Author: Otello Gnaramori
Date: 12:48:17 04/13/02
Go up one level in this thread
On April 13, 2002 at 13:43:58, Roy Eassa wrote: > >If someone were to handsomely pay 5 strong GMs (they need not be the 5 top GMs, >but younger is better in this case) to spend 3 months (8 hours a day, weekdays >only -- a regular job) with 5 very fast Athlon systems... > >...to come up with skills and techniques to dramatically improve their results >against such programs as Fritz 7b, Chess Tiger 14, Gambit Tiger 2, Hiarcs 8, >Junior 7, and Shredder 6 Paderborn... > >...and the GMs would share ideas and results... > >At the end of the 3 months, how much further would the art of anti-computer >chess have evolved compared to where it is today? > > >Is there NO chance that GMs who would go through an exercise like this could >come out of it able to consistently beat today's top programs on today's fast >PCs? > >Or can you can acknowledge that such GMs MIGHT (not 100% WOULD, not 100% >WOULDN'T) acquire such a skill under such circumstances? > >Today's GMs, with their current set of skills and motivations, are indeed >matched and often outdone by today's best programs running on fast PCs -- there >is ample evidence of that! > >But how can one reasonably conclude that this fact RULES OUT the ability for >smart people to study and develop new skills? > >I did not believe (as some did in the '70s and early '80s) that no computer >would EVER consistently beat some GMs. And I do not believe now that no GM will >EVER consistently beat computers. > >I hereby register my prediction that, within the next 10 years, a human player >(most likely an IM or GM) will be able to consistently beat the top chess >programs running on that day's fast PCs. > >Not EVERY IM/GM will acquire these skills, but I think it's over 50% likely that >AT LEAST ONE will. I think that the only sure way to beat comps is memorize the moves of a game where you found a possible weak point (after many trial and errors) and try to force the program to repeat the very same moves. It's an hard (and also unfair job) task since the computer also depend from hash table (and from other eval setting) and can possibly not repeat the very same moves, but with a bit of luck you can carry out this trick. w.b.r. Otello
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.