Author: Jonas Cohonas
Date: 18:52:41 03/22/03
Go up one level in this thread
On March 22, 2003 at 13:07:02, Thomas Lagershausen wrote: >[D]rn1q1rk1/1p2bppp/p1bp1n2/4p3/4P3/1NN1B3/PPP1BPPP/R2Q1R1K w - - DeepJunior had >played the weak 11.Bf3? in game 6 against Kasparov and black was more than >fine.According to GM Yasser Seirawan computers are player with a elo 2400 but >make no easy mistakes.That´s all. The fault in that argument, which makes it quite ridiculous when you think about it, is that a 2400 human GM who never made any mistakes, would be rated much higher (we call them super GM's, they hardly ever make any positional mistakes, but when compared to comps they are 2400 in tactics under normal tournament TC and even worse in blitz :) ELO is not a static way of measureing playing strenght, it measures your performance based on your opponents rating, which he got from the same system, therefore there is no way to take out the 2400 strenght measure and then add perfect play in tactics without trying to guesstimate what the rating of the computer would then be! If we should use static ELO measures loosely just for the heck of it, it would look something like this: Kasparov: Opening strenght 2900 Middlegame strenght 2750 Calculation strenght 3000 (finding a plan that is sound in short and long terms) Tactics 2650 (high for a human) etc. =2825 in total This is of course more an illustration of my point than a scientiffic meassurement, the same should apply to computers when talking about their strenght and weaknesses, the # 2400 in itself has no meaning unless you have the win/loss/draw ratio of games against enough opponents with different ratings. In other words we can say without a shadow of a doubt that Deep Junior's performance rating against Kasparov was 2837, i am sure that if Yasser drew a 6 game match against Kasparov he would agree :) Regards Jonas
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.