I agree with many here that black is completely winning at the end of the game. I guess the easiest would have been 38... Rd8 instead of 38... Re8, to prevent 39. Nd7+ and to threaten 39... d4 when the rook invades. Obviously he can't play 39. Ke3 to prevent 39... d4, and 39. Nf3 can simple be answered with 39... Re8 when your rook will soon eat some yummy white pawns.

But also after 38... Re8 you should be winning. In the worst case you have to play Nd7+ Kf7 Ne5+ Kg8 with still a big advantage. But Kf6-e7-d6 as suggested looks fine as well.

Twilight, I think you have to look at the actual variations. What does Fritz say about these lines:

40. ... Kd6 41. Nf7+ Kc6 (41. ... Kc5 42. b4+ forces Kc6) 42. Kg5 Re2 43. f4

a) 43. ... Rxb2 44. Kg6 Rg2+ (44. ... Rxa2 45. Kxg7) 45. Ng5. These have some promise for white with the passed f-pawn.

b) 43. ... Kd7 44. Kxh5 Ke7 45. Kg6 Kf8 46. f6. In this line white has passed pawns on the f and h files.

Loomis, you cannot proclaim that Fritz 11 is saying the position is -1.6 because of a material advantage. I left my computer on and now Fritz is saying that the position is -2.1

Myself, I'm a 2110 CFC member, and I've done many studies on endgames. Let's evaluate the following:

-Black is up the exchange but down a pawn.

-Black has a active rook, pressuring on the e5 knight, and is threatening to control the 2nd rank.

-Black's pawn structure is superior that white's pawn structure because of the doubled pawns and the 2nd rank pawns that will be easily attacked.

-White's knight is the only piece blocking the open file, and is continuosly being attacked by the rook.

-If the king moves, it's hanging the knight, which would obviously be losing.

-If the knight moves, your rook is going to slaughter the 2nd rank.

From these 6 points, I think that white's counterplay is little, and that progress for black should be made easily.

Twilight, I think you might be a little ambitoius. After 40. Ne5 Kd6 41. Nf7+ Kc5 42. Kg5 you can't just proclaim this position won for black without considering that white is about to have two passed pawns on the f and h-files. Things seems pleasant for black until you start calculating and realize how many moves it will take black to get a passed pawn rolling compared to whites. Fritz's -1.6 advantage is probably all material as the white pawns getting passed and moving up the board are beyond the horizon. But they're probably the most important feature on the board.

And reply to Hicetnuc, 45. Ne2 Kd6,

I guess you mean 40.Ne5 Kd6

It looks like a good idea, and worth a try

Reply to Benjiboy, Bd1! is a novelty invented by some computer IDK. (White is forced to take the bishop and hang the g2 pawn.

And reply to Hicetnuc, 45. Ne2 Kd6,

What can white do here? He will either check on f7, or retreat his knight. If the knight goes to a square such as g6, Re2 immediately wins a pawn, and will easily generate a winning passed pawn. If he checks on f7 which I think is a mistake, the black king will march to c5/b5, and it will be white's turn to move, when his knight is in a bad position.

8.Qa4 Bd1 was a novelty discovered by Fritz (one of the 1st computer's TN) and was published in the Informator. It just refuted the 8.Qa4 variation, and the top players stopped using it.

After 8.Qa4 Bd1 9.Qxd1 Qxg2 10.Rf1 cxd4, white hasn't much to show for his material deficit

I can't see how after 8. qa4 bd1 is "an incredible move" doesn't white just reply by qxd1 ? 

iTwilight, could you give us a sample line which gives black winning chances at the end ?

The final position is draw?

I don't think so, neither does Fritz. Fritz 11 says the position is -1.6, a huge advantage for black.

Kg5 isn't important, you need to free your rook on the file and swipe the 2nd rank.

Instead of repeatedly going to Kf6, you should play a natural Kd6, and if he plays Nf7, you can just walk your king to the queenside and take all his pawns.

Good game ! It's interesting you point difficult to see tactics in your analysis.

In the ending, I wondered about a plan to open some files for the rook, for example 35...b5 to storm the kingside (a5-b4...)

The final position is definitely drawn, because black can't prevent the repetition without allowing white to play Kg5. But black had chances with 38. ... Rd8 instead of 38. ... Re8.

this looks more similar to situations i've experienced than a lot of the ~2500 examples i've seen on here so it's interesting to see your thoughts

I was looking over this game in the first few moves and it reminded me of a game I saw in the postmortem of a recent tournament.  Turns out, its the same game :)  This is David Stamper, btw.   So you and Craig were looking over this game, and he said he thought black would probably be able to win with very careful play, but an extremely hard endgame!

Thanks for sharing!