Author: Stuart Cracraft
Date: 14:48:55 09/04/04
Go up one level in this thread
On September 03, 2004 at 23:01:26, Robert Hyatt wrote: >On September 03, 2004 at 20:38:16, Anthony Cozzie wrote: > >>On September 03, 2004 at 18:26:49, Álvaro Begué wrote: >> >>>On September 03, 2004 at 16:20:03, Rick Bischoff wrote: >>> >>>>How to avoid the "Oh crap I'm getting mated in 8 moves time to start throwing >>>>everything away to delay the inevitable" syndrome when your opponent is not a >>>>computer and might not even SEE the mate in 8? >>> >>>My program doesn't have any of this, and I haven't thought much about it, but, >>>how about making the move that you thought best before the score dropped a lot? >> >>This never occurred to me, but I bet as a computer you could get away with a >>_lot_ ;) >> >>If you are GM playing blitz, and the computer leaves a piece hanging in a >>complicated position . . . do you take it ? >> >>anthony > > >Berliner wrote about this kind of idea. IE the previous best move looked very >good until an extreme-depth search proved it wrong. Often when the best move >suddenly looks bad, alpha/beta comes up with a move that appears to be slightly >better, but to a human it just loses instantly. IE rather than give up a piece >for a pawn at some extreme depth where a human is unlikely to find the move in >real time, the computer just tosses two pawns trivially and loses. > >There Berliner suggested playing the original "best move"... and taking your >chance that your opponent won't see the deep stuff. Won't work against >computers of course... I remember this. It seemed like a good idea. My problem with it is knowing what is an extreme-depth search. Are we to take the PV and classify anything of length >= N ply as an "extreme-depth" search. It is interesting. Lasker would have liked it. Stuart
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.