Author: Robert Hyatt
Date: 15:39:46 06/01/04
Go up one level in this thread
On June 01, 2004 at 18:28:50, Sune Fischer wrote: >On June 01, 2004 at 18:15:31, Robert Hyatt wrote: > >>On June 01, 2004 at 17:47:42, Sune Fischer wrote: >> >>>On June 01, 2004 at 13:55:43, Robert Hyatt wrote: >>> >>>>On June 01, 2004 at 12:59:47, Sune Fischer wrote: >>>> >>>>>On June 01, 2004 at 12:35:18, Matthew Hull wrote: >>>>> >>>>>> >>>>>> >>>>>>It's like pulling the legs off a bug to measure how fast it's wings flap. They >>>>>>are dismantling the chess-playing entity known as "crafty" and measuring one of >>>>>>it's component parts, then claiming the result as the product of how "crafty" >>>>>>plays chess. It is simply "evil". >>>>>> >>>>> >>>>>Yes but you can hardly expect anything else. >>>>> >>>>>I have little doubt that Crafty is in its strongest configuration is when it is >>>>>operated by Bob himself, running on a 4-8 way opteron, playing with learning on, >>>>>ponder on, 15 GB hash, 250 GB ETBs on ultra fast SCSI drives, private tournament >>>>>book, etc.. >>>>> >>>>>So no matter how we choose to play with Crafty we cannot make Bob happy :) >>>>> >>>>>Consider though, that for many of the other engines similar things apply. >>>>> >>>>>-S. >>>> >>>>Bob is "always" happy when crafty is played with everything in "default" except >>>>for the bookkeeping things like hash, hashp, cache, tbpath, etc. That's the way >>>>_I_ play it, that's the way I test it. That is the way I believe it plays the >>>>best it can play. >>>> >>>>Nothing wrong with someone turning anything or everything off. But it should >>>>not be called "Crafty" in that case. Perhaps "Crafty (customized)" or something >>>>to indicate it is not the "normal" crafty... >>> >>>It is called Crafty with ponder off or Crafty with no learning, what is wrong >>>with that? >>> >>>The tournament conditions are usually specified for that exact reason. >>> >>>Nobody is claiming it is Crafty in its strongest setting, just that these are >>>the conditions it must play under, like all the other engines in the tournament. >> >>The point is that the conditions are not "identical". Program A with good book >>and no learning, program B with weak book and learning. Turning off books hurts >>A more. Turning off learning hurts B more. > >So be it, let it hurt B more. > >We wanted to know what the strength relations were with learning off, now we >know. Why? Do you want to know the strength relation with the evaluation turned off? Selective search turned off? Etc??? > >This says nothing about the strengths with learning on, of course, but we never >claimed it would either. But it certainly leaves that impression, wrong as it is... > >>So the idea is flawed from the >>beginning... > >I'm sorry you feel that way. > >-S. That's life. I believe the hand-tuners have their place, and that the programs should be run with their books "as is". I believe that the "learners" should be treated the same way. Can't turn off hand-crafted openings. Why turn off learning?
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.