Author: Uri Blass
Date: 07:35:38 11/08/01
Go up one level in this thread
On November 08, 2001 at 10:09:35, Jonas Cohonas wrote: >On November 08, 2001 at 10:04:32, Christopher R. Dorr wrote: > >>1. Of *course* a GM would play anti-comp. Only his 'anti-comp' stuff might be >>different from *your* anti-comp stuff. You think that Kasparov wouldn't play >>'anti-Anand' stuff that is tailored to Anand? Of course he would. Just as he >>would play 'Anti-Fritz' stuff when playing Fritz. >> >>2. Sure there is a point. if I can make the computer play like a 2000, then the >>computer is *not* a GM. Do you think there is *any* way I could make Kramnik >>look like a 2000? If the point is to evaluate the computer *as an opponent*, >>then the weaknesses of that opponent are fair game. >> >>3. Maybe, but then the result is meaningless. Just as meaningless as asking 'Is >>anyone here capable of playing a King's Gambit and sacrificing a full piece to >>mate toe computer's black king on h8?' and then trying to extrapolate that >>information to something else. >> >>The simple issue is 'Can anybody here beat the 'best' program on a 1 GHz box at >>40/2?' Any other limitations artifically weaken the human, and make the test >>meaningless. >> >>Chris >> >You missed the point: play the comp like you would anyone else! The point of chris is that kasparov does not play against anand like anyone else. I think that we can get information about the ability of computers against humans when the humans do not know who is the opponent by a tournament when the players only know that one of the players is a computer and they have no idea who is their opponent even in cases that they play against humans. It may be interesting to do a tournament between Fritz and the best 10 players in the world under these conditions. Unfortunately I do not know about a sponsor who is interested in doing it. It may be interesting to know if knowing the opponent really help humans It is possible that there are cases when knowing the opponent is counter productive because humans are afraid from the opponent. It is possible that part of the humans can play better against computers when they do not know that they play against a computer. I believe that in most of the cases knowing the opponent is an advantage for humans but it is still interesting to know how much rating points they earn from it. Uri
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.