupid openings and
>>>many sidelines and even openings kramnik never played before his entire
>>>life.
>>>
>>>The match will be forgotten.
>>>
>>>Kasparov - Junior will draw hopefully more attention from the big crowds.
>>>
>>>Of course 'deep junior' the big crowd wil not see as a program called 'junior'.
>>>
>>>They will see it as a successor of deep blue of course.
>>>
>>>First you have 'deep blue' where they needed some old big machine,
>>>then you have 'deep junior'. Very logical. I will not even explain
>>>it too much around me. It's impossible to explain.
>>>
>>>computer = computer. Deep == Deep == Diep
>>>
>>>I stopped already long time ago explaining that Diep != Deep Blue
>>>You can't help it. You can't blame them either.
>>>
>>>If they see Diep with blue interface as Deep Blue, obviously
>>>they will see Deep Junior as the latest Deep Blue version.
>>>
>>>In some respects they are right. It's also focussed upon getting
>>>a shitload of nodes a second, it's searching deep, and it's positional
>>>play is not very good.
>>>
>>>Perhaps Kramnik will understand soon how stpuid he was. We can only
>>>hope Kasparov does by now.
>>>
>>>I guess IBM marketing department managed to inform at least 30% of
>>>the world population about Deep Blue solving chess in 1997.
>>>
>>>The only question i ask myself is: How big of a crowd will in december
>>>2002 be reached?
>>>
>>>Knowing how many TV crews are in Jerusalem, perhaps they reach some market
>>>there.
>>>
>>>>On October 22, 2002 at 11:53:26, Robert Hyatt wrote:
>>>>
>>>>>On October 21, 2002 at 18:24:44, martin fierz wrote:
>>>>>
>>>>>>On October 21, 2002 at 13:12:37, Robert Hyatt wrote:
>>>>>>
>>>>>>>On October 21, 2002 at 10:22:39, Vincent Diepeveen wrote:
>>>>>>>
>>>>>>>>On October 21, 2002 at 08:34:31, Fernando Villegas wrote:
>>>>>>>>
>>>>>>>>It is not valid that they created an awful machine. They didn't
>>>>>>>>play any computerchess world championship nor did they join any
>>>>>>>>other computer chess events where the european programs could measure
>>>>>>>>themselves with deep blue.
>>>>>>>>
>>>>>>>>After 1995 they quit facing european programs.
>>>>>>>>
>>>>>>>>All we know is a few horrible games from both deep blue and kasparov.
>>>>>>>>
>>>>>>>>It is not trivial that deep blue 1997 could show better play
>>>>>>>>than the poor level in these games.
>>>>>>>>
>>>>>>>>It is for sure that kasparov is the person to blame of course. he
>>>>>>>>was not only an idiot, he was also bad for chess.
>>>>>>>>
>>>>>>>>Where the 4-4 from kramnik is a sad reality, he will be able to possibly
>>>>>>>>face other programs again. Kasparov will play junior.
>>>>>>>
>>>>>>>So kasparov made _one_ big mistake in resigning a drawn position, and mixing up
>>>>>>>an opening (if that is really what happened) sequence of moves, and he is an
>>>>>>>idiot.
>>>>>>>
>>>>>>>Kramnik resigned a drawn game, and blew a couple of openings, and he is "ok"???
>>>>>>
>>>>>>not that i want to take sides in any debate involving deep blue, but:
>>>>>>kramnik resigned a drawn game, true, but it was very hard to spot. and i don't
>>>>>>know where he "blew openings" - not even one.
>>>>>
>>>>>I was talking about anti-computer more than anything else.  His first four
>>>>>openings were
>>>>>tame and nearly perfect for playing against a computer.  Then he got more
>>>>>aggressive and
>>>>>left his original plan, it seems...
>>>>
>>>>actually, to me it seemed as if the fritz team changed something in the openings
>>>>after 4 games, not kramnik. kramnik did NOT play anti-fritz chess in this match.
>>>>he played "kramnik as usual", which is quite good as anti-computer-chess goes,
>>>>of course. i had this impression already after the first 4 games, that kramnik
>>>>was not playing specific anti-computer lines he discovered at home (i really
>>>>hope we'll get some insight into his preparation some day - e.g. did he know
>>>>..Bf8?? was coming), but just his usual openings. i don't think kramnik changed
>>>>anything after game 4. personally, i think his super-blunder in game 5 changed
>>>>the match completely - kramnik probably makes such a blunder once in 100 or more
>>>>games at most, and fritz just got lucky. i think that kramnik then wanted to
>>>>"make up for it" with a brilliant game 6 and that backfired too (naturally...),
>>>>and then he was only a shadow of his usual self for games 7 and 8... at least it
>>>>should teach him a good lesson if he ever has another go at the computer!
>>>>i also think your are generally right that kramnik just lost his concentration a
>>>>bit after the first 4 games when everything was going so smoothly. that is just
>>>>very human :-)
>>>>
>>>>
>>>>>>>Somehow your "logic" totally escapes me...
>>>>>>>
>>>>>>>Fritz couldn't beat Kramnik in the match  even after he made at _least_ one
>>>>>>>trivial-to-spot
>>>>>>>blunder that turned a dead draw into a dead loss a piece down.
>>>>>>
>>>>>>kramnik made exactly ONE trivial-to-spot blunder, Qc4??. the position was not a
>>>>>>dead draw without that blunder. i think it's a draw, but if kramnik had been
>>>>>>100% sure that this was in fact a dead draw, he could have gone into this ending
>>>>>>by force - and he didnt, which tells us something about what kramnik thought
>>>>>>about this ending - that it was not *dead* drawn.
>>>>>
>>>>>My point was that in the Kasparov vs Deep Blue match, Kasparov resigned in a
>>>>>game
>>>>>that he thought was lost.  But which deep analysis showed was drawn.  Kramnik
>>>>>resigned
>>>>>a game that was probably drawn, although it has not been subjected to the same
>>>>>analysis as
>>>>>the DB/GK game.  But the similarity is there, he resigned _too early_.
>>>>
>>>>you can say that with the benefit of hindsight. lots of people wrote this here,
>>>>but it's just not true. if you are sitting at the board, and only see losses,
>>>>you resign. which means if you do not see the "miracle save" in that position,
>>>>you resign - there is no point in playing on if you do not see it, because every
>>>>other possible line will just lose. you cannot call resigning in that position
>>>>an "easy-to-spot-blunder", just as you cannot call kasparov's resignation an
>>>>"easy-to-spot-blunder". it was a wrong decision obviously, but not easy to
>>>>spot...
>>>>
>>>>>In the last game Kasparov made what most considered to be a blunder with h6.
>>>>>Kramnik
>>>>>_clearly_ blundered a piece and turned a probable draw into a sure loss with one
>>>>>move that
>>>>>takes most programs a few milliseconds to spot as losing.
>>>>>
>>>>>Yet Kasparov was routinely criticized as playing like a 2000 player, yet Kramnik
>>>>>has not
>>>>>gotten any such comments.
>>>>>
>>>>>Everything is the _same_ except for the opponents.  One was the hated Deep Blue
>>>>>from IBM,
>>>>>the other is a popular micro program...  :)
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>>
>>>>>>Nxf7 turned out to be a mistake too, but much more in the sense that you should
>>>>>>not play this way against computers, and specially not when you are leading with
>>>>>>+1. in a chess sense, it is very far from "trivial to spot"... as it turns out,
>>>>>>Nxf7 was a ?? for kramnik for the rest of the match - he couldnt recover after
>>>>>>game 6...
>>>>>
>>>>>I concsider that to be a strategic blunder, if not a tactical blunder.  But that
>>>>>only
>>>>>highlights the issue here.  Kramnik actually played much worse overall than
>>>>>Kasparov
>>>>>did, but was still able to draw the match.  I think the first four games were
>>>>>more revealing
>>>>>to me, personally.  The last 4 games seemed to be twilight-zone stuff...
>>>>
>>>>i don't know how you can say that kramnik played "much worse". i see that both
>>>>players made one big chess blunder (...h6 for kasparov in game 6, ...Qc4 for
>>>>kramnik in game 5), both resigned a drawn position where it was extremely hard
>>>>to spot that it was a draw, and apart from that i see no real blunders.
>>>>obviously, Nxf7 by kramnik is a very questionable decision, but then you can say
>>>>kramnik made 3 mistakes in 8 games and kasparov 2 in 6, which is very close to
>>>>each other.
>>>>besides, kramnik played excellent anti-computer-chess in games 1-4, and you
>>>>cannot say that kasparov played a single excellent anti-computer game in the
>>>>whole match.
>>>>on the other hand, DB played a great game (the one where kasparov resigned where
>>>>he had a perpetual), outplaying kasparov completely, something which DF never
>>>>did to kramnik in the whole match.
>>>>
>>>>aloha
>>>>  martin
>>>>
>>>>
>>>>>>judging from the games, DF certainly didn't seem to be "much better than DB",
>>>>>>which at least didnt produce such ridiculous moves as DF did :-)
>>>>>>which is not to say that DB would not have been capable of playing such moves
>>>>>>too...
>>>>>
>>>>>I certainly agree.  Kramnik made more bad moves than Kasparov, yet he didn't
>>>>>lose
>>>>>the match. That says something...
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>>
>>>>>>aloha
>>>>>>  martin
>>>>>>
>>>>>>
>>>>>>
>>>>>>>While Deep Blue
>>>>>>>_did_
>>>>>>>beat Kasparov in a match where both made mistakes.  And Fritz is much better
>>>>>>>than
>>>>>>>deep blue?
>>>>>>>
>>>>>>>Somehow, again, your "logic" totally escapes me...
>>>>>>>
>>>>>>>In fact, your "logic" is really just a form of envy/agenda, IMHO...
>>>>>>>
>>>>>>>
>>>>>>>>
>>>>>>>>Perhaps kasparov has LEARNED a bit more than kramnik has.
>>>>>>>>
>>>>>>>>If you lose once from the thing, then only when you are world champion
>>>>>>>>you can play it again. But for sure is that fritz exists as software
>>>>>>>>and you can buy it and play it, and it joins tournaments too usual.
>>>>>>>>
>>>>>>>>that's not the case with deep blue.
>>>>>>>>
>>>>>>>>We just know it sucked ass, based upon its play. Kasparov sucked even
>>>>>>>>more of course, but he always got away with poor chess against programs.
>>>>>>>>
>>>>>>>>In fact it is realistic that he didn't care for getting 2.5-2.5, just
>>>>>>>>game 6 IMHO he was imagining deep blue to be so bad, like 1980 software,
>>>>>>>>that he thought he coudl get back to a draw or something, after playing
>>>>>>>>horrible blunders like b5.
>>>>>>>>
>>>>>>>>We do not know. All we know is that humans when playing computers do not
>>>>>>>>show very good play. Look to kramnik. he plays the first 4 games like
>>>>>>>>he plays rapid games. He gets 3-1 then (lucky machine) and the rest of
>>>>>>>>the games he doesn't care simply.
>>>>>>>>
>>>>>>>>But still 4-4 is acceptible from historic viewpoint.
>>>>>>>>
>>>>>>>>What we do know is that kasparov has on average played 1-1 against
>>>>>>>>deep blue.
>>>>>>>>
>>>>>>>>First match easy win 4-2, second match by some poor games a loss 3.5-2.5
>>>>>>>>
>>>>>>>>Then IBM stopped. Wise decision. AFter so much  marketing that deep blue
>>>>>>>>has solved chess even, they had to quit of course. Or they would look
>>>>>>>>stupid in 1998.
>>>>>>>>
>>>>>>>>>Bob
>>>>>>>>>Feng DO mention problems with the program by Thomas. If they were enough to
>>>>>>>>>speak of "full of.." or not it is a matter of tastes in the writting style. My
>>>>>>>>>impression was and still is that the author did have many problems and even so
>>>>>>>>>they created an awful machine. Of course this does not means the software
>>>>>>>>>problem were more or worst than the hardware problems.
>>>>>>>>>Anyway the core of my mressions is the first: DP could have been absolutely
>>>>>>>>>untouchable if worked one year more.
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>My best
>>>>>>>>>Fernando



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.