A farce
Andreas Mader
mic wrote
>This the internet, you know. Anything you say, yell, or even think
>while you're typing it can be used against you. Please tell us about
>the *couple of details* which made you *change your mind*. If you
>don't, you'll have to withdraw your accusation *this match was a
>farce*.
>
>mic
>
>>Best wishes
>>Andreas Mader
>
>>"You can fool some people all the time and you can fool all people
>>sometimes, but you cannot fool all people all the time."
>
>Is this to be considered a warning against Gari or against IBM lawyers
>which haunt you for slander?
Maybe you are joking...
This is the world of free speech, you know?
I wrote that this is my *very personal opinion*, and not, that it is a
fact. I do not have to take back personal opinions. Maybe you want to
have it handled in another way, but I don't.
Best wishes
Andreas
Andreas Mader
I was on holiday for a few days and when I returned to my office today I
found a couple of private posts asking why I changed my mind about the
DB-GK match. The main reason was a talk with a journalist (who is a
strong chess player, too). He returned to Vienna and gave us many
insights. I do not want to talk about all that I've heard since I have
no evidence. But these things are proven:
1. VERY strange moves in some of the games made by DB that noone can
reproduce with a PC program.
2. VERY strange moves and mistakes made by GK.
3. Noone can prove that all the "computer" moves were really made by DB.
4. Refusal of the DB team to show printouts.
5. DB is (was) unable to play the first move with white. The move had to
be entered manually by the DB team. This was standard approx. 20 years
ago.
6. IBMs stock index went up by approx. 4 percent after the match.
7. Strange behaviour of all participants BEFORE the last game.
8. VERY strange behavior of GK (refuses to explain why he made such
strange opening moves in Game 6) and the DB team (they showed absolutely
no pleasure) during the closing ceremony.
9. DB will not prove its strength in a GM tournament.
10. Rumors, rumors, rumors... (not proven, of course).
Maybe this is a miracle like the WM tournament Lasker vs Schlechter,
where Schlechter only needed a draw in the last game, but he lost in a
won position. Either that or it is a farce!
Best wishes
Andreas
Robert Hyatt
Andreas Mader (ma...@p6.gud.siemens.co.at) wrote:
: I was on holiday for a few days and when I returned to my office today I
: found a couple of private posts asking why I changed my mind about the
: DB-GK match. The main reason was a talk with a journalist (who is a
: strong chess player, too). He returned to Vienna and gave us many
: insights. I do not want to talk about all that I've heard since I have
: no evidence. But these things are proven:
: 1. VERY strange moves in some of the games made by DB that noone can
: reproduce with a PC program.
That's only because DB is much stronger than a PC. Most programs have
found that Be4 (compared to Qb6) gets better and better as they search
deeper. There's no mystery here... just extreme depth and an evaluation
that favors the position after Be4 more...
: 2. VERY strange moves and mistakes made by GK.
Someone reported other such strange moves in other world championship
matches. I rely on the "meat makes mistakes" proverb... It seems to
hold pretty well.
: 3. Noone can prove that all the "computer" moves were really made by DB.
I don't see what this has to do with anything. How can *anyone* "prove"
that a program made a move, after the fact? The log can be doctored. The
program can be doctored. You can't prove such things, ever. Of course,
the big question is *who* would be qualified to help DB? The player that
has been beating Kasparov regularly over the past few years? Who might
that be?
: 4. Refusal of the DB team to show printouts.
the showed printouts to the "arbiter". this was documented. they didn't
want to reveal to Kasparov what they could "see". Kasparov doesn't reveal
what he was thinking during the game. I think their position was perfectly
balanced against this.
: 5. DB is (was) unable to play the first move with white. The move had to
: be entered manually by the DB team. This was standard approx. 20 years
: ago.
this is simply wrong. I've watched this program's predecessors play
chess. The human *can* select a first move, just like you can with Crafty,
by tweaking with books.bin. But it will play games without any human
interaction. "scratchy" did this on ICC a couple of years ago, compiling
the best win/loss record *ever* seen on a chess server.
: 6. IBMs stock index went up by approx. 4 percent after the match.
Perhaps this is the reason they chose to hire Hsu, Campbell, etc, and
then dump money into developing the chess hardware they are using? IE
nothing but "a return on investment?" Their stock would have gone up
had they lost the match, most likely. The P/R was all good for them.
: 7. Strange behaviour of all participants BEFORE the last game.
: 8. VERY strange behavior of GK (refuses to explain why he made such
: strange opening moves in Game 6) and the DB team (they showed absolutely
: no pleasure) during the closing ceremony.
Probably shock. Probably hard to digest that an era had "ended"...
: 9. DB will not prove its strength in a GM tournament.
: 10. Rumors, rumors, rumors... (not proven, of course).
the worst part of the entire cycle, in my opinion. The rumor rate
here has been higher than the "three-headed baby marries half-human
half alligator from Amazon" stories in the tabloids at the supermarket
checkout stands..
: Maybe this is a miracle like the WM tournament Lasker vs Schlechter,
Chris Yee
On Wed, 21 May 1997, Andreas Mader wrote:
> 1. VERY strange moves in some of the games made by DB that noone can
> reproduce with a PC program.
How does this contribute to your argument? If noone can reproduce it,
then anyone can, right? Who is he, anyway?
> 2. VERY strange moves and mistakes made by GK.
OK, I see your point here.
> 3. Noone can prove that all the "computer" moves were really made by DB.
Again, who is Noone, and how does his ability to prove this help you?
> 4. Refusal of the DB team to show printouts.
So?
> 6. IBMs stock index went up by approx. 4 percent after the match.
This is a result, not a cause!! Anyways, it's called...listen
carefully...."stock split". June 1.
Chris Yee
On Wed, 21 May 1997, Andreas Mader wrote:
>
> This is the world of free speech, you know?
And which world is that?
mclane
Andreas Mader <ma...@p6.gud.siemens.co.at> wrote:
>1. VERY strange moves in some of the games made by DB that noone can
>reproduce with a PC program.
and Kasparov thinks the moves were played in style of Karpov.
>2. VERY strange moves and mistakes made by GK.
right.
>3. Noone can prove that all the "computer" moves were really made by DB.
No evidence, rolf and Bob would say. No evidence.
>4. Refusal of the DB team to show printouts.
Right. Thats not very scientific.
>5. DB is (was) unable to play the first move with white. The move had to
>be entered manually by the DB team. This was standard approx. 20 years
>ago.
???? What . It is the first time i hear this. It cannot move with
white? Why ?
>6. IBMs stock index went up by approx. 4 percent after the match.
they earn much with this match. They earned much with the last year's
match .
>7. Strange behaviour of all participants BEFORE the last game.
Aha!
>8. VERY strange behavior of GK (refuses to explain why he made such
>strange opening moves in Game 6) and the DB team (they showed absolutely
>no pleasure) during the closing ceremony.
Maybe because the game was obviously tooooooo unbelievable.
>9. DB will not prove its strength in a GM tournament.
Aha.
>10. Rumors, rumors, rumors... (not proven, of course).
I am pleased that you have now the same opinon I have.
And you have come to the same conclusion although your source of data
was different, not I have convinced you anyway, and altough you are a
totally different person like I am.
Brilliant. We will see others change their mind too.
Thanks for your open views.
mclane
Andreas Mader <ma...@p6.gud.siemens.co.at> wrote:
>>Can you further tell me what you have heard ?
>Not all the rumors. I have no evidence. But some seem to be even worse
>than what you've stated.
Brilliant !! My wildest imagination cannot construct a conspiracy
theory that would be as crazy as the real match was.
>I'm very curios what CSS will write in the next issue....
Of course this magazin would write the earth is a disc if it would be
an advantage for Friedel and Steinwender.
>Because your acting in this way all the time. Certain names (e.g.
>Friedel) and just like with Pawlovs dog you are acting the same way
>every time (conspiration...).
Who was the computer-chess-advisor in the match we are talking about ?
Tell me his name ? Aha.
You see, it works. Wherever he is, conspiracy is not far away.
>In most cases I can foresee your
>reactions. You are acting in a very hard way against the Swedish rating
>list, but when it proves your theories you quote it as if you were
>taking it for granted.
Of course!! The ssdf is an instrument. Why should I allow my opponents
to MIS-/USE this instrument, and I should not use/misuse it?
>You are acting in a even harder way against CSS,
Friedel is a member of this magazine too. Isn't he ?
>but when they publish something you like (e.g. a test where CST gets
>2800 points :-)) everybody should have an abonnement. And so on and so
>on....
No - I would also write a good article about Rebel8 in computer-schach
& Spiele. My goal is: if something is good, the world should be
informed about. Of course I would not further work with them. But e.g.
in the Rebel8 case I though: CSS should write about this program.
Computer Schach and Spiele is not my favourite magazine. But since we
have no better one, we have to comment on it.
Easy reason. I am against what they write. But because it is the only
computer-chess magazine in germany, I have to use it to argue about.
My point with the YAZGAC-test was, that AGAIN a Dipl.Ing. or
mathematician has developed a mad-test !! And has decsibed his formula
on 3 pages just to find out in the end: that the formula is not
working.
We had this before. And it makes me laughing. I am not a fan of them
because they write something positive about CSTal.
In fact, a few days ago Dieter Steinwender asked me for data about
Chess System Tal and I was very amused, because he asked me:
Why didn't we get any information about this program till now ?
And I thought: because I don't wanted you to get the information !!
Some things are easy.
>Thats why I have my own reflex when I read your postings....
Yes. Reflex. Maybe pawlovs dogs when you hear my alias ?!
>I didn't say they have fooled us. It was just a quotation. :-)
Sorry. You are absolutely right. It was just a quotation.
>I personally don't care if GK is a nice guy or not and if IBM makes
>another billion dollars. What really disturbes me is that they use chess
>as a vehicle and discredit it.
They use chess / computerchess as a vehicle. A few paragraphs you
named Friedel. Exactly.
Uses computerchess as a vehicle !!!!
I always tried to use Steinwender/Friedel as a vehicle to get
information about computerchess. But since I have found out that they
use computerchess as a vehicle for making fame/money and idle stuff, I
cannot use them anymore.
Although they swim in the same water, they have totally different
goals.
BTW: I have told Friedel this in person (Aegon97) and he was
disapointed about this. He said: What ? You really think I am shallow
and my goals are these ?
>Sorry, I do not want to tell you (or someone else) all the rumors at
>this time. Nothing personal...
No problem. I understand.
Thanks at all.
>Best wishes
>Andreas
Wayne Howard
hy...@crafty.cis.uab.edu (Robert Hyatt) wrote:
>: 3. Noone can prove that all the "computer" moves were really made by DB.
>
>I don't see what this has to do with anything. How can *anyone* "prove"
>that a program made a move, after the fact? The log can be doctored. The
>program can be doctored. You can't prove such things, ever. Of course,
>the big question is *who* would be qualified to help DB? The player that
>has been beating Kasparov regularly over the past few years? Who might
>that be?
Hi, Bob!
This is precisely the problem I have with the current "human
intervention" theories. Who, exactly, do the theorists suspect of
helping DB to find the moves that knocked off Kasparov? Karpov? Anand?
Benjamin???
Isadore Gunsberg was inside Mephisto. Who do they think is inside Deep
Blue?
I would very respectfully ask for some names here. :)
best,
Wayne Howard
ps. Fischer??????? :)
Don Fong
In article <5lur23$h...@juniper.cis.uab.edu>,
Robert Hyatt <hy...@crafty.cis.uab.edu> wrote:
>Of course, the big question is *who* would be qualified to help DB?
>The player that has been beating Kasparov regularly over the past few
>years? Who might that be?
ah, but the point is that you don't have to be stronger than DB
or than Kasparov, to help. Ingo Althofer is constantly telling us
how a 3-hirn of human + 2 computers can play stronger than any of the
three alone. Bob, you yourself have remarked how human operators "cheat"
on the servers by helping their programs from time to time. even if
the human operator is not stronger than the computer or than the opponent,
the operator can still *compensate for weaknesses* in the computer's game.
isn't it obvious? consultation helps. that's why it's forbidden.
Paul F. Dietz
On Wed, 21 May 1997 18:39:33 GMT, who...@iconn.net (Wayne Howard)
wrote:
>This is precisely the problem I have with the current "human
>intervention" theories. Who, exactly, do the theorists suspect of
>helping DB to find the moves that knocked off Kasparov? Karpov? Anand?
>Benjamin???
As the online magazine Slate pointed out, the accusation is
akin to accusing a Formula 1 driver of peddling.
Paul
Andreas Mader
Thorsten Czub alias mclane wrote:
>WHAAAAAATTTTTTTT ????
>
>Andreas, you are killing me the second time, no the third time.
>First time was when you gave up Modul/PC-Schach,
>2nd time when you thought I am Rolf Tueschen,
>and now, the 3rd time:
>You attack me because I thought about an arrangement (nothing to do
>with conspricary, just that they said: Mr. Kasparov, if you play
>weaker than 6:0, we would give you, in an exchange to substitute your
>lack of reputation WHEN you lose one game, a million-dollar for each
>lost point weaker than 6.). I have not said that MOVES were arranged.
>Only that if he plays weaker for some reasons and loses a game, he
>would get extra pocket money...
>
>So you said: Mclane, you and your conspiracy theories....
Right.
>And now you tell me that you think the match was a farce !
Right.
As I've said before: I've changed my mind. This happens not often, but
it happens sometimes. Do you stick to your own opinions no matter what
happens?
>Can you further tell me what you have heard ?
Not all the rumors. I have no evidence. But some seem to be even worse
than what you've stated.
>We have many heavy rumors here in germany (why here ? Cause Friedel
>lives in germany!) about german Grandmaster commenting on the games.
>Also the german press (chess-magazines) will not be very NICE to
>Kasparov and his team. The comments are without any doubts.
I'm very curios what CSS will write in the next issue....
>So I would be pleased about your NEW feelings. If you don't want to
>publish them in public, please send them to me in an email.
>I mean - why do you attack me for saying: this match is a PR-gag (I
>have said this since I saw game 2 ) and now you change your mind. I
>mean, I have nothing that you change your mind. But you have hurt me
>by always pressing the CONSPIRATION button. Before I post in public
>about the games I speak with many chess-players. They have good ELO's
>and we study the games together with Comuters to find out WHY what
>happens. Why do you think it was just a crazy idea, like a reflex.
Because your acting in this way all the time. Certain names (e.g.
Friedel) and just like with Pawlovs dog you are acting the same way
every time (conspiration...). In most cases I can foresee your
reactions. You are acting in a very hard way against the Swedish rating
list, but when it proves your theories you quote it as if you were
taking it for granted. You are acting in a even harder way against CSS,
but when they publish something you like (e.g. a test where CST gets
2800 points :-)) everybody should have an abonnement. And so on and so
on....
Thats why I have my own reflex when I read your postings....
>Best wishes
>Andreas Mader
>"You can fool some people all the time and you can fool all people
>sometimes, but you cannot fool all people all the time."
>So - they have fooled us for money !
I didn't say they have fooled us. It was just a quotation. :-)
>And all this noise was only to
>make money. IBM got so much money from last years event, they
>extrapolated how much they could get with this match. And - as you
>have seen in the massive public-reaction, television and articles in
>Spiegel/Focus, their was a massive campaign by Friedel and Co.
>
>IBM will make mutch money indirectly out of this match. How many
>million $ do you think ?
>And Kasparov ? His PCA match human-human did not get the high amount
>of $ he thought his organisation would get.
>So this match human-computer could help him making money.
I personally don't care if GK is a nice guy or not and if IBM makes
another billion dollars. What really disturbes me is that they use chess
as a vehicle and discredit it.
>I am interested in further dialogue. What have you heard. Or why
>changed your opinion. Thanks Andreas.
Sorry, I do not want to tell you (or someone else) all the rumors at
this time. Nothing personal...
Best wishes
Andreas
Don Fong
In article <5m06d0$2...@juniper.cis.uab.edu>,
Robert Hyatt <hy...@crafty.cis.uab.edu> wrote:
>da...@taic.net wrote:
>: All that's necessary, as even Bob will admit after some sober
>: reflection, is someone who is strong enough to change a *few* of the
>: intended moves. Even an International Master would suffice for this job.
>
>I'm not sure an IM would be qualified to overrule DB. I don't think there
>is an IM that could even dream about beating it.
cmon Bob. the point is that the IM could be strong enough to
*help* DB even if it could never *beat* DB. i know you understand
this, because you are constantly complaining about operators who
"cheat" on ICC by overruling the program in key positions. you don't
really think all these operators are really stronger than the
programs they are overruling, do you?
hmmm, more to the point. i seem to remember you yourself saying
how CRAFTY would gain quite a few rating points if you were allowed to
overrule it occasionally. are you stronger than CRAFTY?
consultation can allow a team to play stronger than any of
its members.
Andreas Mader
Robert Hyatt wrote:
>
> <snip>
I am pleased that your vacation has ended, even if you cannot share my
opinion!
Best wishes
Andreas
Andreas Mader
Robert Hyatt wrote:
>
> da...@taic.net wrote:
> : Robert Hyatt <hy...@crafty.cis.uab.edu> wrote:
>
> : : : 2. VERY strange moves and mistakes made by GK.
>
> : : Someone reported other such strange moves in other world championship
> : : matches. I rely on the "meat makes mistakes" proverb... It seems to
> : : hold pretty well.
>
> : that are both transparently bad and turn out to be bad? (...b5 is
> : obviously answere by a4!)
>
> He might have blundered. Or he might have thought that he could win a
> piece up with an exposed king. I'd hope he didn't try that against
> Fritz, and win, and then expect the same to work against DB.. There's
> a monstrous difference in the two programs. At least 400 Elo, probably
> more...
>
> : : : 4. Refusal of the DB team to show printouts.
>
> : : the showed printouts to the "arbiter". this was documented.
>
> : I'd have felt a lot more confident if they had been shown to you.
> : Who were they shown to? A savvy computer program user, even?
>
> Ken Thompson. Who I'd think even more qualified than I am to analyze
> and inspect... Also Mike Valvo was on the committee I believe, although
> I am not sure who saw what, when...
Are there any comments made by Ken Thompson concerning the printouts? Or
is he forced to remain silent?
Andreas Mader
Jouni Uski wrote:
>
> > 4. Refusal of the DB team to show printouts.
>
>
> Jouni
I am happy to hear that! Why aren't they published on the internet so
that everybody can study them?
Andreas Mader
Chris Yee wrote:
>
> On Wed, 21 May 1997, Andreas Mader wrote:
>
> > 1. VERY strange moves in some of the games made by DB that noone can
> > reproduce with a PC program.
>
> then anyone can, right? Who is he, anyway?
>
>
>
>
>
>
>
> > 6. IBMs stock index went up by approx. 4 percent after the match.
>
> carefully...."stock split". June 1.
I only have an opinion. No need to be rude.
Robert Hyatt
Robert Hyatt
Don Fong (df...@cse.ucsc.edu) wrote:
: In article <5lur23$h...@juniper.cis.uab.edu>,
: Robert Hyatt <hy...@crafty.cis.uab.edu> wrote:
: >Of course, the big question is *who* would be qualified to help DB?
: >The player that has been beating Kasparov regularly over the past few
: >years? Who might that be?
: ah, but the point is that you don't have to be stronger than DB
: or than Kasparov, to help. Ingo Althofer is constantly telling us
: how a 3-hirn of human + 2 computers can play stronger than any of the
: three alone. Bob, you yourself have remarked how human operators "cheat"
: on the servers by helping their programs from time to time. even if
: the human operator is not stronger than the computer or than the opponent,
: the operator can still *compensate for weaknesses* in the computer's game.
: isn't it obvious? consultation helps. that's why it's forbidden.
Consultation helps in computer vs computer vs human, quite a bit,
because the human can offer the computer some advice at key times,
such as when to move, when to think longer, and maybe even force a
different move when you see one of those ugly Kh1 types of moves.
But against Kasparov? In a game where time was never much of an
issue at all for the computer. I can't think of anyone that would
want to overrule DB's choice of a move in a game against Kasparov.
I certainly wouldn't. I might want to prod it and say "look deeper
here" but that would be about it.
I simply don't buy the human intervention theory, sorry. I've seen
lots of programs make moves no human would consider. On occasion,
they are even brilliant. On others, they are rediculous. I think
it's interesting that suddenly DB makes marvelous moves and then
the human-helping theory arises to account for those moves that are
"just too good." The explanation might be a little simpler than
that. DB might be that good...
Robert Hyatt
da...@taic.net wrote:
: Wayne Howard <who...@iconn.net> wrote:
: : This is precisely the problem I have with the current "human
: : intervention" theories. Who, exactly, do the theorists suspect of
: : helping DB to find the moves that knocked off Kasparov? Karpov? Anand?
: : Benjamin???
: There is a common mistake both you and Bob Hyatt are making here.
: The mistake is that you are assuming that such a person would have to have
: been making all, or most, of the moves that were "played".
: That's false.
: All that's necessary, as even Bob will admit after some sober
: reflection, is someone who is strong enough to change a *few* of the
: intended moves. Even an International Master would suffice for this job.
I'm not sure an IM would be qualified to overrule DB. I don't think there
is an IM that could even dream about beating it. In fact, I don't think there
are very many GM's that can beat the thing, period.
However, I prefer "innocent until proven guilty" and not the inverse. The
biggee move has been Be4 rather then Qb6. We now know that most programs
think the two moves are *very* close. Minor scoring changes could make any
current program play Be4. A deeper search might also do this for some. It's
likely that Be4 is easily explained by their eval and depth, rather than by
human intervention. Ditto for most any other move.
If someone would be good enough to post the PGN for a game of your choice,
with a move that seems to be "too good" for a computer to play, I'll bet
we can get a bunch of us analyzing it with computers and prove that some
program will play that move given enough time... or else prove that the
move played is close to other moves, close enough that eval differences
rather than human intervention made the difference...
just an opinion, of course. No way to *prove* anything...
Don Fong
In article <5m0601$2...@juniper.cis.uab.edu>,
Robert Hyatt <hy...@crafty.cis.uab.edu> wrote:
>Consultation helps in computer vs computer vs human, quite a bit,
>because the human can offer the computer some advice at key times,
>such as when to move, when to think longer, and maybe even force a
>different move when you see one of those ugly Kh1 types of moves.
the same principle applies in human-computer games.
>But against Kasparov? In a game where time was never much of an
>issue at all for the computer. I can't think of anyone that would
>want to overrule DB's choice of a move in a game against Kasparov.
why not? i bet a lot of GM's would want to override Deeper
Blue if it were playing moves like Deep Blue last year! of course
you can say "but this year it didn't make those blunders". but then
of course then the conspiracy theorists can say "that is mighty
suspicious". understand?
>I certainly wouldn't. I might want to prod it and say "look deeper
>here" but that would be about it.
argh !!! but Bob, that would be still be cheating !!! and you
are implicitly admitting that (even) you are strong enough to help
the computer! even though you are certainly no GM. the "human touch"
judiciously applied can give the machine an extra edge.
>I simply don't buy the human intervention theory, sorry.
i don't buy it either. but your arguments against it
just plain don't hold water.
Ingo Althoefer
For me it is a big surprise that a cool-headed man like Andreas Mader
begins to think in mysteries and to believe in rumors.
Andreas Mader wrote:
>I was on holiday for a few days and when I returned to my office today I
>found a couple of private posts asking why I changed my mind about the
>DB-GK match. The main reason was a talk with a journalist (who is a
>strong chess player, too). He returned to Vienna and gave us many
>insights. I do not want to talk about all that I've heard since I have
>no evidence. But these things are proven:
>1. VERY strange moves in some of the games made by DB that noone can
> reproduce with a PC program.
Which moves do you mean, concretely ?
At least the two situations in game 2 ( 37.Be4 instead of Qb6, 45.Kf1 )
have been explained in rgcc by different people ( Ed Schroder, Matthias Wuel-
lenweber, myself, ... ) in satisfactory ways.
>2. VERY strange moves and mistakes made by GK.
GK simply tried to play anti-computer chess.
>3. Noone can prove that all the "computer" moves were really made by DB.
>4. Refusal of the DB team to show printouts.
C.J. Tan announced that "appropriate" parts of the printouts will be given
to the public in the next few months.
>5. DB is (was) unable to play the first move with white. The move had to
> be entered manually by the DB team. This was standard approx. 20 years
> ago.
This is new to me. Did this really happen ?
>6. IBMs stock index went up by approx. 4 percent after the match.
>7. Strange behaviour of all participants BEFORE the last game.
>8. VERY strange behavior of GK (refuses to explain why he made such
> strange opening moves in Game 6) and the DB team (they showed absolutely
> no pleasure) during the closing ceremony.
>9. DB will not prove its strength in a GM tournament.
This has not been decided yet.
> Maybe this is a miracle like the WM tournament Lasker vs Schlechter,
>where Schlechter only needed a draw in the last game, but he lost in a
>won position. Either that or it is a farce! Best wishes Andreas
Sorry, but Schlechters behaviour is not a farce. The match conditions were
that he had to win by a margin of two points to "win" the match. So he
needed a second win in the last game. If the final result would have been
5.5 : 4.5 to his favour he would not have become world champion !
Ingo Althoefer.
da...@taic.net
Robert Hyatt <hy...@crafty.cis.uab.edu> wrote:
: But against Kasparov? In a game where time was never much of an
: issue at all for the computer. I can't think of anyone that would
: want to overrule DB's choice of a move in a game against Kasparov.
: I certainly wouldn't. I might want to prod it and say "look deeper
: here" but that would be about it.
Oh, come now. Don't engage in your ever-more-frequent
disingenuity. I can think of plenty of positions in which I can probably
know better than DB what the best move.
Or, at least, if an intended move *should not be played at all
costs*. That kind of assistance can raise even a DB by a number of rating
points. I'm sure you can think of some positions also.
Andreas Mader
Ingo Althoefer wrote:
>
> For me it is a big surprise that a cool-headed man like Andreas Mader
> begins to think in mysteries and to believe in rumors.
>
> Andreas Mader wrote:
> >I was on holiday for a few days and when I returned to my office today I
> >found a couple of private posts asking why I changed my mind about the
> >DB-GK match. The main reason was a talk with a journalist (who is a
> >strong chess player, too). He returned to Vienna and gave us many
> >insights. I do not want to talk about all that I've heard since I have
> >no evidence. But these things are proven:
>
> >1. VERY strange moves in some of the games made by DB that noone can
> > reproduce with a PC program.
>
> Which moves do you mean, concretely ?
> At least the two situations in game 2 ( 37.Be4 instead of Qb6, 45.Kf1 )
> have been explained in rgcc by different people ( Ed Schroder, Matthias Wuel-
> lenweber, myself, ... ) in satisfactory ways.
>
> >2. VERY strange moves and mistakes made by GK.
>
> GK simply tried to play anti-computer chess.
>
> >3. Noone can prove that all the "computer" moves were really made by DB.
Maybe there was a kind of Dreihirn behind DB...?
> >4. Refusal of the DB team to show printouts.
>
> C.J. Tan announced that "appropriate" parts of the printouts will be given
> to the public in the next few months.
>
> >5. DB is (was) unable to play the first move with white. The move had to
> > be entered manually by the DB team. This was standard approx. 20 years
> > ago.
>
> This is new to me. Did this really happen ?
>
I was told so by Michael Ehn who watched the last 4 games "live" and
gave a detailed report of the event, how he saw it and a lot of
background information. As far as I can see it there are many people who
believe that there was something wrong with the match, but most of them
remain silent.
> >6. IBMs stock index went up by approx. 4 percent after the match.
> >7. Strange behaviour of all participants BEFORE the last game.
> >8. VERY strange behavior of GK (refuses to explain why he made such
> > strange opening moves in Game 6) and the DB team (they showed absolutely
> > no pleasure) during the closing ceremony.
> >9. DB will not prove its strength in a GM tournament.
>
> This has not been decided yet.
>
> > Maybe this is a miracle like the WM tournament Lasker vs Schlechter,
> >where Schlechter only needed a draw in the last game, but he lost in a
> >won position. Either that or it is a farce! Best wishes Andreas
>
> Sorry, but Schlechters behaviour is not a farce. The match conditions were
> that he had to win by a margin of two points to "win" the match. So he
> needed a second win in the last game. If the final result would have been
> 5.5 : 4.5 to his favour he would not have become world champion !
This is new to me.
>
> Ingo Althoefer.
If there had been "only" strange moves by GK or "only" a few moves by DB
that no commercial program can reproduce I would not have been
suspicious at all. Its the coincidence of so many facts (why does it
take a few MONTHS to publish printouts?). To every fact you can find an
explanation, this is true. I changed my mind after I listened to the
report of Michael Ehn. Now he has published an article about the GK-DB
match in an Austrian newspaper and he, too, wrote nothing about a
possible manipulation... ("Weil nicht sein kann, was nicht sein darf!")
Best wishes
Andreas Mader
da...@taic.net
Robert Hyatt <hy...@crafty.cis.uab.edu> wrote:
: : 2. VERY strange moves and mistakes made by GK.
: Someone reported other such strange moves in other world championship
: matches. I rely on the "meat makes mistakes" proverb... It seems to
: hold pretty well.
Have very many world champions prepared a known bad line with TN's
that are both transparently bad and turn out to be bad? (...b5 is
obviously answere by a4!)
: : 4. Refusal of the DB team to show printouts.
da...@taic.net
Robert Hyatt
Don Fong (df...@cse.ucsc.edu) wrote:
: In article <5m06d0$2...@juniper.cis.uab.edu>,
: Robert Hyatt <hy...@crafty.cis.uab.edu> wrote:
: >da...@taic.net wrote:
: >: All that's necessary, as even Bob will admit after some sober
: >: reflection, is someone who is strong enough to change a *few* of the
: >: intended moves. Even an International Master would suffice for this job.
: >I'm not sure an IM would be qualified to overrule DB. I don't think there
: >is an IM that could even dream about beating it.
: cmon Bob. the point is that the IM could be strong enough to
: *help* DB even if it could never *beat* DB. i know you understand
: this, because you are constantly complaining about operators who
: "cheat" on ICC by overruling the program in key positions. you don't
: really think all these operators are really stronger than the
: programs they are overruling, do you?
: hmmm, more to the point. i seem to remember you yourself saying
: how CRAFTY would gain quite a few rating points if you were allowed to
: overrule it occasionally. are you stronger than CRAFTY?
: consultation can allow a team to play stronger than any of
: its members.
I don't "complain" about it. I have mentioned that a human + computer
is a formidable opponent for a computer. Because the human can help with
time management, and can override the Kg1 type of move. However, against
Kasparov, the human would have to be *very* good to (a) overrule a computer
that is clearly a strong GM and (b) not make mistakes that Kasparov would
see and exploit.
I don't believe it is possible at this level, to attempt such a thing and
get away with it. I believe IBM knows this. I can pick out moves that a
computer won't play, just by analyzing a game with a computer program. We
can do the same for the DB games. you simply pick a game, and a move you
believe was human-influenced. We already *know* that Be4 was not human
influence...
Robert Hyatt
Don Fong (df...@cse.ucsc.edu) wrote:
: In article <5m0601$2...@juniper.cis.uab.edu>,
: Robert Hyatt <hy...@crafty.cis.uab.edu> wrote:
: >Consultation helps in computer vs computer vs human, quite a bit,
: >such as when to move, when to think longer, and maybe even force a
: >different move when you see one of those ugly Kh1 types of moves.
: the same principle applies in human-computer games.
: >But against Kasparov? In a game where time was never much of an
: >issue at all for the computer. I can't think of anyone that would
: >want to overrule DB's choice of a move in a game against Kasparov.
: why not? i bet a lot of GM's would want to override Deeper
: Blue if it were playing moves like Deep Blue last year! of course
: you can say "but this year it didn't make those blunders". but then
: of course then the conspiracy theorists can say "that is mighty
: suspicious". understand?
Not in the least. Last year's DB was *new*. This year's has had a
year of additional work. It *had* to be better.
: >I certainly wouldn't. I might want to prod it and say "look deeper
: >here" but that would be about it.
: argh !!! but Bob, that would be still be cheating !!! and you
: are implicitly admitting that (even) you are strong enough to help
: the computer! even though you are certainly no GM. the "human touch"
: judiciously applied can give the machine an extra edge.
I agree it would be cheating. But my point was, many are claiming that
some DB moves were not found by DB. Everyone jumped on Be4, claiming
Qb6 has to be played by a computer. Both Crafty and Ferret have found
that qb6 is a *dead* draw. And either might find Be4 is not, later.
: >I simply don't buy the human intervention theory, sorry.
: i don't buy it either. but your arguments against it
: just plain don't hold water.
nor do yours for it. Somehow you want IBM to prove something that can't
be proved. I say the people that claim cheating have to prove it *did*
happen, not the other way around... At least in the USA.
brucemo
Robert Hyatt wrote:
> I agree it would be cheating. But my point was, many are claiming that
> some DB moves were not found by DB. Everyone jumped on Be4, claiming
> Qb6 has to be played by a computer. Both Crafty and Ferret have found
> that qb6 is a *dead* draw. And either might find Be4 is not, later.
To clarify, mine got a draw score on ply 20 for Qb6. That doesn't mean
it is a dead draw, or even a non-dead draw, it just means that it got a
draw score.
If I had been trying to prove that that move leads to a draw, I would
have done more analysis, I certainly wouldn't have posted that conclusion
based upon one search from the root of that position.
The reason that I felt that it was useful and interesting to post the
results of that search is that DB had just the one search to base its
decision upon as well.
To a computer, I can conceive that Qb6 in that position might appear to
lead to a draw, which may cause it to choose something else that also has
a fairly low score, it MAY not be the case that DB found that Be4 was a
crush.
bruce
chrisw
--
http://www.demon.co.uk/oxford-soft
Robert Hyatt <hy...@crafty.cis.uab.edu> wrote in article
<5m1ej3$i...@juniper.cis.uab.edu>...
> Don Fong (df...@cse.ucsc.edu) wrote:
> : In article <5m0601$2...@juniper.cis.uab.edu>,
> : Robert Hyatt <hy...@crafty.cis.uab.edu> wrote:
> : >Consultation helps in computer vs computer vs human, quite a bit,
> : >because the human can offer the computer some advice at key times,
> : >such as when to move, when to think longer, and maybe even force a
> : >different move when you see one of those ugly Kh1 types of moves.
>
> : the same principle applies in human-computer games.
>
> : >But against Kasparov? In a game where time was never much of an
> : >issue at all for the computer. I can't think of anyone that would
> : >want to overrule DB's choice of a move in a game against Kasparov.
>
> : why not? i bet a lot of GM's would want to override Deeper
> : Blue if it were playing moves like Deep Blue last year! of course
> : you can say "but this year it didn't make those blunders". but then
> : of course then the conspiracy theorists can say "that is mighty
> : suspicious". understand?
>
> Not in the least. Last year's DB was *new*. This year's has had a
> year of additional work. It *had* to be better.
Never heard of verschlimmbesserung ?
Apologies for the spelling.
Chris Whittington
>
> : >I certainly wouldn't. I might want to prod it and say "look deeper
> : >here" but that would be about it.
>
> : argh !!! but Bob, that would be still be cheating !!! and you
> : are implicitly admitting that (even) you are strong enough to help
> : the computer! even though you are certainly no GM. the "human touch"
> : judiciously applied can give the machine an extra edge.
>
> I agree it would be cheating. But my point was, many are claiming that
> some DB moves were not found by DB. Everyone jumped on Be4, claiming
> Qb6 has to be played by a computer. Both Crafty and Ferret have found
> that qb6 is a *dead* draw. And either might find Be4 is not, later.
>
chrisw
--
http://www.demon.co.uk/oxford-soft
Robert Hyatt <hy...@crafty.cis.uab.edu> wrote in article
<5m1esd$i...@juniper.cis.uab.edu>...
> da...@taic.net wrote:
> : Robert Hyatt <hy...@crafty.cis.uab.edu> wrote:
>
> : : But against Kasparov? In a game where time was never much of an
> : : issue at all for the computer. I can't think of anyone that would
> : : want to overrule DB's choice of a move in a game against Kasparov.
> : : I certainly wouldn't. I might want to prod it and say "look deeper
> : : here" but that would be about it.
>
> : Oh, come now. Don't engage in your ever-more-frequent
> : disingenuity. I can think of plenty of positions in which I can
probably
> : know better than DB what the best move.
>
> : Or, at least, if an intended move *should not be played at all
> : costs*. That kind of assistance can raise even a DB by a number of
rating
> : points. I'm sure you can think of some positions also.
>
> I'd be willing to bet you can't find *one*, unless you are sporting a
> new (GM) title I don't know about. Are you qualified to look over the
> shoulder of a Dlugy, or a Ivanov, and suggest better moves to them?
> I doubt it, because *I* wouldn't. DB is right in that class of player...
>
Yeah. YEAH. YEEEAAAAH.
g5, G5, GGGGGG55555555
Chris Whittington
>
Cameron Hayne
In article <338459...@p6.gud.siemens.co.at>, Andreas Mader
<ma...@p6.gud.siemens.co.at> wrote:
>Jouni Uski wrote:
>> > 4. Refusal of the DB team to show printouts.
>> They have sent them (74 pages) to Garry!
>> Jouni
>
>I am happy to hear that! Why aren't they published on the internet so
>that everybody can study them?
Maybe because IBM is a company - i.e. their purpose is to make money -
and therefore they can decide to do what is best for their bottom line.
(This is not a criticism, merely a statement of fact.)
There is no more reason why IBM should reveal details of Deep-Blue's
analysis than there is for any other commercial entity to reveal details
about their own products or software.
--
Cameron Hayne (ha...@crim.ca)
Centre de recherche informatique de Montreal
Ingo Althoefer
Andreas Mader wrote:
:Ingo Althoefer wrote:
:> Andreas Mader wrote:
:> >1. VERY strange moves in some of the games made by DB that noone can
:> > reproduce with a PC program.
:>
:> Which moves do you mean, concretely ?
:> At least the two situations in game 2 ( 37.Be4 instead of Qb6, 45.Kf1 )
:> have been explained in rgcc by different people (Ed Schroder, Matthias Wuel-
:> lenweber, myself, ... ) in satisfactory ways.
I want to repeat my question: Which moves of Deep Blue do you mean ?
Which moves are you unable to reproduce ?
:> >3. Noone can prove that all the "computer" moves were really made by DB.
:Maybe there was a kind of Dreihirn behind DB...?
I don't think so.
Of course, what I believe is that a man-machine combination including Deep
Blue would be stronger than Deep Blue alone:
Take Double-Deep-Blue & Boss ( that means: Deep Blue in a 2-best mode, and
a human who has the final choice between the proposals of Deep Blue ). DDB&B
should win long matches against "pure Deep Blue", if the Boss were not chosen
completely at random. For instance each of the following persons might be
appropriate candidates for the Boss position:
Garry Kasparov or Joel Benjamin or David Bronstein or Ingo Althoefer or
Murray Campbell or Frederic Friedel or Matthias Feist or Thomas Mally ...
( or perhaps even Andreas Mader after some training sessions :) ).
I make this conjecture only under the important assumption that between
the games both Deep Blue and Double-Deep-Blue & Boss would not be treated
or modified differently.
:If there had been "only" strange moves by GK or "only" a few moves by DB
:that no commercial program can reproduce I would not have been
:suspicious at all. Its the coincidence of so many facts (why does it
:take a few MONTHS to publish printouts?).
Again my question: Which DB moves do you think are not reproducable by at
least some of the top PC's ?
Best and second-best wishes, Ingo Althoefer :) .
PS: In the recent days I reread the articles of Frederic Friedel in Computer-
schach & Spiele ( April and June 96 issues ) on the 96-match between GK
and DB. And I was stunned: There are so many parallels in the behaviour
of Garry Kasparov in both matches. Especially, also during the 96-match
he had made accusations against IBM to produce DB moves by human
intervention. And he also was very nervous and under tensoin in the 96-
match, especially after game 4 ( many people, including me, had forgotten
this already ). Perhaps he would also have broken down in game 6 of the
96-match, if the DB team had accepted the draw offer in game 5 with the
consequence of a 2.5 : 2.5 score before the final showdown...
Of course, these are speculations only. But, taking Friedels 96 reports
with a few changes at the end ( concerning the loss in game 6 ) would
give a rather good description of what happened in NY weeks ago, at least
from my perspective of view.
mclane
alth...@pdec01.uucp (Ingo Althoefer) wrote:
>For me it is a big surprise that a cool-headed man like Andreas Mader
>begins to think in mysteries and to believe in rumors.
Yes - maybe I Andreas is now writing in MCLANE-mode !!
We should switch him back into cool-Mader-mode.
It seems also NORMAL people , REALISTS, who normally rely on facts and
evidence, come to the same ideas about the fight. Poor Bob.
mclane
Andreas Mader <ma...@p6.gud.siemens.co.at> wrote:
. Its the coincidence of so many facts (why does it
take a few MONTHS to publish printouts?). To every fact you can find
an
explanation, this is true. I changed my mind after I listened to the
report of Michael Ehn. Now he has published an article about the GK-DB
match in an Austrian newspaper and he, too, wrote nothing about a
possible manipulation... ("Weil nicht sein kann, was nicht sein
darf!")
mclane:
right. And it has nothing to do with conspiracy.
I am pleased that you subscibe this with your own change of mind.
Weil es im Himmel eben keine Steine geben darf!
Jouni Uski
brucemo
Andreas Mader wrote:
> 6. IBMs stock index went up by approx. 4 percent after the match.
I don't know what conclusion you expect to draw from this, but in
any case this is ridiculous. Go to:
http://quote.yahoo.com/quotes?symbols=IBM&detailed=3m
and see if you think the stock experienced anything significant
between May 3rd and May 11th.
It went up, but it had beeng going up pretty steadily for the
previous month, even more sharply, if anything.
A stock will often go up or down several percent in one day, it is
an extremely common occurrence.
In case you aren't following the US stock market, the whole thing
is going up quite a bit these days.
The web is a useful research tool.
bruce
Robert Hyatt
Robert Hyatt
Andreas Mader (ma...@p6.gud.siemens.co.at) wrote:
: Robert Hyatt wrote:
: >
: > : : : 2. VERY strange moves and mistakes made by GK.
: >
: > : : Someone reported other such strange moves in other world championship
: > : : matches. I rely on the "meat makes mistakes" proverb... It seems to
: > : : hold pretty well.
: >
: > : Have very many world champions prepared a known bad line with TN's
: > : that are both transparently bad and turn out to be bad? (...b5 is
: > : obviously answere by a4!)
: >
: > piece up with an exposed king. I'd hope he didn't try that against
: > Fritz, and win, and then expect the same to work against DB.. There's
: > a monstrous difference in the two programs. At least 400 Elo, probably
: > more...
: >
: >
: > : : the showed printouts to the "arbiter". this was documented.
: >
: > : I'd have felt a lot more confident if they had been shown to you.
: > : Who were they shown to? A savvy computer program user, even?
: > Ken Thompson. Who I'd think even more qualified than I am to analyze
: > and inspect... Also Mike Valvo was on the committee I believe, although
: > I am not sure who saw what, when...
: Are there any comments made by Ken Thompson concerning the printouts? Or
: is he forced to remain silent?
I don't know, but apparently a photo of the Be4 output was in a paper. I
*hope* it can be found, entered, and posted here. Data would be wonderful
to see...
Robert Hyatt
Ingo Althoefer (alth...@pdec01.uucp) wrote:
: For me it is a big surprise that a cool-headed man like Andreas Mader
: begins to think in mysteries and to believe in rumors.
: Andreas Mader wrote:
: >I was on holiday for a few days and when I returned to my office today I
: >found a couple of private posts asking why I changed my mind about the
: >DB-GK match. The main reason was a talk with a journalist (who is a
: >strong chess player, too). He returned to Vienna and gave us many
: >insights. I do not want to talk about all that I've heard since I have
: >no evidence. But these things are proven:
: >1. VERY strange moves in some of the games made by DB that noone can
: > reproduce with a PC program.
: Which moves do you mean, concretely ?
: At least the two situations in game 2 ( 37.Be4 instead of Qb6, 45.Kf1 )
: have been explained in rgcc by different people ( Ed Schroder, Matthias Wuel-
: lenweber, myself, ... ) in satisfactory ways.
: >2. VERY strange moves and mistakes made by GK.
: GK simply tried to play anti-computer chess.
: >3. Noone can prove that all the "computer" moves were really made by DB.
: >4. Refusal of the DB team to show printouts.
: C.J. Tan announced that "appropriate" parts of the printouts will be given
: to the public in the next few months.
: >5. DB is (was) unable to play the first move with white. The move had to
: > be entered manually by the DB team. This was standard approx. 20 years
: > ago.
: This is new to me. Did this really happen ?
No. It's a common practice that before starting a game, the operator of
any program can select an opening. With Crafty, you can edit the start.pgn
file, and rebuild the books.bin file, and make crafty play anything you want
for the first move. They do something similar. But DB can play without such
opening guidance. Look at the history of "scratchy" on ICC... that was DT
a couple of years ago, playing "automatically."
: >6. IBMs stock index went up by approx. 4 percent after the match.
: >7. Strange behaviour of all participants BEFORE the last game.
: >8. VERY strange behavior of GK (refuses to explain why he made such
: > strange opening moves in Game 6) and the DB team (they showed absolutely
: > no pleasure) during the closing ceremony.
: >9. DB will not prove its strength in a GM tournament.
: This has not been decided yet.
: > Maybe this is a miracle like the WM tournament Lasker vs Schlechter,
: >where Schlechter only needed a draw in the last game, but he lost in a
: >won position. Either that or it is a farce! Best wishes Andreas
: Sorry, but Schlechters behaviour is not a farce. The match conditions were
: that he had to win by a margin of two points to "win" the match. So he
: needed a second win in the last game. If the final result would have been
: 5.5 : 4.5 to his favour he would not have become world champion !
: Ingo Althoefer.
George Disher
Thanks Bob you are always saving me a lot of typing. I concur with
everything you said in the last post. I would like to know what the
farthest that any program has surchased in possition from move 37 of
game two. Did anyone get past 19 ply and if so what was the eval.
Andreas Mader
Ingo Althoefer wrote:
> At least the two situations in game 2 ( 37.Be4 instead of Qb6, 45.Kf1 )
> have been explained in rgcc by different people (Ed Schroder, Matthias
I mainly thought of these two moves. I do not have the time to read
every posting in rgcc, but from the postings I've read until now I
didn't get much satisfaction. Since we both have no evidence, noone can
prove his opinion, we can only speculate.
In case 1 (Be4) it is said by many people (maybe not by you) that DB
went to such an incredible search depth that it found a better move than
Qb6. On the other hand: Why didn't is see the draw in game 2, when it is
able to get to such incredible search depths?
Best and second best wishes, too (I, the boss, chose "best wishes"! :-))
Andreas
Robert Hyatt
: : : disingenuity. I can think of plenty of positions in which I can probably
: : : know better than DB what the best move.
: : : Or, at least, if an intended move *should not be played at all
: : : costs*. That kind of assistance can raise even a DB by a number of rating
: : : points. I'm sure you can think of some positions also.
: : I'd be willing to bet you can't find *one*, unless you are sporting a
: : new (GM) title I don't know about. Are you qualified to look over the
: : shoulder of a Dlugy, or a Ivanov, and suggest better moves to them?
: : I doubt it, because *I* wouldn't. DB is right in that class of player...
: Here's one idea.
: The kind of position in which black's rook is sweeping up lots of
: +1 pawns on the kingside, in an otherwise winning position in which *it
: does not have to do that*, while on the queenside, White has managed to
: get a position in which after 20+ ply Black will not be able to maneuver
: his rook behind the one passed pawn that is about to be supported to queen
: on the QR file.
: I don't have to be a GM to know to tell Black not to enter such
: complications, and just win normally. And, because I know the pattern, I
: may even be able to calculate whether the pawn will queen more easily than
: DB can.
: And that's just one idea.
In simple endings, as you point out, yes. But in the middlegame? It
would be *much* harder. All of the Kasparov complaints from match 1
and match 2 were middlegame, or endgames with queens+minors still on
the board...
Robert Hyatt
chrisw (chr...@cpsoft.demon.co.uk) wrote:
: Yeah. YEAH. YEEEAAAAH.
: g5, G5, GGGGGG55555555
: Chris Whittington
Was this a request for everyone to look at this position. When the
game was being played, Crafty would have played g5. I believe that
Bruce reported the Ferret liked g5. Others have reported that some
commercial programs like g5 as well.
If this is a legit position to analyze, I'll run it for a few hours
to see? Others can do the same. remember that some GM's said it
was forced. Others questioned it. It obviously is not "clear."
Robert Hyatt
George Disher (gdi...@mindspring.com) wrote:
: Thanks Bob you are always saving me a lot of typing. I concur with
: everything you said in the last post. I would like to know what the
: farthest that any program has surchased in possition from move 37 of
: game two. Did anyone get past 19 ply and if so what was the eval.
I don't have the PGN in front of me, but if this is Be4/Qb6... then I
have run to 21 plies, where I got a score of zero. Someone else has,
I believe, run even deeper with Crafty, and saw something similar.
(the score of zero was based only on searching Qb6. at depth 15-16,
Qb6 is about .3 better than Be4 according to Crafty... but it slowly
drops to zero over the next few plies. I don't remember what it was at
ply=20, but at 21 it was zero... although as Bruce says, not absolutely
a draw by repetition, although it could well be...
Robert Hyatt
Andreas Mader (ma...@p6.gud.siemens.co.at) wrote:
: Ingo Althoefer wrote:
: > At least the two situations in game 2 ( 37.Be4 instead of Qb6, 45.Kf1 )
: > have been explained in rgcc by different people (Ed Schroder, Matthias
: I mainly thought of these two moves. I do not have the time to read
: every posting in rgcc, but from the postings I've read until now I
: didn't get much satisfaction. Since we both have no evidence, noone can
: prove his opinion, we can only speculate.
: In case 1 (Be4) it is said by many people (maybe not by you) that DB
: went to such an incredible search depth that it found a better move than
: Qb6. On the other hand: Why didn't is see the draw in game 2, when it is
: able to get to such incredible search depths?
The two positions are a little different. From analysis with Crafty, to
reach ply=21, takes a *long* time. And it returns a score of 0.0... I did
not let it search Be4, but at a depth of 15-17-etc Be4 was producing a score
of +.3 to +.5, while Qb6 was +.8, until it dropped to zero at 20 or 21.
They should easily find that, which they apparently did. The other position
after Qe3 is horribly deep. Probably requiring a search that can reach a
depth of about 60 plies along the right checking lines, to find the draw.
And now, after more analysis, there is again suggestions that it wasn't a
draw. that after Q takes the bishop, where we were all mainly expecting
either h4/h5 or Qc5, someone has suggested Qc6 and d6 to get the pawn
moving. I have not run this by the computer, yet...
: Best and second best wishes, too (I, the boss, chose "best wishes"! :-))
: Andreas
Brion Brooks
I agree with Bob (who I've never met) that the conspiracy theory or
"fixed match" is a bunch of hooey. As a history buff, I've seen lots of
conspiracy theories. They all have one thing in common -- they are
impossible to refute (that's a hint, Bob). Every piece of evidence used
to defeat a conspiracy theory is transformed into one more part of the
"conspiracy".
As for the IBM stock issue, I seriously doubt that IBM has rested the
future of its business on whether a design team can make one of its
computers play a good game of chess (a game that most people have
absolutely no interest in). Give me a break. Why not claim that Nike
fixes basketball games so that Jordan will look better when wearing its
shoes? (Now that I think about it, even _that_ would be more likely than
the IBM scenario.)
I tell you one fact that might change my mind, though. Its the same
evidence that the SEC looks for to help decide if there's been insider
trading. Tell me if GK and the DB team bought huge amounts of IBM stock
just before the match and then sold 'em shortly afterward when the stock
"skyrocketed" by a whopping four points.
Regards,
Brion.
Martin Borriss
In article <5m2t53$9...@juniper.cis.uab.edu>,
At a time where you, Bob, as one of the last defenders of IM's playing
strength have turned against those by saying something along
"No IM would stand a chance against Deep Blue" I can tell you that IMs
or Masters in general would have avoided getting into a situation where
g5 was forced... ;)
No doubt that it is probably very,very hard to play vs. Deep Blue. But
thinking of game 5 and 6 in the first match, game 1,3,4 of match I believe
that human influence would have helped a lot. Not to mention the idea that
as coordinator one could steer the game into tactical slaughters...
Martin
--
Martin....@inf.tu-dresden.de
Robert Hyatt
Martin Borriss (bor...@inf.tu-dresden.de) wrote:
: In article <5m2t53$9...@juniper.cis.uab.edu>,
: hy...@crafty.cis.uab.edu (Robert Hyatt) writes:
: >chrisw (chr...@cpsoft.demon.co.uk) wrote:
: >
: >: Yeah. YEAH. YEEEAAAAH.
: >: g5, G5, GGGGGG55555555
: >: Chris Whittington
: >
: >
: >Was this a request for everyone to look at this position. When the
: >game was being played, Crafty would have played g5. I believe that
: >Bruce reported the Ferret liked g5. Others have reported that some
: >commercial programs like g5 as well.
: >
: >If this is a legit position to analyze, I'll run it for a few hours
: >to see? Others can do the same. remember that some GM's said it
: >was forced. Others questioned it. It obviously is not "clear."
: >
: At a time where you, Bob, as one of the last defenders of IM's playing
: strength have turned against those by saying something along
: "No IM would stand a chance against Deep Blue" I can tell you that IMs
: or Masters in general would have avoided getting into a situation where
: g5 was forced... ;)
I knew you would pop in here. :) And you made a great point, of
course. Which was also my point. If, in game 1, a problem occurred,
it wasn't at g5. It was much earlier. Black was struggling at that
point, in my analysis. And I first thought they had simply thrown a
wild punch. Then I find that Crafty and other programs (not all but
many) like g5 as well. I didn't pursue g4 however, I don't believe
Crafty liked that.
: No doubt that it is probably very,very hard to play vs. Deep Blue. But
: thinking of game 5 and 6 in the first match, game 1,3,4 of match I believe
: that human influence would have helped a lot. Not to mention the idea that
: as coordinator one could steer the game into tactical slaughters...
I would now discount game 5/6 from the first match. This seemed like a
much stronger mule this time around. Strong enough that even in games 1,3
and 4 of this match, GK had a fierce struggle on his hands. I personally
believe he was quite lucky to win game 1. That got pretty open, and DB
simply misevaluated those two connected pawns...
: Martin
:
: --
: Martin....@inf.tu-dresden.de
Patrick J. LoPresti
gdisher> Thanks Bob you are always saving me a lot of typing. I
gdisher> concur with everything you said in the last post. I would
gdisher> like to know what the farthest that any program has
gdisher> surchased in possition from move 37 of game two. Did anyone
gdisher> get past 19 ply and if so what was the eval.
Funny you should ask; my 10-day run of Crafty run just finished last
night. The results are appened to this message.
Be4 does indeed look better the deeper you go. (Perhaps not
surprising, since white is pretty much winning anyway...) I still
like the theory that Deep Blue liked its queen where it was, because
it wanted to recapture with the queen if/when Kasparov decided to
trade rooks. Or, put another way, it wanted to discourage the
immediate rook exchange from happening at all, which even Crafty
agrees Be4 should accomplish.
This is all just an anthropomorphic way of saying that everything is
easily explained by Deep Blue having a slightly different evaluation
function than Crafty.
Of course, this doesn't "prove" that there was no human intervention,
but it does mean that there is no reason to think that there WAS human
intervention. Then again, if the DB team were going to cheat, they
would surely only pick moves from among the top two or three
recommended by the machine, right? :-)
So does anyone have a copy of those DB "printouts" yet?
- Pat
White(37): search Be4
search Be4
White(37): go
clearing transposition table
clearing pawn hash tables
time limit 166666:39
depth time score variation (1)
6 0.18 -0.254 Be4 Rxa2 Rxa2 Rd8 g4 h5 Ra7 hxg4 hxg4
6-> 0.18 -0.254 Be4 Rxa2 Rxa2 Rd8 g4 h5 Ra7 hxg4 hxg4
7 0.50 -0.247 Be4 Rxa2 Rxa2 Qd8 g4 Rc7 Ra1 Qd7
7-> 0.50 -0.247 Be4 Rxa2 Rxa2 Qd8 g4 Rc7 Ra1 Qd7
8 1.40 -0.240 Be4 Rxa2 Rxa2 Ra8 Rxa8 Qxa8 Qe3 Qa1+
Kf2 Qb2+ Kf3 Kf8
8-> 1.40 -0.240 Be4 Rxa2 Rxa2 Ra8 Rxa8 Qxa8 Qe3 Qa1+
Kf2 Qb2+ Kf3 Kf8
9 4.50 -0.210 Be4 Rxa2 Rxa2 Rb8 Ra7 Ra8 Rxa8 <HT>
9-> 4.50 -0.210 Be4 Rxa2 Rxa2 Rb8 Ra7 Ra8 Rxa8 <HT>
10 13.58 -0.178 Be4 Rxa2 Rxa2 Ra8 Rxa8 Qxa8 Kf1 Qa4
g4 h5 Ke2 hxg4 hxg4
10-> 13.64 -0.178 Be4 Rxa2 Rxa2 Ra8 Rxa8 Qxa8 Kf1 Qa4
g4 h5 Ke2 hxg4 hxg4
11 1:38 -0.099 Be4 Rab8 Ra6 Qd8 Qd2 Qd7 Qe3 Rc7 Ra7
Rxa7 Rxa7 <HT>
11-> 1:38 -0.099 Be4 Rab8 Ra6 Qd8 Qd2 Qd7 Qe3 Rc7 Ra7
Rxa7 Rxa7 <HT>
12 2:39 -0.110 Be4 Rab8 Ra6 Qd8 Kf1 Rb7 g4 Qd7 Kg1
Qe7 Ra8 Rc7
12-> 2:39 -0.110 Be4 Rab8 Ra6 Qd8 Kf1 Rb7 g4 Qd7 Kg1
Qe7 Ra8 Rc7
13 6:33 -0.209 Be4 Rcb8 Rxa8 Rxa8 Rxa8 Qxa8 Qe1 Qa7+
Kf1 h5 g3 Qa2 Kg1 Kf8 Qe3 Qa1+ Kg2
Qd1 Kf2
13-> 6:33 -0.209 Be4 Rcb8 Rxa8 Rxa8 Rxa8 Qxa8 Qe1 Qa7+
Kf1 h5 g3 Qa2 Kg1 Kf8 Qe3 Qa1+ Kg2
Qd1 Kf2
14 13:37 -0.120 Be4 Rcb8 Ra6 Rxa6 Rxa6 Qd8 Kf1 Rb7
g4 Kf7 Ra7 Rd7 Rxd7+ Qxd7 Qe3 Kf8 Kf2
Qf7
14-> 13:37 -0.120 Be4 Rcb8 Ra6 Rxa6 Rxa6 Qd8 Kf1 Rb7
g4 Kf7 Ra7 Rd7 Rxd7+ Qxd7 Qe3 Kf8 Kf2
Qf7
15 21:22 -0.110 Be4 Rcb8 Ra6 Rxa6 Rxa6 Qd8 Kf1 Rb7
Ra7 Qb8 Rxb7 Qxb7 g4 Kf7 Qe3 Qd7 Kf2
15-> 21:22 -0.110 Be4 Rcb8 Ra6 Rxa6 Rxa6 Qd8 Kf1 Rb7
Ra7 Qb8 Rxb7 Qxb7 g4 Kf7 Qe3 Qd7 Kf2
16 121:21 -0.088 Be4 Rcb8 Ra6 Rxa6 Rxa6 Qd8 Kf1 Rb7
Ra7 Qb8 Rxb7 Qxb7 g4 Kf8 Ke2 Qc7 Ke3
Qb6+ Ke2
16-> 121:21 -0.088 Be4 Rcb8 Ra6 Rxa6 Rxa6 Qd8 Kf1 Rb7
Ra7 Qb8 Rxb7 Qxb7 g4 Kf8 Ke2 Qc7 Ke3
Qb6+ Ke2
17 208:51 -0.106 Be4 Rcb8 Ra6 Rxa6 Rxa6 Qd8 Kf1 Rb7
g4 Qb8 Qa2 Kh7 Ra8 Qc7 Qa5 Rb6 Ra7
Qd8
17-> 208:51 -0.106 Be4 Rcb8 Ra6 Rxa6 Rxa6 Qd8 Kf1 Rb7
g4 Qb8 Qa2 Kh7 Ra8 Qc7 Qa5 Rb6 Ra7
Qd8
18 744:37 -0.081 Be4 Rcb8 Ra6 Rxa6 Rxa6 Qd8 Kf1 Rb7
g4 Qd7 Qa2 Kh7 Rc6 Rb8 Qa6 Be7 <HT>
18-> 744:37 -0.081 Be4 Rcb8 Ra6 Rxa6 Rxa6 Qd8 Kf1 Rb7
g4 Qd7 Qa2 Kh7 Rc6 Rb8 Qa6 Be7 <HT>
19 1356:56 -0.081 Be4 Rab8 Ra6 Qd8 Rc6 Rc7 Raa6 Rd7
Ra7 Kf8 Rxd7 Qxd7 Ra6 Ke8 Ra7 Rb7 Ra5
Qc7 g4 Kf8
19-> 1356:56 -0.081 Be4 Rab8 Ra6 Qd8 Rc6 Rc7 Raa6 Rd7
Ra7 Kf8 Rxd7 Qxd7 Ra6 Ke8 Ra7 Rb7 Ra5
Qc7 g4 Kf8
20 4154:10 -0.051 Be4 Rab8 Ra6 Qd8 Rc6 Rc7 Raa6 Rd7
Ra7 Kf8 Rxd7 Qxd7 Qa2 Kg8 Qa6 Bc7 d6
Bd8 Bd5+ Kh7 Be6 Qe8
20-> 4154:11 -0.051 Be4 Rab8 Ra6 Qd8 Rc6 Rc7 Raa6 Rd7
Ra7 Kf8 Rxd7 Qxd7 Qa2 Kg8 Qa6 Bc7 d6
Bd8 Bd5+ Kh7 Be6 Qe8
21 14546:10 -0.041 Be4 Rab8 Ra6 Qd8 Rc6 Rc7 Kf1 Re7 Ra7
Rd7 Rca6 Kf8 Rxd7 Qxd7 Qa7 Rb7 Qa8+
Rb8 Ra7 Rxa8 Rxd7 Ra1+ Kf2 Ra2+ Ke3
21-> 14546:10 -0.041 Be4 Rab8 Ra6 Qd8 Rc6 Rc7 Kf1 Re7 Ra7
Rd7 Rca6 Kf8 Rxd7 Qxd7 Qa7 Rb7 Qa8+
Rb8 Ra7 Rxa8 Rxd7 Ra1+ Kf2 Ra2+ Ke3
da...@taic.net
Howard Exner
Chris Yee <cy...@wpi.edu> wrote in article
<Pine.OSF.3.95q.97052...@wpi.WPI.EDU>...
> On Wed, 21 May 1997, Andreas Mader wrote:
>
> > 1. VERY strange moves in some of the games made by DB that noone can
> > reproduce with a PC program.
>
> How does this contribute to your argument? If noone can reproduce it,
> then anyone can, right? Who is he, anyway?
Noone is a type error. It should read "no one".
Howard Exner
Don Fong <df...@cse.ucsc.edu> wrote in article
<5lvp0q$8...@darkstar.ucsc.edu>...
> In article <5lur23$h...@juniper.cis.uab.edu>,
> Robert Hyatt <hy...@crafty.cis.uab.edu> wrote:
> >Of course, the big question is *who* would be qualified to help DB?
> >The player that has been beating Kasparov regularly over the past few
> >years? Who might that be?
>
> ah, but the point is that you don't have to be stronger than DB
> or than Kasparov, to help. Ingo Althofer is constantly telling us
> how a 3-hirn of human + 2 computers can play stronger than any of the
> three alone. Bob, you yourself have remarked how human operators "cheat"
> on the servers by helping their programs from time to time. even if
> the human operator is not stronger than the computer or than the opponent,
> the operator can still *compensate for weaknesses* in the computer's game.
>
> isn't it obvious? consultation helps. that's why it's forbidden.
>
Bob, a while back we discussed the K Kup game between Rebel 8 and Crafty.
I mentioned the bizarre plan of Rebel's to chase a pawn at the grave
expense of imprisoning it's Queen. Wouldn't this be an example of a strong
program making a wrong choice, one which a weaker human player would
no doubt overrule (as in 3-hirn). I believe that a weaker player can on rare
occasions
spot such a faulty plan. Were not some of Deep Blue's moves in game one
criticised by most every GM annotating the game. (ie: the wasted time spent on
shuffling the Queen and Bishop along the a5-d8 diagonal)
Robert Hyatt
Howard Exner (hex...@dlcwest.com) wrote:
: Don Fong <df...@cse.ucsc.edu> wrote in article
That I agree, although I didn't understand Qa7, period. It seemed to
instantly lose a pawn according to Craftys analysis, and it wasn't very
deep, especially considering this is a 1 hour per move on a P6/200 type of
game.
However, after watching Kasparov vs Karpov, or somethhing similar, would
you really feel confident in telling DB "no, not that."???
I'm a decent chess player, and I'd be hard-pressed to even consider over-
riding a move from that machine, because i couldn't possibly know whether
a move like g5 in game one is a gross mistake, a great move, or merely
"legal"...
Howard Exner
Robert Hyatt <hy...@crafty.cis.uab.edu> wrote in article
<5m78f9$u...@juniper.cis.uab.edu>...
No I certainly would not feel very confident overriding Deep Blue but I believe
a 2500 player would be able to do so on rare occasions. There must be
some chinks in the armour that a seasoned IM or GM can discover during
actual play (as opposed to post game analysis).
>
> I'm a decent chess player, and I'd be hard-pressed to even consider over-
> riding a move from that machine, because i couldn't possibly know whether
> a move like g5 in game one is a gross mistake, a great move, or merely
> "legal"...
I agree but would you not detect weak moves in a program one class above you
(200
rating points say)? I believe it is relative to the program and the player.
Robert Hyatt
Howard Exner (hex...@dlcwest.com) wrote:
: Robert Hyatt <hy...@crafty.cis.uab.edu> wrote in article
The reason I say "no* is that in the first game, we saw at least one "odd"
rook move that was criticized by all of the GM's analyzing the game (I
believe it was something likd Ra7, rather than Rb8 to defend the b-pawn.
After the match, when asked, Kasparov said "that was the *only* move" and
game reasons why.
That would be a big danger... overriding a move that looks "odd" but is
in fact critical...
Robert Hyatt
Robert Hyatt (hy...@crafty.cis.uab.edu) wrote:
: Howard Exner (hex...@dlcwest.com) wrote:
Jeez... *bad* typo. In the first *match*... last year... not the
first *game* of this year's match...
da...@taic.net
Robert Hyatt <hy...@crafty.cis.uab.edu> wrote:
: : : I'd be willing to bet you can't find *one*, unless you are sporting a
: : : new (GM) title I don't know about. Are you qualified to look over the
: : : shoulder of a Dlugy, or a Ivanov, and suggest better moves to them?
: : : I doubt it, because *I* wouldn't. DB is right in that class of player...
: : Here's one idea.
: In simple endings, as you point out, yes. But in the middlegame? It
: would be *much* harder. All of the Kasparov complaints from match 1
: and match 2 were middlegame, or endgames with queens+minors still on
: the board...
Bob, Bob, Bob. You were willing to bet that I could not find
*one*. I found one. It took me about 60 seconds, which was actually the
time I spent entering the words describing the position.And what do you
do? Do you say "Oh, you found one. I was wrong"? Do you say "Oh, tell me
about your new (GM) title"?
No, you just qualify your prior absolute comment, by now saying
"In simple endings, yes..."
Bob, do you really want me to now spend *more* than 60 seconds to
come up with another position, for the middle game? When I do, what will
you do then? Will you pay off then?
Ok, here's another position: You're a pawn up on the queenside.
You got nasty pieces pointing at the black kingside also. Black may be
able to defend. Or maybe not. But, ya know, the real GM decision is to
just trade, and win safely in the endgame. Not so for our intrepid
computer, always willing to push for the max, since he's got pieces
pointing you know where. Put a human at the controls, and he may decide
safe endgame.
Now, please pay up.
Johanes Suhardjo
Maybe useful for others. Comments/suggestions are welcome.
Johanes Suhardjo (joh...@farida.cc.nd.edu)
--
What the hell, go ahead and put all your eggs in one basket.
=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+ CUT HERE =+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=
This is an attempt to follow Bruce Moreland's advice on how to test versions of
a chess program as posted on rec.games.chess.computer newsgroup on May 22,
1997. Bruce is the programmer of ferret, 1995 world microcomputer amateur
champion and 1996 world microcomputer speed chess champion. According to
Bruce, take 30 positional problems (no check mate), run each problem for
3 minutes and compare the time to reach the right answer, the PV (principal
variation), the depth reached, and the time to finish each ply.
This set contains 30 problems taken from:
Andrew Soltis' "Pawn Structure Chess"
Israel Gelfer's "Positional Chess Handbook"
Jeremy Silman's "How to Reassess Your Chess"
Neil McDonald's "Planning"
Y. Averbakh's "Chess Endings: Essential Knowlege"
pos001 to pos010 are opening positions
pos011 to pos020 are middle game positions
pos021 to pos030 are end game positions
pos001
~~~~~~
(Silman, Diagram 202)
R - - Q K B - R Black's plan is simple: He will play for a queenside attack
P P - N - P - - via ...b5, ...Nb6, ...Nc4, ...Qa5, etc. It turns out that
- - P - P - P N Black's undeveloped pieces are quite active while White is
- - - P p - - P unable to challenge Black's coming assault. With this
- - - p - p - p position in mind, we can meditate upon the maxim: No
p - n - b n p - development is better than a bad development.
- p p q - - - -
- - k r r - - -
Black to move
pos002
~~~~~~
(Silman, Diagram 206, Silman-Gogel, U.S. Open 1981)
R - - Q R - K - Bd1 followed by Bc2 eyeing Black's kingside and further
- - - N - P B P controlling the e4 square. f4 taking e5 away from Black's
P P - P - N P - Knights. Bd3 making Black's ...b6-b5 advance impossible.
- - P p - - b -
p - p - - - - -
- - n - - b - p
- p - q - p p -
r - - - - r k -
White to move
pos003
~~~~~~
(Silman, Diagram 218, Silman-Getz, World Open 1990)
- - - - R R K - Black is a pawn down but his active pieces, play against
P P P B Q - P P White's weak pawns, and control of c5 give him lots of
- - N P - - - - compensation. White made use of a tactic and played the
- - - - - - - - surprising 16. Bxe4! Giving up this light-squared Bishop is
- - p n N b - - very risky for White but he accurately judged that Black
- - p q - - p - would have to sacrifice further material to keep his
p - - - p p b p initiative of passively go into a very bad endgame.
r - - - r - k - 16. ...Qxe4 17. Nb5! The c7 pawn can't be defended without
White to move Black fixing White's pawn structure (with 17. ...Qxd3).
17. ...Rxf4 18. gxf4 Qxf4 19. Nxc7 Wins more material and
swings the Knight over to the wonderful d5 square.
19. ...Re5 20. Nd5 Rg5+ 21. Kh1 Qxf2 22. Qf3 Qh4 The battle
that was started in the opening (material versus activity)
is still being fought. 23. Rg1 Seeking exchanges.
23. ...Bg4 24. Qf4 h6 25. Rg3 Ne5 26. Rag1 Black's attack
is running out of steam. White's 'take everything' strategy
has won out!
pos004
~~~~~~
(Soltis, page 21, Maroczy-Charousek, Nuremberg 1896)
R - B - - R K - White's bishops look murderous, but this is exactly when the
P P Q - B P P P blocking effect of the enemy pawns is most useful. Black
- - - - P - - - activated his majority with 16. ...f5! 17. Qe2 Bd6! 18. h3
- - - - - - - - e5, and he has already usurped the initiative. 19. b4 e4
- - p - q - - - 20. Bc2 b6 21. f3! exf3? The correct method was 21. ...Bb7
- - b b - - - - 22. fxe4 Rae8 after which Black's kingside majority has been
p p - - - p p p converted into a domination of the center.
r - - - - r k -
Black to move
pos005
~~~~~~
(Soltis, page 110, Botvinnik-Golombek, 1956 Olympiad in Moscow)
R - Q - - R K - 12. e4 Maroczy Bind. The primary advantage of the Bind is
P B - - P P B P that if Black failed to fight the constriction of his
- P N P - N P - pieces, he can be strangled. After this simple move Black
- - - - - - - - was reduced to passivity. Passivity is fatal.
- - p - - - - -
- p n - - - p -
p b n - p p b p
r - - q - r k -
White to move
pos006
~~~~~~
(Soltis, page 201, won by Geza Maroczy in 1929)
R - B - R N K - White opened a knight file by force: 17. f4! Bd8 18. Bf2 Rb8
- - - Q B - P P 19. g4! g6 20. h4 Na5 21. Nxa5 Bxa5 22. a3 All that White
P - N - P - - - needs to win now is to prevent ...b4 as he prepares to open
- P - P p P - - up the kingside. The inevitability of success in this
- - - n - - - - strategic plan led Black to an error that shortened the
- n p b - - b - game.
p p - - q p p p
- - - r - r k -
White to move
pos007
~~~~~~
(Soltis, page 229, Reti-Spielmann, Marienbad 1925)
R - - - R B K - 16. f4! Qe7 17. g4 Nf6 18. e5! Black has a problem of
- P - B - P P P limited space that is common to the Indian complex. His
- - P P - - - - king knight cannot retreat. If d7 were free he could stand
P - N - - - Q N better by exchanging pawns and dropping his knight to the
- - p - p - - - vacant sqauare. But now he must play 18. ...dxe5 19. fxe5
- p n - - - p p Qxe5 and face the consequences of 20. Ne4 Qc7 21. Nxf6+ gxf6
p b q - n p b - 22. Nf4 which threatens Nh5. The Black kingside cost him
- - - r - r k - the game quickly.
White to move
pos008
~~~~~~
(Soltis, page 264, Larsen-Radulov, Leningrad 1973)
- - - Q - R K - 20. Ba5 Nb6 21. b4! and Black's center disintegrated.
P B R N - P - P There was no real defense (21. ... c4 22. Bxd4!). The game
- - - - - B P - went 21. ...Be7 22. exd5 Bd6 23. Qh6 cxb4 24. Ng5 f5 25. Ne6
- - P P - - - - Qd7 26. Rxc7 Resigns.
- - - - p q - -
- p b b - n - -
p - - - - p p p
- - r r - - k -
White to move
pos009
~~~~~~
(Soltis, page 285, Toran-Korchnoi, 1956 students' olympiad)
R - N Q - R K - While White has accepted the Panov formation, Black marshals
P P - - P P B P his attack on the d-pawn. His knight is headed for f5 where
- - N - - - P - it will join two other minor pieces in the assault on d4.
- - p P - - - - This should have warned White that he must start an active
- - - p - - B - plan such as 13. b4! (13. ...Nxb4 14. Qb3).
- p n - b n - -
p - - - b p p p
r - - q - r k -
White to move
pos010
~~~~~~
(Soltis, page 300, Pillsbury-Showalter, Nuremberg 1896)
R - B - - R - K Pillsbuary, one of the greatest planners, found one in
P P - - - - P P 15. c5!, intending to open up the b-file. He got what he
- Q P - P N - - wanted after 15. ... Qc7 16. f3 Nxc3 17. Rxc3 Bd7 18. Nxd7!
- - - P n P - - Nxd7 19. b4! Rf6 20. b5 Black's kingside attact is
- - p p N p - - invisible, and White had a dominating position after
- - n b - - p - 20. ... Rg6 21. Kf2 h5 22. h4! Rf8 23. Rb3 Rf7 24. Rdb1 Qd8
p p - - q p - p 25. bxc6 bxc6 26. Rb7 Qa5 27. R1b3 Rgf6 28. Qb2.
- - r r - - k -
White to move
pos011
~~~~~~
(Gelfer, Diagram 101, Karpov-Larsen, Tilburg 1980)
- - K R - - - R 14. ...h5 15. h4 g5! Exploiting the fact that this pawn is
P P P Q - - P P immune to capture, Black improves the formation of his
- - - - B P - - pawns. 16. f3 Rdg8 17. Rf2 Qc6 28. Bd2 g4 Accentuating the
- - - - - - - - power of Black's bishop. 19. f4 Bc4 Here Larsen comments:
- - - p - - - - "White is crushed, smashed and finished."
- - p - b - q -
p - p - - p p p
r - - - - r k -
Black to move
pos012
~~~~~~
(Gelfer, Diagram 165, Hort-Hubner, West Germany 1982)
- - R - - R K - 16. ... Bxh4! 17. Rxc6 Rxc6 18. Nxh4 Rxc1+ 19. Bxc1 Qa4
- P - Q B P P P 20. Qg4 Kh8 21. a3 Qc2! 22. Qf4 Kg8 23. b4 Rc8 24. Be3 Qd1+
P - N - P - - - 25. Kh2 Qh5 26. g3 Rc3 27. a4 Rb3 Black won in 38 moves.
- - - P p - - -
- - - p - - - p
- - - - - n - -
p p r b q p p -
- - r - - - k -
Black to move
pos013
~~~~~~
(Gelfer, Diagram 279, Karpov-Spassky, Moscow 1974)
R - Q - - R K - White exchanged his somewhat passive bishop with 19. Bg4!
- - - N B P - P leaving Black with an unenviable position. 1-0 in 35 moves.
- - P - - - P -
P - - - P - - -
p N - - p - - -
- - n - b b - p
- p p - - r p -
r - - - - q k -
White to move
pos014
~~~~~~
(Gelfer, Diagram 299, Petrosian-Lutikov, Tbilisi 1959)
- R - - Q R - K A powerful pawn-thrust serves to pin-point the awkward
- - N B N P B - placing of the black knights. 24. b4! Nc8 (After
P - - P - - P P 14. ...cxb4 25. c5 Black is hard-pressed to prevent 26. c6,
p P P p P - - - since 25. ...dxc5 fails against 26. d6, attacking both
- - p - p - p b knights). 25. bxc5 dxc5 26. cxb5 Nxb5 27. Bxb5 Rxb5 28. 0-0
- p - - n - - p Castling late (or not castling at all) was a characteristic
- - q n b p - - of Petrosian.
- - - - k - - p
White to move
pos015
~~~~~~
(Gelfer, Diagram 465, Yanosky-Portisch, Stockholm 1962)
- R - - - - - - 48. ... Rxc4!! 49. Qxc4 (If 49. bxc4 Rb4 and ...Qa6 is
Q - - - - - P K powerful). 49. ... Rb4 50. Qc3 Qa6 51. Kf2 c4! The c-file
- - - P - P - P is to serve as a base for the final onslaught. 52. bxc4
P - P p P p - p Rxc4 53. Qa3 Qc8! Move by move, Black's threats accumulate.
p R b N p - p -
- p q - - - - -
- r - - - - k -
- r - - - - - -
Black to move
pos016
~~~~~~
(McDonald, page 21, Karpov-Kasparov, London Wch (5) 1986)
R - R - - - K - 20. Nh3!! a5 21. f3 a4 22. Rhe1! Now we see the difference
P - - B - - B - between Nf3 and Nh3. White is able to bring his rook to e1
- - - p P - P P where it defends the e5 pawn. This means that when the
- - P - p P - - knight on h3 begins its journey to d3 via f2, Black's ...g5
- - p - - b - p response is useless, since after White's Bxg5, the e5 pawn
- - - - - - - - is defended by the rook. 22. ...a3 As a rule, passed pawn
- - - k - p p - can be either very strong or very week. In this case, the
- - r - - - n r passed pawn would be strong if it could be supported by
White to move Black's king's bishop, but the bishop on g7 is a mere
spectator. It is completely shut out of the game. Black is
is effectively playing a piece down. 23. Nf2 a2 24. Nd3 Ra3
25. Ra1! Now Black could already resign.
pos017
~~~~~~
(McDonald, page 35, Skold-Boleslavsky, Helsinki 1952)
- - - - - R K - 21. Qh4! Kg7 22. Qd4 Threatening d6. 22. ...c5 23. Qc3
P P - B Q - - P Here 23. ...Qd6, breaking the pin on the queen along the
- - P - P R P - e-file, was necessary. White would maintain a strong
- - - p r P - - position (his pressure along the e-file is intense) but
- - p - - - - - there would be no forced win.
- - - b - - - q
p p - - - p p p
- - - - r - k -
White to move
pos018
~~~~~~
(McDonald, page 50, Kasparov-Yurtaev, Moscow 1981)
R N B Q - R K - 13. g4!! Gains space and introduces the idea of g5, chasing
P P - - B P P P the knight from f6, then winning the d5 pawn. If the knight
- - - - - N - - is chased from f6, then White can also play Bc2 and Qd3,
- - - P - - - - attacking the black kingside and forcing Black to play
- - - p - n - - ...g6, then he would have converted his development
p - n b - p - - advantage into something more permanent -- a long-term
- p - - - - p p king's defence, which White could try to exploit in the
r - b q - r k - middle-game. 13. ... Bd6 14. Kh1 Re8 15. g5 Bxf4 16. Bxf4
White to move Nh5 17. Bxb8!! Another profound positional move. Let's
look at the time saved by White in playing this move.
First, he removed the bishop from attack by the knight at h5
without loss of time; second, he cleared the f4 square for
the advance 18. f4, which discovers an attack on the knight
on h5 and so forces Black to play 18. ...g6, wasting a move;
and third, the white queen goes to the attacking square f3,
at the same time ruling out Black's move ...Bh3, which would
have gained time for Black's development by attacking the
rook on f1. The time Kasparov saved by playing 17. Bxb8 is
just the time he needs to carry out a pawn storm on the
kingside.
pos019
~~~~~~
(McDonald, page 50, Salov-Mcdonald, Oviedo 1992)
R - - - R B - - 30. ... Bxb4 31. cxb4 Kd6 Black has a clear plan. He wants
- P K - - - - - to create two connected passed pawns in the center, which
- - - - P - - P will steam-roller through to the eighth rank. The only
p - - P - - P - barrier is White's f4 pawn. This must be removed. The most
- n - - - p - - important element of this is the use of the h-pawn as a
- - p - - - p - battering ram to attack the g3 pawn. Once the g3 pawn
- p - r - - p - drops, then the f4 pawn will be fatally undermined. 32. Rh1
- k - - r - - - Rf8! Much better than Rh8. 33. Rd4 Rf6 So the rook
Black to move defends the h-pawn, but in a far more active way than after
32. ...Rh8. 34. Rf1 Rg8! Threatening 35. ...gxf4 36. gxf4
Rxg2.
pos020
~~~~~~
(McDonald, page 96, Legki-Mcdonald, Douai 1993)
R - - - - R K - If the center were to dissolve then White's king could find
P P - - P P B - itself exposed to pressure along the a7-g1 diagonal.
- - N P - N P - 14. ...e5!? After any plausible continuation, the center
- - P - - Q - - opens. Black's rooks can then become involved in the
- - p p - p - P attack. If White loses control of the d4-square, then he
- p n - p - p B faces ruin.
p b - q n - - p
r - - - r - k b
Black to move
pos021
~~~~~~
(Silman, Diagram 207, Damjanovic-Fischer, Buenos Aires 1970)
- - - - - - - - To win, Black needs to penetrate with his King. At the
- - - - - - - - moment though, White's own monarch will keep him out by
- - - - - N - P hanging around e3 and d3. Fischer solves this dilemma by
P - K - P b P p advancing his a-pawn. 1. ...Kb5 2. Kd3 a4 3. bxa4 (letting
- P - - p - p - ...a3 makes the a2 pawn an indefensible target) Kxa4 4. Kc4
- p - - k - - - Ka3 5. Kc5 Kxa2 6. Kxb4 Kb2 Now Black is winning for two
p - - - - - - - reasons:
- - - - - - - - 1) His King is deep in enemy territory and as a result will
Black to move be the first to much White's pawns;
2) When pawns are only on one side of the board, the Knight
is superior to the Bishop. This is because the Bishop's
long range powers are useless, while the Knight's
ability to go to any colored square is of great
importance -- nothing is safe from the beast!
pos022
~~~~~~
(Silman, Diagram 208, Fischer-Taimanov, Vancouver 1971)
- - - - - - - - 1. Bb3 Ka7 2. Bd1 Kb7 3. Bf3+ The King can no longer
- K - - N b - - happily mark time on a7 and b7. 3. ...Kc7 4. Ka6 Now
- P - - - - P - White's next step is to put his Bishop back on f7 where it
P k P - - P - P ties the Knight down to the defense of the pawn on g6.
p - - - - p - p 4. ...Ng8 5. Bd5 Ne7 6. Bc4 White wants Black to move his
- - p - - - p - Knight first, since then Bf7 will be with tempo. 6. ...Nc6
- p - - - - - - 7. Bf7 Ne7 8. Be8! Zugzwang! Any King move will drop the
- - - - - - - - b6 pawn. Any Knight move would drop the g6 pawn, and
White to move 8. ...c4 9. Kb5 or 9. Bf7 picks up the e4 pawn.
pos023
~~~~~~
(Silman, Diagram 212, Kuznecov-Silman, Oregon Open 1986)
- - - - - - K - Black's plan:
- B Q - - P - P 1) Place his minor pieces on their best possible squares.
- P P - - - P - 2) Trade Queens since she has little to do with the
P - N - P - - - interaction of the minors.
p - p - p - - - 3) Tie White down to the defense of the a4 and c4 pawns.
- - n - - p - - 4) Advance pawns on the kingside and grad as much territory
- p - q - - p p as possible.
- - - - - b - k 25. ...Ne6 Eyeing d4 and f4. 26. Ne2 Qe7 27. Nc1 Kg7
Black to move 28. Nb3 c5 Ending any White hopes of a c4-c5 advance and
nailing down the d4 square for the Black Knight. 29. Nc1
Bc6 30. Qd1 Qd7 31. Qxd7 Bxd7 32. b3 Nd4 The White Knight
is stuck guarding the sad little pawn on b3. 33. Kg1 f5!
Creating new imbalances. Now White mus either view e4 as a
point to be defended or he can trade and give Black the
possibility of a passed e-pawn via a later ...e5-e4.
pos024
~~~~~~
(Silman, Diagram 215)
- - - - - - - K 1. a3 Black threatedned to steal the game by 1. ...a3!
- - - - - P P P 1. ...Kg8 2. g4 Kf8 3. Kg2 Ke7 4. Kf3 Ke6 5. h4 f6 6. h5 Kd6
- - - - - - - - 7. f5 Ke7 8. Kf4 Kd6 9. g5 fxg5+ 10. Kxg5 Ke5 11. h6 and the
- - P - - - - - game is over.
P P P - p p - -
- - p - - - - -
p p - - - - p p
- - - - - - - k
White to move
pos025
~~~~~~
(Silman, Diagram 217, Mephisto(C)-Silman, American Open 1989)
- - - - - - - - White should retain his Knight by 1. Nd4. The computer was
- - - - - - - - programmed to think that Bishops were superior to Knights in
- - - K - - - - an endgame so it happily allowed Black to capture on b3. A
- - - P - - N - Knight is usually preferable to a Bishop with pawns on one
B - - - P - P - side of the board because a Knight can attack any pawn and
- n - - p P p - can chase the King from any square. The ability to go to
- - b k - p - - any color is more important than long range powers in such
- - - - - - - - situations.
White to move
pos026
~~~~~~
(Averbakh, Diagram 113)
- - - - - K - - 1. Kf1 Ke7 2. Ke2 Kd6 3. b4 Kd5 4. Kd3 f5 5. f5 g6 6. g3 a6
P - - - - P P P 7. a4 Kc6 The King is unable to remain in an active
- - - - - - - - position; it is running out of moves and must retire.
- - - - - - - - 8. Kd4 Kd6 9. b5 axb5 10. axb5 Kc7 11. Ke5 This is the
- - - - - - - - simplest. White gives up his passed pawn but obtains a
- - - - - - - - decisive material superiority on the other wing.
p p - - - p p p
- - - - - - k -
White to move
pos027
~~~~~~
(Averbakh, Diagram 119, Tchigorin-Marshall, Carlsbad 1907)
- - - - - - - - 1. Nd5 Nd7 2. g5 h6 3. Nf6 Nb6 4. h4 hxg5 5. hxg5 Rf8 6. Kc5
- - - - - P K P Na4+ 7. Kd6 and black King cannot get free.
- N - - - - P -
- p - - - - - -
- - - k - - p -
- - n - - - - p
- - - - - - - -
- - - - - - - -
White to move
pos028
~~~~~~
(Averbakh, Diagram 128, Kotov-Botvinnik, Moscow 1955)
- - - - - - - - 1. ...g5!! 2. fxg5 (White also lost after 2. hxg5 h4
- - - - - - - - 3. Bd6 Bf5 4. g6 Bxg6 5. f5 Bxf5 6. Kxb3 Kg2) 2. ...d4+!
- - - - B - P - It is important for Black to preserve his b3 pawn. 3. exd4
- - b P - - - P Kg3 4. Ba3 Kxh4 5. Kd3 Kxg5 6. Ke4 h4 7. Kf3 Bd5+ White
- - - - - p - p resigned.
- P k - p K - -
- - - - - - - -
- - - - - - - -
Black to move
pos029
~~~~~~
(Averbakh, Diagram 130, Levenfish-Rauzer, Tiflis 1937)
- - - - - - - - 1. Nd5! Bg5 2. f4! Bd8 3. Nb4+ Kd6 4. Nd3 g5 5. Ne5 Ke6
- - - - - - - - 6. d5+ Kf6 7. Kc5 gxf4 8. Nf3 Black resigned, since he
- - K - - - P - cannot stop the pawn without suffering decisive material
- - - - - P - - losses.
- - k p - - - P
- - n - B p - p
- - - - - - p -
- - - - - - - -
White to move
pos030
~~~~~~
(Averbakh, Diagram 138, Tarrasch-Rubinstein, San Sebastian 1911)
- - - R - - - - Here passive defence by 1. ...Rd6 loses quickly because of
- - - - - - - P 2. Ke2 followed by 3. a4, threatening 4. a5. Black's
r P - - - - K - salvation is in active play. 1. ...Rd2! 2. Rxb6+ Kg5
- - - - - P - - White has two extra pawns, but Black threatens to construct
- - - - P - - - a mating net around the white King by f4-f3. 3. Ke1 Rc2
- - p - - - - - 4. Rb5! Kg4! 5. Rh3+ (To permit 5. ...f4 and 6. ...Kf3
p p - - - p - p could only lead to a loss for White). 5. ...Kxh3 6. Rxf5
- - - - - k - - Rxb2 7. Rf4 Rxa2 8. Rxe4 h5 Draw agreed 6 moves later.
Black to move
Rolf Tueschen
da...@taic.net wrote:
>Robert Hyatt <hy...@crafty.cis.uab.edu> wrote:
>: : Here's one idea.
Please, I want to share this GM talk with one note.
Remember, that a human *sees* this sort of thing in a period as short as a clip
of his eye. Bah, no 60 seconds.
So, this is another good point by *data*, and, zut alors, I wanna know who this
guy/woman really _is_. He writes almost _my_ thoughts. :)
> Now, please pay up.
Robert Hyatt
: : : : I'd be willing to bet you can't find *one*, unless you are sporting a
: : : : new (GM) title I don't know about. Are you qualified to look over the
: : : : shoulder of a Dlugy, or a Ivanov, and suggest better moves to them?
: : : : I doubt it, because *I* wouldn't. DB is right in that class of player...
: : : Here's one idea.
: : In simple endings, as you point out, yes. But in the middlegame? It
: : would be *much* harder. All of the Kasparov complaints from match 1
: : and match 2 were middlegame, or endgames with queens+minors still on
: : the board...
: Bob, Bob, Bob. You were willing to bet that I could not find
: *one*. I found one. It took me about 60 seconds, which was actually the
: time I spent entering the words describing the position.And what do you
: do? Do you say "Oh, you found one. I was wrong"? Do you say "Oh, tell me
: about your new (GM) title"?
Wait. Maybe I overlooked your "better move"... but let me make sure
it was from the right place:
the game must be played by DB, which pretty much means somewhere in the
6 games we saw. You have to find a "better" move for DB in real-time,
because you'd have to override during the game, and wouldn't have hours
to analyze, like we did after each of the games...
The example you are giving is, I believe, a fictional one. I agree that
with many programs even *I* can find better moves than what they play,
at times. But I've never found one for DB, particularly in "real time."
: No, you just qualify your prior absolute comment, by now saying
: "In simple endings, yes..."
: Bob, do you really want me to now spend *more* than 60 seconds to
: come up with another position, for the middle game? When I do, what will
: you do then? Will you pay off then?
Just don't fabricate on. Find it in one of the actual games. It is easy
to fabricate positions to support anything. We were talking about a human
overriding DB... So what would you override after only 2-3 minutes of
searching, with no help?
: Ok, here's another position: You're a pawn up on the queenside.
: You got nasty pieces pointing at the black kingside also. Black may be
: able to defend. Or maybe not. But, ya know, the real GM decision is to
: just trade, and win safely in the endgame. Not so for our intrepid
: computer, always willing to push for the max, since he's got pieces
: pointing you know where. Put a human at the controls, and he may decide
: safe endgame.
: Now, please pay up.
That's no good... because I can produce a position where I watched
Crafty play yestarday... KQPP vs K, it forced the opponent to take the
queen to see the mate in 19 from the tablebase hits. Was that right or
wrong? Positions like you describe didn't occur in the DB match. It
is saying "if DB would reach a position where there is an obvious win and
a speculative win, and it goes for the speculative win, I'd overrule it."
However, I don't recall such a position happening...
Howard Exner
We have both presented arguments through example (mine from game #1 of the
rematch
and yours from Game #1 of the first match). Both are valid in my opinion and
we seem deadlocked but the question remains, is Deep Blue subject to being
overruled?
My main argument was an individual on rare occasions can overrule a computer
program
a class above them. Why then can't a 2500 rated player do the same at their
level of
understanding chess. I admire the Deep Blue project but am not willing to
bestow on it
the mantle of infallibility quite yet.
Michael Wilson
In article <3384A4...@p6.gud.siemens.co.at>,
Andreas Mader <ma...@p6.gud.siemens.co.at> wrote:
>Ingo Althoefer wrote:
>>
>> > Maybe this is a miracle like the WM tournament Lasker vs Schlechter,
>> >where Schlechter only needed a draw in the last game, but he lost in a
>> >won position. Either that or it is a farce! Best wishes Andreas
>>
>> Sorry, but Schlechters behaviour is not a farce. The match conditions were
>> that he had to win by a margin of two points to "win" the match. So he
>> needed a second win in the last game. If the final result would have been
>> 5.5 : 4.5 to his favour he would not have become world champion !
>
According to the source I'm reading, it is also inaccurate! I'm
getting this from Hannak's biography of Em. Lasker. It devotes an entire
chapter to the Schlecter match, and details the draws in the first four
games, then the nastier fifth. Supposedly Lasker let his king wander
all the way (unprotected) to the queen's wing, while Schlecter went after
it hard. Finally, round about the fiftieth move, Lasker figured he was
safe, and ready to begin his counter-attack. He was wrong, and found
himself in a mating net from which there was no escape.
The next four games, of course, were draws. In the final game,
Lasker found himself in the uncomfortable position of having to win a
game against Schlecter, the man called "The Draw-Wizard." Not an easy
task, especially when (According to Hannak!) Schlecter only needed to draw
the game to win the match! So, the 2 point minimum thing is nonsense, or
Hannak himself got it wrong.
As for Schlecter's behaviour, Hannak offers an explanation for
it that I personally find rather appealing. Hannak theorizes that Carl
(Schlecter) didn't want to win the match on a fluke. Certainly, Lasker
tried to get him to step out of character and play an aggressive game,
but Hannak theorizes that Carl didn't really need that much encouragement.
To quote:
"[Schlecter] too wanted to avoid a draw. He did not want
to win the match on the strength of Lasker's unlucky slip in the
fifth game. He was loath to win the World Championship 'by a fluke.'
He wished to earn the title the hard way, and in trying his hardest
to do so, this frail and amiable Viennese put up a truly heroic
fight."
Anyway, the game was one of those incredible see-saws. Schlecter deviated
from opening theory by the fourth move - his move later became common, about
20-30 years later. Lasker got the upper hand early, then overreached
himself. Schlecter took over the initiative and the game. At that point,
he could easily have made a safe draw of the game. Instead, he played for
the win. On his 35th move, Schlecter made an error that let Lasker fight
his way out. Four moves later, he blew the draw, and Lasker kept the game
solidly in hand from then on.
I've got the game, complete with annotations, right here in hard
copy, but it's a nuisance to type in, and it would probably violate some
copyright laws anyway.
Mike Wilson
(wil...@rpi.edu)
--
Mike Wilson
(wil...@rpi.edu)
Robert Hyatt
Howard Exner (hex...@dlcwest.com) wrote:
: We have both presented arguments through example (mine from game #1 of the
: rematch
: and yours from Game #1 of the first match). Both are valid in my opinion and
: we seem deadlocked but the question remains, is Deep Blue subject to being
: overruled?
: My main argument was an individual on rare occasions can overrule a computer
: program
: a class above them. Why then can't a 2500 rated player do the same at their
: level of
: understanding chess. I admire the Deep Blue project but am not willing to
: bestow on it
: the mantle of infallibility quite yet.
I don't think it is infallible either. But *very* *very* strong.
I hate to ask for replays, but would you re-explain your game-1 move? I
believe that I missed this during our news breakdown last week, or else I
misread it.
A move there should be worth discussing in the context you are talking
about since it happened recently...
Howard Exner
Robert Hyatt <hy...@crafty.cis.uab.edu> wrote in article
<5mam7q$6...@juniper.cis.uab.edu>...
> Howard Exner (hex...@dlcwest.com) wrote:
>
>
> : We have both presented arguments through example (mine from game #1 of the
> : rematch
> : and yours from Game #1 of the first match). Both are valid in my opinion
and
> : we seem deadlocked but the question remains, is Deep Blue subject to being
> : overruled?
> : My main argument was an individual on rare occasions can overrule a
computer
> : program
> : a class above them. Why then can't a 2500 rated player do the same at their
> : level of
> : understanding chess. I admire the Deep Blue project but am not willing to
> : bestow on it
> : the mantle of infallibility quite yet.
>
>
> I don't think it is infallible either. But *very* *very* strong.
Yes in games 3,4, and 5 it played many strong defensive moves making
it difficult for GK to deliver the knockout blow.
>
> I hate to ask for replays, but would you re-explain your game-1 move? I
> believe that I missed this during our news breakdown last week, or else I
> misread it.
>
> A move there should be worth discussing in the context you are talking
> about since it happened recently...
I used the example where in game #1 Deep Blue was shuffling both
it's Queen and Bishop along the a5 to d8 diagonal. This series of moves
seemed wrong somehow.
(11. ... Qa5?
Up to now, the machine has played just like a human. Qa5 is a different story.
A Grandmaster would never play such a move, and neither would a weaker player.
Why? Well, this is by far not the best place for the queen. After all, it
doesn't really do anything on a5, but get in the danger of being threatened by
white's pieces. But how can you tell a computer that? The computer simply looks
at the move, calculates as far as it can, doesn't see anything wrong about it,
and plays it. And he is right - there is no immediate danger, but nonetheless,
a human intuitively senses this is not the best place for the queen, and would
put it on the more natural c7 or e7.
12. ... Bc7?
Everything I said about black's last move applies here too. Bc7 is once again a
move a human master would never even think about. There is no real scope
associated with this move, and the only thing it does, is just close the queen
in its own bad place. The lack of long term vision that the machine is
displaying will most likely hurt it again during this match.) This analysis is
from Gabriel Schwartzman.
Also I've made an assumption on the human who intervenes that he or she
would be at Deep Blue's side viewing the evals of the second and third best
moves and so on.
Perhaps a second move choice would have a slightly lower score but intuitively
to a GM be
a better choice.
Robert Hyatt
Howard Exner (hex...@dlcwest.com) wrote:
: Robert Hyatt <hy...@crafty.cis.uab.edu> wrote in article
: (11. ... Qa5?
I sort of agree, although Qa5 is not gross...
: 12. ... Bc7?
But Bc7 certainly was, as now the queen and bishop are biting into each other,
and the queen is cut off from retreating to help on the kingside, although white
didn't immediately try to exploit this...
: Everything I said about black's last move applies here too. Bc7 is once again a
: move a human master would never even think about. There is no real scope
: associated with this move, and the only thing it does, is just close the queen
: in its own bad place. The lack of long term vision that the machine is
: displaying will most likely hurt it again during this match.) This analysis is
: from Gabriel Schwartzman.
: Also I've made an assumption on the human who intervenes that he or she
: would be at Deep Blue's side viewing the evals of the second and third best
: moves and so on.
This I don't think is possible. DB is doing just like I do, traditional alpha/beta,
which means the second and third-best moves are totally unknown, except for those
rare searches where at the last iteration the first move is 3rd best, and is then
replaced by the next move, which is really second best, which is then replaced by
a move further down which is actually the best. But for most searches, you have a
best move and nothing else... the famous alpha/beta "issue."
: Perhaps a second move choice would have a slightly lower score but intuitively
: to a GM be
: a better choice.
If you had such information, it could be much better to let a human choose...
:
David Hanley
Howard Exner wrote:
>
> I used the example where in game #1 Deep Blue was shuffling both
> it's Queen and Bishop along the a5 to d8 diagonal. This series of moves
> seemed wrong somehow.
>
> (11. ... Qa5?
>
>
> 12. ... Bc7?
>
Don't you think it's odd that after your 'two blunders'
DB got a even or even better middlegame? Remember, DB lost later when
kaspy sack'd the exchange. Actaully he was *forced* to sac, otherwise
DB was going to be much better. DB missed the positional compensation
for the sacrifice.
The number of players who could truly help DB is very small, excepting
the special case you have noted:
If there are several 'good' moves at the end of the search, let a human
player choose the move which conforms to a plan they see. This would
ensure progress along a reasonable plan.
dave
Howard Exner
David Hanley <da...@nospan.netright.com> wrote in article
<338B60...@nospan.netright.com>...
> Howard Exner wrote:
>
> >
> > I used the example where in game #1 Deep Blue was shuffling both
> > it's Queen and Bishop along the a5 to d8 diagonal. This series of moves
> > seemed wrong somehow.
> >
> > (11. ... Qa5?
> >
> >
> > 12. ... Bc7?
> >
>
> Don't you think it's odd that after your 'two blunders'
> DB got a even or even better middlegame?
These were cited by Gabriel Schwartzman. Other GM's analysing the game
similarily criticize these moves. The topic in this thread is Bob's question,
who can overrule Deep Blue. This is to me a clear example of when a player
like Joel Benjamin for example would want to intercede. The topic is not that
Deep Blue is
weak or strong, clearly it plays a strong game.
Having a better middlegame than GK in this game is questionable, but the wasted
tempos on Qa5 and Bc7 are explained nicely by GM GS.
> Remember, DB lost later when
> kaspy sack'd the exchange. Actaully he was *forced* to sac, otherwise
> DB was going to be much better.
The converse of saying Deep Blue forced GK into the win is that Deep Blue
blundered.
>DB missed the positional compensation
> for the sacrifice.
>
>
> The number of players who could truly help DB is very small, excepting
> the special case you have noted:
My post did say that I believe overruling holds for rare occasions.
I just can't accept the extreme view that no one is capable of detecting better
moves than Deep Blue plays (for now anyway). On this we are in agreement.
David Hanley
Howard Exner wrote:
>
> David Hanley <da...@nospan.netright.com> wrote in article
> <338B60...@nospan.netright.com>...
> > Howard Exner wrote:
> >
> > >
> > > I used the example where in game #1 Deep Blue was shuffling both
> > > it's Queen and Bishop along the a5 to d8 diagonal. This series of moves
> > > seemed wrong somehow.
> > >
> > > (11. ... Qa5?
> > >
> > >
> > > 12. ... Bc7?
> > >
> >
> > Don't you think it's odd that after your 'two blunders'
> > DB got a even or even better middlegame?
>
> These were cited by Gabriel Schwartzman. Other GM's analysing the game
> similarily criticize these moves.
Ok. And GK says 'at 14 or 15 DB had a good position' he cites g5-g4
as the
blunder which lost the game. The fact that other gm's thought these
blunders
shows that allowing them to overrule wouldn't help--they would have
passed up good
but weird moves such as these.
> The converse of saying Deep Blue forced GK into the win is that Deep Blue
> blundered.
Yes. A move that many gm's thought was good. Again, overruling
wouldn't help.
> My post did say that I believe overruling holds for rare occasions.
> I just can't accept the extreme view that no one is capable of detecting better
> moves than Deep Blue plays (for now anyway). On this we are in agreement.
Quite possibly, but in different degrees. Would you have overruled the
..h5! move?
Almost any GM would have; it gave DB a good game, however. How about
the prepetual check
save? The machine looked like a f**ing genius; almost any gm would have
lost in the
same position to kasparov.
For example, I can't think of any DB game in this match which a human
would have
been able to improve upon, exception the game (3?) in thich it played
the bishop to e7
instead of g7. Even that turned out ok.
dave
Don Fong
In article <338C6...@nospan.netright.com>,
David Hanley <da...@nospam.netright.com> wrote:
> For example, I can't think of any DB game in this match which
>a human would have been able to improve upon, exception the game (3?)
>in thich it played the bishop to e7 instead of g7. Even that turned
>out ok.
the problem with your reasoning is that IF the conspiracy theorists
are right, the human has *already* improved the games by the time you
see them. basically your reasoning is circular.
Ren Wu
>Maybe useful for others. Comments/suggestions are welcome.
>
> Johanes Suhardjo (joh...@farida.cc.nd.edu)
>--
>What the hell, go ahead and put all your eggs in one basket.
>
Does anyone run this test yet? Or can someone convert this to .fin or
.fen format?