Welcome to the commentary of the Kasparov - Deep Blue Challenge. What you're reading is exactly what Maurice and Yasser are saying at the match - thanks to the fine work of the stenographers at the scene. Please forgive them a typo or two and enjoy the match!!
Commentary by Maurice Ashley and Yasser Seirawan
Game 1 February 10, 1996 Part A
MR. SEIRAWAN What I was thinking, here, though, when we look back on this event in 1997 and 1998, I think that what this event will allow scientists and chess players to do is establish a yardstick, to say okay 1996 this is how many points we scored. Comes the year 2000, you know, have we improved or have we even in fact gone backwards?
MR. ASHLEY Well, the games haven't begun yet and the computer may still show us up. Deep Blue scheduled to begin in a few minutes. No one knows how strong it really is. Usually we like to measure using a rating system and we know Kasparov is the highest rated chess player in the game. But no one knows. This computer is like top-secret. Secret service kind of behind the doors don't let anybody know, hush hush. And this amazing amount of secrecy behind it is making us wonder what is it going to come out with? They mention 32 processors working in concert against this 1 man. It can see 50 billion to a hundred billion positions every 3 minutes. I mean I have trouble putting on my socks in the morning. What can it do to chess?
MR. SEIRAWAN More decisions than you and I are going to make in all of our lifetimes will happen in today's games. I know that they're getting ready to begin, and what we'll have, of course, is at the chess table in the other room there will be an operator sitting opposing Kasparov. So Kasparov won't have to stare at this monster kind of humming away, you know, and we're expecting -- and I'm expecting the move D4. I just want to say the setup.
What we have on this screen to your right is the actual board position. So that will remain constant. So you'll see the game always in Realtime. And what we have here is a special program, fritz 4 point 0. It's 1 of the top commercial chess products. And it will help us also in our own analysis in case we miss something quite obvious. And over here I don't know what is going to happen with that.
MR. ASHLEY Maybe all kinds of stuff. As a matter of fact, Kasparov, there you see the actual room they're actually playing in which is 2 rooms removed from us and Kasparov will be entering shortly. Of course, the entire team blue awaiting in anticipation of the arrival of the world champion who should be come in shortly.
We should tell you in the first game the computer has the white pieces. It was a toss yesterday at the press conference. And Kasparov has the black pieces and I think that Kasparov was actually quite happy with having the black piece. So he says. But I imagine no matter what color he had, he would have put a positive spin on it.
MR. SEIRAWAN Oh, definitely. My view is kind of I don't want to give a misappropriation to a quote. But I recall I believe it was Rod Laver who said against his tennis opponent all he needed to see was his opponent's front happened, backhand and serve, and then he would know how many sets it would take him to win. And I think in this case basically what we have here with Kasparov, all he needs to see is the opening and offensive of Deep Blue and the defensive of Deep Blue. I.e. he only needs to play a black and a white game. And then game 3. It's party time.
MR. ASHLEY In the match against Anand that happened a few months ago when Kasparov won his world championship title, it took him quite a while to get started, really get into the flow of the match. As a matter of fact, he drew 8 games in a row before finally even losing the ninth game. Is Kasparov off step a bit and can he do that kind of strategy against the computer?
MR. SEIRAWAN Well, you have to give Anand a great deal of credit because, of course, Anand is an extraordinarily great opponent, and it took, indeed, as we say, 9 games for Kasparov to get a good measure of his opponent Anand. And I think, of course, what we'll see today --
MR. ASHLEY Garry Kasparov has just entered. World champion in the room. I'm not sure if he will grace us with his presence later. He may do that later. But I guaranty tea if he loses, don't count on it.
MR. SEIRAWAN Good point. Very good point.
MR. ASHLEY The guy gets out of here real quick.
MR. ASHLEY Well, he is all set and Kasparov has had tremendous practice against computers in general. He's played many 30 minute matches and 5 minute speed matches against the computer, but never an over the board match under world championshipconditions.
MR. SEIRAWAN Absolutely.
MR. ASHLEY What does that mean? Is this an advantage for the computer? Kasparov a bit --
MR. SEIRAWAN I think he's concerned about the lights. I mean that's 1 of the things that is disconcerting, of course, because the computer doesn't care what candlelight is going on.
MR. ASHLEY He's smiling, smirking, but I don't think he's too pleased at something. Maybe the lighting on the board. You know, it was interesting to me because people said that there's no psychology involved in --
MR. SEIRAWAN Oh, E4.
MR. ASHLEY I don't think it's an actual move. I think it's simply a test of the lighting and whether or not Kasparov will be disturbed.
MR. SEIRAWAN No he's writing down his move.
MR. ASHLEY He is writing the move down. But I think -- I wonder is there any possibility of psychology against the team. I mean after all if Kasparov were to offer the team a dsraw very early, what would they do? Deep Blue of course would not accept a draw, but what would the team do? It would be a tricky position for the team.
MR. SEIRAWAN They'll go into a committee like huddle and so on. E4, C5, a Scillian. C3, the all bin variation of the Scillian. Yesterday --
MR. ASHLEY Where is Fred Freedo? Is Fred Freedo around?
MR. SEIRAWAN You've got E6 on there?
MR. ASHLEY Yeah.
MR. SEIRAWAN 1 of the things that happened at yesterday's press conference is it was introduced or let known by c j tan, the leader of the Deep Blue programming team that they had engaged the services of grandmaster Joel Benjamin from New York. And I could see Garry on stage and he is tapping his foot and he started looking in the air and he thought to himself what openings does Joe Benjamin play? What openings has grandmaster Joel Benjamin induced the computer to play? So because I think of Joel more as a D2 D4 player, that was why my prediction was wrong and we had E4 today on the board.
You huh. D5. I was going to say that 1 of the things about Garry that is very predictable versus the computer which is a bit unknown because it was only recently that the new chips were even pressed, is that Kasparov loves to play the Scillian defense. This is his favorite defense as black. So this was quite an expected choice by Garry and the computer could conceivably have an check check.
MR. ASHLEY Is it a wise thing for Kasparov to do to play a rich opening like the Scillian defense against a computer that can calculate and brake down those complex its very quickly.
MR. SEIRAWAN I don't know. I think that Garry, he's not afraid. Let's be very, very clear. Garry is not afraid of the computer. So he's willing to mix it up. But I mean in a sense that is the computer's biggest thereat, is its ability to calculate forced variations very quickly. And of course what we see right now is just the very early opening, very, very standard opening choices by both white and black.
I was going to say that we also would enjoy your input, so if at all you have any questions, it's a little -- I'm getting a little bit of a feedback from the lights, but -- yes, sir? Par par.
(Question from the audience.)
MR. ASHLEY Night of 3 and bishop to G4 by black.
MR. SEIRAWAN It's just at the top. But I was referring to --
(Question from audience.)
MR. SEIRAWAN That's correct. It's being communicated remotely. Do we know the answer to that? Is it at core knell? It's in Yorktown New York much.
MR. ASHLEY It's at a research centerment they are connected over network, via network.
MR. SEIRAWAN Okay. The type of that they're going into is what is known as the all bin variation. And 1 of the things that gets distinguished in this opening is the Centre Square D4. The pawn that Garry has just captured, this pawn on D4 becomes is known as an isolated pawn. That is to say White's D4 pawn will not be able to be supported or protected by another white pawn. It has to be supported by its pieces. The isolated queen's pawn's positions have been debated for Part B To Be Updated Soon. Part C .
MR. ASHLEY I'm not sure the computer is raising its own time during a game, but it's already been told by the programmers not to take too long to think because it knows the time limit. I think they report more of -- they even let the average go down so that the computer never loses on time. Well, now it's attacked Kasparov's queen.
MR. SEIRAWAN So we have knight B5 queen on D6 is being attacked. And while we were just talking, I was thinking about the tactics and whether or not the computer has this opportunity. Can we run a variation?
MR. ASHLEY Sure.
MR. SEIRAWAN I was thinking of the variation queen to D7, just stepping back because the queen has been attacked and then the followup move, the knight to E5. Just looking at this, move knight to E5.
Now, the intent here, of course, is look there's a bishop on H5 under attack and there's a queen on D7 under attack. So now Garry's moves will be rather forced. Perhaps a variation like knight takes E5. We'll just take a quick look at that. That would come to recapture. D captures E5. Now, again, the bishop on H5 is under attack. Black could have a move like bishop E2 and then a move like queen captures E2, and now we have a problem for black because there's the threat of knight to D6 check interfering with black's ability to castle and also the knight on F6 is under attack.
MR. ASHLEY Yeah, that wouldn't be too pleasant.
(Laughter.)
MR. ASHLEY And for the first time, Kasparov has stopped to think about the position, which means now that the computer has thrown out the new move and Kasparov has to be very careful because 1 mistake, this thing is ruthless. Computers just put you down. I remember analyzing a position with a friend of mine, and we were looking at the position no problem, whatever. And we put it into the computer to look at, and it came back in our program, which was only Fritz 3 and the thing came back, thought about it for 15 seconds and said make an 8. What? Make what? And it was just exact too a fault, no mistakes. So if Kasparov makes any kind of mistake like that, he can just forget it. There's no pity being taken on him.
MR. SEIRAWAN They're remorseless.
MR. ASHLEY You're out, like the terminator. You're just gone. So this is 1 very interesting variation. Let's get back to the actual position.
MR. SEIRAWAN Which perhaps explains why queen D7 is a bad move. Another thought that's pretty good.
MR. ASHLEY That was good.
MR. SEIRAWAN Another thought that does occur is a move like queen back to D5. Well, the queen just came from D5, you say. So, yes, what black is saying is that you've developed that knight with tempo, so bravo for you. But what are you going to do to disturb the position? So a move like queen D5 is quite a provocative move, I must say. But it does threaten in some cases to play a move like bishop takes F3. If we could just put bishop takes F3 up on the board.
MR. ASHLEY Difficult to do.
MR. SEIRAWAN For example, room C1.
MR. ASHLEY Okay.
MR. SEIRAWAN Black has a threat after the move queen D5 and that would be bishop takes F3. Note that the bishop on E2 protects the knight on B5 and so you cannot play bishop takes F3 because of queen takes B5 losing a piece. So if Garry was just to retrace his steps with a move like queen D5, well, the computer would have to do something creative to follow up the move knight B5.
Any questions? Yes.
(Question from audience.)
MR. SEIRAWAN The question is whose side do I prefer. At this moment in time, I really think it's -- it's even, you know. If I had any slight preferences, it would be for white. And that is because I'm already castled. I mean I like safety. Now, the computer doesn't really understand safety. The computer kind of understands threats, but I love a nice safe king. Everything is okay, this is good, we'll go on to the next step. And in this particular moment black needs to bring his king into safety.
MR. ASHLEY He's 1 tempo behind.
MR. SEIRAWAN He's a step behind. He's a step behind.
MR. ASHLEY Yeah. And that could be a big deal because the computer could start cooking up some threats, and your king is sitting in the middle of the board, and all of a sudden, boom, you're out there.
MR. SEIRAWAN I'd just like to echo something that Maurice has just said. I love to play against computers. I love to play blitz chess, and I was playing blitz chess against a particular program. I think it's chess machine. So I'm playing the program and I'm just cooking. I've built up a winning position, and I know it. But I can't help but my eye keeps straining to see what the computer is analyzing and what it thinks of my brilliant play and sure enough I look at that knight. I say to myself I've got a winning position. I've got the computer. And it was also something like meet in 6. But I just don't see it. It is in a nanosecond. It's pretty unnerving.
MR. ASHLEY It's brutal. And, also, to think about it, the computer is thinking on Kasparov's time right now. Now Kasparov has been thinking for 8 minutes. Now if the computer can do a hundred billion positions every 3 minutes, that's 200 billion every 6 minutes, that's 300 billion by the time Kasparov moves. And if he takes a look at it longer, the computer has the game worked out to meet.
MR. SEIRAWAN Well, that's the idea. But it's very interesting because last night I was very, very fortunate and I had a meal with about 8 premiere scientists, organizers here from ACM and IBM and other institutes, and we were talking about the fact that the computer literally predicts my every move.
When I play against the computer, and I win, it's not a surprise to me at all because I'm expecting to win. But what happens is the computer anticipates my every move. I've never beaten a computer that hasn't anticipated my every move. So it's possible that Deep Blue could anticipate Kasparov's every move. It just doesn't understand how strong his moves are. Sometimes I don't either.
MR. ASHLEY Until it's too late, and then he realizes that Kasparov is indeed winning.
MR. SEIRAWAN That's right.
MR. ASHLEY And that's the advantage. The experience of the top grandmaster to play strong moves without even needing to know what your next move is. I mean, there are many positions you can play where you don't have to think about it. I got you baby.
MR. SEIRAWAN Whatever you do you're going down.
MR. ASHLEY Right. The computer, by its nature, analyzes everything, has to look at everything, and at the moment Kasparov is trying to do his best to look at everything because this new move, knight to B5, has stopped him in his tracks, and his time now at 147 and ticking down 1 hour and 47 minutes ticking down; the computer 146. He is about to step below time. But not critical.
There are moments in a game when a grandmaster will collect his thoughts in a new situation like this and try to figure out exactly what's going on. And Kasparov thinking in this situation he's got a lot of time if you check the average 3 minutes a move. It's move 13. He could have used as much as 39 minutes and still be okay. So he's all right for now. He just has to figure out what to do. Part D To Be Updated Soon. Part E
MR. SEIRAWAN There was a suggestion of bishop F4.
MR. ASHLEY Is that good or what? Now that seems as if the black bishop has left that sensitive side of the board. Bishop F4 seems very interesting.
MR. SEIRAWAN Before I look at the move bishop F4 too far into depth, let's be clear. This is a short-term threat. This is a short-term tactical threat. The move bishop F4 works in conjunction with a knight on B5. What the anticipation is to either play bishop to this square D6 or this knight to this square D6. So the move bishop F4 is really a kind of one move, Charlie. It makes a single threat. And if black meets that threat -- could you put the move rook D8 covering the square D6?
We're getting used to this pad. By Day 3 we're going to be great. Day 1 it's a little tough.
You just stopped the move bishop D6 and I developed a piece.
MR. ASHLEY Put pressure on the d pawn, secured his queen side. Maybe too many positives for black in this scenario.
MR. SEIRAWAN Well, let's go back to the position on hand. And let me just get a -- I love doing this. It's a political year and polls, polls are very important. Polls are very, very important. They determine everything. And right now I just would like to get a quick poll. Who likes Kasparov's position?
(Members of audience raise hands.)
MR. ASHLEY I think you scared them. Only 5, 6, 7, 8 people.
MR. SEIRAWAN Who likes IBM's position?
(Members of audience raise hands.)
MR. SEIRAWAN Who thinks the position is rather equal, balanced?
(Members of audience raise hands.)
MR. SEIRAWAN Well, I think we have more people that actually think IBM is ahead. Well, I don't. I like Garry's position.
MR. ASHLEY You're a minority in the room.
MR. SEIRAWAN Because I'm anticipating that there's really only one way for IBM -- I should be saying Deep Blue -- Deep Blue to pursue the position, and that is the move knight on E3 to F5. This seems to me to be the most consequential move. Should we take a look at it? Knight F3 to E5. Okay. The idea is, because white is slightly ahead in development, what he wants to do --
MR. ASHLEY Oh, before you could even finish the thought --
(Laughter; applause.)
MR. SEIRAWAN We're lucky. We're lucky.
MR. ASHLEY Grandmaster title wasn't in the bubble gum wrapper. Well, it has indeed played knight to E5. And why don't you finish postulating from on high.
MR. SEIRAWAN What I was going to say is this seems to me the only way of building up on a slight development. This forces bishop takes E2.
MR. ASHLEY Which has been played. Kasparov whipping off his move.
MR. SEIRAWAN Queen takes E2 is virtually a forced response. So we can anticipate that quite quickly. And then what we're going to see is that it's very important what's going to happen on the c file. The c file is going to become the crucial theater of battle. Do we all know b, b, c's, 1, 2, 3's.
MR. ASHLEY We hope.
MR. SEIRAWAN As it relates to chess.
(Laughter.)
MR. SEIRAWAN Each square has its own separate address or coordinant. So you have the a file, b file, c file, d file, all the way so on to the h file.
MR. ASHLEY From left to right.
MR. SEIRAWAN And then we have the ranks, the first rank, second, third, fourth and so on. So the lower left hand corner square is the A1 square. The upper right-hand corner square is the H8 square. So what you're going to see is a move queen takes E2. We'll put that on board. And then we're going to see the move castles finally safety. Safety. Then we'll see the move castles.
(commentary from the audience.
MR. SEIRAWAN We'll get to that in a moment. Then we're going to see the move knight takes C6, b takes C6, knight C3.
As a chess grandmaster this is the type of position I would look at, and that's why I wanted to take a poll. Because this is going to lead directly into an end game after knight C3, where you have the situation of the pawn structure. Look at black's pawn structure. He has 2 split pawns, the A7 and C6 pawn. And white has also 3 pawn islands, the A2 B2 pawn, the D4 pawn, F2 G2 and H3.
Now, one of the former world champion's, Emanuel Lasker, wrote a brilliant thesis on these types of endings, and he postulated endlessly and, of course, he was the world champion, so he knew what the heck he was talking about. And in these types of positions he favored black, and it almost seems, as we've just seen, kind of a logical progression of moves to get to this type of end game. It's a long theoretical argument as to why he did. Perhaps we'll get into that a little bit later. But for the moment let us just leave it with the idea if this position is forced, I'm going to favor Garry.
MR. ASHLEY Now, what's interesting too, and you make this point. Another thing, just going from experience of these types of positions, I know that when I have an isolated pawn position, I absolutely never want to give up my light squid bishop. I love that light squid bishop. That light squid bishop attacks like crazy. And to see the exchange of the light squid bishop makes me cringe a little bit because I'm thinking that's a great attacking piece. Now why is that? Because I know that piece is used to pressure the black king in many, many castle positions. That bishop be spinning around and sometimes you sacrifice it, but it aims a big gun at the black king side castle position. That's speaking from experience. The computer could care less what I think. But watching that exchange makes me feel like some of White's attacking potential has been drawn out. The sting has been drawn out watching.
And the more exchanges, another general principle that we know about these types of positions is, the more exchanges for black, the merrier. Because that d pawn is going to be weak for a very long time. That D pawn isolated in the middle of the board will be weak and maybe in an end game white will end up maybe even losing that pawn. So a dangerous situation has arisen for white in that he's going to have to prove maybe tactically that his position has holding up.
(Commentary from audience.)
MR. SEIRAWAN That's a good question.
MR. ASHLEY What was the question?
MR. SEIRAWAN The question was if the recapture queen takes E2 is such an obvious move, well, why don't he just do it? And that is to say why doesn't Deep Blue just immediately make the recapture? And that's again not my area to understand. I don't really know why sometimes the computer will go into a deep thought on a position where I expect the answer is immediate and on other positions where I think oh, man this is a most complicated position I ever saw in my life. This thing is going to spend, you know, 6 days coming to a solution. It's answer is immediate. So one of these days I'm going to have Hans come up --
MR. ASHLEY Maybe somebody can answer this for us. Why does the computer in some obvious situations not respond obviously and instantly?
MR. SEIRAWAN A recapture.
(Commentary from audience.)
MR. ASHLEY Basically the question is answered by the fact that the programmers themselves don't want the computer to rush into what may be a premature decision.
(Commentary from audience.)
MR. ASHLEY So that's the balance that one makes. In what looks like obvious situations there may be something in the long term and you kind of are funneling for that, some kind of hedge factor.
(Commentary from audience.)
MR. ASHLEY Interesting. Another very profound point by Hans Berliner is that maybe the computer will just calculate and store those positions in its memory, and after it makes its obvious move, if you make one of several responses, it will respond instantaneously to your move. So, in fact, it wasn't just randomly analyzing forever. Some very interesting points. Well, Kasparov has castled, and a position you talked about, Yas, may result after knight takes C6. But instead it has played the rook to C1, rook a to C1, and is now, in earnest, threatening to capture the knight on C6 and therefore win a pawn.
MR. SEIRAWAN Right. And this is also something that I've often kind of thought about regarding computers. And I really feel sorry for these poor programmers. I really, really feel sorry. What I mean by that is not only are the programmers trying to give the pieces values and obviously in some positions a rook is not as strong as a rook should be, and it's actually rather impotent. So what happens is it's very difficult for the programmers to know which are the strongest pieces and also it's very difficult for the programmers to know which squares to control.
I just mentioned a few moments ago that the new theater of battle is going to be the c file. And of course we saw the move rook C1. But when you start the game, when you start a game of chess, you say, well, the most important files are the central files, the e file and the d file. So you want to teach your computer program bring your rooks to the center. And sometimes, well, the center files aren't open and suddenly it's the c file becomes the most important file, or in other cases the f file. He might have played F4/F5. So now we have the c file, and it always -- I feel an empathy towards the programmer when the computer doesn't realize that it's the c file and not the d file that is the crucial one.
Here what Deep Blue is doing is he's laying a trap. Okay? He would like to see the human Kasparov play knight captures E5.
MR. ASHLEY Laying a trap.
MR. SEIRAWAN Laying a trap.
There you go.
Knight takes E5. D captures E5. Now that bishop on E3 that wasn't bothering anybody a moment ago, look at this. Suddenly after knight D5 comes the square bishop C5. And the queen and the rook are on the same diagonal and black will lose in exchange.
MR. ASHLEY Ouch.
MR. SEIRAWAN So the idea of knight -- rook C1 was to reinforce control of the square knight D5. Now we're going to see the move, I think, rook F8 to C8. And this is also very curious too. There's always this fight that chess players have, which rook, which rook.
MR. ASHLEY Well, he's chosen the wrong rook. That Garry Kasparov thought he was world champion.
MR. SEIRAWAN Just to finish the thought, always for chess players it's a really tough struggle, one of the hardest chest struggles to make, which rook. And I didn't want to say as an absolute because he ended up choosing the rook. I would have chose the other rook for the other reason. Garry wants to preserve the option of rook on A8 -- rook on F8 to D8 come to the central file. But we're get into that end game that I was speaking about earlier. Can we actually go back a move or two a moment.
MR. ASHLEY Yeah, sure.
MR. SEIRAWAN Try rook on F8 to C8, if you can do that.
MR. ASHLEY No, we don't want to do that. You want to do rook F8 C8?
MR. SEIRAWAN Right. I'll tell you, we're getting better and better.
MR. ASHLEY I'll get it folks. I'll get it. There.
MR. SEIRAWAN You got it. And then the idea was that after knight takes c 6 I'd like to take with the pawn. Would b take C6? And now the reason comes a little bit more clear why I wanted to choose the rook on F8, because the b file is open for black. So that after knight C3 bishop takes C3, rook takes -- whoops.
MR. ASHLEY Go ahead. Bishop takes, rook takes.
MR. SEIRAWAN Bishop takes, rook takes. Rook on A8 to B8. That was the idea that I had, was to reserve that rook for the b file.
But Garry already has his own ideas.
Let's go back to the game position. That's why I say it's always a tough struggle that chess players have determining which rook they're going to use.
Any comments? Questions, inquiries? Sir?
(Commentary from audience.)
MR. SEIRAWAN Actually that's a very good question. And again I hate to do this. Hans is probably not a paying customer. He's certainly an invited person. We'll get our pound of flesh from Hans on this one.
Hans, the question that the spectator had was are programmers chess players? Are they chess grandmasters or does it really matter?
(Commentary from audience.)
MR. SEIRAWAN Has responded by saying that it's actually more important that the programmers be excellent programmers and that when a problem develops they'll simply -- the programmers will scurry to a chess player and say excuse me, could you answer this for me and then they'll go on and program in the answer.
So the answer to the question is, in Hans's view, far more important that they be good programmers than chess players.
MR. ASHLEY This last move, bishop to G5 has been played.
A question down front. Yes?
(Commentary from audience.)
MR. ASHLEY The question is by a young and upcoming Kasparov -- and I think he's answered his own question. He wanted to know why not rook captures knight on C6. Well, there's a pawn protecting it. That would be very dangerous since rooks are better than knights in 95 percent of situations.
Now this move, bishop to G5, pinning the black knight to A6. And the game is starting to get a little tense. Pieces are starting to touch that bishop leaped out to G5 and maybe threatening in some positions to take that knight on F6 and create some sharp dynamic possibilities. Clearly white is being a little bodacious in its position, looks to go do some damage.
MR. SEIRAWAN Let's go back just a moment to before that move and if we see the position before DEEP BLUE's move, the bishop on E3 is actually not in the game. I mean this little trap of bishop C5 that we saw earlier, that's not happening. And the problem is that from White's view the bishop may not get a chance in the future to play a role in the game. So I actually like this move, bishop G5. I think this was the best move because in some positions now white has the threat of knight G4. If we can put say rook D8, we'll move a rook on F8 to D8.
There's this possibility of knight G4 when black's pawn shield is going to get disrupted on the king side. So I like this move, bishop G5. I just think that that bishop could potentially have been useless.
(Commentary from audience.) Part F
MR. ASHLEY Maybe he was just scratching his nose or something. Kasparov does not -- admiration? I'm not sure. From those of us who know Garry, he looks at many people with admiration.
MR. SEIRAWAN I would just like to buttress that with a little story concerning the view of admiration. I've played Garry 5 games. I've played Garry in hundreds of offhand games and in competitions, and the overall score is two wins for Garry, two draws and a win for myself. So out of the five games I get to tickle him every now and then. You see? But I wouldn't have expected the results to those particular games. Because I had 1, I was winning. Those were the two -- one I won and -- no. The two games that I was winning both ended in a draw. The two games that I lost were drawn.
MR. ASHLEY The 2 games that you lost were drawn?
MR. SEIRAWAN I was avoiding repetitions. And the one game I won I was worse. But what happens in 1 of these games typically -- and, again, it's a story of admiration, is I had made a move that Garry really didn't like and rather than, you know, say hey Yas, you know, that was a really good move, you know, I got an evil scowel.
(Laughter.)
MR. SEIRAWAN He was not happy. But that was a look of admiration. Thank you, Garry.
(Commentary from audience.)
MR. ASHLEY They're different operators. There are 3 operators. All the members of the team in the room are rotating and each 1 will be a new -- become a new operator. Theres not one specific. He may be the head of the whole operation if he is in fact c j tan that you're talking about.
MR. SEIRAWAN No, this is not c j.
MR. ASHLEY But even if he was, they're rotating it the operational duties. It's not 1 specific operator, as you'd have in a world champion match. One may work for an hour and then have a break and then another will come in. All the programmers will take a chance to be near royalty. King Kasparov playing in this epic match. They would love to put in some moves. And bishop to king 6, Kasparov's relatively quick response, eying that d pawn, which has lost some of its protection after that last move. But, Yas, this is a very sharp continuation by the world champion. Of course, the computer could carry out an exchange combination on the F6 square, bishop captures F6. And if queen takes, then knight to D7 would win a rook for a knight. This would be a very bad move on the world champion's part. Clearly the computer would be winning in this situation, I think, from what I'm seeing. And if pawn takes, if pawn takes instead, then Kasparov's king has been shattered. King's side pawn has been shattered. How bad is it? Because obviously Kasparov has evaluated the situation. How bad is it for his king to be opened like this? Especially with a computer, you know. It gets your king in its sights...
MR. SEIRAWAN Before I just answer that question, I was just going to say the gentleman up front had said is the operator C.J. Tan. In fact, I had I've seen C.J. walking around, and I'll certainly do my best to rope him up here and to let you ask questions about what they've managed to do. It's truly an engineering marvel what IBM has created, and there's some fantastic things --
MR. ASHLEY That's actually Feng-Hsiung Hsu. He's the top guy working on the computer itself. Tan is the overall manager of everything, but this guy -- it's really this guy's baby.
MR. SEIRAWAN What you see now on our analysis screen is just as Maurice has described. Black has to -- black's pawn shield has been compromised, but the result is that he will win a pawn. There's no way from this position that white will not have to lose that D4 pawn.
MR. ASHLEY Well, check this out. Fritz is loving white's position. It's going ga-ga. Fritz 4, we have analyzing for us, is -- it's just loving white's position with a point 44 advantage, which is closing in on half a pawn, which is quite significant. I would say anything like point 1 or point 2 you could sort of dismiss. When you start hitting point 4.5, it's as close to having a real spatial advantage, positional advantage as the computer is going to get. Point 7, point 8 it thinks soon it's going to win material from you. You're going to have to give up a pawn sooner or later. But just maybe this is a result of the fact that computers put a high premium on some positional factors. For example, double pawns and exposed kings. And if the computer sees that you have double pawns and exposed king and material balance is maintained, it will really like its own position. It will say you're at a big disadvantage. But strong players know that that's not always the case. Double pawns can be very powerful and sometimes an exposed king is not so exposed.
MR. SEIRAWAN Indeed. And this is one of those cases.
MR. ASHLEY And it has captured on F6.
MR. SEIRAWAN This is one of those cases that Maurice was talking about earlier is when he has an isolated Isolani position, he hates to lose that light squared bishop because he knows just how powerful a piece that can be. After so many pieces have been traded, black's pawn shield isn't going to be quit so crippling. It's not going to cause him such great disruptions because his king is still safe enough despite the fact that his pawn shield has been compromised. So again, I think that after g takes F6, a forced response, we'll see Fritz has moved knight C4. There's several moves. Queen G4 check. When I was a beginner, I loved to play movers like. I just loved to check my opponents king. I thought that was the cat's meow. Check, I would say with great happiness. And the reality is that there would be a case of patzer, meaning weak player, sees a check. Patzer plays a check. The idea is after king G4 check king H8 white has no followup.
Okay. You should especially as beginner players be very wary of checking your opponent's king unless you can see a followup. After the move king H8 sometimes the move rook G8 will come. The rook will come and counter and attack white's queen, and suddenly it will be black who takes over the initiative. So queen G4 check is probably not going to be played. Knight takes C6, that's -- knight C4?
MR. ASHLEY Knight C4 is being -- he's thinking.
MR. SEIRAWAN I'm just thinking about knight takes C6, rook takes C6. Remember that when your king is under attack the more pieces go off, the better for you. So the idea being to encourage an exchange of rooks. Rook takes C6. B takes C6. Knight C3. And now we have a situation that is an unclear ending. If you play bishop takes D4, of course then there's the trap queen G4 check and that really does hurt because then you pick up a bishop. Does everybody see that?
MR. ASHLEY For those who don't --
MR. SEIRAWAN Bishop takes D4 would be a no-no because of the move queen G4 check.
MR. ASHLEY This is the kind of thing that the computer nor Kasparov will fall for. You can forget about it.
MR. SEIRAWAN So knight C6 is a possibility and knight C4. I know that Fritz has for some time been busy himself with the move knight C4.
MR. ASHLEY That doesn't look like a bad idea. After all it threatens to capture the bishop on B6 and maybe double those pawns even more, leading to some extra weaknesses in black's position. But I think this is precisely the kind of situation that Kasparov feels the computer will never defeat him, never catch up with the top human mind, which, of course, he always means himself.
(Laughter.)
MR. SEIRAWAN That kind of goes hand in hand.
MR. ASHLEY But he is -- you know, give the guy credit, he is the Michael Jordan of chess. And he feels that this is such a complex problem that you always have to tell the computer something that's bad. You have to say double pawns are bad. Weakened king you want to avoid. Knights are worse than bishops in most situations as computer plays knight to C4. It's better to have a bishop than a knight in most situations. Some kind of parameters to give the computer guidance as to how to play the moves, not just brute force calculating all the way down. But some sense of strategy. Once you give a priority to the computer, that alone is going to hamper it somewhat. The flexibility and thinking that a human can do and say, wait, this situation is not so bad for me, a computer would not rejudge, reevaluate its entire foundation of its thought because it's in there. It's there. This is what the rule says. And so I'll obey it time and time again. Human beings and the top grandmasters find the fact that they can be the exception to the rule.
MR. SEIRAWAN Well, in chess, of course, you have what is known as the paradox. I mean you have a situation where you say, okay, the queen is the most powerful piece on the board. And yet sometimes the best way to win a game of chess is to sacrifice the queen. So there is a great deal of paradox in chess. And I'm just looking and, you're right, Fritz is getting very happy about its position.
Sir?
(Commentary from audience.)
MR. SEIRAWAN The gentleman just asked us if we could recapitulate the last move. I'll just talk our way through it. If we could just go back. What white had just done is play the move bishop to G5 pinning the knight. And then Kasparov had played bishop to B6, attacking the pawn on D4. The knight was captured on F6. And what Deep Blue's intent was to force the pawn to capture because if the queen captured, as Maurice pointed out, there was a fork with knight D7. So the pawn capture was forced. And now DEEP BLUE is quite happy because it said, well, I've broken up your king side shield a bit and he played knight C4. Is DEEP BLUE a he or a she? Knight C4.
PARTICIPANT Got to be a he.
MR. ASHLEY It's so stubborn, got to be a he.
MR. SEIRAWAN Now it's getting very interesting here because the whole fight is going to revolve around whether or not the isolated pawn on D4 is strong or weak, is strong or weak. If white ever gets the opportunity to push the pawn up the board forcing an exchange of pawns, we'll know it's strong. If black conversely is able to blockade the pawn and force its capture, we'll have known it's weak.
What I'm anticipating is the move rook on F8 to D8 and now the move rook on F1 to D1.
MR. ASHLEY Just to hesitate for a moment, Fritz thinks that move is not too good, analyzing alongside with us. It thinks that it has a huge advantage and it's even looking like 15 different ways to kill the move.
MR. SEIRAWAN 41 impact.
MR. ASHLEY Yeah. What am I talking about? 15. But it thinks after a move such as this one it can play knight takes B6 and have a serious advantage. But I have a feeling that this is all based on the fact that it just thinks that double pawns are bad, always bad, and the more you have, the worst your position. And so I don't really have to look at anything else because you just -- you're just going to have an awful position once I inflict this situation on you.
MR. SEIRAWAN Again, as Maurice earlier said, I mean obviously double pawns are a weakness. It's better to have the flexibility of having nondouble pawns, but it doesn't always mean that double pawns are bad.
And, conversely, double pawns can oftentimes be good because they control a lot of squares. So I don't put -- and again, it's the arrogance factor as far as the human being is concerned. I'm contemptuous of what Fritz or DEEP BLUE might think of its advantages because Garry has double pawned.
MR. ASHLEY Let's remind those who are not with us here because we're being broadcast on the worldwide web. All over the world people have taken an interest in this match. I should remind that here we have Fritz 4 helping us analyze the battle going on in the room itself with Kasparov versus DEEP BLUE. So when we refer to Fritz, don't think we're going crazy. Fritz is actually just helping us to analyze.
MR. SEIRAWAN And it's a program.
MR. ASHLEY And he's also a program.
MR. SEIRAWAN But just to continue the thought, so in this type of position, again it's not a judgment about the double pawns. The judgment has to do about the pawn on D4, if the pawn on D4 is weak or strong. And that's just the key question. Again, the double pawns I don't care about. And so I discount the advantage that Fritz or even DEEP BLUE may be awarding itself. The reality is if black can play moves like rook on D8 to D5, rook on C8 to D8, queen to D7 and just take the pawn, white is going to be in a lot of trouble.
Conversely, if white can ever force the break D8 D5, white will have an advantage. And that's the fight.
And just to continue the thought rook on F1 to D1 protecting the pawn.
(Commentary from audience.)
MR. SEIRAWAN We don't know yet. We're just trying to get that in. Now there's all kinds of antics going on, but one of the things that I would like black to do is to play the move F6 F5. The idea is to prevent the move queen G4 check and in some cases simply to play queen to F6 protecting the king a little bit, but also saddling up to the attack of the pawn on D4. And again this is a rather consequent position. There are a lot of alternatives. I'm just saying this is a rather consequent position. Is the pawn weak or strong? That will be the deciding factor in the outcome of the game.
MR. ASHLEY What's interesting too, watching Fritz 4 analyze this position, is that a while ago it had a point 8 advantage for white, as soon as the move was played point 8 advantage for white. But then as soon as a move was made like F7 -- F6 to F5, suddenly it reduced its advantage to only point 4. And that happens again because every move that's made lets it see a little bit further down into the tree what is possible. So while it may calculate only so much in a given situation and think that it has a great position of, once you make a move it goes that little step further and it goes hey this is changing. And it can evaluate it has a great position now and just 2 moves a little bit it will suddenly change its mind because that is a function of what it's doing. It can only analyze so much. And Kasparov's experience tells him I don't need to think that far. That rook is on the d line or whatever and that's all he needs. No need to calculate, well, make sure he's not going to lose in some situation, but still make sure he's going to have a good position and the computer will have to catch up to him and his great ability to know just what's happening instead of having to do all that calculation. At the moment though he is thinking, and I should tell you the time. 124 for the computer left on the computer's time, and Kasparov with 108. So Kasparov down 16 minutes on the clock. But not really relevant at this stage of the game, this move, knight to C4. Again it's knight to C4 was White's 19 move. And Kasparov has in fact played your move.
MR. SEIRAWAN Yea.
MR. ASHLEY It's kind of unfortunate, Yas, that you know all these good moves and he's the one making the money.
(Laughter.)
MR. SEIRAWAN I've often wondered about that, Maurice, I truly have. Thank you for pointing that out.
MR. ASHLEY And this position has arisen and Fritz 4 believes that the advantage is tremendous. Borderline winning a pawn in this situation. If it gets to 1 point 0 it means definitely I will win material. Fritz will not give itself 1 point 0 no matter what compensation it has for the material unless -- no matter what kind of big space it might have. We might see a rosy position and think I'm just killing the guy. Fritz won't think like that unless it sees I can win material concretely. It will give itself point 99 I guess before saying that it has that great an advantage. If it doesn't see how to actually win a pawn. Once it gets to 1, that means I see a pawn. I'm taking it. So right now it's at point 69, dropping back a bit as the depth of analysis goes further and further, and now it's going to have to come up with a pretty strong move. DEEP BLUE will have to come up with a pretty strong move to try to consolidate its position. The question again, Yas, that d pawn. What's up? Is that going to die?
MR. SEIRAWAN That's what we're going to be finding out fairly soon. Because of what Maurice has been talking about and what all of these fractions and 1 point 0 means, as far as Fritz is concerned.
I've invited Frederick Friedel, a chess computer specialist from Germany, to join us in a few moments so he could speak specifically about the strengths and the weaknesses of the computer. And again one thing that I would really like to stress is that the advantage that Fritz gives white in this position is just as you say. It's built upon its horizon, what it can see, how many moves it can evaluate. It may be we evaluate 2 ply. I don't know. Just for argument's sake, 6 moves down the road and it still sees that it has an advantage after 6 moves. Whereas Kasparov is seeing oh, a series of patterns. He's seeing this idea of trimming on the file, rook D5, rook D8, queen D7. He's seeing the idea of playing F5, queen D6. And Kasparov may literally have in his mind a plan of action that might be 20, 30 moves long. So he's not looking so much to say at the very short-term consequences of 1 or 2 moves. He may be thinking, boy, all I've got to do is keep this under control for another ten, fifteen, 20 moves and I've got it. It's in the bag.
MR. ASHLEY He doesn't see all those moves specifically, does he? He doesn't calculate all the possibilities because, for one thing, he really couldn't.
MR. SEIRAWAN That's right. It would just be he'd run out of time on his clock. It would be physically impossible. So I mean basically what black would like to do is simply prevent a pawn on D4 from moving forward. And again that would probably be based on the moves F5, queen F6, rook d 35 and so on.
A spectator had asked earlier why rook D1, the pawn is not under attack because again this is just a type of move that's going to be necessary in the future. One of the things that I'd like to point out, one of my favorite operations is the old rook lift. Are you all familiar with the rook lift?
No, tell us. Are you kidding me? Okay. Go back. Don't put the move rook D1 in. Well, the old rook lift involves the move rook C3. Now, the idea is that the rook lifts itself up the board, usually 2 squares and the idea is it then slides across the rank to a square like G3, check, and then it will set up a beautiful checkmate. So the old rook lift. Knight takes D4 wouldn't have been playable. That's tactical because just knight captures D4 loses a piece.
MR. ASHLEY Kasparov has in fact played F5, not to interrupt you, Yas, but your genius is just undiminished.
MR. SEIRAWAN He is so good. Okay, D5 -- F5
MR. ASHLEY You sure you're not passing moves on to this guy?
MR. SEIRAWAN Does this sound system go directly to his chamber there?
Okay. But the idea of the rook lift was sometimes to bring the rook across and especially against an open king.
MR. SEIRAWAN What I was going to say, it wouldn't have worked because there's not enough pieces on board. But this is one my favorite operations.
MR. ASHLEY One thing I'd like to bring to the attention of the audience is what I was talking about. Just Fritz 4 is obviously not DEEP BLUE and it works on different parameters and different ways of analyzing, but just to show you what I meant, Fritz 4 thinks it has a point 84 advantage. That means decisive. Almost soon to win material. That's my point. Almost bordering on decisive with point 84. Now,
A. Let's see what it thinks.
(Commentary from audience.)
MR. ASHLEY That's possible. The point made, that it hasn't seen that far yet. But we'll let it sit on the position for a little while, and I don't think it's going to reduce what is said because this position is not about depth. Depth in this position would be if I found a checkmate. This is not about how far it's analyzing. This is about how bad black's king side is and black's pawns are. So if it's depth, the computer was counting on the check. Here it's only basing on the pawn structure. It's not going to change for a long time from this evaluation if at all. In fact, it's played very quickly. DEEP BLUE has in response to F5 the move queen to E3. It would be very interesting to know what DEEP BLUE thinks about this position.
But now after the move queen to E3, which Fritz is liking as it brings its evaluation up, maybe these computers can be partial too. You know what I mean? They could be good buddies, Fritz 4 and DEEP BLUE.
Let's introduce Frederick Friedel as Grandmaster Seirawan takes a break.
(Applause.)
A gentleman knowledgeable about a thing or two with regard to Fritz 4. Fred, what other limitations of creations like this 1 as we see that they're very strong in calculating, but we know they have some limitations?
MR. FRIEDEL The limitation of the computer of Fritz and of DEEP BLUE is that there's a horse end somewhere. They can calculate up to a horse end and behind that there is nothing. There is no consideration that this might lead to a win or anything. It just doesn't exist. And so if you can calculate tactics or strategy, strategy tends to turn into tactics if you go very deep, deeper than the horse end, and you can always win. The problem is they're going deeper and deeper. These guys are probably going 12, 13, I don't know how many ply and you know it becomes progressively more difficult.
MR. ASHLEY But what about this point, though, that it it seems to me that the computer always puts -- always has to put some kind of credence on certain possibilities. It will say double pawns are bad, weak kings are bad.
MR. FRIEDEL Yes.
MR. ASHLEY Bishops are better than knights.
MR. FRIEDEL Yes.
MR. ASHLEY And there's some situations where that's not true. A weak king is fine. I mean it's not really weak. A double pawn is controlling a lot of squares. You'll never access to those squares.
MR. FRIEDEL So the computer knows nothing about that.
MR. ASHLEY Will it simply misevaluate that kind of situation?
MR. FRIEDEL Absolutely. But it's evaluating millions of positions in the process. So you don't know -- you know, it might find something else which makes this irrelevant and still play very good chess. You cannot imagine how many -- now Fritz here is looking at maybe on this machine 100,000 moves per second, positions per second. You cannot imagine how much junk is in there. And there are positions you wouldn't believe if you saw them. With a white king all the way on G6 or H6 and white has no pieces left and Fritz is saying uh-uh, bad. And so these things -- you know, it misrepresents, misevaluates a double pawn. So these are tiny little problems which it overcomes by looking at thousands and thousands of other good positions.
MR. ASHLEY Well, does it -- for example, this 1, where Yas feels that double pawns are completely irrelevant. In his estimation double pawns are completesly irrelevant. The isolated pawn is the real thing. But does Fritz have it in its mind's thinking double pawns are bad?
MR. FRIEDEL Yes, it knows that this is bad. It's bad for black. Very primitively it's very bad for black. Look at the double pawns. Look at the pawn structures. It doesn't look good for Fritz.
MR. ASHLEY And that's been programmed?
MR. FRIEDEL Yes, because we're doing this for every single position in chess. For every potential position that might arise from chess. Fritz knows nothing about this position. There are certain exceptions. And that's Garry's, you know, main chance, is that Garry understands this is an exception. Here a double pawn may be good. Here the isolated pawn is weak, et cetera. And if he understands that and if he gets positions in which the computer doesn't understand this but he does, he wins.
MR. ASHLEY Well, the computer never really understands positions anyway. So when you say understand, you mean does not misevaluate based on what you in fact have put into the program?
MR. FRIEDEL Yes. If it doesn't astray, too far astray.
MR. ASHLEY So is the computer's limitation a human limitation of being able to -- just not being able to program it in the best way, in the optimal way because you have to give it these parameters?
MR. FRIEDEL Yes, because we're doing something completely different. The difference between -- we're not trying to teach it how to play, to think like a human being, not at all. It does something which is so different that you can hardly imagine it. Just look at the numbers. Garry is looking at 2 positions per second. How much do you look at?
MR. ASHLEY About half the number weighed.
MR. FRIEDEL Garry looks at 2 point 1 or something per second and it's looking at maybe 200 million. So obviously they're doing very, very different things. The amazing thing is they have a chess board in between and each of them is making moves. So that, looking from the outside, you can see 2 intelligent entities playing chess against each other. Looking on the inside you say 1 is doing something which they call playing chess. The other is doing something which is completely different. It's just, you know, crunching numbers at a stupendous rate. And then saying okay knight G5 and it's translated that into a move.
(Commentary from audience.)
MR. FRIEDEL There are a number of reasons.
MR. ASHLEY Repeat his question.
MR. FRIEDEL Why can't you program some sort of -- for example, to say that the queen side looks weak? Is that your question? And then to play accordingly. A number of reasons. First of all, what do you mean by weak? We've consulted grandmasters. We have them. They work very, very intensely. They spend hours, days, weeks trying to explain to us what that means in technical terms. And we get the impression they don't even know what it means because it's so vague and so completely hopeless to program. But then on the other hand you see 2 grandmasters talking about the same position and 1 says weak and the other says weaker and they've understood each other. Even if they don't speak the same language, they just look at the queen side and do this. And these people understand each other. So obviously they're talking about something very real, but in a way we cannot, you know, put into our thinking. We've tried and tried and tried.
The second reason is if we do find certain things which apply in certain positions we put it in and Fritz, for example, is looking at millions of positions and every time it's saying to itself does that apply. Let me check. Then it goes and starts to check whether that applies. And what happens is it slows down. And, you know, putting in more knowledge means looking at less positions. And it sounds very seductive to put in huge amounts of knowledge. But the problem is that by going a little bit deeper, you usually are stronger by putting in a lot of knowledge. Part G
(Commentary from audience.)
MR. FRIEDEL
MR. ASHLEY The question has there been any application of chaos theory in chess.
MR. FRIEDEL I have done no work personally on chaos 3 in chess or on neural networks and so on. These are which I've heard a lot about, but I've never seen any.
(Commentary from audience.)
MR. FRIEDEL I haven't looked at it and I don't know of it.
(Commentary from audience.)
MR. FRIEDEL He says he's done some work on neural networks, but he's seen some work on neural networks and this is the same here. People have proposed projects to us, to the Fritz people. You know, let's -- I have a neural network system and so on. We'll try something. But it's very basic research which I think should be done at universities and they should tell not only the chess community. The chess community probably won't be interested to know because it's not strong. It doesn't play chess. It doesn't play good chess. But I suppose other scientists would be very interested to know what the results are.
MR. ASHLEY Let me ask you this question What is the basic difference between DEEP BLUE and a program like Fritz or the ones that are commercially available? I mean I realize it's quicker.
MR. FRIEDEL 1,000 times faster.
MR. ASHLEY But are they using a different type of technology also than the average computer.
MR. FRIEDEL Slightly different, but the difference between DEEP BLUE and Fritz is very, very small compared to the difference between Fritz and some other PC programs I know. So they're very close. I mean, all these programs have in the end got down to doing brute force searches, doing the same -- basically the same thing. Openly some of them are a little bit troubled at doing it. And all the ones that tried completely alternative strategies do not exist anymore. I know of none. Foreign name tried for years to -- he was the person who said it had to be done in a completely different way. But he delivered nothing after 15 years or 17 years. The program never made a move. It never publicly analyzed a position and so on. So that it didn't translate into a real program.
MR. ASHLEY That's interesting because Mikil Badvinick was chess world champion for many years and then he ventured to try to develop this computer playing program and then his star pupil, Garry Kasparov, now playing against the best,hits he has claimed that it can not be done. It will never beat the best player. So we have this clash of the generations.
MR. FRIEDEL You mean Garry? Garry knows it's got to beat human beings.
MR. ASHLEY Not him?
MR. FRIEDEL It will beat him if he's still world champion. The question is when.hits you can do a poll here. Everyone should guess and we could put money on to this. For example, I did a question. I asked all the experts and it ranges from 1993, it's going to happen. This was somebody who I asked in 1992. Shoe told me that it will be 1995. So he's off by at least 1 year. Campbell, the other DEEP BLUE, person told me it will be 1996. So he has to win now. I said 1999. Garry said 2010 or maybe never. Because what Garry said was there is a possibility that as we go deeper and deeper and deeper there's a curve which goes up like this and at some stage it hits Kasparov's level and then it goes higher.
Theoretically, if you calculate if they're looking at 300, 400 or 500 million notes per second, there's a theoretical curve that goes up. You can calculate how deep is it going to go, how much did each play of search in the past increase playing strength? You can calculate that exactly and then you can look at the curve. The curve is a little bit snake formed. So you don't know exactly how it continues. Optimists say this is all noise, you know, with little squiggles in it, draw a straight line through the average then at 500 million notes per second it's 2,000950 below points which means it blows him off the board. He gets 1 draw. If the pesimists say look at the end of the curve, it's sort of flattening out, then he is -- he has a long, long time, like 20 years, 30 years, the realists say. It may not happen now, but it's going to happen very soon.
MR. ASHLEY Well, I should -- very interesting points being made.
MR. FRIEDEL We can come back to them.
MR. ASHLEY But at the moment Kasparov played queen 2 F6 and DEEP BLUE responded instantaneously with the move D5, which at first blush may look like a pawn is sacrificed, but in fact it is not. The pawn on B6 is in the eye of the white queen, the glare of the white queen attacking the B6 pawn and also the B7 pawn may also fall. So the complications begin here. The question is how to capture this pawn on D5. Kasparov with 54 minutes left. DEEP BLUE with 119, 118 left. This sacrifice, well temporary sacrifice is going to create more weaknesses in black's position, depending on the sequence of moves that Kasparov chooses.
Let's try 1 variation, rook takes D5 which seems very logical. Maybe rook takes D5 in response. After this black will have to recapture with a pawn. This, of course, creates a new situation on the board. Now I think queen takes b --
MR. FRIEDEL Try queen takes B6.
MR. ASHLEY Now, in this analysis queen takes B2 by black is possible and this looks actually very interesting for black. Many attacking possibilities, but even the computer is, Fritz 4 helping us analyze is not finding anything really strong for white, up to and including this move rook to C3, which it gives as nothing really special. So this sacrifice still not absolutely clear.
MR. FRIEDEL Let's go back and look at what Fritz was --
MR. ASHLEY Because the b pawn --
MR. FRIEDEL -- suggesting for white.
MR. ASHLEY The b pawn might become a problem. Black queen attacking the b pawn, but now the complications really begin. This is where the computer excels. This is the kind of situation we've been talking about. It's not whether or not it's relative equality on the chess board. This is where the computer is just excellent. Now it's analysis time because all the connections are there. It's attacking something concrete, black's attacking something concrete. Exchanges are concrete. This is not strategical maneuvering. This is just brute force calculation. Here Kasparov has indeed captured rook takes D5 very quickly. It is not necessary for black to take the rook, but as I suggested before, it did capture and Kasparov now recapturing. This is a critical situation. Taking the pawn on B6 is possible, but after queen B2 it would seem not to lead to anything unless DEEP BLUE has found something surprising, something that we are not seeing here. It gives the move rook to c -- it's giving many moves. That is, Fritz is giving many moves. But a confident Kasparov has walked away from the chess board.
(Laughter.)
MR. ASHLEY I mean in the middle of the madness, one really has to wonder whether or not he is that confident or he's just nervous, needs a break. But at the moment now Fritz looking at the position, Fritz 4 ability ing with us thinking maybe that rook to C3 is a very strong move, but many, many complications occurring here. The b pawn may drop by force in a situation like this. So Kasparov is going to have to really be careful. But look at the situation on board, folks. The pawn, black's pawns will never be worse. Dead worse. If you look at these pawns, double b pawns, isolated d pawn, double f pawn, isolated h pawn. This is the kind of position where, you know, I tell my students don't ever, ever do. No one does this. Of course the world champion is doing it. Don't tell my students, please. This is the kind of position, though, that could potentially be horrific for black, potentially. But although I mention that black's pawns can never be worse, there is one pawn on the board that could become significant in this position and it's black's d pawn. That pawn is isolated, yes, but also past, because it's passed, it could divert White's attention, forcing white to deal with this potential powerhouse racing down the board trying to become a queen. It may not get there. It probably will not get there. But it will be a source of worry for white for some time to come. It is something white is going to have to control in the position, make sure that this pawn goes nowhere. 1 famous maxum goes pawns can be weak, sometimes it's not how many pawns you have, however, or how healthy your pawns. It's whether or not I have 1 good 1. If I have 1 strong 1, you sometimes can even take 3 or 4 of mine, but that 1 good 1 is thinking about becoming a queen and you will end up losing the game. So here that d pawn is black's main consolation because there's nothing else he can hope for to compensate for this bad position.
MR. FRIEDEL I'd like to save the game.
MR. ASHLEY So a critical position on the board. It looks like rook to C3 is necessary if white wants to retain any advantage. Kasparov is still away from the chess board and I don't know if he's walking back stage confidently or banging his head on a wall or what. But I'm sure that he is concerned. Kasparov is a great bluffer. I mean a great, great bluffer. During the Anand match he pulled faces, held his nose like this as if to say that move stank. I mean, he really shows his feelings at the board. Even against a computer opponent he will show his feelings. Because he's just an animated kind of guy. He really just likes to show his feelings at the board. Now he's back calculateing the situation. So he knows that it's critical. He's aware that this is it right now. He could live or die in a position like this. He better control the tactics or else he will be out.
I had a question over here.
(Commentary from audience.)
MR. ASHLEY It indeed competent deed has. B3 has been played by the computer. That b pawn couldn't go. Black needed to keep the b pawn on the board. It wouldn't allow that to drop from the board. Now it is, in earnest thinking about, capturing the pawn on B6. So the question is can black save that pawn. So my question would be what if a move like knight to E7 defending the pawn -- passive but greedy. How does this lose a piece? Rook C8 and queen c -- well, you see how even an international master could overlook simple tactics. Rook takes C8, knight takes C8. Queen E8. Wins a piece. The plus of the computer is it doesn't go for, oh, that looks passive; oh, that was stupid. I win a piece. So 1 plus for the computer.
So without the ability to move his knight and then the b pawn is history. That was a critical variation to analyze. Without black's ability to move his knight then that pawn has no way of being defended because there's no way for black to bring anything to bear on that b pawn. That B6 pawn so the computer by force, DEEP BLUE will win back the b pawn and the question is what is the state of affairs after it recaptures the pawn since it will.
Now we see Fritz 4 saying it thinks that the advantage is really not that much, just still point 2 in White's favor. But still the question is which move should the computer play. But without being able to defend the b pawn because of that tactic. Somehow it's going to have to consolidate.
Queen D8 as an offense has been suggested, but I have a feeling that you're weakening everything in this position. A check, for example, followed by knight D6 would be extremely dangerous. Queen G3 check for starters. A move like king F8 maybe. Knight D6. Now a rook and 2 pawns are under attack. Boy, the f pawn is going to create some meeting possibilities for whites. This would be extremely dangerous for black to go into history. You have to accept when a pawn is dead, it's dead. Now the question is if he gives back the pawn, how can he give back the pawn and keep a good position?
Queen B2 has been suggested, which is definitely a funky move. This is not really attacking anything. It's not attacking either pawn on the queen side, which is why the computer did what it did. It would force some kind of move, but look also how it makes the king's side, black's king side a little bit suspect. Strange moves like rook C3 come to mind. I'd be a little worried about moving my queen away when there were no real threats, a move like queen B2.
The move D4 -- F4. F4 is certainly an interesting move, although I'm not sure if it helps white; helps black in the long run because of a simple move like queen takes B6.
Fritz is already saying that it's point 94 advantage. The pawn move just attacking queen does not secure black's position. Black needs to secure its position quickly, and I tell you it needs to somehow figure out how to advance its d pawn. For example, in this position here D4 is possible. Well, it's possible although it loses the d pawn instantly. But remember black is up a pawn. My point is that black is up a pawn for the moment. This just --
MR. FRIEDEL Wins a pawn for white.
MR. ASHLEY This gets the pawn back.
MR. FRIEDEL Wins another.
MR. ASHLEY Well, Fritz is saying that it will win a pawn. My problem is could we consolidate this position in some way for black. I just want to point out the tactic that knight takes knight is impossible because of rook takes C8. Just for those who are not aware of this tactic. The reason for D4 would be to buy time and maybe play a move like rook to D8, maybe somehow consolidate the position. This is something that Kasparov will think about. When you have a pawn advantage when defending a good idea sometimes is to sacrifice the pawn back to consolidate the position. Even sometimes when you lose a pawn, the position will go towards equality. Here black would very willingly, very willingly give away his 3 pawns on the left side of the board for White's 2 pawns. Gladly, gladly. Of course. Because then it would be much simpler for him to defend his position because the position would have pawns on only 1 half of the board,hits those positions tend to be easy to defend because it's limited room to move. This is the kind of thing facing Kasparov now, but I can't say that, knowing Garry Kasparov, the slashing, attacking player that he is, that he's happy with this kind of situation.
Fred, to ask you this question, do you think that now the position is getting towards those types of situations where computer really excels and DEEP BLUE could be menacing in this kind of game?
MR. FRIEDEL All right. I'm not neutral on this. I hope not. But possibly. It looks very complicated. Look at Garry, the way he's working on this position. But you'd never know. He might have it all under control. It might be all planned.
MR. ASHLEY Well Kasparov now with head in hands as we welcome back Grandmaster Yasser Seirawan, who will be back in a moment.
But certainly a very exciting position on the chess board and very problematic for Kasparov. Getting somebody a time pressure is part of chess. You know it's going to just move. It's losing, but it's still going to move. Kasparov sitting here frozen in thought, 44 minutes left to play, all his moves, feeling the pressure, wondering what is the variation that helps him here. He doesn't want to lose this first game. That would be a definite setback for him. So he's going to be careful, but look he's down below 43. He'll make a note of it when he's down below 30 because that's when it really starts getting tense. You'll see him moving a lot in his chair trying to keep with the program.
(Commentary from audience.)
MR. ASHLEY The question has been posed what if rook to D8,hits, Yasser, I'll give the floor to you.
MR. SEIRAWAN Thank you, Maurice.
MR. ASHLEY What about a move like rook to D8, preparing with good reason, preparing to try to make that d pawn work, that D5 pawn, the main for black's position, what about a move like that? Would that be a position?
MR. SEIRAWAN Before I answer that, it's poll time. I just came from some of those back rooms and talking to some of the people and getting their insights. So, first of all, poll time. All those who like IBM's, DEEP BLUE's position?
(Members of audience raise hands.)
MR. SEIRAWAN You guys are easily convinceable.
Who believes in Kasparov's.
(Members of audience raise hands.)
MR. SEIRAWAN Maybe one-third. I think Garry has got a great position.
(Applause.)
MR. SEIRAWAN Okay. I think Garry has got a great position and I've been looking at it and the gentleman who suggested rook D8 take a bow. That was right. That's the right move. So what we're going to see is a move rook D8, queen takes D6. It's kind of a little bit complicated, but we're going to get there. Now the computer has won back its pawn so it's a happy camper. Rook D7. I was analyzing this with a reporter Stefan from Germany. We have a great deal of media interest, by the way, in this match worldwide and Stefan said, well, but this is great. I said because look at all of black's pawns how shattered they are, the structure. I said yes, but look at that pawn on D5. That's the key. Stefan suggested the move rook E1 and we've been looking at rook E1, pawn to D4.
Rook to E8 check. Rook to E8 check. The king has but 1 square, king G7. Now again that pawn on D4 I think is the decisive pawn. He, Stefan, suggested queen C5, which looks right. Right? I mean this is the thing to do. You're getting black's king. But if you don't, watch out. D3. Go ahead.
MR. ASHLEY You're just going to ignore the guy slamming down next to your king like that?
MR. SEIRAWAN Go, go.
Queen F8 check. Now king to G6. Now, unless I've missed something and Fritz is going nuts.
MR. ASHLEY No, Fritz likes you, boy.
MR. SEIRAWAN Oh, look.
MR. ASHLEY Fritz 4 is liking black. That's bad when a king is on G6 and the computer likes the king, boy that's bad for the other -- I mean just really bad. It's getting worse. That d pawn. I'm glad that you came and justified my comment about the d pawn because I was telling them bad news with that d pawn.
MR. SEIRAWAN Maurice has been suffering. He's been working hard. We're going to give him a break shortly.
MR. ASHLEY I was mentioning to them before that the d pawn was extremely strong. That's the potential in black's position. If he can get that d pawn rolling down the board, it would be fabulous. In this position the d pawn is a monster.
MR. SEIRAWAN Exactly. This is what Garry talked about yesterday at the press conference, who was a wonderful -- Garry is a wonderful spokesman for chess and one of the things he was saying is that really what gets him, Garry, furious is that when he plays the computer, you know, he tries to gravitate all of his pieces towards the computer's king and it does not seem a threat, but he just thinks to himself where there's smoke, there's fire. He's just going to be very happy getting his computer's king.
The intent there is that the computer doesn't care.hits in that final analysis position that we saw that's what the computer says. Okay there's a couple of pieces around my king, but there's no threat. Then the d pawn becomes the most powerful pawn in the position. So again I think that what's going on with Garry at this moment is he's thinking to himself, gee, there's going to be 2 very powerful pieces next to my king. I better be sure that I'm not losing this position because 1 of the things, as you were mentioning, he doesn't want to lose. He's not going to take any risks.
MR. ASHLEY The other thing is your position, you analyzed -- he's played instead king to H8. Bizarre is not bizarre. Wow, that's a wowser. The thing -- well, clearly, he's given himself the opportunity to get counter play along the g line. He can now move his rook to G8 which is a completely different idea and may indeed be a viable one, but still you've got to wonder what was wrong with the variation you mentioned. But it's easier to call those variations when you have a chess set in front of you and you're making the moves and you can make sure you're not getting than to look at it in your head and say I'm playing DEEP BLUE it looks like 500 billion positions, in 3 minutes. It's queen is right around my king. So its rook. I'm not sure if it's mate. It looks like it's not, but I'll give it the opportunity. I mean it's like a pit but when it gets ahold of you, so farce playing a safe move, as we said before, the computer doesn't believe in safety. I don't have to worry about safety, is it good or bad. Kasparov worried about safety, a critical game. First game, you lose it, it's pretty tough going. Your confidence might flag. Kasparov doesn't want to go through that. So he plays a line which may be much safer, though seemingly a lot less safe than the variations you suggested.
A great prize at stake, $400,000 for the winner and his honor as the best player in the world. He's really defending the human race. That's what he's doing from the encroachment of computers. He doesn't want to be the world champion that finally loses to these things. It wouldn't be a dishonor, but for him it would be.
MR. SEIRAWAN Well, I'm a little bit puzled over this move because it wasn't the move that I was thinking about. So I'm kind of like Fritz, seeing a new move, I'm suddenly...
Well, I was just going to say that the variation that I had proposed seems to suggest that I'm not afraid about losing the B6 pawn. I think Garry took the same perspective. Now also in the variation I proposed, what I was attempting to do was lure White's pieces on to the offensive and then just use my d pawn. Now, what Garry has taken upon himself is he's also mentally already made the adjustment forget it go ahead and take my B6 pawn. But he wants to go on the attack. Well, let's see what he might have had in mind, shall we? Okay. Let's take a look at queen takes B6, although Fritz seems to be liking queen D3. I'm not quite sure why. But after queen takes B6, what Kasparov has in mind is a more like queen G5 attacking the rook on C1 or the immediate rook G8. Perhaps rook G8, with the anticipation of playing queen G5 threatening queen takes G2 meat and queen takes C1. Okay? So what does Fritz think about --
MR. ASHLEY Well, DEEP BLUE has indeed ripped the pawn off of D6. Kasparov with 38 minutes left on his clock, down to the computer's 112 left on the clock. So a tremendous time advantage, half an hour time advantage for the machine. But still within manageable complications for the world champion much he is not in time pressure, although you see the advantage on time for the computer. This is not time pressure for the world champion. It all depends on what happens in the next few moves, how quickly he will play. Will he whip out his next move rook to G8? He has to be very careful.
MR. SEIRAWAN What you're seeing there by the way was just Garry shadowing the pieces. A shadow is a French term that means to adjust. What happens is amongst the top professionals in the world, they like the pieces to form squarely in the middle of a square. The operator was just laxidaisically making the move, showing that he's not a deep chess player.
MR. ASHLEY Well, Kasparov has gone over with rook to G8. This is Garry Kasparov in his element, the kind of move 1 would suspect of Kasparov. He does not like to defend. He does not like to sit and hold the position. If he sees the opportunity to go over on the offensive you can bet he will play it in a second and is leaving the board again. In the meantime, our analyst Fritz 4, our computer analyst Fritz 4 believes that the advantage is only point 34, within manageable bounds for white, which means that black is not in any deep trouble as far as it's concerned. But again DEEP BLUE does see deeper and may see more than this program. Now, the question is how strong is this counter play? What is black doing that's so serious with his rook to G8 move?
MR. SEIRAWAN Well, I be deed.hits is it serious?hits the d pawn shield again is -- again, it's very nice for white to have such a nice pawn shield. But if he -- if it can be disturbed this pawn on F5 can be quite and annoying 1. For example, again we realize of course the threat is queen G5. That is --
MR. ASHLEY We should make that point. If white were to be greedy and play the move queen takes B7 --
MR. SEIRAWAN Watch what happens to Fritz.
MR. ASHLEY -- then queen G5...And it's over. Good knight. Attacking the pawn on G2, threatening meat and also the rook on C1. can expect that not to happen.
MR. SEIRAWAN Okay. So because this threat of queen G5 is so powerful, 1 idea is to play the move G2, G3. Okay? Just blocking that g file for just a moment because of that rook. Queen C5.
MR. ASHLEY Instead of the response to the threat defending to the rook, playing queen C5.
MR. SEIRAWAN Instead of that position, I was going to say the pawn on F5 can be utilized to play the move F5 F4.hits so this weak double pawn suddenly becomes a crucial factor in the attack. Instead we've seen the move queen C5 which seems very logical to me. Sometimes white can play queen C3 and just come come back and try to get that black king in the corner. But more importantly, white just takes a moment to defend a rook on C1.
MR. ASHLEY Also possibly maybe here, Yas, is queen takes D5 for white because that defense the G2 square and by defending that square then the possible meating threats disappear. So certainly that is now a threat for white. Queen captures pawn D5 and that wins a pawn. So black has to be careful about that next move.
MR. SEIRAWAN Okay. 1 of the things that you really ought to do as a chess player to improve your games is to say to yourself what is the direct threat? Find the direct threat and then you say what happens if I play it. So let's take a look at what would happen if Kasparov was to threaten meat and 1. Let's take a look. Queen G5.
Meat in 1?
As Maurice was saying, white does have the defense queen takes D5 but of course that would leaf the rook on C1 unprotected.
MR. ASHLEY Right. This position is definitely --
MR. SEIRAWAN So G3 would be provoked.hits then we have the move that I was kind of suggesting exists is a move like F4. I know that Fritz is happy about check check.
MR. ASHLEY It looks somewhat dangerous, admitted.
MR. SEIRAWAN Knight D6.
MR. ASHLEY Knight D6 ignoring those threats and coming up with a big threat. Knight takes D7 for exampleing a king and queen.
MR. SEIRAWAN Let's have a little bit of fun. Let's try a move like rook G7 and let's see what Fritz thinks of this. Now you see the reason I'm not really happy about this, if white could, he could play the move queen F8 check, right, the queen on C5. You have to be a little bit -- what does Fritz think about this.
MR. ASHLEY Fritz think thinks this is completely winning for white, with different moves.
MR. SEIRAWAN Let's go back just a few moves. Let's go back a couple moves and say instead of the move queen G5, let's try queen G7. Then we've got queen takes D5. I apologize. Queen takes D5 we'll just protect. I apologize. So that is why queen G5.
MR. ASHLEY This is the thing too, about -- you know, we talked about king safety and Kasparov was a bit worried about his king in your variation. It seems as if it holds up that his king was safe in that line. The computer not worried about its king in this variation says come on attack. I don't see meat. I don't see how you get through. So you can attack. One thing you know when you play these computers at speed chess they're brutal when you think that, you know, let's sacrifice, let's sacrifice another piece. We're killing you now. Then all of a sudden it makes a move and all of a sudden where are my pieces? I want them back. Because it just refutes attack so coldly. Here Kasparov might have felt typical Garry Kasparov attacking, beat you up. I'll get you now. I'll move my king over rook G8. What do you do know? The computer says go ahead. I'm going to take all your pawns and beat you.
MR. SEIRAWAN This is the blood lust, yes. This is very, very human too. Let me say this, that it's so nice. Again, I refer to my comment when I was a beginner oh, boy I just love making that check. The check was maybe even a bad move. But it just thrilled me to be on the attacking side of a position. Even if it was short term. So I would really go for the attack as a beginner and oftentimes I would lose because as Maurice was saying my attacks were easily rebuffed. This is a case where that pawn shield, that would make me worried. A pawn shield like F2 G2 and H3 is a very, very stable 1. So I wouldn't have thought of king H8 and rook G8. It is my first choice because that pawn shield is very stable. Now Garry probably had calculated a series of moves that had encouraged him in this direction.
MR. ASHLEY Maybe now noticing that it's not working out to what he thought. It's interesting, you know, if the computer played queen C5 in less than 2 minutes. Queen C5 was just quick. Why? Because it's a position that demands calculation. Am I getting mated? No. Okay. So I'll play a move. A player, a human against Kasparov without question would have had to study this position for a good 20 minutes.
MR. SEIRAWAN Right.
MR. ASHLEY Am I really going to let this guy attack me like this? I know how he sees everything. Tony Miles, when he played a match against Kasparov where he was wiped out, he called Kasparov the 27-eyed monster who sees everything. But Kasparov never had to play this 32 microprocessors seeing 50 billion positions and in 3 minutes. I think if you want to say something, everything, this is it. So there's no intimidation factor. There is no psychological intimidation. That in and of itself gives Garry Kasparov his advantage. Part of his strength is not just his ability to play great chess moves. Part of his real strength is his ability to sit down in front of you and make you get scared. He plays a move and you get terrified. You think what is this guy going to do to me. I played him once in a move where he played 3 international masters and a grandmaster. I was one of the international masters. It was actually 4 international masters. He was assembled against the 5 of us. All right? You think 16 has a shot and he was just coming to every board and going bang, bang bang. You're frantically searching for a good move,hits he's in front of you and you think oh, God, I'm dead. I whipped out a move and wiped out all 5 of us like it was a walk in the park. Regular day. To problem. You don't want to be intimidated, but you just play weaker moves as a result. Now go home and think what a fool normally I play this move and that move and this move. But no, it was Kasparov in front of you and somehow your knees got weak and you got scared and you played substantially weaker moves.
Of course the computer is not going to be intimidated by this big charasmatic factor in Kasparov that made him dominate the world of chess for so many years. So that advantage that he has is reduced completely and there are no such doubts in the computer.
(Commentary from audience.)
MR. SEIRAWAN Yes, this is what Fritz is suggesting. I don't -- I think it would be in White's favor. I think it would be in White's favor to exchange queens. There's 2 factors involved in that. The first is the fact that black's king is much further away in any ending. Now we have a material balance. The second is the black D5 pawn.
MR. ASHLEY Kasparov has advanced his pawn to D4.
MR. SEIRAWAN To D4.
MR. ASHLEY Ignoring his attack and instead trusting in that d pawn, which as we said, would be a factor in the long term, but Fritz immediately is spotting this, calling it a blunder that loses a pawn instantly to knight D6. What is going on?
MR. SEIRAWAN Well, I don't want to double guess Garry too early, but if he's going to push the pawn on D4, there's an old cardinal principle. Rooks belong behind past pawns. It's almost like a mantra. You can just repeat it time and time again. In this case we can say the rook on G8 clearly would have had a stronger preference.
MR. ASHLEY It's like he's playing with 2 plans because instead of going behind a past pawn and pushing that he moved his king over feigned an attack and now he's changed his mind when his rook would have been better placed behind the king. So very definitely I shall play by the world champion. Now knight D6 looks like serious threats to the f pawn certainly.
MR. SEIRAWAN Well, obviously knight D6 does threaten knight takes F5 winning a pawn clear enough. But also there's that knight takes B7 pawn and then that knight on C6 has suddenly lost its support. So --
MR. ASHLEY Clearly sacrificial possibility here by Kasparov and I mean making his move and getting up from the board again. This guy never sits down at the board, no matter what. But leaving the board instantly and I have to imagine that his nerves are getting to him right about now. Kasparov, as strong as he is,hits definitely the best in the world, I've seen this all the time. He hates to sit at the board. When asked why you get up at the board, he has said he doesn't want to say that he tries to do it to intimidate the opponent but he has said I'm nervous. I can't just sit there forever. I mean what am I supposed to be doing when I know what's going on and I'm just waiting for the other guy's move. So he likes to go away, do whatever to relax himself, go away from the board. This move is certainly an aggressive 1. It is a clear pawn sacrifice and he is counting on that d pawn and maybe his counterplay on the king side to get him through. But maybe boy it's looking shaky.
MR. SEIRAWAN I've just turned my attention to the screen on the left. This is quite a funny scene. What we had there was the arbiter, Mike Valvo, peering behind the blue screen, the DEEP BLUE screen. Mike, by the way, is international master, so very, very competent chess player as well as arbiter. If you had kind of seen him, he was sort of like -- his eyes were getting bigger. So I think what he was seeing was kind of this assessment. There you go. You see, Mike? He's looking and saying, oh, my goodness does Garry know he's in trouble?
(Laughter.)
MR. SEIRAWAN I mean what we have here is a neutral arbiter, you know.
MR. ASHLEY That's a serious move. No question Kasparov is aware of the move knight D6.
(Commentary from audience.)
MR. ASHLEY Well, knight D6 has been played and we're going to find out if it's a good move or it's a blunder. All of us looking at this saying, Garry, what did you do? Well knight D6 and the whole position is critical. A pawn is lost. That's clear. A pawn is gone. We don't have to ask whether or not it's gone. It's gone. It's history, dead. Not only is the pawn gone. Kasparov's jacket is off.
(Laughter.)
MR. ASHLEY Now he's wiped out -- oh, man, I've seen this from Kasparov before. He's wiped out, the move F4 and has left the board. It worked against Anand. Boy, Anand was scared when he did this. Hello. Look at those moves. But now the computer has got to be thinking I'm winning pawn. What are you doing? What exactly do you think you're doing? Now --
MR. SEIRAWAN Actually, so am I.
I don't get it, Garry. I don't get it.
MR. ASHLEY Now, you look at this position and yes, it's menacing. It is scary. A human would stop and start wondering what Kasparov was doing. Should I just take his pawn? Can I just take this guy's -- when he's so confident he's taken the jacket off. He's making the moves quick. He's getting up from the board. There's no way under that kind of psychological pressure you can just grab his pawn and say you're an idiot. I'm going to win. No way. But the computer is sitting there going jacket, what jacket? Pawn, thank you. Got it.
You lose.
MR. SEIRAWAN But looking at the position for a moment, one of the things that makes -- if we could just go back to the move F4, one of the things that makes black's attack more difficult is the move knight E5. I mean black would like to get queen rook and knight all involved. Remember, chess is a game where you use your whole Army and that knight on C6 isn't participating. So let's give Kasparov the move knight E5 instead of the move F4.
Thank you. You're very sharp.
MR. ASHLEY An eagle eye from the audience, saying this move was actually a bad one. 1.
MR. SEIRAWAN The idea is you want to bring the knight toward the king's side but it has a drawback. The drawback is queen takes E5. So after queen takes E5, there is a fork, a knight fork and a deadly one at that.
MR. ASHLEY Weren't these great to do back in the younger days?
MR. SEIRAWAN Oh, man we loved this.
MR. ASHLEY Never again.
MR. SEIRAWAN Now, let's look at the position on hand. After the move F4 again we would like to involve our knight,hits also the same trick exists. Let's see the move like, for example, B4. B4 for white now. It's a silly move I know. But then comes the move rook G5, again trying to get rid of this queen on C5, but we fall victim to the same trick. Queen takes G5.
MR. ASHLEY Hello.
MR. SEIRAWAN Queen takes G5. Our favorite friend knight takes F7.
MR. ASHLEY So that knight on D6 is really powerful, extending in many different positions.
MR. SEIRAWAN That knight on D6 plays several roles. One is a defensive role because this knight on F7 tactic is always there. Also, it prevents that knight on C6 or even that rook on -- rook going to G5. One of the things that Kasparov's last move did, F4, was introduce the move F3. I'd also like to take this --
MR. ASHLEY Wow. After all that, just ignoring the attack knight takes B7, now a move like knight E5 and pieces could hover around the computer's king. But this is precisely the kind of situation where either the computer wins or loses because of its insistence on how it thinks and its insistence. Maybe it's look at the position and said I've gone 9 moves deep. I've gone 9 moves deep and it doesn't lose. I'm fine. Now Kasparov makes a move and now it goes to that tenth move and it goes oh, God, I'm getting mated. This is the trick. Will this happen in this given situation? Very dangerous for the computer, or maybe it sees nothing and Kasparov is down a pawn and dead. Completely critical situation. Every move must be let's go Garry. You can start chanting, if you like. But, boy, he's in deep --
MR. SEIRAWAN Deep thought, yes.
I just wanted to say that the man I was going to introduce was David Levy of the ICCA. I was going to invite him to talk about what Maurice was just saying, the computer's strength. The computer's strength is that it calculates openly concrete threats,hits since it couldn't calculate the mate --
MR. ASHLEY But Garry's great experience must be telling him, Yas, that this is a fabulous attacking position. The white knight on B7 is not an impressive piece at all away from the king's side. Black is concentrating massive forces against White's king. More forces than we saw in the Gulf War is being directed at White's king. What is white doing? I mean knight E5 now is very possible as an alternative.
Has white really gone crazy? But if it doesn't see mate, if the computer literally doesn't see mate or you winning your pawn back very soon, it's grabbing the material. It's just going to take a pawn and say prove it. But if that mate is 1 move beyond its horizon, the game is over. All Kasparov has to come up with now is aggressive continuations in order to press that home.
MR. SEIRAWAN What Maurice is exactly saying, exactly that, is that he, the human being, has the potential, perhaps we'll find out, of outcalculating the position,hits somewhat Maurice has been saying is that the horizon, how deep the computer can see into a position -- let's say for argument's sake the computer can see every permutation of the next 7 moves. If after 7 moves it thinks that it's winning, it will do what it's done. But if Garry could see 1 move beyond the seventh, the eighth move and he determines that he's winning, he could actually outcalculate the computer.
MR. ASHLEY In addition though, if you look at this position, you would say things about the setup, black rook on G8 on the same file as the king. Black pawn on F4 near the king. Black queen could sneak around the king. You don't even calculate so much. Part H
MR. ASHLEY Sometimes it's not even calculation. Sometimes it's just the feel of a big attack going on and you just don't want to have to face this. And one the greatest attackers of the age, Garry Kasparov, trying to figure out whether or not he's actually mating.
MR. SEIRAWAN Well I haven't played blitz for $400,000 in my life, so I wouldn't want to play this game -- this for a game of 5- minute chess. But this kind of reminds me that chess has got a wonderful history, a 1600- year history. And one the beautiful parts about chess is the wonderful anecdotes. And one those is from Monte Carlo, where Dr. Alexander Alekhine had sacrificed a rook. And every chess grandmaster had crowded around Alekhine's table to say, my goodness, he's a rook down. He's a rook down. You know, is Alekhine going crazy? What's going on? And Miguel Najdorf had walked up to Alekhine and said, what are you doing, Doctor? And he said, My nose tells me something. Alekhine hadn't calculated everything. He had just taken that he had enough compensation. And that's a beautiful romantic story. But now we're in the age of computers, where a little bit of the romance is maybe going out of it. Because if the computer can just say, nice, I'm happy for your nose --
(Laughter.)
MR. SEIRAWAN But as Maurice has been saying, I mean, black has a lot of threats and we'll probably get to that in a moment.
Sir?
(Commentary from audience.)
MR. ASHLEY The question from the spectator is, in a position like this, would the calculating power of the computer actually be a disadvantage because it would -- his point being it would start analyzing completely irrelevant things in a position when only certain things are of critical importance. Maybe we can ask Hans Berliner if he has thoughts on this position.
(Commentary from audience.)
MR. SEIRAWAN Were we accurately saying that --
MR. ASHLEY Knight E5 by Kasparov has in fact been played.
MR. SEIRAWAN Were we saying that Kasparov had calculated beyond that? ?
(Commentary from audience.)
MR. ASHLEY So the computer is selective in that sense. It's not just crunching out --
(Commentary from audience.)
MR. ASHLEY So a point being made by Hans Berliner that the computer is not just calculating, calculating, calculating. It's really saying, well, you know, precedence is given to checkmate positions, to make sure it understands that, and it will give precedence to those lines and analyze those lines so that it is not checkmated.
And 2 moves have been played, 1 for both sides. Knight E5 by black, a very strong centralizing move. But instant response by the computer as it anticipated this move, centralizing the queen and defending the g pawn on G2, and now it is definitely up 1 pawn. Fritz 4 feels that it's 1.44 up. Not only the pawn, but more. Look at white's 2 pawns on left side of the board.
MR. SEIRAWAN Yes.
MR. ASHLEY Those things are passed. Those things could potentially become very strong. And the computer now threatens very strongly to play the move knight to D6. And when that knight gets back to the middle of the board, if Kasparov doesn't have something concrete, we will be saying that this game will go to DEEP BLUE. So a critical situation here. Has Garry overstepped the limit? Has he sacrificed and gone into what Hans Berliner calls the main strength of the computer, defending? Maybe it will just stop his attack and he will lose this game. Kasparov thinking. Now he's dropped below the 20 minute mark down, to 19.25 and ticking, need to go make 10 moves, 11 moves in 20 minutes. This is starting to approach critical time pressure for Kasparov and he is now shaking his head. I've seen that before. Not too happy with the turn of events.
MR. SEIRAWAN I'll get to the gentleman with his hand up in a moment. I just wanted to say that before the move knight B7 what I had anticipated black having that Garry being able to play was the move F3, if we can put that on the board, followed by the move F3 threatening rook takes G2 check, F3 G3.
Rook takes G3? No, no, I was actually anticipating the move queen F4. And I thought, this looks very good for black because the queen cannot be captured. I'm threatening rook takes G3 and I'm threatening the rook on C1, but I think the computer had calculated the move rook C8, which stops me cold because you cannot take the rook. You cannot take the pawn on G3. And so I think maybe Garry had also missed move rook C8. He looks quite nervous now, as I would be.
The gentleman who had his hand up.
(Commentary from audience.)
MR. SEIRAWAN Okay. The gentleman has made a very good point. What he said is that there's very few pieces on the board. It's easier, therefore, for the computer to calculate. The 2 passed pawns on A3 and B3 are so strong it means that in any end game situation Garry is a goner. Any end game of this game means that black is going to lose because of that. And therefore black's only chance is to have a knockout blow with a checkmate against DEEP BLUE's king.
MR. ASHLEY But Garry, I mean, if we do take up this issue of analyzing this position, it seems to me that I can't find a strong continuation for black. I can't find how he can press this advantage. It just seems to me that knight D6 for white is going to shut everything down and all mate threats -- what mate threats it? It just leads nowhere. There are no strong attacking situations. I promise you, if I could find 1, I'd suggest it. But what to do? The main variation was F3 G3 queen F4. It's not working. So what do you do? What do you do?
Queen G5 has been suggested. It threatens basically nothing. The move knight D6 in response. I'm always going to play this move.
MR. SEIRAWAN And then again, if F3, there's always this beautiful queen takes E5 and our favorite friend f knight 7 is just in time to save DEEP BLUE's circuits.
In the back.
(Commentary from audience.)
MR. SEIRAWAN Okay. That's a good suggestion. Thank you. The intent is knight to D3 now. F3 G3, 4 sequence of moves. Knight to D3. Now, what , this move does is it attacks the rook on C1 and it also prepares a very nice shot, knight to F4. But I think the problem is rook to C6, as Fritz has predicted. Rook C6 gains a tempo against black's queen. And also keep in mind queen q 4. I think we're running -- I think queen F4 --
MR. ASHLEY It runs into rook C8 automatically. No.
MR. SEIRAWAN And now rook C8, probably. Oh, we're overlooking something?
MR. ASHLEY No, we're not. No. I mean, the variations we've suggested are the salient ones and everything here is so desperate for black that I daresay that Kasparov has sacrificed a pawn for no reason that he can find right now. Look, his head in his hands with only 14.18 left on the clock.
MR. SEIRAWAN I have 14 minutes for Garry, 14.11.
MR. ASHLEY Yeah, 14.11 to DEEP BLUE's 1 hour and 3 minutes.
MR. SEIRAWAN Well --
MR. ASHLEY Can Kasparov get a draw? He's played the move F3. He's gone for it with F3.
MR. SEIRAWAN The computer thinks that it's going to win.
MR. ASHLEY And G3, the instantaneous response by the computer, not at all intimidated by Kasparov's move. It's been ready calculating is this possibility. Sorry, let me just get that in.
MR. SEIRAWAN G3, indeed. Okay.
MR. ASHLEY And now the challenge is on Kasparov. All avenues of attack have been reduced. It's cut off now. The g line is nothing. Knight D6 again the main threat for white. There's nothing to do to break down the g line and the computer is simply seemingly safe. What is there? Can he advance maybe the d pawn? No, that's not going to work. Knight D6 is constantly the major threat in this position. And I have a feeling that we're seeing a loss for the world champion.
MR. SEIRAWAN I just want to add that white --
MR. ASHLEY Knight to D3 has been played very quickly by Kasparov.
MR. SEIRAWAN I think that this was necessary because the threat was in fact that white would just play queen takes D4 because there was no longer this knight F3 shot, so now knight D3 was the only way to protect that pawn on D4.
MR. ASHLEY Now rook C6 looks awesome. Fritz 4 is saying it has a 2.47 advantage. And you know the thing is that Kasparov versus human, it comes up with some tricks. Kasparov versus human, tricks are for kids.
(Applause.)
MR. ASHLEY What is he to do? I think Kasparov is as aware as anybody. He's practiced against computers more than anyone because's had to play several exhibition matches against them and he knows the main strength of computers is precisely where it is. And he has allowed himself to get into that type of battle. Maybe he couldn't help it, but he went over on the attack. That alone should have made him a bit wary because attacking the computer often doesn't work because it's such a genius at defending. Now it's showing it's not afraid of anything and Kasparov's attacking possibilities are being reduced move by move. And a prayer may save him now, but it looks terrible. It really looks awful.
MR. SEIRAWAN I couldn't but echo Maurice's words any further. I thought that -- again, there should have been a warning bell going off in Kasparov's mind when he said I'm going to play king H8, rook G8, and I'm going to sacrifice my queen side. Because the reality is this is a beautiful pawn shield in front of white and it's very, very difficult to break down a pawn shield like that. So there should have been that flash, that red stop sign that said wait a minute. Let me just reconsider that. The approach of rook D8 I think was very good. Rook C6, if you played a move like queen G7 I can always play knight D6.
MR. ASHLEY Even queen F3 is possible. Knight D6 is extremely strong.
MR. SEIRAWAN What Maurice is saying is that the pawn on F3 becomes extremely vulnerable after the move rook C6. So -- are we down to some time here?
MR. ASHLEY I'm sorry to say the computer still has an hour left on the clock. So forget about time pressure. Kasparov with 12.49, 12 minutes and 49 seconds to make 8 moves in a tough position.
A comment made. The machine could crack. Yeah.
(Laughter.)
MR. ASHLEY You know, I could always go back, you know, and pull the plug. Unfortunately, DEEP BLUE's little brother is waiting in reserve just in case something like that happens. They do have a backup system in case there are any technical difficulties. So that may work once, but it won't work a second time.
MR. SEIRAWAN I'd like to just suggest the attempt that after rook C6, if we can put that on the board, because this is undoubtedly what we're going to see. I mean, Fritz has just stopped thinking.
Queen F4, as we've seen -- just a moment, if I can interrupt you. Queen F4 is bad, as we'll see in a moment. But I think queen H4 also is bad. No, I think that even stronger would be the move king H2.
MR. ASHLEY Ooh.
MR. SEIRAWAN Just stop everything.
MR. ASHLEY Well, no, king H2 is a little scary maybe. I don't know if it's --
MR. SEIRAWAN Does it say rook C8? What does it say, king H2?
MR. ASHLEY It's playing rook C8. It's doing everything.
MR. SEIRAWAN The reason I think queen H4 is necessary is because, if we try queen F4 instead of queen H4, we run into rook C8, as we'll see.
(Commentary from audience.)
MR. SEIRAWAN Thank you, sir. I'm actually a chess grandmaster. This might come as a surprise.
(Applause.)
MR. SEIRAWAN Rook to C8, queen to D2.
MR. ASHLEY Queen D2, my friend, allows queen takes D4 check. Excuse me. I've stopped your story.
MR. SEIRAWAN What happened?
MR. ASHLEY I'm sorry. The computer has. I'll get back to it in 1 second.
MR. SEIRAWAN What I was going to say is that I thought queen F4 would run into a forced mate, that Garry would get mated by force.
(Commentary from audience.)
MR. SEIRAWAN Yes, that's exactly what it means, a pound sign -- that's why I was saying I think the move queen F4 would result in a forced mate against Garry. So queen F4, just to complete the thought, rook to C8, rook to C8 -- oh, it's okay. We understand. Rook C8, queen to D2, just so that everybody catches up with us. This is our first day at the office, so we'll get better. Queen takes D4 check. And now F6 is forced.
MR. ASHLEY Then it's just over.
MR. SEIRAWAN And it's mate after queen takes F6.
(Commentary from audience.)
MR. SEIRAWAN I'm saying rook C6. Now, okay, after queen E5, we get that hopeless end game that we're talking about, once we get into this end game with these 2 passed pawns as the gentleman suggests. Black is going to lose at the time ending. But after queen H4, I just want to see how -- whether queen H4 changes the equation. Does it say what -- well, queen --
MR. ASHLEY Whoa, rook to C7.
MR. SEIRAWAN Okay, rook to C7.
MR. ASHLEY Making an even stronger move. Rook to C7, an aggressive attacking move.
MR. SEIRAWAN Don't laugh. It happened recently. Garry was playing, I think, a chess genius.
MR. ASHLEY No, he was playing this guy, Fritz 4.
MR. SEIRAWAN Well, did Frederick tell that story? What happened is the operator entered the wrong move and the 2 players were playing happily their positions and suddenly the operator realized that the position on the board wasn't the same that the computer was analyzing. And the rules say that in such a situation they can't take it back. I think that the operator just had to say this is the position on the board.
MR. ASHLEY No, no, no. What they did was, the mistake in that situation was actually made. They went back to where the 1 change in the position was, because no illegal moves are possible, everything couldn't be played as it was. And so they inputted the new move, but with all the same moves played for the computer as if Kasparov had played the other move, which was completely wrong because, as you know, in chess if you make 1 move and I respond to that move and you make a different move, I want to make different responses because it just becomes better. And they forced the computer to play the same moves it did which were bad under those circumstances and gave Kasparov a big advantage and he managed to win the game very easily.
MR. SEIRAWAN Let's take a look at the new situation.
MR. ASHLEY But rook C7 is not a bad looking move. Rook C6 was an obvious tempo gaining move. That F7 pawn looks like history now and where is the attack?
MR. SEIRAWAN I think the intent of rook C7 was to force the move queen F4 because queen F4 now would transpose right back into C8. So maybe the move rook C7 cuts out some of the other nonsense of perhaps queen H4, for example.
MR. ASHLEY So rook C7, clearly Kasparov cannot give pause to defend this f pawn. The question is does his attack work? Does he get something going way down the line past the computer's horizon. But it doesn't feel like that kind of position. What can he do? Just to show you that this sacrifice here does not work, rook takes G3 check is possible and pawn takes G3 and now the f pawn looking for a glorious career goes to F2 with check and most unfortunately white's king can just step in the way and no checks anywhere game over.
MR. SEIRAWAN That's all she wrote.
MR. ASHLEY So nothing glorious for the world champion in that line. So that doesn't work. If the direct attack doesn't work with the rook, then the question is maybe some attack with the knight. Knight to F4 for example has to be considered.
MR. SEIRAWAN Well, let's give another shot here. I mean, this looks ridiculous even to my eyes.
MR. ASHLEY Rook to E8.
MR. SEIRAWAN Rook to E8.
MR. ASHLEY Well, this is a continuation we did not think about, bringing the rook --
MR. SEIRAWAN I think knight D6 is dead.
MR. ASHLEY -- to the e line. And the question is, is there a mating continuation. But knight D6 looks --
MR. SEIRAWAN Deadly.
MR. ASHLEY -- looks scrumptious.
MR. SEIRAWAN The idea of Garry's last move, by the way, rook E8, before we find an anecdote to it, I mean the intent is wonderful, is to bring the rook into an attack because obviously on the g line there's nothing further to talk about. The g line is close. So the idea is to be able to play rook E1 check and then to play knight takes F2 and set up a mate. However, DEEP BLUE's on move. Let's take a look at knight D6.
MR. ASHLEY Knight D6 looks devastating and Fritz 4 agrees by giving an advantage of 3 point 91 to white.
MR. SEIRAWAN Okay. So --
MR. ASHLEY That's brutal. And Kasparov is shaking his head. This turn of events he could not have anticipated. The world champion looks as if he's being crushed by the computer.
MR. SEIRAWAN He's so delicate, isn't he?
(Laughter.)
MR. ASHLEY I don't really quite know how to feel about this. Kasparov, arguably the best player of all time. We're looking at the best of all time, having this kind of miserable position against the computer. It has happened to him before that he lost a game against the computer and then proceeded to pummel it into submission. Will it happen with this latest generation computer? I do not know. But certainly the programmers at -- DEEP BLUE's programmers are going to feel quite ecstatic about their little creation today.
MR. SEIRAWAN Well, let's take a look at some of the forcing lines. I mean, what does he do after the move knight D6?
(Commentary from audience.)
MR. SEIRAWAN This game, just to repeat, Garry has played against the computer lots and lots of 5 minute games, quite a number of game 30. That means that both players have to play the entire games in 30 minutes. But this is an historic match in the sense that this is the first time that the computer is playing really on a human being's I have to, where it's like several hours for the game.
MR. ASHLEY What it's doing is the first time against a world champion because it has played other human beings and won in some games and lost also. But this time against a world champion, definitely a first.
MR. SEIRAWAN Well, let's just take a look at rook E1 check. We're going to see king H2. And then knight takes F2. Let's just take a look at knight takes F2, with the threat of rook H1. Now, does Fritz see a forced mate? But I'm afraid that -- of the move knight takes F7.
MR. ASHLEY And knight D6 has been played. Don't be surprised if Kasparov resigns against this move. He might just throw out his last effort with rook E1 check. Look at him shaking his head. He is devastated under the cold, cold calculation of the computer. And he knows it. He knows more than anyone that this -- there's no bluffing here. The computer has seemingly refuted his attack. He could very well resign. He could resign.
MR. SEIRAWAN No, I think we're going to see the move knight takes F7, knight takes F7 check, queen takes F7. Queen takes F7. Now I think we're going to see the move queen D8 check, keeping in mind you don't take the queen. There's a mate in 1 on the board. Queen to D8 check. No, queen G8 would allow queen takes D4.
MR. ASHLEY Or queen F6.
MR. SEIRAWAN So king G7. But now white could take the queen with check, rook takes queen. King takes F7. Queen D5 check.
And then white's next move in this variation would be to play a move like queen takes F3 and to keep the attack or even G4.
MR. ASHLEY Taking on F3 looks pretty good.
MR. SEIRAWAN Watch out for that d pawn. No, that's over. So this is a loss.
MR. ASHLEY That's just 1 random variation.
MR. SEIRAWAN This is what Kasparov is calculating. And again he doesn't have a lot of time left.
MR. ASHLEY He has played rook to E1 check. The king has gone to H2. Only move. And now Kasparov -- I wish he could pull the rabbit out of the hat, but I'm afraid the rabbit is dead.
(Laughter.)
MR. ASHLEY He has taken on F2.
MR. SEIRAWAN Knight F2.
MR. ASHLEY Knight takes F2 quickly. Kasparov with 5 moves to make. Time control. He has enough time, but he knows it has nothing to do with the position right now. We're seeing something very dramatic here, in losing -- you know, before the match I thought that Kasparov would win the match, you know, pretty standard, Garry with his strength. And I said it would be very exciting if the computer won the first game. That would make it interesting. Well, boy, Garry, I'm real sorry. Don't come looking for me.
It's interesting now, boy. It's real interesting. And the computer has just shown its superiority in this game, although the variation you showed earlier, rook D8, rook D7 allowing it to get off that attack looked very strong. And it agreed, at least Fritz 4 agreed, that that was a very strong variation.
MR. SEIRAWAN Exactly. And I was just going to say that for Garry what's very surprising is, and I do mean this in the greatest sense of the word, as a great compliment to Garry. Garry is one those humans that really fearlessly plays against computers in international competition. A lot of humans do not like to play against computers because of the psychological drama or lack thereof. And the fact of the matter is that Garry is a proponent. He'll put himself at risk and in this particular match, you know, really he is. He's putting a lot at risk with very little rewards, in the sense that if he was to win this match everybody would say, well, you're the big favorite. If he loses the match, everybody says what does that mean now? You're no longer the world's best player and blah, blah, blah, blah, blah. So he's put himself at considerable risk. And I think because he felt he was going to win very easily. What he did in this particular game for me was the most surprising, was he knows, he knows very well the computer's strengths. The computer is a great defer. And yet he chose to challenge it in its greatest strength.
MR. ASHLEY And like you said, he's so aware of this fact, you know, after the game you don't go up to him and say, Garry, you know that the computer played really well.
MR. SEIRAWAN You don't pitch on the inside.
MR. ASHLEY Why did you do that? He knows. But you can know. That is what makes us human. You know what the right and wrong thing is and, duh, you do it anyway. And you think, why did I do that? It was stupid. It happens to world champions too.
Unfortunately it's white's turn to move. Knight takes F7 is --
MR. SEIRAWAN But just to point out, the gentlemen are excited, that black does have a very important threat.
MR. ASHLEY Very interesting.
MR. SEIRAWAN But again Fritz is analyzed to a perspective of 5 pawns, which by the way in international grandmaster competition the victory of a pawn or even 2 is quite sufficient. 5 is overkill, if I can use that word.
MR. ASHLEY 5 is go home.
MR. SEIRAWAN 5 is go home. Let's pack up the pieces. So knight takes F7 and --
MR. ASHLEY And it has been played. Knight takes F7.
MR. SEIRAWAN Oh, boy.
MR. ASHLEY Well, do not be surprised if the world champion gives up in this position at some point. At some point it seems that he will. He will not finish this game.
MR. SEIRAWAN I think that after this game, and I do agree with Maurice and others, I think that now Garry realizes that it's over. But I think he's going to be really furious with himself because undoubtedly he was debating should I play rook D8 and just play the positional way and push my pawn home for the victory or should I just score a win?
MR. ASHLEY But he still has the problem of knowing whether or not that attack was good in hindsight. We'll be looking at it and saying, well, the computer didn't really have anything in that variation. But at the board is it mate, is it mate, oh, god is it mate, does the computer have something? It seems like it was much more difficult for him to do.
MR. SEIRAWAN Indeed. And it has always been Garry's nature when playing against human opponents to be the 1 on the attack.
MR. ASHLEY Yeah, always. True to himself and maybe at fault here. 1 more point, we should mention Garry Kasparov is at his best after losing a chess game. Kasparov rarely gets demoralized. He is at his absolute best after losing a chess game. So tomorrow's game you want to see. Believe me he lost to Anand and then he played his most spectacular game of the year and crushed Anand. Kasparov is the kind of guy who shakes off defeat very easily and is energized by those losses, not by somebody who might be demoralized by I'm playing bad, how stupid I am. He will kick himself, but he will also bounce back. That's the kind of person he is. You want to see tomorrow's game for sure.
MR. SEIRAWAN If you look at Fritz' serve, any move that isn't a check allows black to mate in 1. Any move that isn't a check -- oh, sorry. Thanks. Oops. But it's very interesting. You see there at the computer, you know, just analyzing a lot of moves.
MR. ASHLEY So the computer will continue to do that even when it should be focused on the task at hand, which is to get at the black king. And now different moves are being suggested. 1 move is a sacrifice queen which actually doesn't look that bad. Another possibility is the very simple knight G5 check and then rip off the pawn. And then once the f pawn goes off the board, then there are no attacking possibilities left to black. It just has a naked king ready to be mated. So I think knight G5 is a very simple solution to the problem and Kasparov once his f pawn goes with no binding on the white king, he looks like he's gone. And he's sitting there, you know, we know it, guess that he does. He knows it. He's been in this game. He's really been demolished. He was completely crushed for 1 bit of rash decisionmaking.
MR. SEIRAWAN That's right.
MR. ASHLEY And was it a flaw of Garry? Well, maybe it's a flaw of human -- I don't think it had to do with understanding in this position, a spectator mentioned. Clearly Yasser Seirawan saw another possibility and it happened. I think faced with a choice, Garry Kasparov went with his personality, which was not to be on the offensive, but to be on the attack. Typical human response. And that is our weakness, we dare say, and the computer is showing it here. We go with what we feel is good. We kind of feel, you know, this is what we want to do, instead of what may be true in front of us. So here looks like a very simple way of winning knight G5 check, rip off the f pawn, pieces just hung out to dry, black king completely open.
MR. SEIRAWAN Let's see it. Well, you see what's going to happen when white captures the pawn on F3, that pawn on F3 is a critical ingredient in Kasparov's mating attack. He loses that ingredient, there's no longer a mating attack and his king gets exposed. We'll see it, for example, after a move like king H6. Then knight takes F3. And what I have -- there's also rook takes H7 check, king takes G5. Let me just make my point. Knight takes F3 and then in this position I have 2 threats. I have the threat of rook G6 and I have the rook of c rook to C6, winning the king for the queen. So I'm just saying that that pawn is a crucial, crucial --
MR. ASHLEY There may be a variety of ways to win. The only thing about rook A7, it's a bit weird. Okay, maybe it's a little bit unclear for us but the for the computer it will also realize that's a winning continuation and it looks very strong in fact because on the next move it can just rip the pawn off on F3. So it looks like an embarrass the of riches for white, which way to go, how to beat the guy up. And I mean look at this. Plus 7.16. You know, is there someone with a towel back there? Just throw it, Garry, throw it. We don't want to see anymore of this. This is over. We may mention Garry's chances with the white pieces, let's talk about tomorrow. We kind of know that barring incident, like pulling the plug, the computer will win this game. It's within 99.99 percent certainty. Wrong time to bet on this game.
MR. SEIRAWAN You know, I've always fancied myself in terms of being able to predict things accurately and the crystal ball is now completely cracked. I had predicted that Garry was going to win 2 of the 6 games and that 4 would be drawn. That is to say he would win the match comfortably 4 to 2. I thought that it would be years, literally years from now, before a world champion like Garry Kasparov was to ever lose a classical game of chess. Well, okay, rethink. Tomorrow, obviously Garry will be very upset about this loss today. I think he really -- I don't want to say threw it or gave the game away or something, but he just played into the computer's strength and got punished.
(Commentary from audience.)
MR. SEIRAWAN Thank you. You are true. But the idea is, I think what Maurice has said earlier is correct. Garry is great. After losing a game, he really comes back very, very strong, unlike, you know, most of us. We take it to heart. And I really feel that Garry will come with a different attitude about this match.
MR. ASHLEY Well, tomorrow will tell, will it not? Tomorrow will tell. After losing this first game, if he loses again, boy, he might lose all of them.
MR. SEIRAWAN No, no. Don't be scaring us like that.
MR. ASHLEY But how good is a computer? Were you impressed? That's the question. Were you impressed by the computer's play in this game? Did you see anything new from Deep Blue in this game? Did it say to you DEEP BLUE is really substantially stronger than anything you've every seen?
(Commentary from audience.)
MR. SEIRAWAN Thanks, Hans, I love you too. I don't know how to answer your question.
MR. ASHLEY Rook A7. And Garry Kasparov has indeed resigned. Garry Kasparov has lost the first game.
MR. ASHLEY Wow, what kind of finish was this? And Kasparov no doubt will be upset. And don't expect him to come back here now to take questions. But as we said Kasparov, one the great comeback kings of our time. Certainly tomorrow's game will be critical to the whole complexion of this match. So I would advise that you take time out of your afternoon and come check in with us.
MR. SEIRAWAN Just 1 last thought. I know the match is over. I know the game is over, but I see a very pensive Garry at the board and it's conceivable I know he's not happy with himself right now. He may be coming out. I will ask our host ACM whether or not Garry will make an introduction or a speech.
MR. ASHLEY Well, we will see if he does come out.
MR. SEIRAWAN Maurice, I will go see if I can speak with Garry and invite him.
I just wanted to say that I've gone back to Garry's dressing room. The door was closed. I spoke with Bob Rice, the Commissioner of PCA, and Frederick Friedel, who is here also to assist Garry. And it's a little bit unclear whether or not at the moment whether Garry will be joining us. I certainly don't want to say that he will because that's not the case. I haven't spoken with him and it's indeed more likely that he won't be coming. But there's still an outside chance as he digests the new situation that he'll be magnanimous and join us. But other than that, you're welcome to stay. If he were to join us, it would be in another 5 minutes. If he wasn't, you'd lose 5 minutes of your life. Thank you. Go back to your conversations.
(Proceedings concluded.)
- - -
[ The Game | The Program | Commentary | Grandstand | Deep Blue ]
[ IBM home page | (C) | (TM) | IBM Small Planet Pavilion | Internet 1996 World Expo ]