Commentary for white move 29:

And also just one more question, it's related to the second one. On the search tree, do you guys reevaluate the search tree after each move, or would you use -- I would imagine a part of the search tree since it considers --

DB MOVE: 29 Rd1

YASSER SEIRAWAN: By the way, I will defer and I will say that Angela did outplay me there, you had a better position in the end. You got me, you got me. We did have a move --

MIKE VALVO: Re1-d1? Is that right? An interesting move, an intriguing move. , voluntarily allowing the doubling of a pawn. Not so bad. Concerning your computer-related questions about hash tables and whether you can combine Deeper Blue and Deep Blue and things like that, I must say that I'm not a part of the IBM team --

MIKE VALVO: And they aren't talking too much. All I know is that Deep Blue and Deeper Blue has played each other, and Deep Blue loses to deepr blue by a ratio of 3-1 or 4-1. The question is what is the difference between the two? Basically positional evaluation, the new machine is twice as fast. But in terms of the other final details, they haven't really talked about it, that I know of.

YASSER SEIRAWAN: Regarding pawn play and things like that, one of the very biggest weaknesses, especially in the early days, and Mike you can speak better to this issue than myself, is that the computers did play endgames very badly. They play endgames very badly and I've seen just a very large increase in the ability of computers to play better endgames, and maybe Deep Blue understands these tricky pawn endings a little bit better than its predecessors and colleagues.

MIKE VALVO: They didn't understand how passed pawns become queens in the old days.

GK MOVE: 29...Qc7

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