Best game you will ever see!
I've beaten 2250 players on MSN before, so, yes... I believe they are indeed inflated
About the game itself, the opening requires a lot of comments:
A 1900 playing 3...h6 and 4...Qf6? hmm...
positionally, 6.Ne1 is a big mistake. You have a very large lead in development and you can use it naturally by opening the position with d4 (...Bg4 isn't very dangerous. You have many ways to maintain a big edge).
You actually justified his suspicious development with Ne1. the only real serious problem with the opening he chose is lack of development and you play an anti development move like 6.Ne1? why? whatever it is trying to achieve, it gives up a lot more. You shouldn't be afraid of anything in such a position, and if you were thnking about irritating his queen in a couple of moves by opening with f4, it takes too long with that maneuver... It's like giving a few moves for one.
Not sure what was the point of Ne1, but after it, you probably lost all your advantage, or most of it.
6...Qg6 luckily for you, your opponent cooperates. Simply Nge7 and he's fine. Be later and o-o-o... (or just a more modest Bd7)
I have a feeling that 7.Qc2 is a bad move. He did not use the extra time you've given him to develop. Blast everything open. It's complex, but it looks like it can't be bad for you: 7.d4!? and if he tries Qxe4, 8.Nf3 followed by Re1, Qb3... you have wonderful play for the pawn. If he doesn't play Qxe4, you just played a very powerful central move.
7...Bh3 Just a bad move by him. So what if the pawn is pinned? he won't be able to add pressure in the following moves, and the bishop will have to retreat anyway. The square is already defended, and once again, 8.d4 looks much better than your d3.
I really don't understand 10.Rf2, and I believe it is an error.Using a misplaced rook to free the knight looks like a bad idea to me. You can play Kh1 whenever you want to force his bishop to retreat. I think black missed his chance to take the initiative here with 10...d5 11.exd5 Nxd5. Looks to me like all of black's pieces have easy and immediate entrances to the game while the white pieces are badly underdeveloped, uncoordinated, and blocking each other's paths.
That would be a sharp contrast to how the game began.
h5 is not without a bit of poison too though, but I think at least that can be dealt with, with accurate play. Perhaps Qb3 to force something like d5 with mass exchanges in the center. (This time, because of immediate temporary threats, he would be unable to gain a lead in development...) I don't know, but it's an option.
So much for the opening, and the middlegame fight begins
The middlegame is extremely flawed in this game. By both sides.
instead of 14..Nh5, I believe things could have gotten bad for you quickly after 14...a6 15.Na3 and you'll be playing without the knight for many moves to come, without compensation (Nxd6 doesn't work here...), so black should play for the central break d5 somehow. Perhaps with Be8!?, I think black just has a big advantage there. You'll be effectively a piece down and under pressure for a few moves, which should be enough to generate a lot of play.
Nxd6 has to be an error. You should not have sufficient compensation for the piece. Not even close. White should be lost if black plays even just reasonably well from here.
16...cxd6? what?? why in the world not Bxd6? Still black should be winning, but why make things more complex unnecessarily for him?
17.Bxa6? he COULD just take. A queen alone does not constitute an attack.
17...bxa6 18.Qxa6+ Kc7, and everything is covered, while black has time to bring in more defenders.Re3 Rb8 achieves nothing, and it will take you quite some time to bring any other attacking units.
So two big mistakes in a row by black (cx instead of Bx, and fear of taking on a6). BIG mistakes, and now the position is complex. Black is no longer necessarily winning. White isn't winning either though. After Be3, it's tricky, but I believe black is STILL with the better chances after simply Re8.
I've made some really long analysis and it looks as if black can reach ok positions. Even if no longer easily winning (though he MAY be winning... I have no idea).
19... Be6?? allows 20.Qxc6 followed by mate.
You went for a very big advantage instead.
22.Qb6+? a big mistake. Why are you trying to attack with only one piece? the queen alone does not constitute an attack when there are no immediate justifying tactics. 22.Bb6+ is much better. Actually, Qb6+ Kd7 reaches a very complicated position, whereas after Bb6+ you have a much better game.
25.Nxg3 is... interesting, and is not that bad, black may be able to hold after it in my opinion, BUT black had a much better option in: 25.f5! crazy position. I like black's chances better than white's here.
There are many options for black on move 27, and I believe his choice of h2 is a bad one. The black attack loses most of its sting.
After that there are white already has a pretty nice position I think, but black makes sure he helps white with a few small mistakes....
Interesting game.

Clearly those stating that MSN ratings are inflated dont know how ratings work!
they are a way to distinguish between a pool of players -- the numbers used are completly arbituary and mean nothing outside the of that system.
Clearly those stating that MSN ratings are inflated dont know how ratings work!
they are a way to distinguish between a pool of players -- the numbers used are completly arbituary and mean nothing outside the of that system.
True, but the meaning of inflated ratings is not that something is wrong with the system, it's that the scale have grown very quickly.
There are various reasons for it to happen.
Some claim this is happening in the Fide rating system as we speak, and has been happening for quite some years.
It doesn't mean anything bad about the system itself, since the ratings only hold their original relative mathematical meaning in that same system.
However, scaling methods have been developed to create regressions in order to get a reasonably accurate rescaling between different rating systems, given enough players from both system, so stating that ratings lose ALL meaning outside their own system is incorrect. They just lose their original meaning.
That said, a couple of years ago I found it very difficult to beat 2000 players on MSN, and, while I hardly improved since then, these days it's extremely easy beating most of them. That shows the ratings have undergone massive inflation in that system, so that if not long ago you thought your rating there was 2000, it's higher now:) That's all. One possible reason is a big sudden change in the number of players.
The reason that comparison and scaling methods are relatively easy to work out, is that the differences at different strength between two given systems are constant!
Since if two players, whose ratings are x and x+40 in one ELO system, will have a result of about 55%-45% in the stronger player's favour in a long match, no matter in what system they're in.
Simply, the ratings can be ~1900 and ~1940 in one system and ~2234 and ~2274 in another system (the ~ sign is because it's not 100% accurate, it's a very close approximation).

^ The problem I was elluding to is that lots of people are going to think 1900 msn is comparable to 1900 here, they see the game and suddenly then the msn 1900's are "over-rated"** -- and that is not the case.
** I think most people in here are actually refering to players rating "being Inflated", rather than the actual problem of "rating inflation.", But that might be an incorrect assumption on my part.

I feel 16....BxNd6 or even taking with the queen is superior, hoping to get your king's bishop to c5. Then his Bxh6 sac is met with ...Nb4. As it went, he had
18.Bxb7+ and if Kxb7 then
19. Qb4+ making the bishop sac justified.
Your 19. Be6 was obviously a blunder as has been elaborated by others.
And it seems to me that 37....Rg8 might have won for you.