Author: Walter Koroljow
Date: 04:31:56 01/18/01
Go up one level in this thread
On January 18, 2001 at 02:18:57, Dann Corbit wrote: >On January 18, 2001 at 01:52:31, ERIQ wrote: > >>wow, I think we have gone to a whole other level now. But I get the point :) >> >>my last words on this topic is that it does amaze me how some don't think that >>a comp. is on the same level as the "average" IM or GM this really does seem >>odd, but not totally surprising though as alot of players think that their >>hero is on a level all his own and I think just hate the thought of a comp. >>being there too. > >You might find it surprising (for instance) to discover what many of us are even >arguing about. The point of contention has nothing to do with whether or not >people think chess programs play on the GM level. It has to do with whether or >not it has been proven. > >I think that they probably are at GM level. I don't think it has been proven. > >I also don't believe in the composite computer player that has been manufactured >out of a collection of machines and programs, but I could certainly be wrong >about that. (IOW -- there may *not* be any weaknesses exposed by the same >program playing on the same machine hundreds or thousands of times, but I think >that probably there will be *some* effect). > >In any case, the object of contention is most assuredly *NOT* one of belief, but >(rather) one of proof. I think it has been proven that comps are at the GM level. Someone posted a while ago that the average GM rating was 2503 (I may misremember by a very few points). My post of yesterday shows that the 95% confidence for the average of comp ratings is 2503-2594. One could call that proof. But let me also give a counterargument. What the above shows is that if nothing changed and no one learned anything, then the average comp would be better than or equal to an average GM with high probability. But how much better? From the numbers above, one might guess 50 points or so. That is practically unmeasurable in a match of reasonable length. And, most important, how hard would it be for the average GM to learn enough anti-computer play to pick up 50 points? Remember, a kid with any promise at all picks up 200 points a year. Well, GMs are still kids when it comes to computers (almost a quote from Bob Hyatt, who gives specifics). There is plenty they can learn, and I have seen enough poor anti-computer games to believe that. Much as I hate to say something so "pie in the sky by and by", I think GMs could beat comps on the average if they devoted effort to it. By the way, I do not understand what a composite computer player is. Cheers, Walter
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.