Commentary for white move 25:

YASSER SEIRAWAN: Yes, Garry has been very provocative, and as Angelina said, when he played Nh4 in game one provoking g5, we see the same type of thing here. To answer the gentleman's question from my perspective, first of all I do have a much different perspective than most of my colleagues. My first one is that I actually enjoy very much playing against the computer. To my mind, the computer has been a great aid to me. I've learned a great deal with middle game positions --

DB MOVE: 25 Qd3

YASSER SEIRAWAN: Not offering the exchange of queens as I thought.

ANGELINA BELASKAYA: Probably thought that after Qe3, and if play Qe3, after Qxe3 there is a problem, because if Rxe3, d4 with the pawn, and the pawn attack's both pieces, the knight and rook. And if pawn takes the queen on e3, now we have doubled pawn for the computer, and I think the program just doesn't like to have those two pawns, because probably pawn on e5 will just lost immediately after moving the Nc7 or moving Nd8.

YASSER SEIRAWAN: Very good. Now we see why she's women's U.S. champion, right? (Audience applause.) All right, after the move Qd3 maybe the computer has a clear idea of playing Nc3-e2, and then -- let's just see this. So the computer, let's say Garry --

ANGELINA BELASKAYA: The first idea is to capture pawn on g6.

YASSER SEIRAWAN: Didn't even realize that. Obviously the first idea is to capture the pawn on g6. Probably the computer would be eyeing that pawn very greedilly. I was just going to say that Garry were to play a consequent move Nd7-b6 trying to come to the c4 square, a positional response would be to play Ne2 -- Nc3-e2, attacking the queen, the black queen would be perhaps say forced back to the f7 square, and then /THAEURTZ a possibility of Nd4, establishing the knight in the center of the board. Something like that we might be seeing a variation like that. Finally, to answer the gentleman's question. I love playing computers. I think of them as a tool. They're great database engines. I can call up all of Mike's games or Angela's games, my own games, and I can learn a great deal by analyzing databases. Also for chess programs let's say I have a difficult position I don't know how to answer the position, I can just stick it out computer, it will analyze all night. I'm col back, I can print out its analysis, and I may learn something new that may bring me a victory. From the perspective of when the computers will become the word's best players I think it's inevitable but not for a lot of the reasons that we've been discussing. My view is one that deals solely with the issue of memory. What I mean by that is more chess games are being recorded than ever before in the history of man. I mean millions of games in a database, all of the opening moves, even the tricky ones like Garry, Garry's moves today. These games will be cataloged so that for the future games what you have is a number of moves in the opening. Let's move that the average chess game lasts 60 moves. So today, most of the databases go 20, even 25 moves deep. So the opening, if you think of a time graph, a move graph, the first 20-odd moves are let's say nonoriginal moves by the computer, it's just playing its computer library. On the other end of the sprect rum -- spectrum, you have the endgame. And the databases for the computers in the endgame are going deeper and deeper and deeper, and I refer you to the mate-in-68 moves that was announced by a computer in a recent computer vs. human event, because it had this endgame on a CD-ROM. So if you imagine this graph, you have the openings with more and more moves being cataloged, you have the endgame with more and more endgames being cataloged and what you have in the middle is the middle game, so this window of opportunity for players like myself and others to outplay the computer is getting squeezed. As the years pass, that window is getting smaller and smaller. I remember my first computer had 20 megabytes, and I thought that was great. My computer today has a gigabyte. And what's the next one, a teraflop or something?

MIKE VALVO: Terabyte.

YASSER SEIRAWAN: Terabyte? So you can stuff tons and tons and tons of data into a computer, and its recall is perfect. And that's the diff. The recall of a computer is per. So that's my view, and it may take ten years for that to happen and I think that's a reasonable amount of time before it's a memory game.

MIKE VALVO: Looks like Garry's got a problem in this position. He doesn't have much time, and the clock is ticking away, and it's his move, and the computer is attacking a pawn. This must be annoying for him.

ANGELINA BELASKAYA: Let's go back to the position.

YASSER SEIRAWAN: Another question from our audience, please? If we go go down to our juniors.

AUDIENCE MEMBER: This is a good thing to speak with?

MIKE VALVO: Grandmaster William Lombardy.

YASSER SEIRAWAN: Grandmaster William Lombardy.

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