You know, it's strange how the ELO system has created the impression that a struggle between human beings is a matter of weighing two quantities.

Will deeds of heroism soon receive numeric expression? Really, Oswald Spengler must be rolling in his grave with laughter over the whole idea of ratings.  You could quantify the relative strength of composers (surely composers today must be better--humans progress, after all!).

I think humans enjoy watching a fight, as Lasker points out, and of course in a fight you want to know who will win. And it is a fundamental fact of nature, as Wittgenstein might say, that wolves eat lambs; and so we speak strength or ferocity considered apart from a particular fight. But this is very far from a numeric idea.  Numeric ratings are well-ordered in a mathematical sense, but "who will win" is not; A beats B beats C beats A, as is well known.

Furthermore, it is unclear what is being imagined when we say Bacrot (#37 today) plays better than Karpov did, or to get right down to it, that Carlsen is stronger than Fischer. We want to know who would win, or who is the more ferocious animal, but the way the world is, a great white shark would eat a grizzly in the water but a grizzly would eat a shark on land.

Which of the following scenarios tests whether Carlsen's rating is inflated relative to Fischer's?
(1) Carlsen and Fischer are given databases and eight months to prepare, but sequestered. They play an unlimited match to 6 wins. Fischer goes up 5-0, but Carlsen exhausts Fischer with a long series of draws, and the older Fischer's physical condition becomes the main determining factor in the match. With no Campomanes in the wings, a physically collapsing Fischer loses.  (2)Fischer and Carlsen play a match, but they get to exert psychological pressure on the organizers and each other before the match. Fischer effectively shatters Carlsen's nerves before the match begins. Fischer wins. (3) Carlsen's play develops to where it is seen as more beautiful even than Capablanca's. He wins everything in sight. But when the two play a match after a year's preparation, Carlsen cannot tolerate the almost incomprehensible psychological pressure of a fight with Fischer. He loses (Alekhine-Capablanca). (4) 4-year-old Fischer and Carlsen are transported and taught chess in the same technological era (which one? Perhaps their styles are better suited to different eras). They play a match at age 28. Fischer works much harder and wins.

Really, the notion of "the better player" only makes sense where there is a gross disparity in results. For the rest--competition is something we do, not something we are, and descriptive language is far superior to numeric ratings in capturing what transpires when we do it.

--So what is the ratings race? The opacity of the rules, the mathematical priests presiding over the proceedings, and the fact that it includes very many of a given players' games give it the aura of a divine stock price. It is simply ONE more, different, competition, not the distilled essence of a player's capacities.  Human beings are not stocks.

I am not complaining about the existence of ratings; in and of themselves they are harmless. But I believe they are grossly overrated.

Uruk wrote on 03/24/12 at 14:12:52:


The article didn't really get into specifics about the technicalities of his work, which I'm sure were immense.  I'd think that a large part of it has to do with variance, i.e., how much the computer's evaluation of the position changed throughout the game as opposed to using the computer's evaluation as an absolute measurement.  But I'd be curious to know a little more about the methodology used.

Uhohspaghettio wrote on 03/22/12 at 04:24:08:



The goal isn't to find out who played most like an engine. It's to find out who plays better. 

To ignore the improvements in the opening is to ignore ~1/3 of the game. The authors made the right choice to include the openings, not just because it's easier but because improvements in the opening are important!

Willempie wrote on 03/22/12 at 10:16:11:


I see, we are on the same page. Sometimes a problem when discussing things at a forum.

fling wrote on 03/22/12 at 09:42:23:

I am not saying that at all. Just that I feel that progress is overrated and is often due to circumstances that are not directly related to the actual performance. It isnt a coincedence that in athletics the records in technical disciplines have stood that long. There the technique of the athlete itself is a bigger factor and the rest will just get you mm's. 
Same goes for chess, preparation has skyrocketed, but endgames less so.

Excluding openings would be awfully hard though - when do they end? Very late on quite often. 

I mean theoretically even people our sort of strength can catch someone in detailed preparation and play a near perfect game. Very rare of course, but it'd have been basically impossible 40 years ago!

Willempie wrote on 03/22/12 at 09:05:26:

You're probably talking about air humidity, as I don't believe that today's water is any more or less humid than 40 years ago!? 

But seriously: The introduction of huge data bases and computers (and the possibility to connect them) can easily be considered as a paradigmatic change in chess training methods.