[V-2]

There's also rating inflation, though, which makes cross-era comparisons flawed.

And besides, many players have their nemesis. Sheer numbers don't reflect that.

Eg. Anand used to be one of the strongest players in the world, but overshadowed, some say intimidated, by Kasparov before the latter retired.

Given Anand's rating, he performed quite poorly against Kasparov.

[CSMastermind]

Are you aware of any attempt to compare cross generationally based on game trees? Find game positions that both players have faced and see who made the better move?

[V-2]

Yes,

http://en.chessbase.com/post/computers-choose-who-was-the-st...

I remember there was a heated debate as for whether or not Crafty was a strong enough engine for the task.

I don't have all the links at hand, but it shouldn't be difficult to find various parts of this discussion.

http://en.chessbase.com/post/using-che-engines-to-estimate-h...

Authors of the research argued that contrary to what common sense implies, even if it Crafty is weaker than some evaluated players could have been, results are still legit.

Of course this approach doesn't give full justice to players - while some might have strived for winning by perfect play, others could deliberately play inferior moves just to create complications and provoke their opponents into making mistakes worse than their own, by getting them into time trouble, making them uneasy etc. (eg. Tal was famous for that).

A game - oversimplifying a bit - is won by the player who made less mistakes than the opponent, not by the one who made less mistakes on an absolute scale. So both strategies are valid as far as sport is concerned.