Author: Sune Fischer
Date: 14:10:05 08/21/03
Go up one level in this thread
On August 21, 2003 at 15:42:39, Uri Blass wrote: >On August 21, 2003 at 14:48:55, Sune Fischer wrote: > >>On August 21, 2003 at 14:35:25, Omid David Tabibi wrote: >> >>>>>"Assumption is the mother of all f**k-ups" :) >>>> >>>>Then you can disable nullmove as well, that f**k ups as well some times. >>> >>>I use verified null-move pruning, it doesn't f**k-up :) >> >>Then it comes at a cost. >>You can't make a strong program without doing certain assumptions. > >Your assumption is wrong. >You can do it. > >Note that verified null move pruning does not cover all zugzwangs. >I believe that it is possible to cover all the problem without costs or at least >without significant costs(being 5% slower is not a significant cost) > >You can decide every time that you search to depth x with pruning to search also >to depth x/4 without dangerous pruning and if the score at depth x/4 suggest >dangerous zugzwang to stop using dengerous pruning. >By this idea you can practically solve everything after enough time. Most techniques carry with them some artifacts, hashing (rep. detection problem) and nullmove being two examples. My rule is simple, if the net gain makes it worth it, then I use it. :) >I do not say that it is the best solution but only demonstrates a possible >solution to the problem. > >The reason that programs do not solve simple positions is only the fact that >programmers decide that it is not important for them. Partly, they know very well it is important to a handful of computer chess enthusiasts, most if not all of whom read this forum, but I believe the average chess player in the end prefers a stronger program. To him a program is dumb if it fails, period. Personally I'd rather fail in a artifical position than in a realistic position. >generating a strong program does not contradict being able not to do stupid >blunders. If you had a pruning rule that gained 50 elo in all games but made blunder moves every time there was a quadrople pawn on the board, would you do the pruning? -S. >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.