Author: Sune Fischer
Date: 00:56:18 02/10/02
Go up one level in this thread
On February 10, 2002 at 01:47:04, Uri Blass wrote: >On February 10, 2002 at 01:25:30, Jeremiah Penery wrote: > >>On February 09, 2002 at 22:31:20, Robert Hyatt wrote: >> >>>I simply say there is no evidence to support a diminishing return theory. It >>>might happen, or it might not. Experimental data from today's programs, >>>searching to depths they might reach in the next couple of years, suggests that >>>no such diminishing return occurs for at least the next few plies. >> >>I think you're confusing diminishing returns in playing games with diminishing >>rate of finding a new best move at higher ply depths. Just because a program >>picks a new move at ply 15 over the move it had at ply 14 doesn't mean the game >>outcome will change. Sometimes it will, but it is not as likely as a new move >>found at ply 4 changing the game outcome over a move found at ply 3. > >I already said that it is irrelevant I do not believe it is irrelevant, there probably is a correlation, just hard to pridict exactly what kind because chess is so strange a game. I hope we can agree, at least, that the player who makes the largest number of bad moves will be the lower rated player. When we get close to the "perfect play" barrier, one bad move in 50 is enough to lose the game, so there is some strange behavior. >and in theory it is even possible that >there is an increasing returns from depth inspite of the fact that there is no >increasing returns in finding new moves. Maybe, but probably not :) >I believe that there is diminishing returns because of the fact that programs >have big book and the book has increasing returns but I doubt if there is >diminishing return in the nunn match results when programs often start to >calculate 20 plies after the initial position. Books would of cause invalidate the tests done, this is like shortening the game and would not help to prove/disprove DR. >An experiment of ernst heinz supported diminishing returns based on games of >Fritz6 against itself. > >I suspect that the book is the main reason for the deminsing returns but I do >not know because I never looked at the games. > > >Other people may think that the main reason is the fact that Fritz played >against itslef and the programs were not different. >I guess that it is not the reason and you will get similiar results between >different programs. > >In a different experiment of the nunn match of Fritz against Tiger I found no >evidence for deminishing returns. > >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.