Author: Soren Riis
Date: 10:38:20 12/14/98
Go up one level in this thread
On December 14, 1998 at 10:29:47, Thom Perry wrote: > If the sacrifice is the correct move and forces a winning position, it seems > to me that this should be fairly easy for the good chess programs to > demonstrate: just let the program play out the game after the sacrifice. > I'm going to do just that with MCP8 tonight and with the learning enabled. > Then I'll replay it several times until MCP8 can find no better alternative, > regardless of which color tends to win. I find this idea very exiting! My understanding (but I am not an expert) is that so far such ideas goes somewhat against the provisional wisdom. However machines are getting so fast and there playing strength so hight that is certainly believe Thom Perrys idea could be extremely successful. The strong chess computers still remarkeble weak if judging the value of an attack. We one could use half the computation time in isolating and investigating sacs (and other unusual moves) which leads to unbalanced positions modern programs could get a quantum leap forward in playing strength. The whole idea of testing conjectures, like "does this attack works", "is it worth winning the pawn" etc. seems very sensible given the enormous speed and strength of modern computers. Did anyone try to implement Thoms Perrys idea? Soren Riis
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.