Author: Robert Hyatt
Date: 06:25:59 01/13/00
Go up one level in this thread
On January 13, 2000 at 08:02:22, Vincent Diepeveen wrote: >On January 12, 2000 at 23:49:43, blass uri wrote: > >>On January 12, 2000 at 18:59:49, Vincent Diepeveen wrote: >><snipped> >>>Deep blue searched only 11 ply here, so we must assume it didn't >>>see the manoeuvre in which the white passed pawns nearly queen (and >>>searching deeper you do see them queen, causing black to prevent that >>>and give up another pawn). >> >>We do not must assume nothing because we do not know exacrly the extensions of >>deeper blue > >Yes we do know the extensions exactly. Forced moves are first 8 ply of >search seen deeper, but not the last 4 ply of search. >Further we can assume the usual extensions (recaptures, check), and >before the qsearch sometimes 1 ply extra if a pattern is there. > > > >>Uri Sorry, but Uri is exactly right. We don't know everything about DB. IE until I told you, did you know that they extend when two moves are better than the rest? I doubt it because I didn't now it until Hsu's book review. Next, it is very difficult to decide whether a move is singular by their definition. What margin of S are they using? That makes a difference. What things are they seeing that we don't expect? That makes a difference. What is their eval scores like? They may have enough big bonuses to override a line where you say "anything but X loses a pawn." It is _impossible_ to predict what they saw, or didn't see, without having the program handy to check the important lines. And I don't see why you are so obsessed with Deep Blue. If you spent as much mental effort making Diep better, it might be 100 points stronger. :) I am not particularly interested in DB. It isn't mine.
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.