Author: Bruce Moreland
Date: 20:04:36 12/18/02
Go up one level in this thread
On December 18, 2002 at 21:35:27, Rolf Tueschen wrote: >Thanks. Now you give the two exceptions as if you could forget them. But exactly >the LONGER plans are characteristic for good GM, so, how about a rethink of the >whole trick? - Would you try to hold it up or is it just one of these tricks to >let LOOK a prog quite good while it has still the known weaknesses? In this case a "long" plan could be white pawns f2, g3, h2, white king g1, black pawn f3, black plays Qh3 while down a rook. Normally it takes three plies to see this: Qh3, any, Qg2#. Three plies doesn't seem like a lot, but when you do null move pruning with depth reduction of two plies, you will have to enter this node with a six-ply search in order to see this. Without null move, you need four plies to see this is threatened (your own move, plus three plies to kill you). This seems incredibly stupid, but in practice, adding null-move forward pruning is like giving the program two swords and a heavy dose of amphetimines. The difference is amazing. It is possible to stand off and shoot the thing Indiana Jones style, but if you don't have a gun, you are screwed. >Therefore my idea to start a new chapter with real knowledge. > >To me it looks as if programmers are always looking for another trick to pretend >something that is _knowingly_ not there. Isn't that simply impostering, Bruce? - >Or is it in a (anyway) helpless situation, like Huebner already stated in the >seventies after he visited the MIT? Is your only chance the hardware? I have been not doing computer chess for a long time. It seems like nothing much has been going on since I've been gone, I'm seeing articles about how everyone has hit a wall. That's fine, I think it's good for computer chess. If everyone is out of ideas, then the one who takes a hammer to the whole thing first is the one who comes out ahead. It's not a particularly easy thing to do though. bruce
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.