Author: Mark Young
Date: 11:29:02 06/20/01
Go up one level in this thread
On June 20, 2001 at 13:53:47, Robert Hyatt wrote: >On June 20, 2001 at 10:54:47, Mark Young wrote: > >>On June 20, 2001 at 10:30:00, Robert Hyatt wrote: >> >>>On June 20, 2001 at 09:58:44, Mark Young wrote: >>> >>>>On June 20, 2001 at 08:57:24, Chris Carson wrote: >>>> >>>>>On June 20, 2001 at 08:16:36, Mark Young wrote: >>>>> >>>>>>On June 20, 2001 at 04:38:01, odell hall wrote: >>>>>> >>>>>>>HI CCC >>>>>>> >>>>>>> Since I believe it has been established that the Conflict Concerning The >>>>>>>Computer GM question boils down to a question of semantics, or the relationship >>>>>>>betweeen words and their meanings, i would like to add a thought. >>>>>>>Perhaps what Doctor Hyatt and others are saying is that Computers UNDERSTANDING >>>>>>>of Chess is at the 2350-2400 level, Although they may, or may not be Grandmaster >>>>>>>Strength. Personally i would agree with many here if they formed the statement >>>>>>>in that Context, i believe computers understand Chess actually at the 2100 >>>>>>>Level, but they play chess at the Grandmaster LEVEL, this is because they have >>>>>>>certain talents that Humans Lack, mainly the ability to accurately count >>>>>>>variations. So maybe we are all agreeing, but not to the wording, or meanings >>>>>>>of defintions. I am sure, mark and chris carson would agree that computers >>>>>>>understanding of chess is at the 2100 level or lower. But they are able to >>>>>>>produce Grandmaster level play, because of other talents which is unique to >>>>>>>Computers? Does this make sense to anyone? >>>>>> >>>>>>Good post, I agree with most of what you said. It is clear that computers are >>>>>>playing at a GM level. Titles and Understanding don't mean anything. All the >>>>>>understanding in the world means nothing if you can not beat the "idiot" >>>>>>computers. no matter what the excuse may be. Results have always been the >>>>>>standard of understanding in chess. No one gives a rats ass if Chris, Bob, or I >>>>>>understand something more about chess then some GM or IM, because we are not >>>>>>winners at a high level. In chess it always comes down to results. >>>>> >>>>>Mark, >>>>> >>>>>You have done a great job with your research and analysis on this topic. I am >>>>>surprised that some of your opponents would not even do the analysis, but asked >>>>>you to do it, which you did. :) >>>>> >>>>>One other thing bothers me, perhaps this bothers you to. No progrma has the GM >>>>>Title, but no program has an IM or FM title from FIDE either. Why do so many >>>>>say that progrmas are IM? >>>> >>>>It is worse then that Chris.... >>>> >>>>I. Why do Bob and others say computers are Grandmasters at 5 min. chess? >>>> >>>>II. Why do Bob and others say computers are Grandmasters at 30 min. Chess? >>>> >>>>III. Why do Bob and others say computers are International Masters at 40/2hr. >>>>Chess. >>>> >>>>I will tell you why for I. and II. and its the RESULTS. The beat many >>>>Grandmasters and have a high rating. >>>> >>>>The answer for III. is simple Hypocrisy. >>>> >>>>And that is what infuriates me, and its intellectual hypocrisy, the results show >>>>them to be Grandmasters at 40/2 hour, but instead of admitting this they commit >>>>intellectual hypocrisy again by calling them International Masters. >>>> >>>>It is blatant hypocrisy for them to call computers GM?s at 5 min and 30 min >>>>chess, but then site Fide standards for 40/2hours, but Fide has no Titles at all >>>>for 5 min and 30 min chess, you can not earn titles at fast time control games. >>>>They only base this on results?..and that is Hypocrisy when computers have >>>>proven themselves in the same way at 40/2hours. >>>> >>>>Q: Does anyone know what standard the chess program Belle was awarded $5000 for >>>>being the first computer to reach master strength? >>> >>> >>> >>>First, I don't remember it being awarded any $5,000 prize. I can ask Ken if >>>someone thinks that really happened. >>> >>>Second, it _did_ receive the "USCF Life Master" certificate at the 1983 WCCC >>>event in New York City. I was there. It earned that by playing in USCF rated >>>tournaments and producing an official USCF rating of 2208, without "excluding" >>>any games or events where it did poorly. IE it was just like Cray Blitz, and >>>all the others.. official members of USCF, playing in official USCF sanctioned >>>events, and producing the requisite 2200 or above rating. >>> >>> >>> >>> >>>> >>>>Q: How does this standard compare to my standard for saying that a computer is >>>>now a grandmaster level program? >>> >>>The rules are different. In the USCF, to become a master, you simply have to >>>get your rating over 2200. Nothing else. In FIDE, to become a GM, you have to >>>get your rating over 2500 _and_ produce a 2600+ TPR over a bunch of games. >>> >>>pretty simple, really... >> >>Bob what Fide standards do you use for calling computers GM's at 5 min and 30 >>min chess? You want to site the Fide standards for your claim on this, and I >>will retract my own standards for 40/2hours, and what Fide standard has a >>computer made for you claim that computers are International masters. > >I don't claim computers _are_ "Grandmasters" at 5 minute blitz games. I say >"they play like grandmasters, or even super-grandmasters," based on watching >many many games where they trounce GM players handily. > >I'm not aware of any FIDE "blitz GM title" so that isn't an issue here. If >a computer consistently produced 2600 TPRs against GM players, I would also >say "It plays like a GM at this time control" too. > >Do I think a computer could actually qualify for a GM title today? I think >it might be possible, but based on a circumstance that FIDE didn't consider >when setting the requirements. A computer is pretty consistent in how it >plays. No "bad days" no "good days". Just "consistent days". If you play >consistently, and you play in enough events, you will have a few very good >results just because you caught some players at a "down time". Since the >computer won't ever get caught in a "down time" where it is just playing >poorly because it is tired, ill, or distracted, it is more likely that the >statistical probability will kick in from time to time and give it a good >result. > >All you have to do is look at the 1980's USCF events where a company like >Fidelity would enter 4 or 8 copies of the same program. One would produce >a 2200 rating, the other three would produce 1700 ratings. Which was more >accurate? Statistically, 1700. But marketing types said "2200". :) > >That kind of 'spike' is the thing I believe _might_ let a computer actually >earn a GM title. If FIDE were to change the rules to say "two events of at >least 12 games each with a 2600 rating or better, and the two events must be >from a consecutive set of no more than six events." then that "luck factor" >would be played down a lot and it would become much more difficult to produce >the "norms". We _know_ a computer can produce a norm here and there. But >can it produce "enough"? And can it do this without crashing in a tournament >and producing a horrible result that kills the 2500 requirement? > > > > > >> >>Is Bob Hyatt the only one who can come up with his own standards. Very >>Hypocritical. > > >No... you simply want to be disingenuous. For a GM title, I go by the >FIDE requirements explicitly. For a pseudo-title like "blitz GM" then >I made up my own standards years ago since there are none that are formal >and recognized. > >Nothing hypocritical about that. BTW, you should check the definition of that >word. It has _nothing_ to do with making up your own standards. It has >everything to do with saying one thing and doing another. Doesn't sound like >me at all... hypocritical: Meaning Two faced or insincere, or Having double stadards, being artifcial or disingenuous, I know exactly what in means. No offence, but I find your arguments in this matter to be just what I have said. I admire your stubbornness, and your intelligence, it is obviously of the highest order, but I think you will go to any extreme to maintain your point of view. Even if this means contradicting your own logic and standards that you use to counter others points of view. > > > > > >> >>> >>> >>> >>> >>> >>>> >>>>I know Bob Hyatt knows the answer. >>>> >>>> >>> >>> >>>Sure do... given above. >>> >>> >>>> >>>>> >>>>>Also, if we were to make a truly Human Title comparison, then I think the >>>>>Fidelity Mark III/IV has earned the USCF title of Master. I think this is the >>>>>highest title any machine has earned. The Mark IV must then be the champ of all >>>>>the titled machines and my Mark III second (wow, no need to compare ratings, >>>>>Titles rule). Obviously the Mark IV has 2300+ knowledge, it has the official >>>>>title, this machine must obviously be years ahead of any other program that has >>>>>not recieved the title and no un-titled human or program can compare to the Mark >>>>>IV. >>>>> >>>>>I ofcourse disagree with the improtance of Title comparison, results count. >>>>>Average GM 2521. Average program on 486 to super SMP hw over 2525, >>>>>programs on 500Mhz and faster are over 2550, 866Mhz and faster are above 2625 >>>>>and fastest SMP's are above 2650, top performance 2702 against 2702 competition >>>>>(I would love to see a 2100 club player do that). :) >>>>> >>>>>Best Regards, >>>>>Chris Carson
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.