Author: Robert Hyatt
Date: 08:06:44 07/14/98
Go up one level in this thread
On July 14, 1998 at 01:48:00, Shaun Graham wrote: >On J >> >>Your logic is broken and you are using a circular argument here. You *can not* >>play games against a GM "anonymously". It isn't done. Not in FIDE, not in the >>USCF nor in any other chess federation. > >No offense Mr. Hyatt but you don't even know what a circular arguement is. It >is an arguement with a structure similar to the following. this is also circular: if we could play games against a GM without the GM knowing that he is playing a computer then fritz could be a GM. But it is impossible to play a GM without his knowing his opponent. But if we could play games against a GM without the GM knowing that he is playing a computer then fritz could be a GM. But it is impossible... looks like a circle, smells like a circle, feels like a circle... by golly it is a circle. Your pre-condition is impossible, so your conclusion is improbable. > >"God exists. We know that to be true,since the bible plainly tells us that god >exists. And we know that what the bible tells us is true, since the bible is >the word of God." -- A circular arguement >> >>So that argument is no good. > >Please > > The GM is going to know he is playing fritz or >>whatever computer, and he will employ anti-computer strategies, and he will >>win most if not all of the games. And no amount of hand-waving can solve that. >> >No one claimed that such an experiment wasever going totake place. The point >was that if one was to carry out a (VSE)Valid Scientific Experiment to find out, >how a chess program would perform against normal grandmaster play then the GM >could not know that he was playing a computer, because the inherent bias would >invalidate the results of the experiment. Find any text on conducting valid >experimental science in the world and you will find no difference. > typically valid scientific experiments are founded in "fact". IE who would want to try to simulate a weather system in a universe where water freezes at -10 degrees C, or where it boils at +70 degrees C? Because those conditions don't exist naturally. Ditto with your GM scenario. Chess has *never* been anonymous. It is doubtful it ever will. I maintain that after playing fritz 2 games, a GM will *know* what he is playing. If you play a 12 game match, you might win the first (unlikely), but you will very likely lose the last 10, with maybe an occasional draw thrown in. So what's the point in arguing about something that won't ever happen? >>Yes, *if* he didn't know, the game might be different. > >The game would be different and that is the whole point. Finding out how a >computer would perform against regular grandmaster play(most specifically as it >would occur in a swiss system event) the game would be different if white always had to play without his h-pawn too. What exactly does a "variant" of chess prove in this context? > > And if a frog had >>pockets, he'd carry a gun and never worry about snakes again, either. But it >>isn't going to happen as you suggest, ever. > >Strawman sir, please no emotion use evidence this kind of thing does nothing to >support an arguement. > >What kind of scientist are you? We don't say never. > I'm the kind of scientist that bases experiments on real conditions, which are the conditions most of us are most interested in. IE you want to play a computer vs a GM in a rock concert, or in a totally black room, or where the GM doesn't know who/what he is playing, or whatever, feel free to do so. But *that* would not be evidence of a GM program. Because the title "GM" is awarded in a specific way, and that does *not* include the conditions you have set forth. > >> >>As to why I'd play "anti-computer" it's simple for a chess-player. If I know >>you don't like tactical games, expect to face a gambit when we play. If I >>know you don't like strategic games, expect a blocked/closed position when we >>play. I play the best way I can to beat my opponent. I *know* computers can't >>handle closed positions, so why not play what they do the worst in? Makes sense >>to me. Makes sense to IM/GM players too, apparently... > > >If you are trying to say that GM's can sometimes beat computers there is >certainly no arguement against that. What is of concern, is can programs such >as fritz play at a GM level. Lets qaulify what is meant by GM. Arthur Bisguier >is a GM, Arthur Dake is a GM, both are elderly, and rated a little more than >2300 USCF. Even if they knew that they were playing fritz they would probably >have serious trouble, and if they didn't know, they would definitely be toast. >So what is important to analyze is the ELO rating. If Fritz can achieve a elo >rating 2450-2500 that's really all that's needed to show that Fritz is the >strength of what we call a GM. Does fritz have some qaulities such as >flexibility, the answer is no. But does it have the strength? The theoretical >point of the earlier statement was not that Fritz would play anonymously in >swiss tournaments, but that if it did, that it would probably be ably to acquire >the GM norm, because without the the anti-computer chess strategys being pressed >upon it Fritz would win a considerable amount. If a grandmaster plays Fritz a >regular training game, for the purpose of beating humans, thus he/she t played >fritz as if it were a human. I doubt that anyone would say that Fritz would not >put up what you would call Grandmaster resistance. I would. And I'll be happy to point you to a couple of GM's that agree with no reservation. I can think of several programs that would do better against human GM players than Fritz. Fritz is quite strong tactically, but it's positional understanding is very weak. I'd much rather see Hiarcs, or Junior or even Crafty take on your GM, because they are not primarily "materialistic" and would offer the GM some positional challenges as well, that would at least make the game more interesting. You simply underestimate what a GM chess player is capable of, probably due to not having interacted with any. I've met several and am good friends with a few, and I still realize that programs have a *long* way to go. When a GM plays "chess" programs have a chance. When a GM plays anti-computer chess, there is virtually no chance. Some of us are working on this problem. Some are not. Fritz belongs in the latter category. It is fearful against other programs, It is not against GM players.
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.