Author: blass uri
Date: 15:49:44 09/11/98
Go up one level in this thread
On September 11, 1998 at 18:07:03, Serge Desmarais wrote: >On September 11, 1998 at 12:10:29, blass uri wrote: > >> >>On September 11, 1998 at 09:03:30, Bernhard Bauer wrote: >> >>>On September 11, 1998 at 03:02:18, blass uri wrote: >>> >> >>>>Sometimes computers play stupid moves that no human in the level of at least >>>>2000 elo could play. >>>>I cannot say it about grandmasters >>>> >>>>Uri >>> >>>In the last Rubinstein memorial the following position occured after blacks >>>move 36...Rb8. >>>FEN:1r/4kp/4pn1p/1P3p1P/5P/1P2KPN/8/3R w >>>White played Ke4?? and black answered Nd5?? >>you mean 37.Kd4 >>>These were stupid moves, but were it computer moves? >> >>No sometimes humans do mistake that no computer could do >>I think that both sides were in time trouble >>but this is rare even in time trouble. >>In more than 99 out of 100 cases grandmasters do not do these mistakes. >> >>I define a program with a human style as a program that guess grandmaster's >>moves better than other programs >>In this position it is a bad idea to guess the mistakes in the game because the >>probability of doing these mistkes is very low. >> >>Uri >> >[Snip Karpov's game] > > I do not follow you well! A chess program that would guess a high percentage >of Petrosian's moves would be qualified as having a more human style than >others, according to your reasonning. But that same program gussing most of >Petrosian's moves would never guess most of Tal's moves... Now, would that mean >that Tal was LESS HUMAN THAN PETROSIAN??? You seem to view the "GM moves" as a >whole, a block. But 2 different GMs could play very differently from a given >position... But they are still "human players"... I thought about the average but you are right because a program can guess one player's move and if the style of the player is different from others it will not be good in the everage. Uri > > >Serge Desmarais
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.