Author: Amir Ban
Date: 13:47:31 07/15/98
Go up one level in this thread
On July 15, 1998 at 11:08:34, Danniel Corbit wrote: >People seem to want to discuss whether or not a computer program is a GM or not. > >The FIDE list does not contain any computers. > >Computers are not FIDE GM's. > >Hence, if we want to talk about computers being GM's or not, we need some new >kind of definition. > >What _exactly_ is a computer GM? > >Being able to beat one seems pretty irrelevant to me. I'll bet once in a while >GM's lose to their kids because they are not paying attention or whatever. > >If we do not know what we are discussing, then the discussion seems rather >pointless. I think the definition that we can agree on is this: A computer can be considered a GM if it performs in tournaments in a way that for a human would gain the GM title. The only reason that computers cannot practically achieve a GM title this way is that FIDE doesn't award titles to computers, but that's a technicality that we want to work around. We want a definition that does not discriminate against computers. On the other hand, we don't want to discriminate in favor of computers. I think many of the posters here want to make convenient shortcuts for computers, and declare them GM's for performance that for a human would not even make a single GM-norm. (The Fredkin committee made a big shortcut for Deep Thought to declare it a GM for performance that would hardly make an IM, but at least they set the rules out in advance). It's tough to be a GM. You don't only have to be very good at chess, you have to play a lot and succeed. No matter how good you are, you'll never be a GM until you play regularly in tournaments for 3-4 years at least. I told here recently that Junior only makes high 2300's after about 4 years of (quite successful) tournament play. And this in spite of the fact that Junior is one of the few programs that takes its playing career with some seriousness and consistency. Maybe Junior plays at GM level and maybe it doesn't, but to be fair, why should it be considered a GM if it doesn't have the rating and the norms, when we would not consider a human a GM unless he's actually gone through the long cycle of gaining the title ? You never hear the argument "John Doe is a GM" between humans, even if John Doe seems to be very strong. If he's GM strength, let him gain the title by the book, and only then everyone calls him GM John Doe. This is not academical. I know of several players who are certainly playing at GM strength, but still have a year or two, with luck, before making the title. Nobody in his right mind calls them GM's. I agree that some programs may be of "GM strength" today, but it will take them some years to prove it (according to my definition). This system was designed to prove over many boards that you really are "GM strength". So I would say: Fritz is not a GM. Perhaps it's performing now at GM level, but that's something completely different. Amir
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.