Author: José Carlos
Date: 15:55:57 06/09/99
Go up one level in this thread
On June 08, 1999 at 17:31:47, Christoph Fieberg wrote: >Humans are far away from playing perfect chess but computers will come more and >more closer to it. At the turn of the century the best computer (Deep Blue) and >the best human (G. Kasparov) are roughly at the same level of about Elo 2800. I >assume that this will not last very long and that in the next 20 years the best >computer (X1) will score 80% against the best human meaning that will see chess >at Elo level 3040. > >I could imagine that if chess software and computer hardware will steadly be >improved one of the next chess computer generation (X2) will beat the >predecessor (X1) finally also with a 80% score pushing the Elo level up to 3280. > >X3 will do the same with X2 (Elo level then 3520), X4 will do the same with X3 >(Elo level then 3760) and so on and so on. X5 reachs Elo 4000 and after some >hundred years X30 will gain Elo 10000! > >If it was ever possible that computers played absolutely perfect chess the >MAXIMUM Elo number would be determined. > >The gnawing question is: HOW HIGH IS THE MAXIMUM ELO NUMBER? > >Or in other words: How far are humans and computers away from playing perfect >chess? > >Regards, >Christoph Fieberg There's a fatal error all along this thread. Every one seem to be assuming that a constant playing strength leads to a constant ELO rating. 1.- ELO funtion is not linear. 2.- New players that enter ELO list introduce a variation in the rest of players ELO's. Suppose player A enters the list with 2350. Player A learns and soon reaches 2400. If player B is a "static strength player", his ELO will go down as he plays against player A, whose strength does not corresponds to his ELO. Only in a planet with "static strenght players" would make sense trying to guess what ELO corresponds to a perfect player. Even in SSDF list we can see rating variations of programs, and they don't improve their strengths. So, perfect player's ELO depends on the rest of the players ELO, on the history point we choose. José C.
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.