Author: blass uri
Date: 12:20:26 07/15/00
Go up one level in this thread
On July 15, 2000 at 15:01:44, Jorge Pichard wrote: >On July 15, 2000 at 13:51:41, Jorge Pichard wrote: > >>On July 15, 2000 at 13:28:34, Jeroen Noomen wrote: >> >>>On July 15, 2000 at 13:18:28, Jorge Pichard wrote: >>> >>>Two points to argue that this doesn't work: >>> >>>1. The 5 Black pieces aren't developped yet. So how to count? >>> When is a piece on the queen's side and when not? Is a >>> non-developped piece a threat or not? >> >>1a. A piece does not have to be developped to be considered dangerous on the >> Queen's side or King's side of the board, the point in case here, is >> weather to Castle so soon on one side or the other without realizing the >> future potential of those pieces located on the same side of the castled >> King. >>> >>>2. Castling is not the only mistake in this type of position. >>> A computer program simply has no idea what plan it should >>> follow. And it will start moving pieces around, with no >>> purpose at all. >> >>Jeroen >> >>2a. Therefore, a subroutine would be very helpful on the earlier stage of the >> openings, where the center is closed. When a computer program play against >> a GM, it should not play the same opening so blindly as a human. what is a >> good opening for a human is not always a good opening for a computer. >> > 2b. The other game where Kramnik launched a king Attack, the same scenario the > center was locked, therefore Deep Junior did not have to Castle on move 6. > it could have waited, since the king was not in immediate danger it could > had played Nf6 and then develop the Queen bishop. The point here is that > Kramnik was just waiting to see where the program was going to castle, and > then mobilize his pieces slowly toward the king, preparing a king side > attack, now if the program did not castle, with the center closed it > would have been a hard wall to break with all DJ pieces surounding the > king. > > Pichard. The problem in the game of Junior against kramnik was the moves g6 and Kh8 and not castling. I guess that 100 times faster hardware could save Deep Junior against Kramnik but could not save Junior against piket(I am of course may be wrong about it). Junior5.9 played 25.Ka1 with 0.00 evaluation at depth 22 after 100 minutes on my pIII450(time control 9 hours/moves 25-30) and I guess that Deep Junior is not clearly better. I think that Deep Junior needs to know to evaluate king attack better and I do not like simple solutions that do not give Deep Junior the right evaluation. Uri
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.