a 2200 is a very good player, but could hardly have hope of beating a 2700. take the world team cup, 2500 players get crushed by 2700's no sweat at all. There is some though that could win, but i really doubt it

Here Kasparov outplays an almost FIDE Master in a simul:

Annotations mine though I didn't look at this game for nearly as long as I should have. 

The time pressure blunder was obviously 31...a4

I've been surprised that even "regular" GMs seem to be awed by the ones at the top. I remember Joel Benjamin talking this way, perhaps about Kasparov (in Benamin's book "American Grandmaster"), and some other examples by GMs I can't recall at the moment. One would think a GM would be in awe of no one. OTOH there have been those proposals to create a Super GM title or similar.

Regarding the strength of top players, I find it remarkable that they are able to play credible games against computers.

Also, I believe someone in this thread mentioned that above a certain level the K factor decreases, so that a 500 point gap at the high levels is more significant than at lower. Many posters seem to overlook this -- or is it not true for FIDE ratings?

If Carlsen breaks the 2900 rating, there will be 200 point gap to a weak 2700 player

2200 vs 2700 is the same as

1200 vs 1700

In terms of winning odds... maybe.

But in terms of work, study, talent, understanding, and experience... there is a much bigger difference between 2200-2700, than 1200-1700.

I would think natural ability and a passion for the game would take a person from 1200 to 1700 with very little book reading and memorization required.  @2200 you probably know all the fundamentals, theory behind the openings as well as several main and side line varients.  Basicaly strong in from opening to endgame with a style of your own developed.  I was surprised to find out Dan Heisman lives close to me and with all his accomplishment and opportunity had reached 2200.  I wonder what it is that seperates the 2200 rated player from the 2499 rated player.  

Fewer holes in their understanding, better at preparing against specific opponents (a big deal at master level and above), and more chunks committed to memory (a GM has 200,000 on average whereas a FIDE master has 25,000 or so) so less of a need to analyse variations as the source of their "intuition" is better developed.  Also better at time management. 

The 3200 player doesn't think the 2700 player has a "profound knowledge" of chess. Really, asking a question with an empty word like "profound" in it is kind of pointless.

I can say ,in my point of view that somehow 1800 and lower ,sometimes the ratings are over exaggerated, you expect a 1800 to crush a 1300 or lower without a sweat, well, I started chess in 2013, and in 2015 my chessa rating (Chess south africa) had just been published and it was at 1020, I got paired against a 1800+ Fide rated player, obviously everyone knew that my opponent had just got a walk over,I can't lie,I was pretty scared my self, the rate of play was 60/60 and we played out through the game,and we were that last board to finish that round, until somewhere when I was left with less that 10 minutes and he was somewhere just below 20, in a position where a CM Rated 1999 said my Drawing Chances were sealed, I even had some slight advantage because of a passed pawn (which even up to today I regret not pushing coz I thought he would some how gang on it with his R+B and my R+B wouldn't save it) ,at the end of the game, my opponent said that he was only playing for time and waiting upon me to blunder (which I surely did when I got scared and played an exchanged Sac, my Rook for his bishop), he added that if I had offered a draw he wouldve accepted as the spectators could see that the game is drawn...I'm still rating 1116 but I've beat players rating up to 1600, well I've seen an 1800 secured draws against 2300+ and also a win , I can say this can be as a result that in South Africa, we have few Fide rated tournaments , and if there are there,its hard for players to access them since they are too far from where they are,so a player gets more knowledge but has no rating to prove it, thus you find a highr rated player drawing or even losing to lower rated player, but as for a 2200 vs 2700? Those 2700+ players aren't "ordinary" chess players, those guys invest more time than a 2200, I mean the k factor or a 2700 is probably 5 times lower than a 2200, their rating is hard to maintain,let alone to achieve, we have about 35 2700+ active players currently (excluding 2800) in something of which I could take a guess and say in a pool of 1000 active GMs, I mean even making it to 2500+ is not child's play,but some cases can be of a rising star who's rating is just rising none stop, like how by the time Carlsen was an IM he could draw against Kasparov in a position where he had a slight advantage, that Gap between a 2200 n 2700 means a whole lot a things and I think a 2200 chance of a draw is somehow dependent on if the GM makes a blunder which if he's in form, he won't make such blunders,and it would have to be a very terrible one which a 2200 can easily exploit,the blunder are hardly those that My Systems Teaches you about or you find in Think/Play like a GM, these guys go about 30moves in prep of a variations,with strong Seconds and strong engines ,infact preparing what to an engine can appear as a blunder until an hour later of the engine running which will say its a great move

Elo gaps have countless different meanings, but they always have one thing in common: They predict the probable outcome of a chess game. Of course a world class GM knows scary much about chess. But on the other hand, chess has a high drawing margin, and a 2200 player usually knows how to hold drawn positions like rook endings 4 against 3 and others.

A 2700 would push any 2200 around the board—dismantling the 2200's pawn structure, pushing the 2200's pieces back onto defensive squares, squeezing little advantages from the position here and there, until the game is completely won.

Most of the time something like that would happen, like with any 500 point gap. And sometimes it wouldn't, like with any 500 point gap.

I realize this comment is 3 years old, but seeing as this thread seems to have taken off after it's been dug up again, I'll respond anyway. What I want to say in response to Ubik's comment is: that's not necessarily the case. It's highly unlikely that a 2200 player would have obtained his rating primarily by playing against 2700s. If his rating has been stable for a while, then it's most likely that in his recent games he's been playing people around his rating, and scoring roughly 50%. Of course, because of how the rating system works, it's possible to estimate what his score against 2700 players would be in theory; however, if he hasn't played all that many games against 2700s, it's impossible to know how accurate that estimate really is.

A grandmaster acquaintance put the difference between, say, 2200 and 2500 (not talking here about the really super-GMs) to me down largely to consistency rather than readily discernable differences in raw chess ability, knowledge or understanding. Off-the-board concentration, focus and stamina seem to make quite a difference in determining rating varations between IMs and GMs in that general range, he was implying. Not playing myself at those dizzy heights I can't support or deny that claim from any kind of personal experience, and only put it out there secondhand for your consideration.

On Rumo's old post from 3 years ago discussing the Meulders-Kasparov game: the reason I was allowed to play in that Swift tournament was not that I was at the time 2nd to Winants in the elo ranking, but that organiser Bessel Kok had promised a place to the winner of the Belgian championship, which was I in 1985. Winants won in 1986. My best ever rating was 2385, which is nothing in comparison with a super GM. But it would be naive to think that someone who takes his job so seriously as Kasparov would not prepare even against a mere good amateur. Besides, as Larsen once said, a top player plays 45 out of 50 games like a grandmaster, a good amateur plays 5 out of 50 games like a grandmaster (or words to that effect). Only an arrogant fool would NOT prepare and that, clearly, Kasparov was not.

Hello Richard, nice to meet you here and thanks for correcting me. I did not know this Larsen quote, but I could not agree more. 

I personally know someone who drew a ~2700 at the World open (I think ti was world open) and I believe their current rating is around 2200-2300