Author: Robert Hyatt
Date: 20:17:35 12/29/03
Go up one level in this thread
On December 29, 2003 at 08:51:19, Anthony Cozzie wrote: >On December 28, 2003 at 22:47:38, Robert Hyatt wrote: > >>On December 28, 2003 at 22:06:59, Anthony Cozzie wrote: >> >>>On December 28, 2003 at 19:58:42, Robert Hyatt wrote: >>> >>>>On December 28, 2003 at 16:48:55, Vincent Diepeveen wrote: >>>> >>>>>On December 27, 2003 at 11:58:58, Slater Wold wrote: >>>>> >>>>>It's not clear who is +5 here, >>>>> >>>>>but whatever the side that is +5, let's consider >>>>> >>>>> a) hardware dated 1996/1997 >>>>> >>>>>playing hardware dated 2002/2003 >>>>> >>>>> b) a public book used for junior >>>>> >>>>>this where everyone knows the junior style is unique and requires certain >>>>>positions to get a full point. >>>>> >>>>> c) everyone testing at home junior is completely killing it with their >>>>>tournament books and newer versions. Even falcon scores 60%. Diep 95% (one >>>>>draw). Others reported similar % at home against it. Crafty gets about 50% >>>>>against the public version having a 10 times hardware advantage against a >>>>>crippled junior as it doesn't get through the tactical barrier. >>>>> >>>>>So let's first consider that at world champs >>>>> a) crafty gets killed in book >>>>> b) it is equal hardware, despite all BS stories he tells here it >>>>> won't work in 2005 either at those supercomputers of course >>>>> c) junior gets above the tactical barrier >>>>> >>>>>End of crafty. >>>>> >>>>>And *that's* why Bob doesn't show up at world champs, whatever nonsense he tells >>>>>here. Everyone who has a chance to win there will show up! >>>> >>>>Then should we not ask "Why do _you_ bother showing up?" You clearly >>>>don't have a chance of winning. Based on lots of past tournaments. >>>> >>>>So I suppose I _completely_ fail to understand your point, assuming you >>>>actually _have_ a point beside just knocking a program you can't beat... >>> >>>I have to more or less agree with Vincent here. The simple fact of the matter >>>is that every circumstance was stacked in Crafty's favor, and it barely won. >>> >>>If >>> >>>10x speed difference + Opponent runs at 40/5 min + public book = barely winning >>> >>>then >>> >>>4x speed difference + Opponent runs at 40/2 hrs + private book = crafty crushed >>> >>>anthony >> >> >>Let's try the math again. >> >>It beat rebel 12 handily at 8-1. It barely beat junior at 10:1. It beat >>shredder 6-4 at 8:1. Hiarcs is "in progress". If you slow crafty down by >>1/2, it would _not_ get "crushed". We can easily test that hypothesis however. >>A factor of two does not go from "winning" to "getting crushed". > >I agree: if Junior played on a 266 celery and crafty on a 1G athlon (instead of >a 3G P4) the score would probably be about equal. But if Junior played on a >4xP4 and Crafty played on a 16xOpteron, I think the score would be heavily in >Junior's favor. I don't see how, nor do I agree, but it isn't worth arguing since it isn't "testable" other than as Peter/Slate are doing. > >>The "circumstances" were not exactly in Crafty's favor either. I don't use >>that wide ICC book for tournaments. I don't play those opentings for >>tournaments. The commercial programs come with a very good default book, that >>has been hand-customized by a professional book writer. My book is just a >>collection of games combined into a big book. The matches were played with >>book learning off, which further hurts a pretty randomly-created book more than >>it does a hand-crafted book. > >We both know guys like Noomen & Kure don't put their best novelties & lines into >their commercial book. They hoard them for important events. Crafty simply >does not have a hand tuned book. Neither does Zappa: we both have to live with >it, but it is clearly worth 30-40 elo to have a good human tuned book. We also know that guys like myself are _not_ going to get caught in book traps. I've been doing this too long. :) There are solutions to the problem, but I will leave it as an exercise to the reader to figure them out. :) > >>But the point was, and still is, Crafty _would_ have been competitive in the >>2003 WCCC event. There has _never_ been a claim by me that I would win. I >>would _certainly_ have chances to win. I've won CCT events with commercial >>programs present and accounted for. Vincent had already declared himself the >>winner of the WCCC due to his hardware. I _know_ better than to make such >>stupid claims... I've been around _way_ too long. > >I do not think Crafty would finish last. But neither do I think it would beat >Fritz, Junior, Brutus, or Shredder, even if it had a 4x hardware advantage. > >anthony Note that it has done it in the past. At WMCCC events, in fact. I don't agree with the basic hypothesis, for that reason...
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.