Author: Gian-Carlo Pascutto
Date: 15:03:49 06/15/01
I witnessed a little flamewar below, and there's several comments I'd like to make so as to increase understanding between the members of this forum. Or maybe it'll get you all to hate each other and me more, whatever... I've snipped out the relevant parts: >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> >>>The ELO rating is only one measure, but not the only one. For example, >>>there's no ELO for Lasker, Capablanca, Rubinstein. How then do you know how >>>strong they were? >>You are incorrect..... you can calculate Elo for all past players, Because we >>know the games they won, lost and drew....If you would like to see them go to >>www.chessmetrics.com >No.. that shows you don't understand statistical sampling. You are >saying Elo is "absolute". I can prove it is _not_. What the hell are you talking about... I never said that. When Did I ever say ratings were """"""""ABSOLUTE""""""""""". I said you can calculate ratings for past players....and Ratings is the only measure we use in chess. I NEVER said RATINGS are ABSOLUTE.... Ratings are calculated for that pool of players, If you think the rankings are incorrect you better take a look at the past list because you will find what history tell us were the strongest players of the day are also RANKED #1 on the repective lists. The Rating numbers themselve don't mean anything 2700,2800,2900 etc. What matters is the point spread from one player to the next. Don't put words in my mouth and then tell me I don't understand >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> Mark, you don't understand. The discussion was about establishing an objective (absolute) bound as the what strength a GM should be. Jorge pointed out that ELO could not be a good criterium because of several reasons, on of which being that there are no known ELO ratings for old players. To that, you replied that you could calculate ELO ratings for those older players. Robert corrected you that ELO is not an absolute measure. Now, you claim you have never said that ELO was an absolute measure. You did not say it in words, but it is directly implied from the context in which you posted the list. The problem is not that ELO is relative measure, I see you understand that. The problem is that because of ELO being relative to other players it is also a relative measure in time. Hence, it cannot be used to compare players from different timeframes in any objective way, which was that the discussion was all about. You may not have used the word 'absolute', but by posting that ELO for older players can be calculated in direct response to someone making a point about ELO being useless to compare players from different timeframes in order to get an absolute measure, you are DIRECTLY inferring that ELO can be used to compare players from different timeframes in an absolute way. Your failure to grasp this is clearly illustrated by the following sentence: >What matters is the point spread from one player to the next. So if Aljechin was 100 ELO points better than Euwe, that means he is equal to Kasparov which is 100 points better than Shirov. You cannot _only_ just compare the point differences and come up with something meaningfull. You have to have a base to compare to. And that base is not portable between timeframes. Putting that base at a fixed ELO number is not going to work because ELO is a relative measure (gee, thats what you've been saying, right?). Comparing a player to the peers of his time (what chessmetrics does) is also not going to work because a weaker player will get a higher rating if the overall playing strength drops (for example in timeframes where there is relatively little interest in chess), and a strong player will get a low rating if his peers are also very strong. ELO cannot be used to set an absolute mark to be a grandmaster if you expect them to be a minimal strength. That was Jorge's point, the win-loss-draw performance of a player is of no use if you are going to compare him to players in other timeframes. Jorge pointed out the lack of ELO for older players to illustrate this. By saying you can calculate the ELO for them, you imply that has any meaning in the discussion. It had not. -- GCP
This page took 0.01 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.