What is it...

. . ., do you suppose, that differentiates a 2800 player from a 2600, or even a 2600 from a 2400.  What is it that a 2800 player has learned that a 2600 player has yet to discover?

Or is it all about something less tangible, and not about knowledge or understanding but about perseverence, killer-instinct, will-power, the ability to sustain concentration, physical fitness. . . or simply natural talent?

Maybe confidence, lack of self-doubt, an ability to handle the nerves?

Sense of humour.

But wouldn't a 2600 player also need supreme confidence?? Or can confidence be quantified and just .05% more be enough to give you an edge?

I don't think there is some magic factor, stronger masters just know a little bit more of everything

You are talking about 2800 elo, and I do know what 2000 is all about.

What can that "little bit" be?  It would seem that a 2600 (elo) player must already know and understand everything pertaining to chess.  I know the difference between myself and say, a 2000 player is very much knowledge and understanding, I can't see that there is some piece of nformation that a 2800 has that is denied a 2600 player.

In an article from a master I read about this. He mentioned the importance of the psychological aspect, which increases on higher levels.
I think in any sport on top level you need to master every aspect, including your state of mind.

As an example he mentioned Tal. That even though he played rather inaccurate (but also creative/aggressive), is opponents feared him and played uneasy because of this. I suppose this edge is enough to make a great player stand out.

Well, there's no doubt that competitors have been spooked by various players such as Tal, Fischer, Kasparov, even Morphy and Menchik and played far below their skill level in those games.  Can we deduce that such an effect is so univeral and consistant that it allows certain players continual higher ratings?  One of the 2800 players today in Kramnik (2811).  I can't picture him as psychologically intimidating - other than his record - but what advantage does he have over, say Peter Heine Nielsen (2658)?

I suspect this may be difficult to answer - just as it is hard for us lesser mortals to grasp the difference between, say, an everyday math PhD and an Euler, Gauss or Newton. 

I think so too, but, like many things, it's interesting to contemplate on and speculate upon.

In a broad sense I believe there is mental aspect there as well, as in really wanting to win, or keeping your cool when facing an intimidating opponent. Also in the game itself, not losing your clear mind in a cramped or uncomfortable position.

When I read an article about the Timman - Kasparov finals for the world championship, the difference in personality was strongly emphasised throughout the story, and its importance over the board. The way Kasparov enters the room or shakes hands, in a subtle way it translates itself to the game. It probably sounds a bit far fetched, but I think these differences matter a lot on top level.

i think it is RAW natural talent

I don't think it's far-fetched at all, though probably of exaggerated importance in match play. 

there is no aweser!

That's very interesting ( to me, anyway) to speculate upon. I wonder, how a game would play out if two players were set up to play a match where there was a 200 - 250 point difference in their ratings, but neither player was told who the other was. Neither would also be told anything about the rating of the other player. Behind the scenes, one is a 2200, the other a 2400. One player makes their move, exits the room, then the other is allowed to come in, be comfortable at the board, think for a similar amount of time as the other player. Make it like an experiment, you know. I would enjoy hearing of such experiments, because the first thing I do when I get matched up in a team match game, is look at my opponent's rating, then play. 

What differentiate them are the openings of Na3 or Nh3, or f6 or c6. Seriously, the level below calculates fewer moves than the level above.

Sometimes the answer is simple but not simpler than Columbus egg.

Especially endings such as B vs B or N vs N, or R vs R, or R vs N or Q vs R+N, etc.

They are just better, why differs from player to player I suppose.

Maybe they can analyze a little further, maybe the positional understanding is a little better: 

At these levels the difference doesn't have to be that much, since they play a lot tighter than the lower levels even the smallest advantages can be converted into a win.

That's seems like a fascination experiment. How much does perception, pre-conception and self-expectation play into competiton... I like the way you think.

I tend to agree with FancyKnight: being little bit better at everything is what ultimately makes a big difference. 2600+ players and even much weaker obviously know all the standard stuff you are likely to find in your average book but there's also lot of silent knowledge hidden in players minds that may not be found in books and may actually be quite hard to put in written form in any case. Intuition plays very big part in chess.

Of course this also varies from person to person. I could see it being entirely possible that 2600 player could play certain positions in 2800 level while others only 2400 level.