Commentary for black move 25:

YASSER SEIRAWAN: Well, let's take a look. Go ahead. We're trying to get that Deep Blue will play a4xb5, and that Garry will respond a6xb5. And now Deep Blue has doubled its rooks on the only open file, the a-file. Obviously a very good strategy by Deep Blue Benjamin /PW-PBG well, in this situation black is now threatening to exchange all the rooks. But an idea for white could be to play Ra5, and now if a capture is made, then white could take back with the pawn. By the way, in the course of the last eight months, I've learned to use ChessBase very well. White would recapture and get a strong passed a-pawn, which could get support possibly by bringing the bishop into b6. So that's a possible continuation for white, just one of the many that Deep Blue will consider.

YASSER SEIRAWAN: We have a question from our audience.

AUDIENCE MEMBER: I was impressed that Deep Blue was in book for so long, over 20 bytes. I was wondering how deep does it go, and how many terabytes does it store for the book?

YASSER SEIRAWAN: Yeah, Joel? JOEL BENJAMIN: The second part of the question I can't begin to answer. What was that word again?

YASSER SEIRAWAN: Terabytes.

MIKE VALVO: A big number.

YASSER SEIRAWAN: It's a big number. JOEL BENJAMIN: Well, as long as the position has occurred before, it has some information.

YASSER SEIRAWAN: What is the longest series of opening luves in the -- moves in the library of Deep Blue? In other words, could Deep Blue play 20 ply, 30 ply, 40 ply, 50 ply of opening library moves were Garry to play such a line? JOEL BENJAMIN: Yeah, it could certainly play 30 moves deep.

YASSER SEIRAWAN: 30 moves for both sides, deep? JOEL BENJAMIN: Yeah.

YASSER SEIRAWAN: And how many -- do you have any concept of how many opening moves may be in Deep Blue's library? Storage? A million opening moves? 400,000 opening moves? JOEL BENJAMIN: Well, basically if it's been played, it's there.

YASSER SEIRAWAN: So you're saying anything that's been recorded is there, anything that's been played is in their database? JOEL BENJAMIN: Well, I don't think we use a completely comprehensive database because I think that could be self-defeating ultimately, because you're not really interested in games that are played between very weak players.

YASSER SEIRAWAN: Exactly. JOEL BENJAMIN: But, you know, any important game that's been played, I believe it should know about.

YASSER SEIRAWAN: We have another question from the audience. Yes, please. Yes, sir?

AUDIENCE MEMBER: What is the nature of fine-tuning, if any, since yesterday's game by the Deep Blue team? JOEL BENJAMIN: I don't know the answer to that, but I could say that probably there aren't going to be too many major changes during the match, because there's always the danger you could change something that will hurt in some area that you hadn't thought of.

YASSER SEIRAWAN: We have a new move by Deep Blue. We have the move f2-f4, the old switcharoo. White up at this moment -- to this moment had been playing everything on the queen-side flank with the moves a4, B 1910 Ra1, Ra3, and so on, and now he's switched to the move f2-f4, pressuring black's e5 pawn. We see that this is a very standard strategy actually in a lot of the classical Ruy Lopez positions. I like the move. JOEL BENJAMIN: Yes, it's kind of a classical strategy of opening up a second front, because the queen-side by itself may not offer enough winning chances for white. The black pieces might be able to defend just one side of the board. But if you open a second front, and you get -- you put pressure on black to sort of stretch the defense, you know, much like you might in a basketball game by spreading out your players so the defense has to guard players all over the court. When you start to spread out the defense you might achieve a break through. So sort of going with a classical type of strategy. It didn't rush the move f4. It waited until it had made some progress on the queen-side and gotten black a little bit tied up there, and then it goes forward with f4.

MIKE VALVO: It seems that the computer does very well with incremental improves just make a couple of yards on every play, just keep squeezing. But yesterday it had no clear objectives and it didn't know what to do. It kind of vacillated until it was able to throw the whole position into chaos for a while. JOEL BENJAMIN: Well, it certainly played some strange moves yesterday, but also there was a part of the game where it played extremely well. Around the moments when it was playing moves such as Bc7, back to d6, and it got that bishop on the -- to c5 and started putting pressure there, and then of course that terrific move Bc5-e7, which was totally expected from me but as soon as I saw it I knew it was a good move because it just had to be.

YASSER SEIRAWAN: Absolutely. I just wanted to say that I took a moment today to go through yesterday's game. And we criticized some of Deep Blue's moves, and rightly so. Do we have another move? Indeed we do. But I just wanted to say that in yesterday, in hindsight, Kasparov played an absolutely magnificent game. It was a really, really tremendous effort. And what you're saying there, Joel, is that Deep Blue came back later in the game, but it was already too late, and that Garry played -- most Mort always might have gotten destroyed after the move f5 and the whole position blew up. It was very easy for Garry to go wrong? JOEL BENJAMIN: I think there are very far Grandmasters in the world that would have actually survived that situation. So, you know, you have to take that into account in assessing Deep Blue's strength.

Real-time text commentary is made possible by LiveNote, Inc. and Vincent Varallo Associates