Commentary for black move 18:

(Applause.) When I was your age, I didn't know what a bad bishop was. What the young buckaroo refers to there is that the dark-squared bishop is a bad bishop because the pawns b2, c3, f2, g3, h4 are all dark squares.

DB MOVE: 18...Ng4

YASSER SEIRAWAN: Deep Blue immediately offered the exchange of bishops, played Nf6-g4. Maybe it's a similar kind of game, the same kind of transposition. I'm not a big man of the move Bc1-g5. We'll just have to see how it pans out. Some questions directed at Matt. Again, Deep Blue is a phenomenal, phenomenal calculating machine, and there are some great things you may want to know, some insights. In the back, sir? Yes?

AUDIENCE MEMBER: Hi, my question is directed towards Matt Tennis. Deep Blue has been a great advance in chess computing but it hasn't really addressed the issue of time management at all. It has a very static approach on how much time it takes whereas the human tends to go into a deep think at certain points in the game. At some point Deep Blue is going to probably have to play a match with a Fischer clock. I was wondering if the IBM team had anticipated this in any way?

MATT THOENNES: Okay, once again you've got me off in a region -- I'm not one of the developers of -- actual developers of Deep Blue so that's a hard question for me to answer.

MIKE VALVO: Deep Blue does have some time management in a sense. First of all it saves 20 minutes in case it has to start the backup. That's the reserve time that it has. Secondly, it thinks on Garry's time, and that's a value -- if it guesses vary's move correcting, which it does 50 percent of the time. Also, it moves very quickly in the opening, and it saves up all that time. If it goats a critical situation where it's doing a lot of move extensions which consume a lot of extra time, it will use the time it takes to do that. So it's not completely static.

MATT THOENNES: Yeah, let me take a couple minutes and maybe give you a broader view of what I've been doing here. Maybe that will help steer some of the questions in the right direction. If you think about it here, I can identify really three areas which we've worked on to do this, and we've been working on this for approximately the last year, one of which is Deep Blue, which you've heard a lot about over the last week. Second area is really infrastructure. How do we make this happen. The video that you see here, the cameras you see in the auditorium, the information that's going out to the Internet from this site. And then a third piece is really public relations. How have we been helping the world understand Deep Blue, the technology, and what the complications are for the future with that technology. And with that, we've been working with the typical things such as press, schools, and museums, and we've interacted with several thousand media outlets. And a number they gave me as I calm out here, was they had over a million visitors on the web site. And the interesting thing is, Deep Blue is not the only technology here on the site. I'm going to use my cheat sheet a little bit. We've wired the building for fiber, to provide high-speed networking on-site. The press room upstairs can surf the Internet on high-speed network access. We've got a prevideo production event going on here. If you think about it, there are eight different cameras that are being switched in to the feed that you see here on my right. We've touched a little bit on the Deep Blue machine itself. In reading the press one thing that they didn't get quite right is that the machine that played last /TKWRAOER isn't -- year isn't the machine that played this year. That machine is still in Yorktown acting as one of our backups. The machine today is an R S 6,000 S P from positive /KEUPSsy. The backup that we've looked at, being computer scientists, things like that, we want to make sure that we have things that we can fall back on in case the machine fails. We've actually got a machine back in Yorktown which we can fall babbling to -- back to as a first backup and actually have a second small machine on site that's sort of the final backup if we lose, you know, everything we still have a machine here on site that we could play with. Besides that, this side is feeding most of the live and generated content that you've seen added to the web in the last week. It's coming back from back stage. The corporate Internet team is here providing that information. We have a live feed from Deep Blue going to the web site. So if the watch the web site, the moves that you see happening are actually coming out of one of the log files that we use for Deep Blue.

YASSER SEIRAWAN: Matt, let me interrupt you. You're saying there's a direct connection between Deep Blue feeding the moves to the World Wide Web, but it's not showing it's analysis, it's just showing the moves?

MATT THOENNES: Just showing the moves. You ask how would you input moves for the World Wide Web. Well, you can sit there and type them --

MIKE VALVO: Is there any chance that the analysis that Deep Blue has already shown on the web after the match, at least?

MATT THOENNES: That's really a question for C. J. as to, you know, where that goes. Besides that feed we have the stenographer here that --

YASSER SEIRAWAN: Two stenographers.

MATT THOENNES: Two stenographers, that are typing down every word I say. That's being fed out live to the Internet through some technology that's been developed over the last year. Last year we had like a half-hour lag, 15-minute lag before you saw what was going out there. We're also feeding live-stream audio out to the web. So my voice is also being carried out onto our web sites. And then off of this site there's servers in Schoenburg, Columbus, and actually in the U.K. which is what you interact with to work on the Internet to work with the chess site for this event.

YASSER SEIRAWAN: How many people worldwide are coming to IBM's site over the Internet? Do you have any idea, is it a hundred countries? Is it monthr?

MATT THOENNES: Given that the Internet goes all over the world, if you can access Internet, then you can access this event from all over the world.

MIKE VALVO: Do we have a question over here?

AUDIENCE MEMBER: I've been trying to get this question in for a while so hopefully you won't think I'm going too far back. Many moves ago when black played Bb4 and then white subsequently made the move c3, at first I thought he might play Bd6 instead of Be7. And Yasser Seirawan said what black was trying to do was win the f4 square. But also a subsidiary idea of that was to play e5. Now, I suddenly discovered maybe -- h4 was the first idea. The second idea was he plays Bd6 and he captures on e5, he didn't want to have the pin so that's why he played Be7. I just was wondering what your thoughts on that was.

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