Posted By: Koles (nikam nejdu) on 'CZchess'
Title: zaznam z posledniho dne zapasu :)
Date: Tue May 13 17:58:12 1997
varovani: tento prispevek je dlouhy :)
obsahuje nejprvnejsi komentare k partii a rozhovor s Garrim Kasparovem.
---
ROBERT J. T. MORRIS: Good afternoon. As you know, we're at the
last and determining game of the Kasparov vs. Deep Blue match.
Today promises to be very, very exciting. We can take as
evidence what we've seen during the last few days.
Now, before we get into the match, I would like to run a couple
of minutes of scholarly discussions about parallel processing
and how far it has come.
Will you run the tape, please.
"From New York -- do you want cream and sugar with that? -- it's
the Late Show with David Letterman.
YASSER SEIRAWAN: I love stuff like this. Over the weekend world
chess champion Garry Kasparov was defeated by the IBM computer
Deep Blue. Deep Blue defeated Garry Kasparov. In a related
story earlier today the New York Mets were defeated by a
microwave oven.
Letterman: This computer Deep Blue is a very, very intelligent
machine. This machine, to give you an idea of how smart it is,
it has the good sense not to pick up a drag queen on Santa
Monica Boulevard!
So far, ladies and gentlemen, this is the very best rehearsal
we've ever had.
"It's the Late Show with David Letterman.
Letterman: Garry Kasparov, the world chess champion is playing a
computer, programmed to make 200 million chess moves a second.
And I believe the series is one win apiece and one draw, and
it's the best of, I don't know, what is it, Paul, 18, 20, 24?
Best of six? Whoever wins the most after six games? Or are
you just making up a number?
But any ways, I've been reading a little bit about this
computer. They're calling it Deep Blue. Not only does it play
chess. This is an amazing piece of modern technology. Look,
we have some tape. Look at what else the computer does, in
addition to playing chess. (Picture of computer with a
ping-pong paddle.) (Picture of a computer driving a sled dog
team.) (Picture of a computer vaulting a high jump.)
"From NBC studios in Burbank, the Tonight Show with Jayleno!"
Len owe: Over the weekend, Garry Kasparov playing chess matches
against the IBM supercomputer. We were at the bar knocking
down brewskies, shouting, guys are going nuts watching this
chess match against the computer. I guess Kasparov won the
first match and then the computer won the first game. Kasparov
did not take it well. Did you see today how he reacted? I
thought this was bad sportsman ship. Here he is playing the
computer. There's the other man representing the computer
here. Now, watch when he loses.
"Check" watch what Kasparov does here? (Shows man unplugging
cord.) (Audience laughter.)
Len owe: I just hope that brings closure to this! (Audience
laughter.)
"From the NBC studios in Burbank, the Tonight Show with Jay Len
owe."
Len owe: And world class chess champion Garry Kasparov playing
the IBM computer Deep Blue in game three today of their
best-of-seven match. Deep Blue, that's an odd name for a
computer. Sounds more like some Smurf porn star, doesn't it?
I guess today was a tie. I guess Kasparov won, right, then he
lost, and today was a tie. We've been glued to the TV
watching! The sitting with the guys, the chips, doesn't get
any better than that.
If you saw today, did you see what happened? Kasparov switched
games on the computer. Very clever strategy. Show this.
Look, Garry is playing. Show the game he switched. Look, they
switched to Candyland! (Audience laughter.)
People think Candyland is a piece of cake!
"From the NBC studios in Burbank, the Tonight Show with Jay Len
owe!"
Leno: Chess champion Garry Kasparov says very impressive is the
IBM Deep Blue computer. Have you been following this thing?
Showing signs of actual intelligence. You know, why waste time
building computers that can beat Grandmasters in chess? Why?
Why doesn't IBM put some time and effort trying to make a
computer that won't break down while you're in Sears trying to
buy a package of underwear? (Audience applause.) (Audience
laughter.)
Have you been following this chess match? People just on the
edge of their seats watching this! It's getting really
intense. Did you see Kasparov today? It looked like it was
getting to him. It looked like the pressure -- show the
match. Show this clip from the match. There's Kasparov.
(Fans yelling from the sideline, Kasparov in a grimace.)
Once those fans start trash talking like that...
ROBERT J. T. MORRIS: I would like to introduce to you Monty
Newborn. Monty is the chairman of the ACM chess committee and
Monty is overseeing the match. Monty?
MONTY NEWBORN: On behalf of the ACM, the Association for
Computing, it's my pleasure to be here and participate in this
exciting event. This is going to make chess history. We're
anticipating an incredibly exciting afternoon. We have the
world champion with 2 1/2 points. We have the IBM chess
program, Deep Blue, with 2 1/2 points. And we're about seven
or eight minutes from starting this game, which will last about
four to five hours.
Perhaps at this time if anybody has to go to the bathroom, this
is a good time. Nobody will leave once we begin. Not that you
can't, but nobody will leave.
We have three commentators that will bring this game to life in a
way that won't be clear by the time it's finished whether it's
a football game, a baseball game, or a chess match. I assure
you, it's a chess match.
Our first commentator is Mike Valvo. Mike?
Mike has been the arbiter at the last match between Kasparov and
Deep Blue in Philadelphia last year. He's an International
Master. He's one of the toughest guys at playing chess with
his eyes closed. He can play between ten and 15 and maybe as
many as 20 people with his eyes closed, and beat most of them.
MIKE VALVO: All these flowery words, and a few minutes ago he
said that we were reduced to weather forecasters in that last
game.
MONTY NEWBORN: We were talking off stage, and this last game was
such an incredibly complex game that it was my understanding
that nobody quite understood the whole thing, although the
commentators were pretty close, I'll give them credit.
Our second commentator is Maurice Ashley.
Maurice is an International Master. He's a New Yorker. He was a
commentator at our last match. And he's the setup man in this
combination. You watch how he sets up everybody on stage
here. He's terrific.
Our third commentator is Yasser Seirawan. Yasser has been the
United States chess champion three times. He's been one of the
top players in the United States for well over a decade now.
He served as a commentator as well in the last match in
Philadelphia, and I'm sure that he'll add the last dimension in
chess expertise to this panel of outstanding chess players and
commentators.
I wish you all a very exciting afternoon and I'm going to turn
the floor over to Maurice Ashley, who will continue and
entertain you for the rest of the day. Maurice?
MAURICE ASHLEY: Thank you, Monty. Welcome once again to game
six, the final game of the Kasparov vs. Deep Blue Rematch. I
don't know about you, but I'm excited, we're excited, the whole
world is excited. And no plug-pulling will happen today, I can
assure you. There's no tomorrow.
MAURICE ASHLEY: There's no tomorrow. This is it. And Kasparov
has it seems been rattled by what is going on here. All of us
are amazed by how Deep Blue has played. The score is tied 1-1,
that is in wins, and three draws. So the winner today takes it
all home. And Kasparov today is Mother's Day, which I wish a
happy Mother's Day to you all. Happy Mother's Day. Kasparov's
mother Clara is here. We often watches Kasparov play, "here"
meaning the Equitable Building. I would like to welcome those
over the WebTV and the Internet who are watching. We are in
the Equitable Building now, the auditorium. On the 35th floor,
Kasparov is preparing to enter to play against Deep Blue, and
operator Joe Hoane, I believe it is, is awaiting Kasparov's
arrival. And about Kasparov's mother watching, there's got to
be some pressure on Mother's Day, Yaz, knowing that her son is
going into battle. And in addition to that pressure, Kasparov
has set this whole thing up as he's defending mankind. Mankind
against the ever-encroaching speed and complexity and problem
solving ability of the computer.
Well, true justice in the human race is enough pressure for
Garry, Yaz. What do you think his chances are for today?
YASSER SEIRAWAN: Well, that's the 700-thousand-dollar question.
That's the winner's purse. And quite frankly, Clara, Garry's
mom, is nervous not only for herself and her son but it seems
for just about everybody on the whole team. I'm nervous with
her. Before the match I thought Garry was going to win very
easily. I really thought that the computer hadn't made as much
progress as it has, and, I don't know, but I have a gut
instinct that Garry is going to win it. A real gut instinct.
But he has put himself under a lot of, lot of pressure, playing
the black pieces in the final game. He's only one once against
the computer in 11 tries -- actually not in 11 tries -- but
when he was black, he only managed to beat the computer once in
the whole series of games. And by the way, Garry does appear
to be late, and I don't know if it's a psychological ploy --
(Audience laughter.)
MAURICE ASHLEY: That ain't going to work with this thing. Mike,
are you as stunned as everybody else as how well Deep Blue is
doing?
And before you answer, Garry Kasparov has seated himself at the
table.
Are you stunned that he is taking his watch off preparing for
battle? Are you amazed that Deep Blue has done this well.
MIKE VALVO: I'm amazed and I think the computer has taught us
new ways to do some things, too. It's just incredible how it
drew that game. It just came out of nowhere.
MAURICE ASHLEY: "That game" meaning yesterday's game?
MIKE VALVO: Out of the blue -- a bolt out of the blue! That was
supposed to be a joke.
YASSER SEIRAWAN: A bolt out of the blue.
MAURICE ASHLEY: Kasparov here you see readying himself. The
press has been tremendous this match, and there are several
press members here on the 35th floor. They also have a press
room on the 49th floor covering the match --
MIKE VALVO: 50th floor now.
MAURICE ASHLEY: 50th floor now. Hundreds of journalists. From
just all over. This match has taken on epic proportions and
this game will decide it all, Kasparov readying himself.
The last game Kasparov had, Yaz, he had a good position, he had
some strong winning chances. Do we expect to see some of what
happened in the last game?
YASSER SEIRAWAN: Well, Garry has been playing what I would best
describe as second-rate openings. He seems to be quite
cautious or is simply afraid of the preparation that Deep Blue
and his team, and it seems to me that there was a whole group
of Grandmasters helping Deep Blue, and so he's played some
second rate openings, not his first lines of defense. We are
expecting e2-e4.
DB MOVE: 1 e4
MAURICE ASHLEY: And in fact the first move of this epic first
game has been played. Deep Blue has played e2-e4.
GK MOVE: 1...c6
DB MOVE: 2 d4
MAURICE ASHLEY: He has repeated the move c7-c6 which caused Yaz
to grown because usually he plays c7-c5. Deep Blue has
prepared and played d2-d4.
GK MOVE: 2...d5
DB MOVE: 3 Nc3
MAURICE ASHLEY: This move d5 is in fact different from what he
had done in game four. Kasparov had played d7-d6, a bit more
cagey, a bit more cautious, and now he's going into what looks
like a Caro-Kann.
YASSER SEIRAWAN: Exactly. The first two moves for Garry, c6 and
d5, it's an opening I play all the time, in fact. What we saw
in game 4 was not the move d7-d5, this really obscure unusual
move d7-d6. And like I said, Garry got a good game in that
particular game. But, again, it's not first-line openings by
Garry Kasparov. Now this move d7-d5.
And let me just talk about this for a moment. Garry Kasparov is
not a Caro-Kann player. The Caro-Kann is an opening that is
very quiet, that is to say it's a very positional-based
opening. Tactics usually come much later in the game and it's
very often that in a Caro-Kann style of play, the whole game is
a strategic, positional game. I had hoped as you saw me, I
tried to make a prediction that Garry is going to play his
favorite Sicilian, c7-c5, which he didn't do. So I don't know
what Garry was expecting with the Caro-Kann.
GK MOVE: 3...dxe4
MAURICE ASHLEY: Kasparov has captured, which is the most
frequent move in this position.
DB MOVE: 4 Nxd4
MAURICE ASHLEY: Deep Blue has responded instantly by playing
Nxe4, and now Kasparov --
GK MOVE: 4...Nd7
MAURICE ASHLEY: That's a move that aze arch-rival Anatoly Karpov
often plays.
Mike, Kasparov has not played one opening that we would call a
real opening, a dip cal Kasparov opening.
DB MOVE: 5 Ng5
MIKE VALVO: By the way, Garry has played the Caro-Kann before.
I have seen some games that he has played. So it's not totally
new to him. It's very interesting that he is playing this line
because this could be a very complicated line, very topical.
So the computer is going to have a lot of what we call book
moves, a lot of storeed positions. This is the most topical
thing in the Caro-Kann today, this particular line. So the
computer could play quickly for the next 15 moves.
MAURICE ASHLEY: Well, Yaz, this last move, Ne4-g5, to many
beginners' eyes, we know the principle don't move a piece twice
in the opening if you don't have to. And here this knight has
moved from e4 to g5 it seems without any provocation. Why
don't you explain to us why this is.
YASSER SEIRAWAN: Do I have to? (Audience laughter.)
MAURICE ASHLEY: You're the Grandmaster.
YASSER SEIRAWAN: I don't want to reveal any of my secrets now.
No, chess theory has evolved over a long, long, long period of
time. And this move Ne4-g5 has been played after thousands and
thousands of previous games where they played Nf3, Bc4, Qe2,
and other such moves.
But the idea of the move knight g5, in principle, is to create an
early attack against the f7 square, forcing black to play
e7-e6, and then the knight usually tries to gain control of the
e5 square.
It's become a favorite weapon, and especially in the mid-eighties
it was played almost constantly for white.
As you mentioned, Anatoly Karpov is a great defender of the
Caro-Kann position, a very illustrious career with the black
pieces, and I think it very likely that we will see one of
those openings that are analyzed out for 15 or 20 moves,
because it's going to be now very hard for Kasparov to avoid
those lines. In these types of positions, you don't want to
play anything original, because you could get into a lot of
trouble early. I think that he's going to play one of the main
lines and be satisfied with the resulting position.
MAURICE ASHLEY: But isn't this the kind of position that often
tends to a draw? A lot of people say the Caro-Kann is a
drawish opening. If you want to win, you've got to play
something sharp like the Sicilian. Does he want to draw today,
Mike? Is Kasparov happy with a draw today and result in a
drawn match?
MIKE VALVO: Obviously he's not going to be happy, but he doesn't
want to lose, either! (Audience laughter.) So he's going to
play a nice, solid opening. He's noticed by now that he's
doing the best in the endgame against this machine, although
the machine seems to be getting away at the last moment every
time. He still is doing better in the endgame than any other
phase of the game. Yesterday he did well in the beginning,
then the computer really fought back hard and, gosh, it looked
like the computer was getting an edge, and we went into an
endgame, and it looked like Garry was just creaming the
computer, going to queen a pawn, and all of the sudden they
agreed to a draw right in front of our eyes.
Nonetheless the endgame offers the best chances for Garry and
that's where he's heading. He's hoping to steer through a
middle game, beat back the attack that Deep Thought -- Deep
Blue -- used to be Deep Thought -- that Deep Blue is presenting
in front of it, and hoping for the endgame. So we may have a
very clear-cut, easy-to-follow kind of game, and that's what
the audience seems to appreciate. There was one game, I think
it was game four, where everybody followed from beginning toned
the whole game and I suspect this will be an easy game to
follow, very strategic in nature, simple. It will be Yasser's
kind of game, he likes this kind of game, he's going to enjoy
it, he's going to say yes, the Caro-Kann is vindicated, but the
rest of us know it can't be true.
YASSER SEIRAWAN: Thanks for that set up. Let me just say on
behalf of myself and other Caro-Kann players, yes, in general
we take the perspective that white with the opening move has
the opportunity of building up an advantage. So the Caro-Kann
really is an equalizing weapon. But it doesn't necessarily
mean that just because I play the Caro-Kann I'm playing for a
draw. The Caro-Kann is a very solid setup for black and if
white overpresses, he easily ends up a victim.
GK MOVE: 5...Ngf6
MAURICE ASHLEY: Kasparov has in fact moved, playing his knight
on g8 to f6.
MIKE VALVO: You know, it's interesting. People may be
wondering --
DB MOVE: 6 Bd3
MIKE VALVO: -- why Garry is taking so long to make almost forced
moves. Why do you think this is the case, Yasser?
YASSER SEIRAWAN: In this particular situation the opening is now
become established. He was unsure of what Deep Blue was going
to play on the third move. Deep Blue had lots of choices, the
advanced Caro, the Panov Botvinnik, with the exchange of pawns
on d5. So he wasn't absolutely sure he was going to get this
position. He's now got the position. And he's preparing
himself for the kind of battle that we've been talking about, a
strategic battle.
MAURICE ASHLEY: After Ngf6 Deep Blue has responded instantly
with Bf1-d3, developing the bishop, putting it on a very solid
square. Potentially Kasparov might castle king-side, so the
bishop is well placed for that.
GK MOVE: 6...e6
MAURICE ASHLEY: Opening a line for his bishop. And again Deep
Blue is clearly in its opening book because it is playing very
quickly.
DB MOVE: N1 f3
MAURICE ASHLEY: Kasparov trying to get his bishop quickly into
the action, we anticipate the bishop on f8 moving shortly.
We would like to note that those who are following this over the
Internet. Some who are unable to follow live but must follow
using a web server or just follow using a chessboard. So we
will try to be as visual as possible, be your eyes and ears, so
to speak, because they're just following it in the written
text.
At the moment, then, we should say to our left we are in an
auditorium in the Equitable Building. To our left is a video
screen that shows the current position at all times. Behind us
is a Fritz 4 computer playing program, very strong, and it has
helped us to do the analysis over time and I guess to sort of
understand Deep Blue in a way. But we find that Deep Blue is a
bit biased in its opinion.
MIKE VALVO: Not today, it likes black better.
MAURICE ASHLEY: Today it's liking black. We will explain what
these mean, some bar graphs and evaluations that we will
explain shortly. And to our right is another video screen
showing Kasparov and a variety of things including the Deep
Blue team, the audience, a number of things at different times.
So, back to the game position --
GK MOVE: 7...h6
MAURICE ASHLEY: Instead of bringing out his bishop with Bd6,
Kasparov has instead --
DB MOVE: 8 Nxe6
MAURICE ASHLEY: Capturing on e6 instantly and Kasparov shook his
head for a moment --
GK MOVE: 8...Qe7
DB MOVE: 900.
GK MOVE: ...fxe6
DB MOVE: Bg6+.
GK MOVE: ...Kd8
MAURICE ASHLEY: Kasparov is shaking his head as if something
disastrous has happened, his king being chased around the
board. Is it possible that Kasparov has played incorrect
theoretically?
YASSER SEIRAWAN: Yes, he has. He blunered. What he did is he
transposed moves. What I mean by that is this position is
quite well known, and you had witnessed me playing the move
Bf8-d6. The idea being that after Bd6, it's standard for white
to then play Qe2, and then after h6, this sacrifice Nxe6
doesn't work because black has the move Kf8 later.
MAURICE ASHLEY: You mean after Nxe6?
YASSER SEIRAWAN: Capturing the knight, there's the check, the
king can go to f8.
But playing h6 one move earlier, the sacrifice that we've now
seen, h6, is possible. As far as I recall, there was a famous
game between Granda Zuniga, Grandmaster from Peru, vs. our very
own Patrick Wolff. And it was a very difficult game for black
to play and it became recognized that the move h6 was wrong.
And Gary, assist -- Garry, as you saw his reactions, the moment
that Deep Blue played Nxe6 so very quickly and reached the
position they now have on the board, he was in just terror,
distress. Because he's -- he recognizes that he's fallen for a
well-known opening trap.
MAURICE ASHLEY: Is this over? Is it that simple?
MIKE VALVO: No.
MAURICE ASHLEY: I mean he's up a piece for a pawn.
YASSER SEIRAWAN: Right.
MAURICE ASHLEY: His king is in a sorry state right now on the d8
square, to be sure. Is it just over?
I remember another game between wolf and Epishin in the same
line, and I believe it was somewhat difficult to just press on
the attack necessarily. I know it's a strong position for
white, it looks like a dangerous position, and a lot of players
would be licking their chops to have a position like this
against Kasparov with his king so disgusting on d8. (Audience
laughter.) But I mean a piece is a piece after aall -- after
all and maybe he can work his way out and not only that, one
thing I'm intrigued by is Deep Blue, in what is a so-called
trappy line, standard opening trap programmed into it, did not
play instantly but spent a lot of time before finally deciding
to play this move.
DB MOVE: 11 Bf4
MAURICE ASHLEY: This traps the king on the d8 square. Now the
king cannot move and Kasparov immediately has to defend.
Kasparov cannot be a happy man, Mike. He played an opening that
is normally not his mixier, and now he's going to have to
suffer fo who knows how many moves.
MIKE VALVO: I don't remember Bf4 being a correct move in that
position.
YASSER SEIRAWAN: In fact, I believe it is. The idea is the
bishop on g6 is a very powerful bishop because it ties down
black's ability to develop his king-side. If black had the
chance, he would love to play the move g7-g5 and then Bg7.
What white has done is sacrificed a piece for a long-term
initiative. Okay, what I mean by that is the initiative means
the ability to make threats. Deep Blue has an ideal attacking
formation, quick development, the king is safe. Black has all
kinds of problems. How is he going to develop his army? The
idea of the move Bf4 is, if we go back for a moment and we
consider another possible idea for white, is what Garry would
like to do is play Nf6-d5, followed by Qe7-f6, getting out of
the way of the bishop on f8 and trying to get rid of that
bishop on g6.
So the idea of Bf4 is to immediate Nd5 with Bg3, when Qf6 is no
longer possible because of Bh4, winning Garry's queen and the
game.
MAURICE ASHLEY: How is this possible, Yaz? I know we often try
to play different openings to fool our opponents, but how could
the "best by test" in the world, Garry Kasparov, make such an
academic blunder? I'm trying to understand it. This guy's
knowledge is encyclopedic, much less -- for him to play so
basic and so wrong.
YASSER SEIRAWAN: And indeed Garry spoke about that yesterday.
He spoke about his memory as being one of the best chess
memories in the world. One of the things that -- and in fact I
find most upsetting about this particular position is, if Garry
Kasparov were to lose today's game, it's entirely conceivable
this whole sacrifice and so on is just in Deep Blue's library,
opening library, and it's done nothing -- it may turn out it
won't even have to play an original move if Garry chooses one
of the variations that it has been programmed as a win for
itself. Which would be very unfair, not only to the Deep Blue
team and its research, but to Garry Kasparov as well, because
all he's doing is losing to analysis by his own colleagues.
MAURICE ASHLEY: To be fair, though, Garry did not have to choose
an opening that he doesn't normally play. I know a friend of
mine, whenever I go into a big competition, my mentor, fellow
name Willie Johnson, always says, "Maurice, play what you
know." And it's good advice. It's served me in good stead.
You go into situations that you're familiar with, instead of
going into something new and you don't know what's going on,
you start thinking for a long time as Kasparov is now. He's
shaking his head. He's perturbed already. Already, the game
just started. We are only on move 11, and Kasparov can
normally whip off 15 moves in a flash, we're on move 11, and
he's suffering already for no good reason.
MIKE VALVO: You know, and it's exactly the kind of position that
he didn't want the computer to have.
MAURICE ASHLEY: How so?
MIKE VALVO: Wild, complicated, tactical. He just didn't want
this. And I wouldn't say that Garry's forte is defensive
chess. He's a good defender, but he's a much better attacker.
MAURICE ASHLEY: Right.
MIKE VALVO: And the thing that occurred to me is the two games
where we had extended opening lines, game two and this game, in
both cases he used Anatoly Karpov's lines, which is kind of
strange. Why not just play c5, like Yasser said, play your own
stuff, go with what got you there, as you say.
YASSER SEIRAWAN: What brought you to the party.
MIKE VALVO: That's what we came to see. We didn't come to see
him trying to trick the computer. We came to see him take the
computer head on, and I had hoped that he would do it in this
final game, he would realize that up to now it hasn't work,
this is the time he has to do it, he's pulled all his tricks,
now let's go with our strengths, your strengths to my strength.
MAURICE ASHLEY: Is he so terrified by Deep Blue and what it
might know that it's just completely thrown him off his game?
YASSER SEIRAWAN: It seems --
MIKE VALVO: It seems so. He said yesterday "I'm not afraid to
say I'm afraid." You said are you going to play powerful in
the last game, and he said, "I'll play the best moves that I
can."
MAURICE ASHLEY: Garry is known for his bravado --
GK MOVE: 11...b5
MAURICE ASHLEY: He's going to put me at a loss for words in a
minute, and that's not easy to do. b7-b5 has been played.
b7-b5, Yaz.
MIKE VALVO: Could we see the computer's clock? I want to see if
this has all been book.
YASSER SEIRAWAN: I think the computer has played very
quickly, --
MIKE VALVO: I was wondering if they were out of book.
MAURICE ASHLEY: Deep Blue has used only six minutes to play 12
moves. Kasparov has used 15 minutes, but the time spent was
really on only one move and that was the last move, b7-d5.
Every other move he actually played quickly. But h6 took him a
little time to play, a little time, and then suddenly the
sacrifice, Nxe6?
A. This mover, Yaz, b7-b5, what's this about?
Seems to me to develop -- he should be wanting to develop
pieces, not push pawns.
YASSER SEIRAWAN: What Garry has to do obviously is develop his
pieces. If you look at his army, the two bishops on the back
row, the two rooks, the misplaced black king means that Garry
has to make a great deal of repairs to his position. I had
just put on the board the move Nd5 that comes with tempo. Then
the idea was Bg3 Qe7-b4, trying to develop the queen, preparing
the knight retreat Nb5-e7. He wants to make sure the knight
when it comes to d5 stays there, and there will not be the move
c2-c4.
So what he did with this last move, b7-b5 is to try to secure the
square d5 for his knight so that he won't have to worry about
the move c2-c4.
MAURICE ASHLEY: But, Yaz, can this position tolerate more pawn
weaknesses? His king on D eight won't find a home for a very,
very long time. Of course the point of the game is to
checkmate the king and the best thing you can do is have pawns
behind it. These pawns look suspicious as defenders of this
king on d8. I don't even see how the king will ever get back
to the king-side. This could be a long trek indeed. b7-b5
seems a little loose to be playing in this kind of situation.
YASSER SEIRAWAN: Well, Garry's got problems, and -- (Audience
laughter.) Garry does have problems in this position. The
first thing he's got to do is address his development. He
hopes that he's not going to get run over in the center and his
king. For example, we could try to understand Garry's hopes
after the move Ne5 by Deep Blue. In general it's a bad idea to
trade attacking pieces for defending pieces, so the move Ne5 is
an unlikely choice. It does, however, threaten Ne5xc6
checkmate. So that after Nxe5 again we would see an unlikely
choice in dxe5.
DB MOVE: 12 a4
MAURICE ASHLEY: Before that could even settle itself as a
positional concept, Deep Blue has instantly moved, seeking to
undermine the b5 pawn with the move a2-a4, and that pawn will
have to make a decision sometime soon.
GK MOVE: 12...Bb7
MAURICE ASHLEY: Instead Kasparov has left the pawn on that
square and has moved quickly with Bc8-b7. And now Kasparov is
moving a little bit quicker than one would suspect in such a
dangerous situation, Yaz. Bc8-b7 was probably seen by Deep
Blue in response to its a4 idea. I guess he's just settled
down, I guess he's just figured, well, I made a mistake, got to
live with it, let's play chess.
YASSER SEIRAWAN: Exactly. In this case he's going to have to
live with it, so he's already reconciled himself to that.
The idea of a2-a4 is the struggle. White wants to knock the b5
pawn out of the way so that c2-c4 becomes a possibility. At
the same time, the move a2-a4 will open up the rook on a1 into
active service, and the rook will get drafted.
MAURICE ASHLEY: It's interesting to me, though, Yaz, what's very
interesting about this position is that, if we look first at
Fritz, the way Fritz is looking at the position, Fritz says
right now that black only has a .28 advantage, .28 of a pawn.
Now, one point is one pawn, that's 1.00 is one pawn.
Now, mathematically, what's been programmed into Deep Blue is
that a knight is worth three pawns, 3.00 and a pawn is one
point. Now simple subtraction, that means that right now,
black has the advantage of two pawns.
Now, we see the compensation immediately. The bad king, the
developed pieces for white. We see that white has tremendous
compensation and could work to try to win the position.
But what if Deep Blue sees the material disadvantage and thinks,
for example, "Maybe I can win the e6 pawn back and start to
equalize material somewhat, and maybe just play like Qe2, gang
up on the e-pawn, and if Kasparov gets Deep Blue to take this
pawn, which is what Deep Blue might want to do considering it's
down material, he might suddenly be winning the game.
YASSER SEIRAWAN: How many of us think that?
MAURICE ASHLEY: My point is this. What would prevent, Mike,
maybe you can answer this question. What would prevent Deep
Blue from seeing the e6 pawn and just taking it if Garry leaves
it that so that it can get closetory redressing the material
imbalance? After all this sacrifice it played was not played
on its own, on its own volition, it was programmed in. Maybe
by now Deep Blue is thinking when the new moves started on the
board "Who sacked my knight?" (Audience laughter.)
YASSER SEIRAWAN: Good point.
DB MOVE: 13 Re1
MAURICE ASHLEY: Re1, eyeing a weakness.
YASSER SEIRAWAN: Which is what with you just pointed out. Maybe
he wants to play Qe2 to gang up on this --
MAURICE ASHLEY: In fact Qe2 in this position practically wins a
pawn because it attacks the b5 pawn and the e6 pawn. That
would actually show a flaw in Deep Blue.
MIKE VALVO: I would like to address this. I think in game two
we saw evidence of reasons why Deep Blue will not settle for
just winning that pawn. Remember that it didn't play Qb6?
YASSER SEIRAWAN: Correct.
MIKE VALVO: It could have won two pawns but it didn't do so
because its king would be exposed.
Now, it's aware of the other guy's king being exposed, too, and
it won't settle necessarily just to win a couple of pawns when
the king's exposure is worth more to it. I don't think it'll
settle for a couple of pawns.
Most computers would. I don't think this machine will.
MAURICE ASHLEY: Well, that would be disastrous indeed if that
did occur and it would show a flaw in the computer's estimation
and valuation of the position.
MIKE VALVO: If it did.
MAURICE ASHLEY: If it did. At the moment Deep Blue has played
Re1. Kasparov is thinking about how to finish developing his
pieces. It's a very tricky task indeed. We should say to our
in-house audience that we love it for you to participate. We
will be sending ushers around with microphones to ask several
questions, and we will do our best to answer those questions.
MIKE VALVO: Who is going to win?
MAURICE ASHLEY: We would also just like before we do that to
welcome some students who have been invited by IBM -- every day
IBM has given tickets for students to come and watch the games,
and we have people from everywhere. First, I would like to
introduce the ridge way public school from White Plains, New
York. We give them a plan. -- we give them a hand.
Who's champ of Ridgeway? Who's the best player in ridge way?
Two hands went up. Can we get a microphone over to the two
youngsters? I see two hands, so I'm going to have to get two
names. What's your name?
AUDIENCE MEMBER: Oscar.
MAURICE ASHLEY: And next to him is also the champion of ridge
way?
AUDIENCE MEMBER: And the nation.
MAURICE ASHLEY: They are --
AUDIENCE MEMBER: National champions.
MAURICE ASHLEY: They are national champions? (Audience
applause.) Who do you think is going to win today?
AUDIENCE MEMBER: Deep Blue. And Oscar, what do you think?
AUDIENCE MEMBER: Kasparov.
MAURICE ASHLEY: All right! Okay, we've got two sides here.
All right, we would also like to welcome Port -- are they from
Port Washington school district? Port Washington?
We also have kids from all over the Port Washington area, not a
particular school. We would like to welcome you for coming.
And as a group, who thinks Kasparov is going to win? Let me see
some hands. And who thinks Deep Blue is going to win? Just
one lonely camera. Okay.
And we also would like to welcome community elementary school 70
from the south Bronx Bronx. Are you here? One of program
C.E.S. 70, they have a strong coach, David McNuety. They have
won many, many titles. Who is the champ here today?
YASSER SEIRAWAN: Two hands.
MAURICE ASHLEY: No consensus today? All right, still strong
players. Welcome to students. We love it when kids come to
watch. After all, they are our greatest fans. So welcome.
(Audience applause.)
YASSER SEIRAWAN: By the way, maybe you can help me on my chess
history. We talk about Garry Kasparov as being the 13th in
line of world champs. We go back to the original world champ,
or the original recognized world champ, Wilhelm Steinitz, and I
believe he was in a banquet with Zukertort, and it was a
closing banquet of a great tournament, and the master of
ceremonies had said something to the effect "Now, please, an
applause for the world's best chess play," and both Zukertort
and Steinitz stood up. And here we have it again, the best
school player.
MAURICE ASHLEY: Well, we have Kasparov deliberating on the
position trying to figure out exactly what he was do. It's not
easy to make a decision here because so much is going on, Yaz.
YASSER SEIRAWAN: And he's been pulling some strange faces,
Garry. Garry is not a happy camper.
MAURICE ASHLEY: And he's known to pull those faces, too. We
like to watch Kasparov because he is so expressive. We can
count on him to let us know exactly what he thinks about the
chess position practically at all times.
We would like to take questions now from our audience, and we
have a question over here to our left.
AUDIENCE MEMBER: Yes, hi. Perhaps Mr. Valvo when he goes off
stage could inquire, it's curious because I was wondering if
you could address the possibility that perhaps Joel Benjamin
prepared this entire line in advance as a potential cook of the
Caro-Kann?
MIKE VALVO: Well, there's two possibilities. First of all, they
have pretty much but in -- put in all of the current games of
Grandmasters, and it could be part of that. That's why I was
curious about the Bf4 move was part of their book or not.
AUDIENCE MEMBER: Right.
MIKE VALVO: And of course there's the possibility as you suggest
that it was prepared in advance and it was analyzed, but that
Bf4 delay bothers me. It makes me wonder if they actually did
go into it deeply or if the game that they followed went a
different way at some point. And I intend to go up and talk to
both Patrick Wolff and to whoever I can find. I'm about to
leave and find out some information.
AUDIENCE MEMBER: That would be very interesting to me because of
course the question then becomes, if this was a line that was
essentially forced upon Deep Blue, you folks alluded to that
before, if it got this nice positional improvement, it may not
realize it and may then revert back to a more materialistic
strategy and allow Garry to equalize. You know, that's why I'm
curious if you folks think this has been fixed on Deep Blue or
if it's really been seeing this advantage all along.
YASSER SEIRAWAN: Well, what you're saying is the sword cuts both
ways. If in fact Deep Blue wins today and it wins solely
because Joel Benjamin or other Grandmasters have prepared the
game, and Garry Kasparov is an unfortunate victim, that's one
way the sword cuts.
The other way the sword cuts is, like the gentleman suggests, the
game hasn't been analyzed by Deep Blue to the nth degree, it's
been given a position that it doesn't like, spoils it, and it
says, "Hey, I didn't lose the game. My jerks who prepared me
for it did." (Audience laughter.)
MIKE VALVO: Also considering the fact, the possibility that
Garry did this deliberately. You know, Garry is a great expert
on the white side of this line. It's amazing to me, he -- to
me that he doesn't know about this Nxe6 and that he just fell
into it.
MAURICE ASHLEY: I am sure he knows about Nxe6. The question is
whether or not he just kind of transposed moves in the heat of
the moment and allowed it to happen. In fact, after Nxe6 it
was clear from his expressions, he shook his head, and played
his response instantly. It's not as if he played Nxe6 and said
oh, where did that come from. He immediately played Qe7,
because he knew that to be the only move in that position. The
way he's acting, he'd have to be an academy award winning actor
to be pretending that he's not upset by this position, getting
into this situation. I'd be shocked if he came out and said,
"Yeah, of course, I had it all prepared." (Audience
laughter.)
YASSER SEIRAWAN: Well, there's no reason to pull faces in front
of Deep Blue. It's not going to work. (Audience laughter.)
And I agree, Garry Kasparov has not won any Oscars, but the
reality is that Garry Kasparov is a very passionate person, he
does reveal himself openly, and I really think that h6 was
simply a failure.
MAURICE ASHLEY: Well, Mike we know it is your break so we'll let
you get the low down on the situation.
We're talking today because someone asked Joe, how do you feel
sitting across from Kasparov? And his reply was "I was
terrified." So what Garry does does work to some extent.
MAURICE ASHLEY: Well, it made him not touch the right bishop
earlier!
MIKE VALVO: He almost moved the wrong piece.
MAURICE ASHLEY: He almost moved the wrong piece.
MIKE VALVO: But, you know, there's an old rule in computer chess
that computers are not responsible for human error. I'll leave
you with that thought. We will thank our colleague, Mike
Valvo.
YASSER SEIRAWAN: We'll take questions from the audience in just
a moment.
MAURICE ASHLEY: Yaz, I'm looking at this position, and I'm
thinking I like white, but I would hate to see Qe2 and Qxe6.
That would just ruin everything. But there are other threats,
aren't there? I mean isn't a move like Bf5 to be thought about
here? White doesn't have to do anything agrees at the moment.
YASSER SEIRAWAN: No, exactly. He can just sit on the position.
It's not easy for black to decide what he's going to do. Black
could make a number of quiet moves, including moves like
Qd1-Qd2, menaceing Qa5+. Qd3, coming over to b3 to hit the e6
pawn. There are a number of really juicy attacking options
here for white. And Garry is going to be under the gauntlet
for a long time.
MAURICE ASHLEY: The worst part is this bishop on f8 and his rook
on h8. And they're just wondering when they ever will get in
the game. I mean this king completely spoils all harmony.
This queen blocks this bishop. This bishop on b7 wondered who
put it there behind this pawn.
I mean the pieces just look horrible. How could Garry Kasparov
play that? It's almost sick looking, this position.
YASSER SEIRAWAN: Well, the only thing that could be said in his
defense is he's got a piece for his troubles. I mean it
counts, it counts.
Let's take some questions over on my left side, please.
AUDIENCE MEMBER: Does Qb4 win a pawn?
MAURICE ASHLEY: You want to win more stuff? Qb4, does Qb4 win a
pawn.
YASSER SEIRAWAN: I think the gentleman is thinking this is a
double attack against the pawn on a4 as well as the pawn on b2,
and indeed this would provoke white to do something on the
queen-side. However, Qb4 also leaves a pawn hanging.
AUDIENCE MEMBER: (Gasp.)
YASSER SEIRAWAN: Can you recognize what you've done? That e7
pawn would allow that rook to come crashing down, and just to
show you some of the problems that might occur is after Rxe6,
if you were to play, for argument's sake, b5xa4, then after Qe2
you've got to pay attention to Re8 mate. You say that?
MAURICE ASHLEY: They say that. They're national champions.
They see everything.
So, Qb4, it's doubtful Kasparov will try to make a move like this
now. The problem is, Yaz, and, you know, you haven't said it
yet. You've been making these wonderful generalizations,
evaluations about the position.
YASSER SEIRAWAN: Thank you.
MAURICE ASHLEY: But you don't want to say what black should do,
do you, because it doesn't look like black has many moves
here?
YASSER SEIRAWAN: Well, indeed, Garry's options are quite
limited. I would reduce them to just a few candidate moves. I
noticed that Fritz says that there are 31 legal moves in the
position for black. And let's just talk about our little
toolbar here. This is a very nice visual aid. What we see
here is that -- okay, this is Fritz 4.01. Next to it is this
little equals-over-plus sign and this minus .44. What that
means is, in the view of Fritz, black has a slight advantage.
Equals over plus means a slight advantage for black. But the
numerical value of .44 refers to its material bias. It thinks
that 1.00 is worth a pawn. So when it's .44, it thinks 44
hundredths of a pawn, the position is better for Garry
Kasparov. That is the view of Fritz.
Below that we see Fritz's suggestion. Fritz says that the move
bxa4 is Garry's best move, and this is a horrible move.
GK MOVE: 13...Nd5
MAURICE ASHLEY: Kasparov has played Nf6-d5, centralizeing the
knight, and, and Deep Blue had anticipated this response, and
has immediately --
DB MOVE: 14 Bg3
GK MOVE: 14...Kc8
YASSER SEIRAWAN: I think that these were the right moves for
Garry, by the way. He didn't have many options. If we just go
back a couple of moves after for example Re1, what really could
he do? If he moves his queen, as we've seen like b4, then he
loses the e6 pawn. The Bf8, the rook on h8, they can't move.
If the knight on d7, for example, was to move to b6, Nd7-b6,
well, this would allow white to bring his knight very
powerfully with Nf3-e5. There's the fork on f7. So, in a
sense, Garry's options, defensive options were extremely few
and far between, and the move -- one of the things I have to
say about the move Nf6-d5 is at least it's consistent. His
whole idea was to play b7-b5, keep the knight on d5, and fine,
he's established it.
We saw the response Bg3. Again, this is a little bit of a
problem because of the move Bh4 could make life very unpleasant
for black. And Garry played Kd8-c8. Again, a good move.
Because this makes room for his queen. I think Garry is
anticipating that he's going to have to give up a second pawn.
I think Garry is getting himself ready for the moves either
Qe7-d8 so he can develop his king-side, or at least Qe7-f6,
because we know that bishop on g6 is just so powerful.
MAURICE ASHLEY: And he has shown that there is a way to unravel
the position a bit. He's planning to develop -- there's still
the long-term problem, Yaz, of this rook on a8. This rook on
h8, as you said, the queen is ready to move. It could drop
back to d8. This pawn, though, still has to be watched.
He could also think about a more aggressive posture, like Qf6.
Then the bishop on f8 would come out and the rook on h8 would
be able to come out. That still would not solve the rook on
h8's problem and hopefully in Kasparov's mind the development
of the forces for white will not reach proportions that will be
destructive to him. So Garry is trying to solve his problems.
What else can you do?
YASSER SEIRAWAN: Exactly.
MAURICE ASHLEY: You've got a position, you have to play it. So
he's going to show his human fortitude and tenacity, and we can
only hope that he doesn't get blown off the board.
YASSER SEIRAWAN: While showing his fortitude and tenacity. A
question from the audience, please?
AUDIENCE MEMBER: My question backtracks to move 1 both yesterday
and today. Yesterday's Indian opening and today's Caro-Kann
showed that Kasparov is choosing his best feel for the
position, it seems he's a better positional player than the
computer. So deciding on an Indian opening with the Caro-Kann
is a conscious decision outlining a strategy.
Now, what kind of terse move does the computer? Does he operate
entirely at random, or when the computer makes its first move,
like today, a king pawn move, he wants open play, its combative
moves. What makes the computer make that option? Is it
entirely option? Can it be programmed at level one?
MAURICE ASHLEY: Oh, most definitely. And Deep Blue -- they have
chosen e4 -- the programmers have, because it is their belief
that it will lead to the kind of positions that Deep Blue will
be able to use its skills the best in. It's not that Deep
Blue -- obviously not what Deep Blue wants or what Deep Blue
favors, because Deep Blue can't do any of that, but they know
that -- at least they feel that with the kind of ability it has
to calculate so many moves per second, which is much, much
better than what Kasparov can do, or any human can do, that
those kind of positions with -- where those complications can
occur will certainly favor Deep Blue, and it should lead toward
those kind of situations.
In an opening like d2-d4, that could lead to several blocked
positions and computers historically have shown that they're
not really very good at blocked positions. Maybe Deep Blue is
an -- has an improvement on previous computers, but even though
it would prefer to go into situations that historically have
shownoids -- otherwise.
Well, Kasparov --
YASSER SEIRAWAN: Just to buttress that point, in both
Philadelphia and in New York, every game that Deep Blue was
white it's chosen e4. So that's not random.
We have another question over on our right side.
AUDIENCE MEMBER: Just a comparison with last year's match, in
game 6 Kasparov trapped Deep Blue's rook and bishop in the
corner, and now it looks like the tables are turned.
MAURICE ASHLEY: Very good point indeed.
YASSER SEIRAWAN: The gentleman recalls for us that in the sixth
game of the Philadelphia match, Garry Kasparov won decisively
by outmaneuvering the computer and forcing its rook passively
in the corners, and here we have something similar.
Just looking at some random variation in the game.
Following the move by Kasparov of Kc8, I'm looking at the move 15
Qe2 Qf6 Qxe6, what Deep Blue might play because it wants to win
its pawns back, its material back. Qxe6 Rxe6 and now because
of the threat of Re6-e8 checkmate, I was just looking at Nc7, a
further sacrifice this, time not of a piece but of a rook.
Ra1-e1 Nxe6 Rxe6.
Kind of a crazy position arises. I mean white's a whole rook
down. I think he may have a pawn or two for it. But this
threat of Re8+ is quite powerful. For example, Nd7-f6 gives us
this opportunity for Bg6-f5, setting up Re6-e8 double-check
mate. Not just one time, but two times. And you played king
out of that checkmate with Kc8-d8 now we follow up with Nf3-e5,
and again similar threats of Nf7+.
So kind of an intriguing way of winning a pawn and continuing the
attack. And this might be what Deep Blue will do, especially
if Garry has chosen a defense that wasn't preprogrammed.
MAURICE ASHLEY: Well, that is interesting, and it just shows, it
just goes to show that the evaluation -- that an exchange of
queens alone --
DB MOVE: 15 axb5
MAURICE ASHLEY: Because the exchange of queens it seems there
may be some attacking chances still.
Deep Blue has captured instead axb5, and Kasparov has recaptured
cxb5.
GK MOVE: 15...cxb5
MAURICE ASHLEY: We should note that Kasparov has a dressing room
in the back and he often retires to it. There is a television
in there so that he can see if a move has been played on the
board. So he likes to go back there either because of nervous
tension or just give himself a break from the chessboard. Now,
Yaz, there are a number of moves in this situation for Deep
Blue, different ways to carry out the attackment at the moment
he could actually win back a pawn it seems with the move Qe2.
But you're exploring a different variation here, the
possibility of Qd3 seems to attract you.
YASSER SEIRAWAN: Well, yeah, this is -- first of all, I like
this move axb5 in some ways because the computer hasn't just
gone ahead and played Qe2 in order to win this pawn on e6. It
seems that Deep Blue recognizes that it wants to keep queens on
the board, so that already shows sophistication, a
sophisticated understanding at least of this position.
MAURICE ASHLEY: Well, Qe2 could still be played -- sorry to
interrupt -- it's still possible. And this makes it clear that
he wins the pawn back.
YASSER SEIRAWAN: But then the idea would be Deep Blue would say,
"Well, the move axb5 cxb5, that trade, that exchange actually
helps black in some ways because that bishop on b7 is no longer
hemmed in by that pawn on c6. So when you're attacking and
especially a direct attack against a king, one of your
principal ideas is to open the position. By opening the
position we mean forcing the exchange of pawns. So we're just
looking now at a possibility of Qd1-d3, attacking the pawn on
b5. Garry may play the move Qe7-f6, but I'm just wondering
what would happen after a7-a6 c2-c4, and then we can see the
effect of what Deep Blue wants to do.
bxc4 Qxc4+, Garry's king is in deep trouble, deep trouble.
And this is exactly what Deep Blue wants to do is open up the
position for his attacking pieces.
MAURICE ASHLEY: This is not the kind of position Kasparov is
going to want to play, and he'll have to figure out a way to
keep the lines closed and not have to come under the attack of
Deep Blue's converging pieces.
YASSER SEIRAWAN: We should also talk about the three-minute
rule -- and it's not a rule, so let me correct myself -- but
Deep Blue -- the time control is 40 moves in two hours, meaning
that Deep Blue has three minutes, on average, per move. So the
programmers have said -- the programmers have said --
DB MOVE: 16 Qd3
YASSER SEIRAWAN: -- that Deep Blue should play a move every
three minutes and the computer should think on Garry's time as
well. This ensures that the computer will never lose on time.
So when the computer played the move axb5 cxb5 came from Garry
right away, we can expect the computer to play within three
minutes.
MAURICE ASHLEY: And it has done so.
YASSER SEIRAWAN: It was getting close to the three-minute mark.
MAURICE ASHLEY: Well, Qd3 shows Deep Blue avoiding the forced
win of material. Very big point, was my concern earlier that
Deep Blue might have played the move Qe2, seeing the attack on
the e6 pawn and seeing the attack from the b5 pawn. That's
materialistic.
That's as far as we understand, typical for computers. They see
a pawn, they want a pawn, why not, it's a pawn.
This move is very atypical. Qd3, Yaz, I dare say, it's a very
human move, showing a complete understanding, complete
understanding of the situation, and it knows -- forget the
material -- that's going to lead to bad positions, I'll end up
losing, I don't want the pawn back. I'm going to mate your
king. That sounds nice.
YASSER SEIRAWAN: It does. But I don't like the -- I think it's
just a sophisticated understanding. And we saw Garry just a
moment ago very sternly shaking his head and not a happy
camper.
MAURICE ASHLEY: It has surprised him with so many decisions. I
mean we have to really give kudos to the programmers because
they have this computer playing such wonderful chess strategy.
Game two was indeed a masterpiece of chess strategy, and you
yourself said that's the best game you've ever seen a computer
play.
YASSER SEIRAWAN: Absolutely.
MAURICE ASHLEY: And just time and time again it plays these
moves, makes these decisions that you just really have to be in
awe of the work the programmers have done to get it to play
chess on this amazing level.
YASSER SEIRAWAN: Absolutely. Kudos to IBM and its staff.
Absolutely.
(Applause.)
But the thing that's making Garry annoyed is that in his practice
match strategy, all of the computer specialists that were
helping him have told him that he can expect a computer that
will go after material. And when you see a decision like Qd3,
Garry shakes his head and he says, "Darn it. The computer is
not playing like a computer, and I want to see the printouts to
make sure that there's not any intervention, because this
computer is playing too darn good."
MAURICE ASHLEY: And he needs to be surprised at this moment. It
looks like he hasn't been able to get his composure back
because of it.
YASSER SEIRAWAN: Well, game two came as a shock, but we do have
a move now by Garry.
GK MOVE: 16...Bc6
MAURICE ASHLEY: Defending his pawn, and again getting up from
the board, walking away and he's prepared it looks like a
little cubbyhole for the king on b7.
YASSER SEIRAWAN: Nice, nice.
MAURICE ASHLEY: Which would solve the problem we're talking
about of the a8 rook and slowly but surely we see Kasparov
trying to solve all the problems. And if Deep Blue doesn't
come up with a concrete plan soon, it could very easily find
itself down a piece. But in the meantime there's so many
possibilities for attack, Yaz, and again you're busy exploring
one just now. I can't keep you under control, you just want
to --
YASSER SEIRAWAN: It's exciting possibilities here for Deep
Blue.
MAURICE ASHLEY: What was that last one about? You're thinking
of maybe the move Ra6?
YASSER SEIRAWAN: Well, after the move Bc6, you're absolutely
right, Garry is preparing that nice little box for his king on
the queen-side with Kc8-b7 and I was just looking at Ra1-a6,
just going all out for the attack.
MAURICE ASHLEY: Attacking the bishop on c6, which is currently
undefended.
YASSER SEIRAWAN: Now, the bishop can't drop back, the bishop
can't go back to b7 because of that e6 pawn would be munched.
Ra6xe6.
So Garry would have to make one of several choices. He's either
have to bring his king to b7, which is the first and most
obvious choice, and then I was thinking a sacrifice, just Rxc6
Kxc6 c4.
This is probably not going to work as far as an attack is
concerned, but --
MAURICE ASHLEY: Looks tempting.
YASSER SEIRAWAN: There's some dangers in this position. For
example bxc4 --
MAURICE ASHLEY: I should just note for the moment that Fritz 4,
despite being up loads of material --
YASSER SEIRAWAN: A whole rook.
MAURICE ASHLEY: -- a whole rook, is saying it only has a tiny
advantage as black which is a concession on Fritz 4's part,
which is saying I'm only up a rook, which is five points, but
here it says it's only up a pawn, which is a certain begrudging
admitting that there are certain dangers here.
YASSER SEIRAWAN: bxc4 Qxc4+. Garry's king would have to drop
back, and --
MAURICE ASHLEY: Either b6 or b7?
YASSER SEIRAWAN: Probably b7. And unfortunately now in this
particular case, if our bishop on g6 were to assist our queen,
we could do something wonderful like Qa6 checkmate or things
like that. So these are the things that Garry has to worry
about.
And also I want to say that for the defender, the defender has to
look at so many threats that it's much easier when you're
playing against the computer to be on the attack than you are
when you're on the defense, because on the defense, you have to
keep Des Moines mind -- keep in mind everything that's
possible. Pieces swinging over, pieces coming up and down the
board.
DB MOVE: 17 Bf5
MAURICE ASHLEY: I suggested this possibility earlier but for me
I see it as a surprising move. And Yaz, you can help me here.
First of all, Deep Blue is down a piece for one pawn. It wants
to attack, more so than anything else. Recapturing the
material especially the e6 pawn is not in the program, at least
should not be in the program.
YASSER SEIRAWAN: Unless it's done under favorable circumstances.
MAURICE ASHLEY: Unless it's done under fantastic circumstances.
Here, though, it comes as a big surprise to me. Because here
the possibility is capturing the bishop and after Rxqueen,
maybe even just bishop takes, maybe. The reason being that
now, Kasparov has gotten a knight -- in this case you have to
give up the f5 pawn, but he's gotten a knight, rook, and
bishop, which is numerically 11 points. A rook is five points,
a bishop is three and a knight is three. That's 11 points.
And what he's giving up is a queen and two pawns. That's also
11 points. The queen is nine and the two pawns, right, is 11.
So that equalizes the material balance. It does bring about a
strange situation on the chessboard. But it seems to me to
solve some serious problems for him. His rook on h8 is now
ready to come in the game. The bishop on e7 is already in the
game. And after one move, Kb7, the other rook will be in the
game. Now, those problems will be solves. The question is,
how ferocious is the white queen? Will it cause too many
problems with the distribution of material? To my eyes, I
don't know. I'm really not sure. I sort of like black -- I
didn't like black at all a second ago, but now I'm starting to
feel a bit more comfortable, just a little bit.
YASSER SEIRAWAN: Well, I'll tell you, in general, it's better --
GK MOVE: 17...exf5
MAURICE ASHLEY: He has taken the bishop.
MIKE VALVO:
YASSER SEIRAWAN: And in fact he had no choice. I would say in
general, three pieces for the queen favor the three pieces. A
rook and two pieces for the queen really favors the three
pieces.
However, that's not the overall determining factor in this
particular position. First of all, Deep Blue will have two
pawns for the queen. But far more importantly, that black king
is exposed on c8. Even if it gets to b7.
Let's look at the situation we were just looking at --
MAURICE ASHLEY: I'm sorry, I hate to interrupt, but Kasparov is
doing some very strange things right now.
DB MOVE: 18 Rxe7
GK MOVE: 18...Bxe7
MAURICE ASHLEY: Another unusual thing is he's put the watch back
on. That's as cryptic as ever when he puts the watch back on.
Usually he puts the watch back on when he thinks the game is
over.
YASSER SEIRAWAN: Usually for himself.
MAURICE ASHLEY: Not right in the middle of the game. And he has
played Rxe7 Bxe7 --
YASSER SEIRAWAN: We are anticipating Qxf5 and after a move like
exf5 there's all kinds of problems. Notice the bishop on g3
covers -- the bishop can easily fall victim to what we call an
overload tactic, when one piece has to defend so many pieces,
there's always a problem that there's an overload tactic and he
may go down.
MAURICE ASHLEY: Garry does not look happy. He looks disgusted,
in fact. He looks like he can't believe what's going on right
now. And I don't really see a move. Maybe we should go back a
step in this position. Maybe there's another move for white.
Fritz, of course, is now changed its opinion and is giving
white the advantage. And Kasparov is still shaking his head.
YASSER SEIRAWAN: He almost seems to be talking to himself,
almost talking to the programmer. I mean it's not disturbing
Deep Blue, but it's scaring the heck out of me. It almost
seems to be giving up.
MAURICE ASHLEY: He's looking off stage, it looks like he's ready
to -- he's looking at someone else. He does have a coach who
is in the room with him but who cannot interrupt in any way.
YASSER SEIRAWAN: Absolutely not.
MAURICE ASHLEY: Cannot speak to Kasparov. This look of his is
certainly one, almost of resignation on his face.
YASSER SEIRAWAN: IM Michael Valvo.
YASSER SEIRAWAN: We're seeing some incredible reactions by Garry
Kasparov. He really is an unhappy camper.
MIKE VALVO: Well, there's some interesting things from
upstairs.
YASSER SEIRAWAN: All right.
MIKE VALVO: One thing I wanted to find out was, was Bf4 actually
calculated or not, but I couldn't find that answer out.
Although Jonathan Schaffer, who we talked to yesterday, who was
the programmer of the world champion computer chess program --
computer checker program said that he believes that it was
calculated. I talked to Patrick Wolff, who has played the
white side of this line and I said do you think Garry did this
on purpose? He said, "Absolutely. Garry did this on purpose.
And it was very brave for him to do."
MAURICE ASHLEY: And he should get an Oscar?
MIKE VALVO: --
DB MOVE: 19 c4
GK MOVE: Kasparov resigns.
MAURICE ASHLEY: Whoa! Kasparov after the move c4, has
resigned! It's over just like that! We should say that Deep
Blue has upon the match. IBM computer Deep Blue has defeated
the world champion Garry Kasparov in an absolutely stunning,
stunning 19 mover! And Kasparov has just simply stormed away.
We should say congratulations to Deep Blue and their
programmers. (Audience applause.)
Yaz, Fritz four now is going nuts, saying that Kasparov in fact
has a huge disadvantage. In fact, white has a winning
advantage after c4, and we will attempt to analyze this
position. I know you guys didn't come here and expect to be
out this quick, but we will try to get as much as possible
done. There will be ceremonies taking place on stage. We hope
that Kasparov will come to the stage. It will be a difficult
thing for him to do after such a loss. But there are indeed a
lot of questions. We're not prepared for this. I thought I
was going to get a break for a second, an hour into the game,
and it's over.
Yaz, what are your impressions? This is stunning. We never
expected this to happen, never, never at all.
YASSER SEIRAWAN: With all due respect, the final position, let's
just try to understand what Garry saw. What Garry saw was that
after this move c4, remember what we were talking about, that
overloaded bishop, how it would suddenly become vulnerable. If
Bxb4 -- Bxc4 Qxc4, queen mate. The whole strategy had been to
keep his knight on d5 with b7-b5. So c2-c4 disrupts his
defensive formation. I think his resignation looks still
premature. There are several moves in this position. There's
Nd5-b4 -- well, let's just see the most -- Nd5-b4.
What does Fritz see that it thinks is the best? Fritz likes to
take the pawn on f5, and he gives himself a valuation of a pawn
and a half -- a pawn and a quarter. It's growing, it's
growing.
MAURICE ASHLEY: Well, after this -- if bxc4 is not possible,
Yaz. One has to admit that black's position is going to be in
trouble very quick. Because these two pawns are not funny.
Having these two passed pawns coming down the board, you can't
like what's about to happen.
YASSER SEIRAWAN: Well, let's take a look. Okay, you're going to
play Qxf5. And I agree, I don't want to see those pawns any
more than you do, so let me take one. bxc4.
Now what we recognize --
MAURICE ASHLEY: Now there's a lot of moves. I was about to
throw Ne5 into the mix. Ne5 has different ideas. One, I could
always get the pawn back here. The threats to the bishop.
Queen is thinking about penetrating into e6 with a double
attack. A lot of threats, Yaz; a lot of threats.
YASSER SEIRAWAN: That bishop on g3 --
MIKE VALVO: And already Fritz 4 is projecting a 3.66 advantage
for white in this position, which Fritz 4 is just evaluating a
huge material loss for black and no way for him to prevent the
flood from coming in and washing those pieces away.
YASSER SEIRAWAN: Wow. Well, with all due respect to Patrick
Wolff, I think the move by Garry h7-h6 was a simple figure
failure and Garry spell into a known book trap.
We hope to get C. J. Tan and his IBM team to address us and at
the same time we'll take some questions perhaps from the
audience. Because all I can say is I am stunned. I absolutely
did not, did not expect this result.
Question in the back? Wow. Holy cow.
AUDIENCE MEMBER: I'd like to take a look at sort of the overall
Deep Blue vs. humanity thing that we were talking about on
Mother's Day to begin with. And as we know we're not as fast
as cheetahs, we don't do math as fast as Deep Blue does. I
don't see any intelligence here in Deep Blue. I see a lot of
heuristics, I see a lot of pattern recognition. I don't see
anything here that says that if Kasparov had 30 games or 40
games, he wouldn't figure out a way to win consistently.
So, should we really be worried about if there's an intelligence
shown here?
MIKE VALVO: Well, I don't think they ever claim intelligence.
They didn't say that --
AUDIENCE MEMBER: Right, but that was on you humanity aspect, the
last bastion about what makes this human is under threat.
MIKE VALVO: I think you're implying if machines had intelligence
we'd be in trouble in some fashion.
AUDIENCE MEMBER: So maybe the Deep Blue team -- if Deep Blue can
just sit in the corner and get smarter by itself, then we'd all
be in trouble.
YASSER SEIRAWAN: I'd just like to say that for myself, I agree,
there wasn't any claims by IBM that it was showing
intelligence. But from my perspective it is absolutely
stunning to me that Deep Blue wins this match. I thought it
was going to be years and years and years into the future. All
I can say is what I just witnessed is a landmark achievement in
the history of computers.
MAURICE ASHLEY: But Yaz, Yaz, what happened we just witnessed?
To me, I have a slightly different spin on this.
If you show this game to any Grandmaster on the planet, any
Grandmaster and say, "Who do you think is playing black," not
one would say Garry Kasparov.
YASSER SEIRAWAN: Right.
MAURICE ASHLEY: To me it seems that this goes beyond just
whether or not the computer is intelligent. We really just
witnessed a human failing here, and it touches -- as it touches
even someone as great as the Michael Jordan of chess. This guy
has fallen apart in front of ghosts, practically. There was no
reason for him to play chess like this. He never plays chess
like this. Do you remember a 19-move loss by Garry Kasparov
just blundering right from the opening?
It doesn't happy. And clearly he's been rattled. He's been
rattled by the computer showing its ability to play chess. Why
need he be so rattled? Why not just -- we've said this this
whole match. He was messing around in all kind of random
openings instead of playing his thing, instead of having
confidence in why he's the greatest human chess player on the
planet. And he's let the computer throw him off his game.
MIKE VALVO: May I jump in here a little bit? Yasser accepted --
Yasser excepted, of course, people that have a lot of
involvement with computers that play them quite often begin to
respect them more and more so that their play actually gets
worse. I'm a living example. (Audience laughter.)
YASSER SEIRAWAN: Your study of computers --
MIKE VALVO: I'm just wondering if Garry started seeing ghosts
off his own mind and he was reacting to them, instead of what
was in front of him. I think he should just be Garry and play
Garry against this machine. Instead, he played like Karpov.
MAURICE ASHLEY: Don't tell Karpov that he played like Karpov in
this game! (Audience laughter.)
YASSER SEIRAWAN: Well, I have to agree that Garry came to this
match with some very strange ideas, strange preparations. We
didn't see the Garry Kasparov we all know and love. Instead we
saw some different version which says, "This is my anticomputer
version," and the anticomputer version just didn't look very
good. With all respect, however, I mean with the exception of
this game and game two, four of the six games he held an edge
he just couldn't win them. I must say that for myself I'm
still so stunned, it's very hard for me to absorb what this
means.
The gentleman's point is that "Hey, look, we're not as fast as
cheetahs and blah, blah, blah, we're just going to have to
accept that machines can catch up with us in some areas of
intellectual pursuit."
Yeah, yeah, yeah. It still bugs me.
MIKE VALVO: I feel cheated somehow. I feel like --
YASSER SEIRAWAN: Well, especially the game is over in one hour.
I mean, gee, I haven't even gotten a break!
AUDIENCE MEMBER: First of all, congratulations to the Deep Blue
team.
YASSER SEIRAWAN: Yes. (Audience applause.)
AUDIENCE MEMBER: And my second question is, to my eyes, when
Kasparov gave up his queen, I'm not sure what move that was,
where he took the bishop. At that point I said he's lost. And
I'm wondering if -- what your conclusion is at that point.
MAURICE ASHLEY: Before we go there, we have a shot of Garry
Kasparov in the press room.
AUDIENCE MEMBER: Looks like a DMV photo.
MIKE VALVO: Does that look like a mug shot or not?
MAURICE ASHLEY: This has to be the lowest moment in Garry
Kasparov's chess career.
YASSER SEIRAWAN: Garry Kasparov is now on the 50th floor. He is
in the press center. He is going to be addressing a press
conference. I believe that we will be able to get a direct
feed to hear some of the comments and questions. Garry is of
course shell-shocked.
MAURICE ASHLEY: He looks completely devastated.
This has to be a total low point in his career, and now Monty
Newborn of the ACM --
MONTY NEWBORN: We're waiting for the Deep Blue team to arrive
and they should be upstairs at -- in about five to ten minutes
at the most. So please be patient for a few minutes.
MAURICE ASHLEY: So we will be able to hear the press conference
upstairs as Kasparov is sitting, not saying anything to anyone,
but questions will be addressed to him, and no doubt he will
respond. He has represented himself quite well this match.
YASSER SEIRAWAN: Just to go back to the gentleman's question, he
said did Garry lose the game at the point that he won Deep
Blue's bishop?
Obviously Garry is facing at the moment that he took the bishop
on f5 the threat of Re1xe6. That is a devastating threat,
because then it would have a double attack against the queen on
e7 and the bishop on c6. At the time that Garry took the
bishop, I thought it was forced, that is, I thought he had to
give up his queen. I don't know --
MAURICE ASHLEY: But there's no physical way to defend that pawn
without creating some kind of --
YASSER SEIRAWAN: No "physical" way? Ha-ha.
MAURICE ASHLEY: I'm just saying how do you do it. I'm just say
Nc7 is one way you could try. But it seems as if Bxc7 Kxc7 and
still maybe Rxe6 is very strong to me.
YASSER SEIRAWAN: I agree.
MAURICE ASHLEY: Just as a quick possibility.
AUDIENCE MEMBER: I was just thinking, to my eyes, when the
queen -- you mentioned that black does have compensation for
the loss of his queen. But to my eyes, I see that in that
book, that's like a Cuisine Art position for the computer to
play. There's too many tactical possibilities for Garry to
hold.
MAURICE ASHLEY: You're right. And I think the point was for
us --
AUDIENCE MEMBER: The Deep Blue team has arrived on stage.
MAURICE ASHLEY: The point was for us we thought Deep Blue was
materialistic, and it was in fact materialistic when it played
c4 and realized it won more stuff. And of course Kasparov's
resignation.
So now the Deep Blue team has arrived. C. J. Tan is sitting next
to Kasparov, and he's being miked up, I believe. But there
will be some ceremonies -- ceremonial duties performed. We do
have a winner.
MONTY NEWBORN: I would like to extend my thanks to the many
people that have been involved in this exciting, historical
event in the history of computing. (This is the post-game
press conference.) I would first like to discuss our stage
agenda for the next ten to 15 minutes and then go on to thank
these people. I would like to -- I'm going to thank a number
of people. I'm then going to call on Dr. Tan. I'm then going
to call on Garry Kasparov. We will then award the prizes. Joe
Deblaz of the ACM will award the prizes and then there will be
a question and answer session from the press.
As I said, there's many people to thank for this exciting
historical event. Certainly the audience downstairs and the
press is one of the largest press events in the history of
computing. First and foremost to thank is Garry Kasparov for
being willing to compete in this historical event as the
greatest chess player in the history of the game, he is
extremely -- he has been most gracious to participate. It is
an be event that we hope -- (Audience applause.)
Garry has been assisted by two people in particular. His coach,
Yuri Dokhoian. Yuri, are you here?
GARRY KASPAROV: My team doesn't need a presentation. Just the
Deep Blue team.
MONTY NEWBORN: The Deep Blue team, the head of the Deep Blue
team is Chun Jen Tan, C. J. for short. C. J. has been the
leader of the project for the last five years and has brought
together one of the outstanding scientific teams that we've
seen to date. It's my pleasure to welcome and congratulate
C. J.. (Audience applause.)
I'll introduce your team, C. J.. It's my pleasure. Your team
consists of Joe Hoane. Joe?
I think a little quick perspective of each of these, of the
responsibilities. Joe is involved in the multiprocessor side
of the program. The Deep Blue program runs on several hundred
computers, and Joe is the one that coordinates the activities
between the computers, primarily.
Murray Campbell. Murray, would you like to stand? Joe, why
don't you stay standing. Murray Campbell is a is a Canadian.
He has been involved in the critical aspect of the scoring
function in this program in particular and much of the testing
that's been done. This program is tested and tested, and it
needs a strong player who understands what the mistakes are.
Of course by this point Murray is getting help from even
stronger players, but Murray is the one that's the chess expert
in the bunch.
Jerry Grotte. Jerry? Jerry is the one that makes sure that the
day to day problems associated with the computer are ironed
out. Somewhat of a technical expert with the hardware of Deep
Blue.
Joel Benjamin. Joel is the real chess expert for Deep Blue, not
to take anything from Murray. Joel is an International
Grandmaster and has been one of the top chess players in the
United States for a decade, at least, now.
Last but not at least is F. H. Hsu. F. H., if you were involved
in baseball, would be the winning pitcher on the team. F. H.
is the whiz behind the hardware and has been involved in this
project from the very beginning. He's been one of the
outstanding scientists in the United States. He's been
recognized by the ACM for -- at that time he won the prize for
the outstanding doctoral dissertation, which is awarded to one
person in the United States every year.
Last, but not least, is Miguel Illescas. Miguel? Miguel has
been helping the team with some of the testing and opening book
preparations.
In addition to the Deep Blue team and Garry Kasparov's team, this
match was run by the ACM, and there were several people
involved in running it. It was a difficult chore, and there
were a lot of compromises and things that had to be worked
out.
Carol Jarecki. Is Carol here? Carol was the match arbiter. Is
she here? She may still be downstairs. Carol did an
outstanding job staying on top of the problems from move to
move, watching over the clock, watching over the opponents, and
should be congratulated for doing an outstanding job.
Ken Thompson, Mike Valvo, and myself had a difficult
responsibility. We were in charge of questions related to
problems that went beyond Carol. Ken is sitting in the front
row. Ken, would you like to stand up? Mike Valvo is
downstairs.
We had some complicated issues to resolve regarding questions
that Garry had raised on what information would be available to
him during the course of the game about the Deep Blue program.
As well, Garry had some serious questions about moves that
seemed beyond what the computer was capable of doing. And the
responsibility of examining the computer print out during the
game was the responsibility of Ken Thompson.
Throughout the games Ken monitored the TV screen, watching every
move that Deep Blue played, and Garry couldn't believe a couple
moves. At one point he requested a printout from two
particular moves.
Ken analyzed the printout and reported back to the Kasparov side
that he saw no irregularities, and the issue seemed resolved.
I would like to point out that the question of determining
whether there's a spirit in the computer that came up with
those moves which none of us could understand is a very
difficult one. The amazing thing for the many of you here that
aren't intimately involved in computers is that it would be
almost impossible to expect the computer to play the same game
again. The interreaction of the many computers will cause one
computer to talk to the other one slightly before the other one
talks to the other one if the game is played again, and
information will not propagate throughout the computer system
almost ever again in exactly the same day.
And the small differences of sending the information around the
system will result in different moves being made, if one
attempts to repeat. Maybe one in ten or maybe one in 20 moves
will be impossible to repeat.
So we face very serious questions here, and I hope that we've
resolved them satisfactorily at this point.
I would like also to thank the match commentators, Maurice
Ashley, Yasser Seirawan, and Mike Valvo. They were one step
way from being weather forecasters in game five. The analysis
was very difficult. The game was beyond those of us watching,
and I would like to thank the three of them for having done an
outstanding job.
At this stage I would like to introduce C. J. Tan, who will
address you for a few minutes. It's been a most exciting thing
to be a part of this, and I'd like to thank C. J. personally
for his keeping the ACM involved. C. J.?
C. J. TAN: (Pulling out a prepared statement.) On behalf of the
IBM Deep Blue team, I am indeed very proud to have played a
role in this historic event. And this is a match that will
benefit everyone, from the students, to the audience, to Garry
and the computer Deep Blue, to many students outside this
building who will be deeply affected by this advance in
technology. And we would like to thank Garry Kasparov, one of
the world's most brilliant minds, for participating in this
great experiment. Garry is a man who sees the future, who
understands where technology can take us. Playing with him
gave meaning to this match.
We would also like to thank Carol Jarecki, the arbiter, for
spending hours in the match room and never stood up or left her
desk.
And Ken Thompson, the inventer of UNIX and a great pineer in many
technologies in use in computer chess today. Who at in our
private communications room hour after hour, to chat with us
and discuss how the games are going.
Why is there such global interest in this match? Because it
visibly shows the world the technology, what technology can do
for man, and how far we have been able to push technology, and
what does this game mean for technology?
The computer played Grandmaster level chess using strategy and
speed. We learned how this nimble and powerful technology can
be shared with the world and translated into real-time
applications.
Now that the rematch is over, where do we go from here?
Well, we will be -- continue our partnership with Garry, but
perhaps on a less competitive livil.
We will be working with Garry in the development of his newly
launched web sight, Club Kasparov, where he will share his
chess brilliance with the world, and especially students all
over the world.
The match was tough on both of us. There have been highs; there
have been lows. And we even had to take Deep Blue for a walk
yesterday morning.
What we have left to do now is perhaps to program Deep Blue to
see how it can learn to take off its watch in the next match.
So again, I would like to thank Garry and all of you that
participated in this event with us. Thank you. (Audience
applause.)
MONTY NEWBORN: At this point it's my pleasure to introduce Garry
Kasparov, who will address you, and I want to say -- an
interesting end to the match. I would have loved to see both
players win, but, Garry, you have my admiration for a long time
to come.
(Applause.)
.
GARRY KASPAROV: Enough. Sorry, I haven't deserved that. I have
to apologize for today's performance. But I don't think it had
anything to do with chess and with the match. I think Maurice
Ashley made a very good statement yesterday saying that I
sounded as if the match was over. And for me the match was
over yesterday, and I have to tell that I had no real strengths
to fight, and I think the result of the game today was quite
justified.
But that's probably not about the result of the man vs. machine
competition. I don't think you -- I don't hope that it will be
taken as granted. The match was lost by the world champion,
but I think there are very good and very profound reasons for
such a result.
I was a bit surprised to hear from C. J. that now they would like
to cooperate on a less competitive level. The cooperation just
stopped (loud, roaring applause.)
And I have no doubt that the spirit of the event will be no
different from the one that took place in Philadelphia one year
ago.
Soon I recognized it was a grave mistake, with all the
consequences that I have to pay at the end of the match, and in
the middle of the match.
It was nothing to do about science. It was nothing to do about
furthering the investigation of computer potential of chess.
There was one zeal to beat Garry Kasparov. And when the big
corporation with unlimited resources tries to do so, there many
ways to do that.
I resigned today. I think the crucial game was game two. And
again, Mr. Newborn, I have to tell you that this is not up to
you and Mr. Thompson to make a judgment whether computer can
play these moves or not. This is obviously beyond our
understanding.
Deep Blue is so complex, and I recognize the complexity of this
machine, the old interconnections that it will never come up
with a same result even if it were under test, again, and
again, and again.
But what is most amazing, that it's -- Deep Blue as we saw in
game one and a couple of other games still has generic computer
problems. And I'm sure that this is not up to people in this
room, not to me, not to Deep Blue team to say it was absolutely
correct and perfect.
I believe that these printouts, if they are available, wanted by
all chess fans, all computer and chess fans around the globe,
and I think that two or three under powerful computers will
tell us whether any other machine can do the same thing as Deep
Blue did in this match. My personal feelings, I doubt.
But again, we faced a machine that had no comparison to make
moves that were beyond anybody's understanding. And I couldn't
have anticipated it before I started to play.
I have to tell you that game two had dramatic consequences and I
never recovered after this game. Not because I lost this
game. In fact, I could make a draw just instead of resigning.
But because there were two major issues that are not yet
resolved. Whatever people are saying here, I still do not
understand how the most powerful and great machine couldn't see
a simple perpetual check at the end of the game.
I'm sure there will be answers provided. I'm sure there will be
a lot of analysis later on. I'm sure I'm in the wrong position
today to complain, because it will be written tomorrow that
Garry Kasparov couldn't lose properly, couldn't be a sportsman,
to accept his defeat, I can even name the newspapers that will
write this.
Yes, so be it, you know. Again, I understand, I fully understand
all the consequences of the result of this match.
But I think it's very important for all of us to state today that
Deep Blue now must intercompetitive chess, competitive chess.
You know, have the team play a normal event, play a world
championship match, under proper conditions, and the scrutiny
that every chess player has to go through. Play competitive
chess, and we shall see whether this man is a prodigy, is a
unique piece -- when this machine is a prodigy, is a unique
piece, or is a lot of human weaknesses shown in one particular
event.
I think it's time for Deep Blue to prove that that was not a
single event it could play. I think it's time for Deep Blue to
start playing real chess.
And I personally assure you, everybody here, that if Deep Blue
will start playing competitive chess, I personally guarantee
you I'll tear it to pieces -- some of them probably too shy to
show up, they can hire the entire GM force of the United States
of America, it will not help, because we know how the
machine -- how a machine plays. Put it into competitive chess,
put it in a fair contest, not that one, make IBM a player, not
a sponsor at the same time, and we will see what is going on
happen.
And I think it is just the beginning. And I have to apologize
again, I am ashamed by what I did at the end of this match.
But so be it. I feel confident that the machine hasn't proved
anything yet. It's a much better machine than the
Philadelphia. It was clear from day one. But it's not yet
ready, in my opinion, to win a big contest. That's my belief.
And again, you can trust me; you can defy me, as a loser, I
deserve that to some extent, but I think it's just the
beginning. Thank you. (Audience applause.)
MONTY NEWBORN: ...Garry we will get a chance to sit down at the
table at least one more time. It's provided the entire country
and the entire world a week of the most exciting chess that
I've ever been a witness to and most of the rest of us.
At this time I'd like to introduce Joe Deblazi, the executive
director of the ACM who is holding a lot of money in his
pockets and I did everything I can do to talk him into take
taking a quick trip to Bermuda but it didn't work.
Joe DeBlazi: I just want to say a few moves, not prolong this
much longer, because I'm sure that Garry and C. J. want to move
on and do other things they have to do.
It's a pleasure that the ACM has been a sponsor of this year,
like the match in Philadelphia.
I was there when Garry Kasparov won. I'm here this year when
Deep Blue won.
One thing is constant over two years. The greatest chess player
in the world is Garry Kasparov, period. (Mild audience
applause.) A computer, even in parallel, cannot approach the
capabilities of a human being.
Understanding that, this was not a test of the human being. It
was not a test of the greatest chess player in the world. What
it was was a test to see how far we have taken this basic
technology and what we can do with it. The winners in the
future are the young people sitting downstairs who are
interested now in chess. Or the young people and many of you
in this room where this technology is going to be applied
against your well-being in medicine and transportation and so
many other fields. It is all of us who have won this match
today.
And the person that's made it possible is Garry Kasparov. So I'm
not going to hand out a winning and losing check. I want to
ask both people to come up and receive at least the financial
reward of their efforts. So I'd like to ask Garry Kasparov and
C. J. tan to come up. I thank you all for having been here.
Thank you very much.
MONTY NEWBORN: We'll have questions from the audience in 30
seconds. We'll give the photographers 30 seconds and then
we'll have questions from the audience. 20 seconds. Let's
just wait one minute until these -- let's sit down, please.
Question over here? (Question not heard.)
GARRY KASPAROV: I suggested that there were things in this match
well beyond my understanding and the understanding of many
people, and I can assure you that probably there is no way to
prove that Deep Blue is making this move or that move, but I
think it will be wise to run -- for everybody, who is curious,
to run the tests. There are very specific positions, very
similar positions in one game, just take only one game, game
two, and I would like to run it, it will take maybe a week or
two weeks, but then everybody can come up with a conclusion.
If, at the end of the day, in two or three weeks' time, no
machine in the world will not come up with the same answer.
Unfortunately, it still means nothing. But, it will all be
very interesting to hear explanations. Because unfortunately
if I heard correctly, even Deep Blue team made some
contradictory statements at the stage, about what machine saw
or didn't see.
But again, it's computer, you know, it's well beyond our
understanding. Has a very different mind. It can come up with
one decision and then change it, come up with another decision,
as Mr. Newborn said, you know, if you run the test, Deep Blue
can come up with a different answer. No doubt about it.
But still as I said, there are some common things in many
computers, and what's most important is the way of evaluation.
And when Deep Blue goes as deep as 25, 30, 35, 40 ply, at one
point it still should give the evaluation. It should evaluate
the position. Because, you know, if you ask a very small
computer about a given position in 25 or 30 ply, and it tells
you that white is better, and if you ask Deep Blue before it
goes all 30 ply, you know, Deep Blue with all respect to its
power, it's not necessarily, you know, can anticipate this
position better than a small computer. Now, again, even if
we -- if nobody in this world will come up with the same result
in game two as happened in the middle of the game.
So it will mean absolutely nothing. But I can have my own
opinion, and I was surprised very much, and this game had
profound consequences and effect on my success in the match.
Reporter: A lot of people are seeing the best chess player in
the world beaten by a machine. Does that diminish the human
spirit? -- human spirit?
GARRY KASPAROV: I don't think so. And from my point of view,
that was my mistake in my statement. You know, I was taking it
in Philadelphia as a scientific experiment. That was a very
competitive match for one side. And I was not ready to see
what's -- you know, what was happening in this match starting
from game two. I had mistakes in preparation. I mean probably
it was difficult to prepare normally for an opponent with no
games, with no ideas.
And, what's most important, the opponent was constantly
changing. I think it's another great achievement that Deep
Blue team was able to change priorities during the match. I'm
really amazed to see that you just change such fundamental
things as bishop vs. knight and suddenly it becomes equal in
game five. Yeah, it box equal because otherwise it doesn't
take on f3. Maybe, you know, again I have no idea what's
happening behind the curtain with Deep Blue. Maybe it is
absolutely an outstanding accomplishment. Maybe, but I know a
little bit about chess and a little bit about chess computers,
and I don't think this machine is unbeatable. I think the
machine has too many weaknesses, and in competitive chess, in
real competitive chess, where you play a match, it will be a
different story. (Unheard question.)
As I say, I am ashame shamed. I am ashamed that I couldn't
prepare myself properly for such an event. But again I would
like to look at the results of the match in two or three weeks
time when we can analyze the games and we can look at the
printouts. I want to understand how Deep Blue won the match.
Unfortunately you cannot do it before you look at what was
produced by Deep Blue's mind during its hours and hours of
calculation.
(Another reporter question by Daniel Slater of ICC.)
C. J. TAN: We will be publishing our technical work in technical
journals and conferences and so forth.
(Another reporter question, not heard.)
GARRY KASPAROV: Again, I think that printouts will be
available. Specifically we are talking about game two, moves
that -- you know, the move Qb6 that was not made, and the final
position. I think there are similarities in these two cases.
Whatever and whoever says, there are clear similarities. And I
want to see for instance, where Deep Blue stopped its
calculation in game two. Because after Qc4 it was in check,
couldn't stop its calculations when it's in check.
We have to look at this, how can you play this thing where it's
changing every day so rapidly and you have no idea and no
control of -- let's look at this, I'm sure we'll study that.
And there is a possibility that, you know, it was a great
computer mind that came up with this -- with these ideas.
There are many questions. For instance, why -- obviously the
Deep Blue team will be willing to answer, what is the principle
that computer -- for instance, in game one. We play game one.
In game one computer spends three minutes per game -- each move
and it spends six minutes at the end of the game where the
position was bad. It saw that the position was losing.
Interesting question, why Deep Blue didn't think longer before
when position was dramatically, you know, going down. Probably
didn't understand it.
And then probably the computer's regime has changed. Interesting
question, when computer spends 15 minutes or, you know, any
time more than five minutes, that's a moment to ask for the
printouts because it's interesting to see, what is happening
inside the machine which is programmed to make moves within a
certain period of time.
(Reporter question about Deep Blue playing tournament games --
are you willing to have the computer enter regular Grandmaster
tournaments.)
GARRY KASPAROV: I don't think Deep Blue is too weak to play in
regular tournaments. I believe that Deep Blue team now, it's
time for them to claim the world champion. There are three or
four players in the world that could Deep Blue who could play a
candidate tournament and if they want to skip it and go
straight to me and play a normal match in competitive chess,
under conditions that will be imposed by an independent
sponsor. I'm willing. I'm willing for this machine to play
real event. And I think they must do it. They must do it one
way or another, either to enter the competitive chess with
other players -- I say, there's three or four in the world that
can afford the luxury to play with such a powerful machine,
because I'm not here to doubt the integrity of the machine
which was at least two times stronger than Philadelphia.
(Another report question, ending in "How could IBM cheat?" "
GARRY KASPAROV: Again, I'd like me and you and everybody else to
look at the printouts especially of game two, also game five,
for instance, and to analyze what's happened. You know, I mean
everything can happen in this world, but again at the end of
the day I'm sure will come to the conclusion that Deep Blue is
well beyond anybody's understanding in this world. But I would
like to see first printouts because there is no information
available to make a judgment. What's happened was outstanding
from my point of view.
Now,, you know, anything happens, you know, and different things
happening, not only in Hollywood movies.
(Reporter question: Game to game a general feeling for is the
computer frozen or what's changed between games.)
C. J. TAN: We said many times before this match, many things are
improved this time around. Number 1 the machine is twice as
fast, more computing power -- (another reporter question.)
C. J. TAN: Yes, I'm answering your question. And the second
thing is we're adding more chess knowledge. The first thing we
have done, we have developed programming tools to allow us to
adjust the parameters faster between each games.
Well, the parameters usually are involved in the evaluation
function for certain strategies and forth. I'm not prepared to
get into the details at this point.
(Reporter question: When the computer, Deep Blue took ten
minutes to make any move, was there any hope from my program,
any assistance of any programs, was there any suggestions from
any of the programmers when Deep Blue took longer than shall we
say three minutes or six minutes or whatever that the program,
at the beginning of a particular game?" "
C. J. TAN: Once the clock starts -- the answer, first of all, is
"no." The second thing I would like to emphasize, once the
clock starts, none of us can interfere with the Deep Blue
system itself, and all the rules that are preestablished before
this match and overseen by the arbiter Carol Jarecki and the
committee meded by Monty Newborn, we followed their rules and
in the printouts of several games as requested were given to
the arbiter, Carol Jarecki.
(Reporter question: Could you respond to Kasparov's earlier
comment to which he basicly suggests that this was not at all
about computer science or a machine and chess, that nothing was
proved by the computer at all throughout this match.)
C. J. TAN: Well, what we wanted to prove is to show that this
technology and computer can indeed play at this Grandmaster
level.
GARRY KASPAROV: No, sorry. Probably you understood me wrong.
No, I never proposed that we stop playing. I said there are
three or four people in the world that could compete with Deep
Blue. This is not about -- computer beating definitely, but
computer is playing definitely by itself, as game one showed,
on a very, very good level.
C. J. TAN: In order to test out this technology, there's no
other person better than Garry Kasparov to be our partnership
in this experiment, and we did a very fortunate as I said a
while ago to have Garry as our partner in this event. And if
we have just played any other Grandmaster, it wouldn't be as
meaningful as it would have if played with Garry.
(Reporter question, not heard.)
C. J. TAN: That's a very interesting idea. You understand so
far we have been doing science, been in the laboratory
constructing computers, and we want to take us away from that
and become a professional chess player. It's a very
difficult. We'll have to think about that.
MONTY NEWBORN: I'd like to just have one quick introduction.
Carol Jarecki has been somewhere over here. Carol here?
Carol, would you just take a bow for a second? Carol did an
outstanding job in the game room and was never seen by anybody
for a whole week. Carol? Take a bow. (Audience applause.)
(Reporter question: Comment on what this means to computer and
chess..., inaudible.)
GARRY KASPAROV: I don't think it's even close, that chess is
taken over. You know, there was one man who was a very good
player, maybe the best in the world, you know, cracked under
the pressure. But that's nothing to do with computer being
unbeatable. . This machine is vulnerable and I have no doubt
in the proper competitive chess it will be beat. Now,
that's -- you know, again you can -- you can say say it's a
postmortem statement that's carrying no value, but I learned a
lot during this match, and I know what you can do with the
machine. You have to play on a very have I high level. I mean
I have to show, you know, the best of my ability, but I have no
doubt that this -- this machine with all these tools, you know,
adjusting it, making it better during the match, this machine
would be badly beaten if it's a proper competitive chess.
(Reporter question, inaudible.)
GARRY KASPAROV: Today's game don't even count as a game because
probably it was even published somewhere else. I was not in
the mood of playing at all. Because, you know, I'm a human
being, you know, and after game two, I had, you know, some
major, major problems of coming back to the match. You know, I
proved to be vulnerable. So, you know, when I see something
that is well beyond my understanding, I'm scared, and I know
something well beyond my understanding.
(Reporter question.)
GARRY KASPAROV: Yeah, plenty of psychological effect. But, you
know, as long as I could keep under pressure, you know, forget
today game, I mean Deep Blue hasn't won a single game out of
the five because again game two resigned when I could force a
draw. Now, force a draw. Now, if someone has another position
stand up and tell that the position was not a draw.
Game two was resigned in a completely drawing position. Is that
a correct statement? (To the Deep Blue team, who are all
shrugging and shaking their heads.)
The final position was a draw. We recognize that Deep Blue made
a bad strategical position blundered and made a perpetual
check.
What I'm saying, before today's game, Deep Blue couldn't win a
game, and I was playing on a high very level. I was proud of
my games game one, game four, game five. You know, it means
that machine is vulnerable, just add more energy, you know,
just more resilience, and machine has no chance.
(Reporter question, inaudible.)
GARRY KASPAROV: One condition, IBM play and not a sponsor. I
think that if it's competitive chess, there's no room for
friendly relations and nice talks. I have excellent working
relations with IBM, I hope we'll continue our work, but --
continue our war, but, you know, competitive chess has no room
for these kind of friendly relations. And that was probably my
mistake that I didn't understand understand before the match
and it was not about simply playing, but it was only about
winning for man --
YASSER SEIRAWAN: By the way, I've been told that Garry is going
to be answering questions for about half an hour, but we don't
expect him to join us here.
(Reporter question.)
GARRY KASPAROV: You know, when you play Deep Blue, you have a
choice. Either to play sort of crap, you know, just very rare
openings, or to play something -- the best lines you know. But
to play the best lines you have to know your opponent. I
cannot study all the openings. I mean I have to know what my
opponent is playing, because the depths of the preparation, you
know, it will be different.
Now, if I have to play next time with Deep Blue, there is no
doubt there will be an opening due and I will play a proper
move, e4 with white and c5 as black, there is no doubt about
that.
But, you know, I played probably what was recommended by every
computer specialist. You know, you don't start confrontation,
and in game one proved perfectly a success. Deep Blue played a
couple of moves, I think Patrick Wolff described it as, you
know, Deep Blue was playing as a numbskull. You know, it was
working. But suddenly it stopped working, suddenly Deep Blue
found its way to break the pawn chains and start confrontation
at a very convenient situation.
Probably, no, with Deep Blue the normal computer strategy doesn't
work.
MONTY NEWBORN: We'll take three more questions from the
audience.
(Reporter question...and that seems to upset you. Did you make
demands -- the same demands of human beings, say
Mr. Karpov...would that upset you?)
GARRY KASPAROV: You know I'm not that stupid to be upset if
Mr. Karpov plays like Mr. Anand or Mr. Anand plays like
Mr. Kramnik, but even human beings with human flexibility are
not able to make change their style dramatically. You will
never make the mistake of saying Garry Kasparov's games look
like Anatoly Karpov's games.
Unique in this match is the computers are flexible and are doomed
to make similar mistakes in similar positions. This machine
didn't make similar kind of mistakes. This machine adjusted
itself during the game to situations that just arised.
(Reporter question, inaudible.)
GARRY KASPAROV: When you allow this piece sacrifice you can
resign and there are many games in competitive chess that this
line happened, but I mean hardly explain what I did today
because I was not in a fighting mood.
MONTY NEWBORN: Take one more question from the audience.
(Reporter: It is usually that the chess competitors sit down
after the game to sit down and discuss the game and analyze
what they did. What was your incentive to keep your...so
hidden and...show us the evaluation of Deep Blue, just lay open
your strategy?)
C. J. TAN: Well, analyzing the moves like chess players do, at
another location we didn't have the opportunity, but to reveal
the inner thinking of Deep Blue is giving away whatever it was
thinking and preparing for the match -- for the game, and we
would rather do that after the match and not during the match.
That's like revealing everything we have programmed inside the
computer. We don't have -- it's very difficult to tell
computer "Give me this piece of information, but not everything
else." So after this match there is certain information that's
interesting to the public, we will be publishing that in the
technical journals and also share that with many other people.
(Reporter question, inaudible.)
C. J. TAN: I believe we will be sharing some of the specific
interesting information such as Mr. Kasparov, which perhaps the
press in some appropriate form.
GARRY KASPAROV: C. J., I think you understood me wrong, you
know. I believe that there is rule in the game of chess that
when the game is over, we sign score sheets and the score
sheets are given to arbiter. Deep Blue's score sheets,
printouts, I think that was no conditions, these all printouts
from game one to game six must be published, somewhere, on
Internet, and anybody who has interest in chess or in chess
computers will study them. This is not our analysis, you, me,,
or Mr. New bone or anybody else. Anybody who has interest.
Because it's a great contribution to the game of chess and
computer science. I believe it's your obligation to print out
everything that Deep Blue has been considering during long
hours of calculation.
C. J. TAN: We will publish what is appropriate for an
appropriate manner, because for 99.9 percent of the people will
not understand what 101001 means, and especially the public.
We will be glad to do that in the appropriate time.
(Reporter question, inaudible.)
C. J. TAN: We just like any other Grandmaster when you go into a
match, you involved many friends and people for advice and so
forth. And we have Joel Benjamin with us since August of last
year, and since a month and a half ago, Grandmaster Miguel
Illescas came over to help us analyzing and testing the
program, through IBM of Spain.
Joel Benjamin a Grandmaster has many other friends. Probably he
has many other advisors talking to him. And they have their
confidential, normal corporate confidential nondisclosure
agreements, so it's not up to us to talk about it.
MONTY NEWBORN: I'd like to close this press conference.
(Reporter question, inaudible.)
(Would you say that you were too nice to IBM...not have access to
previous games as they had access to all of your previous
games? In another chess match where any Grandmaster plays
another Grandmaster they both have ack stoce many, many of
their games. And here you were limited to almost no games and
only had games as you played, whereas they had access to all
your games.)
GARRY KASPAROV: Unfortunately, I was playing too well in the
last year, and I -- after Philadelphia, I didn't take it -- I
mean I took it seriously, but I believe that my biggest mistake
was not to demand certain conditions that would make this
contest fair.
Now, first, I think there must be some games available. This is
number one.
Number two, I think in the future matches we will consider more
opening -- openness of Deep Blue even during the match. And
I -- also, I have to confess that the biggest mistake was that
I followed the advice of computer specialists that all
recommended to play this way.
I think this is the biggest way. And I said, if I have to play
again, we'll play normal openings. This is no doubt.
But in order to play normal openings with Deep Blue with a
machine that has unlimited memory, has a great deem of -- team
of Grandmasters, we don't know even the number, preparing that,
I have to take it as a world championship match. I have to
take it as competitive chess. I did not. I mean I played a
friendly match. I was sure I would win, you know, because I
knew that computer would make certain mistakes. And I was
correct in game one. Suddenly it stopped making these kinds of
mistakes, maybe at the beginning of game three. But my
strategy failed. And maybe if I was in a better mood today I
would survive but, you know, after yesterday's game, which was
a very tough match, I lost my competitive spirit.
Now, fair conditions, all the games available, and, you know, we
take it really, you know, to win or lose, not to study computer
implications.
No, you have to -- to beat this machine you have to play proper
chess. No, it's clear. Whatever happened in game two or game
five, even to beat machine in game one, it takes a lot. It's a
proper opponent, and I have to mobilize more my resources to
play even through the match.
But I -- you know, my preparation was so weak that, you know, I
had to consider what to play before each game, because I
decided not to go -- intentionally not to go to the main
openings, and it was a mistake, because, you know, during the
match we don't have time to come up with something that you
play regularly.
There is obviously these guys are studying very deeply.
No, you have to come, start it out -- but it takes probably a
couple of months preparation. My preparation was maybe ten
days, and that was not enough, even close.
(Reporter question, inaudible.)
C. J. TAN: Since some of the annotated lines of other people
before the match is probably okay, but if I give you the whole
dump or whatever of whatever computer is thinking obviously you
don't have any match at all. So again, I said we will publish
some of that probably. Everybody is tired and I think we all
need to go home and celebrate.
I also want to answer once again, I've answered several times,
about why Deep Blue's games are not available. The previous
incarnations of Deep Blue, again, many games are available.
Many, many of them are in the books. And if this version of
Deep Blue took one year to develop, it is a very young system,
and I would like to have many competitions so many games are
available, so we can (word) as we go along. If we would have
provided the games in November it would not be the same machine
come February. And same thing, if we had provided games that
it played in February, it would not be the same games that it
would play in May. So since it's a developing system, those
games become meaningless while we are doing development.
(Reporter question about openings.)
GARRY KASPAROV: I don't think that when I played -- you
shouldn't be mistaken by looking at the first moves of game two
or six. What I played, it was -- you couldn't consider it an
opening, you know. But also in game two I played something
that is -- you know, it's a main line, but you play sot decent
moves, not the maneuvers that a couple of things happened in
the games of the chess players.
But in game two I believe I can afford certain things. I wanted
to test how it moves in closed positions. I was very much
surprised to see that Deep Blue didn't play on b5 because
normally the computers don't keep the pressure, it starts
taking an advantage. It didn't. It's a smart machine now. I
learned something.
Now, as for today, you know, in game four I successfully
implemented something. But, you know, in game five, I
recognized that, you know, even if I play some tricky openings,
you know, the machine reacts very often like a human player,
you know. There was no -- the strategy didn't work. Deep Blue
didn't make the same mistakes I expected it to make. That's
why, you know, I -- after e4 there is a very limited choice of
dumb openings like I played in game four, and I tried to play
something else. I didn't expect this main line, but again, I
wouldn't like you to take this game as a serious one because my
ability to fight was very much down.
(Reporter question, inaudible.)
GARRY KASPAROV: No, I don't think yet. I think eventually
machine will prevail, but I don't think that you can take
today's day as the day of doom. As I said it's just the
beginning, and I have no doubt that personally I can beat the
machine even if it has a new version in one year's time. But
obviously it's a historical achievement that machine was even
able to play in competitive -- on such a level with the world
champion.
Reporter: In another match in which you have unlimited access
to...would that be more fair in this match?)
C. J. TAN: Garry and I have talked about this new concept of man
and machine playing chess together. What we have in this match
is a precursor to that. We both use computers in different
fashion. And Garry would very much like to propose the new
form of chess he talked about with advanced chess where
Grandmasters would have access to computers while you play
chess against each other, either during a regular match, also
the Internet. Those are many, many possible advancements that
we could see for chess in the future.
(Reporter: Asking question about using computer as an assistive
database, not for calculating moves.)
GARRY KASPAROV: It will help, but, again, I would prefer that,
because that will relieve a lot of energy before the game. But
I don't think it's yet needed, but it will definitely make my
performance much better.
Yeah, I was playing against myself and against something that I
couldn't recognize.
I believe I'm the best in the world and if I lose it's a result
of my mistake. While I wasn't in good shape in this match,
Deep Blue couldn't do anything.
(Reporter question, inaudible.) Garry a world championship match
is a world championship match. Now, you know, if I have to
take it as serious as a world championship match, as defending
my title, preparing properly for the opponent that I can
identify, I will play very, very different. I will play
differently, and again, if you want to check how confident I
am, I can bet the entire prize fund of the next match, whatever
it is, that I will beat the machine. (Audience applause.)
GARRY KASPAROV: Yeah, because then advantages of machine are
growing because, you know, I will be tired. I think we should
play every second day. You should give a human being time to
rest. You know, 20 days, ten games, proper match, you know,
and I'm really taking the challenge and I believe that some
other players would like to participate as well. It's not
about becoming chess player. Again, I don't think there are
many players to compete. I think there are very few that are
are capable of fighting Deep Blue, but I think it's time to
prove that the machine can do a little bit better than this
match.
(Reporter question, inaudible.)
C. J. TAN: While we have the normal databases available,
whatever Grandmaster games are meaningful, we update.
(Reporter question for Mr. Newborn; the first thing that was said
in the press conference was let's do it again. Why don't I
hear that now?)
MONTY NEWBORN: You've already heard it. You're hearing it
from --
(Reporter inaudible.)
MONTY NEWBORN: Well, these things get negotiated with the people
involved and it takes a bit of time. We went away from
Philadelphia, we've all been discussing the question about
whether there's a potential rematch. It certainly is a subject
of discussion. It's something that will get discussed between
IBM and Garry over the next couple months, and if the ACM is a
participant, we certainly would like to be involved.
(Reporter question inaudible.)
C. J. TAN: Well, obviously we are very flattered by this
invitation to play with Grandmasters at their level, the
highest level. And it's something that we will have to discuss
and obviously -- personally I would like to see that happen.
(Reporter question, inaudible.)
C. J. TAN: Oh, I think in the sense that, as I said also that
Deep Blue is a new system, it's not fully tested, and this is
the first time we play six regular games. Obviously if we play
more games we will find out new things, either bad or good. We
don't know.
(Reporter question, inaudible.)
GARRY KASPAROV: This match has no rating implications. If you
ask me to give a rating to Deep Blue, I think it's very -- it's
almost impossible because you have to evaluate something that
makes different moves. If you look at beginning of game one or
game three, you will be flattered even to give any rating to
this machine. Now, if Deep Blue at the end of game five or
game four you will give it say 2800, or maybe higher. But I
still think it's very difficult to make any rating evaluations
today without machine playing the proper competitive chess.
Now, it obviously plays many positions at the level of 2800 or
above. But there are still many weaknesses, and probably the
average will be somewhere around 2800 today.)
YASSER SEIRAWAN: Did Garry say 2800 today?
AUDIENCE MEMBER: Yes.
(Reporter question, inaudible.)
GARRY KASPAROV: Before the match I would say it's not flexible.
Now I doubt, the machine is plex I believe. It knows how to
change priorities, even during the game. There is obviously
one big disadvantage of the machine, but obviously -- also it
looks to be overcome in the match, that machine -- machine has
a limit of calculation and at the end of this calculation it
has to make evaluation.
Now, we discovered that Deep Blue can make unbelievable
evaluations of positions 20 moves deep, you know, 40 ply. Now,
interesting, one machine's problem was in game one, which is
machine doesn't understand positions where it has a material
advantage or disadvantage, and if the implications are very
long, now, in game one Deep Blue recognized it too late.
MONTY NEWBORN: I'd like to close this press conference. I'd
like to thank those that participated. We have an audience
downstairs. The participants are not planning to come
downstairs. But I'd like to thank them for their participation
over the afternoon. I myself thank everybody that's involved
in this point and hope to see everybody on stage one more time
next year. (Audience applause.)
MAURICE ASHLEY: Not much to say. What do you think the
implications are going to be from today?
YASSER SEIRAWAN: Well, obviously this is -- they're 1-1, if you
will. I think it's going to be in IBM's interests to see a
third. I liked the things that Garry was saying that he'd like
to see Deep Blue tested against other top Grandmasters.
Garry has already predicted the sore-loser newspaper articles,
but I think that he really carried the foint far too far, that
there with the intervention and so on. The fact of the matter
is this is an extraordinary computer and it played very well
and it had a lot of problems that Garry couldn't solve the
riddle.
Finally, I was a little bit disturbed by what Garry was saying
there about IBM. I just want to say again, for myself, this is
some of the most exciting chess I saw. I think IBM is a
terrific sponsor, and please, once again, join me in
congratulating them. They have every right to sponsor a
match.
MIKE VALVO: I'd like to make some obvious comments. Garry
implied that there might be some cheating. And I say that
word. Is that too strong?
But, think about it. Is it in IBM's interests to cheat at
something like this? No, I don't think so. I think what was
happening here is something Garry couldn't explain was in the
machine, and because he couldn't explain it, he said, "There
must be something funny going on." I think there are
explanations. I think that -- I hinted at them when I said
that the computer had some extreme flexibility with king
safety, that it seems to -- seemed to allow two or three pawns
worth of value for king safety. And I was told that the
computer considers both sides when it looks at king safety, not
just its own and a small regional effect. It looks at the
entire board. And if it's dangerous for them, it's okay if
it's dangerous for the other guy, too. And it takes that hand
in hand.
And Garry didn't understand that idea, or didn't think of that
idea, and -- well, maybe there are other explanations for what
happened than what he chose, and I wish he wasn't so sore about
it, but I suspect that if there is a match next year, and I
hope there is, that he will become aware of some of these
considerations and try to deal with them in other ways.
MAURICE ASHLEY: Well, this match has been rich with stories.
The controversy will not end. Last year's match 4-2 in
Kasparov's favor. This year 3.5-2.5 in Deep Blue's favor. And
we will see, I'm sure, more of the same. It has indeed been
exciting. We thank you all for joining us, and we hope again
that this is not a match about human vs. machine. It is a
match that will answer some basic questions and help us to
understand our world a bit better.
MIKE VALVO: Thank you very much.
Garry is coming down!
YASSER SEIRAWAN: No, no, there is a possibility.
MAURICE ASHLEY: A possibility that he will join us downstairs.
YASSER SEIRAWAN: I'll go check it out.
MAURICE ASHLEY: Yaz will scout that idea.
YASSER SEIRAWAN: I beg your pardon. We now have a confirmation
that IBM's team, the Deep Blue team will be joining us
shortly.
MAURICE ASHLEY: And we'll see if there's any word of Kasparov,
but I doubt Kasparov will actually come, but the Deep Blue team
will be here shortly.
Just the IBM team will be coming shortly and we'll allow you to
ask them some questions.
MIKE VALVO: Do you think there was enough drama in this match
for you all?
YASSER SEIRAWAN: The question is, I think we may have lost some
of our roving mikes. We will try to bring those in our
audience at least shortly, at least before the Deep Blue keem
comes out. -- team comes out. We will of course be welcoming
your comments and questions. And we're expecting them any
moment now.
I just wanted to see that one of the things that Garry did
mention in his press conference, he preferred to it quite
often, is he double-guessed his own opening work. He said that
"Look, guys, no more Mickey Mouseing around. I want to play
real openings.
MIKE VALVO: I wonder who these computer experts were that he
conferred with.
YASSER SEIRAWAN: Well, of course we do know that Frederic
Friedel was one of his most influential advisors.
MAURICE ASHLEY: You have to go beyond that, though, Yaz.
YASSER SEIRAWAN: Yeah, I think there were many.
MAURICE ASHLEY: There were lots of statements Kasparov made that
obviously could be taken different ways. There were a lots of
hints and suggestions.
AUDIENCE MEMBER: He didn't want them to be introduced. It seems
he was very upset with his pregame preparation.
MAURICE ASHLEY: But what about the statement that he's only been
preparing for only for ten days? What's up with that? These
people have been preparing for him for 14 months, and that was
their only goal. The idea that he could only prepare for a
match like this for ten days and then play those kind of
openings.
YASSER SEIRAWAN: Yes, again, poor preparation, I would say, in
the final analysis by team Kasparov.
The other thing that was very interesting, and it only started to
come out as the press conference went on, and that was you're
getting the idea that, hey, Miguel Illescas, Grandmaster from
Spain, had been putting a lot of energy for IBM Spain.
Joel Benjamin, they hinted, has a number of Grandmaster
colleagues here in New York, and how much work did that group
of people do.
So it's tough. It's conceivable, it sounds like there was a
whole group of Grandmasters. I must say that I honestly had
nothing to do with Deep Blue's preparation.
I always wanted to see Garry win this match and I'm just as
stunned as can be that he hasn't, and the final result.
MAURICE ASHLEY: Is it such a big deal that Grandmasters were
helping Deep Blue? I mean it's teaching Deep Blue how to play
better chess, so what's the big deal if they hire a few
Grandmasters?
MIKE VALVO: It's to be expected.
YASSER SEIRAWAN: Well, I mean it's to their credit that they
did. I must say that many chess computer programmers do not
hire Grandmasters to teach their programs, they just have the
programs do it all themselves. So it's an honor from the IBM
research staff to go out and get expert knowledge. And of
course it's in their interest to do it, but as far as I'm
aware, they only do D it for the first time in Philadelphia
with Joel.
MIKE VALVO: One thing that struck me is that Garry just seemed
to have a single strategy, play strange moves, try to get the
computer out of book. You would think that he'd have an
alternative strategy where he'd switch to some other approach,
another type of an opening, another type of play, and he didn't
try it once in the entire match.
MAURICE ASHLEY: Didn't he say that he couldn't prepare his main
openings as well because it was just a short match, and they
didn't really have time to make these adjustments, and yet I
hear that and I think that Kasparov has such an incredible
repository of knowledge --
YASSER SEIRAWAN: Repertoire, yeah.
MAURICE ASHLEY: -- yeah, and you're thinking, what? You can
play almost anything. This is Garry Kasparov.
YASSER SEIRAWAN: Well, first of all, two quick things. First,
along with what Michael Valvo has suggested and what Garry
Kasparov's overall strategy was for the match, I think also
Garry was trying to get into endgames.
The next thing is, let's accept Garry at his word. That game two
shook him up so badly that he didn't seem to get control of the
emotions thereafter. I don't know how he is sleeping, what
have you, but one thing is for very, very sure, that loss was
more than a loss. It devastated him. It set the tone of the
match. He questioned the "hand of God," the intervention, he
wouldn't let it up, you saw him very stubbornly cling to it
throughout the press conference. So, yeah, he was very
emotionally disturbed --
MIKE VALVO: Yasser, you've played matches against very strong
players.
YASSER SEIRAWAN: Of course.
MIKE VALVO: Do you allow yourself to get that upset about
anything?
MAURICE ASHLEY: Yeah, I was about to ask that question in the
sense of, why is it that when the machine does it, it's so much
more unnerving? It's not like -- you know, you lose to one of
your buddies, you might get upset, but you think okay, tomorrow
I'm going to take care of you.
In this case, it's like somehow it beat him, and now it seems
like some kind of unstoppable force, and why is it such a
ghost, almost?
YASSER SEIRAWAN: Well, in this case Garry clearly acted
unprofessionally when he allowed his emotions to get the better
part of him. I mean as a professional, he's been challenged in
incredibly tense positions -- tense situations, I should rather
say, and he's resolved them advantageously. He's won virtually
every challenge he's ever faced, with this exception. So for
him to allow this to distract him from the game was a very
unprofessional, unwise thing that happened, and I can say for
myself I'm stunned that he allowed it to take over the match
for him.
MAURICE ASHLEY: Well, we would sincerely like to apologize. The
Deep Blue team in fact is not going to come for whatever
reason, they have decided not to come on stage.
We would like to thank you again for coming --
YASSER SEIRAWAN: Shall we take a question from our audience?
MAURICE ASHLEY: And if you have further questions --
YASSER SEIRAWAN: We'll take two, and then we'll let you all get
going.
AUDIENCE MEMBER: Yasser, this is mainly towards you. You made
the comment very early after the match was over that you're
surprised at the computer winning, and was it expected, you
didn't think computers were ready. And I'm surprised by the
whole comment Garry -- I mean obviously with Garry's statements
and everything. But when you have computers have been around
since 1946, and the fact that they have programmed this with
ultimate skill, using Grandmasters and funds and the ability
which computers have today, to me it is not surprising at all.
It is only a question of time and the time happened to be now.
Humans make mistakes and computers don't make mistakes, they will
make only mistakes if what they were programmed are mistakes.
So the fact that they can analyze so many more ply and so many
more moves ahead of what any one human being can do, it's
really not all that surprising.
Now, having said that, allow me just to say one more thing, is I
don't believe that this has to be taken as a serious blow to
mankind, because time has just come for this to happen, it's an
event, it's a very exciting event, but the fact is that a human
isn't facing another human who might be able to make a mistake,
and I think that's all it is. It's just a great event.
YASSER SEIRAWAN: Well, I will just say that I agree with you,
the computer is a genius, but the question is has the time
come? I think the time is coming. I think it shouldn't have
arrived. The point that I'm trying to make is that Garry is a
better chess player than Deep Blue is and that he lost control
of himself during the match, he allowed himself to get
disturbed, and he was very, very poorly prepared. And, yeah, I
think there are a lot of other players in the world that could
beat Deep Blue today. So I'm just saying that, yes, we know
it's coming from 1946 onwards. It shouldn't have arrived this
early. This is an early bird! This is an early bird.
Question there, yes. And this will be our last question, and we
thank you.
AUDIENCE MEMBER: How did the six-game format come to arise?
Because obviously Garry wasn't pleased with -- he wants Deep
Blue to play competitive chess, and I guess it's ten games, or
20 games, or first to ten points. How did six games come about
to be the point --
MIKE VALVO: This is part of the negotiations between IBM and
Garry. This was last year's format. They just repeated it
this year. I don't know the inside story why six was chosen,
why eight or ten wasn't chosen. But I'm sure time is money,
and that would have involved a lot more money.
AUDIENCE MEMBER: And the second part is, with the match tied
2-2 -- I mean 2.5-2.5, why -- it seemed that there was a
tremendous amount of coverage about the fact that there was a
great deal of pressure on Kasparov to win this final match.
What would have been the tragedy -- not the tragedy, but
certainly the great to-do about drawing against Deep Blue? Why
couldn't he have played an opening that he knew obviously a lot
better than that. (Audience laughter.) To draw the final game
in order to draw.
MIKE VALVO: I think Garry took the pressure off of himself
yesterday when he kind of said that everybody else is more
concerned than he is. I forgot how he worded it, but he said
he's just going to go and try to make good moves. But
obviously the pressure did get to him. He couldn't shut it
that easily.
MAURICE ASHLEY: Well, I think the point is that at some point
we're afraid that computers will do something that we think
we're the only ones -- should be able to do. I mean we're
intelligent. We like to think of ourselves as intelligent.
The superior being in the universe, so to speak, although we
don't know what else is out there. And to think that there is
something coming along that may someday do something that we do
and go beyond just computing and start to maybe intuit, to
feel, to do these kinds of things that are supposed to make us
uniquely human and put us in a central place in the universe, I
think this is the big issue and maybe we just defined the
question incorrectly, and so everything else becomes this
battle, becomes more magazine nide than it really ought to be.
I think this is what's happening for us, and maybe if we just
accept that, look, it's just crunching numbers. It doesn't
have a life, it doesn't have these thoughts and feelings, and
we'd be all right whatever it does.
YASSER SEIRAWAN: I'd just like to add one last comment, and that
is, I agree with you, sir. Garry should have chosen an opening
A, he was comfortable with, B, he didn't mind drawing, and yes,
C, if the match had been a 3-3 result, kudos to both, you know,
it's unresolved, and let's come again. So, yeah, that would
have been a smart, wiser action, course of action, but at the
end, Garry again said that somehow after this fifth game he had
put a lot of energy to win the game, he didn't do it, and he's
nervous, he's tense.
Again, I don't know what it was. Maybe he wasn't getting sleep
the night before, but he clearly after the game didn't look
like a guy that was really himself. He looked just shattered,
evidently.
MAURICE ASHLEY: And showed human qualities. And we hope he'll
show another human quality of fortitude and bouncing back in
the face of lost. And he said it already, he's going to rip
this thing apart when he sees it again.
YASSER SEIRAWAN: Tear the thing apart.
MAURICE ASHLEY: Tear it apart. That said, we have been here,
and we'd like to thank you again. And the excitement is not
over. Thank you very much.