Actually I can't remember why I made this move either, I suppose Nb5 was the plan.

With the center locked up, black will be able to scare up a K-side attack. White's pieces on the Q-side don't seem to be sufficiently well coordinated to counter-attack. As white, I'd cut my losses and pull the Q back to c2 or d1, where it can swing over to the K-side.

Maybe the idea was b4?

The purpose of this move seems to be to keep an eye on b6, so to blockade black's queen. But it is kind of passive, while black has concrete possibilities on the kingside very close. Better for white was b4, trying to distract black's attention on the kingside. This is one of this crucial moments in which a sharp intuition shows the way. In fact, with just two unclear moves, white is suddenly in problems. The kind of things you last learn in chess (including myself!  

How about 16...Qe8, getting the Q involved in the K-side action? White seems to be stuck for a plan (Qb3->Qa4->Qb3), and after 16...Qe8 black's K-side initiative looks scary.

If I recall correctly, I tended to think of ...Qe8 as somewhere on the agenda, but not yet, being a little concerned with the protection of b6.

I remember to have seen this very position in a championship's or book's game. somewhere. I would be able to check it, since I remember a very similar discussion, and I recall the move was Ng6, with Nf4 in mind. But Bc8 came into the agenda anyway.

If Ng6 would have been played, now Nf4 leads to 18. BxN, gxf, (or even exf!?) and no matter if white tries to attack the h pawn (19. Nh2). Black may well let white take it and keep on building his attack.

Yep b2-b4 was the plan.

That plan is too slow: black's K-side attack is already well underway. I'd play 18.Nd2 Nf4 19.Bf3, and see how black plans to make progress.

White should have choosen another plan for the only reason he spent two moves without doing this advance (being possible before). This is a third move, that implies a fourth - and b4 is only a little worry for black. As white, would I have felt an urge on the kingside that makes impossible to wait two more moves before initiating an attack? Who knows,,,

To me all I saw was my Bishop who was doing very little and your well posted knight who was doing a great deal.

This gives white another pawn with which to pry open the white K's position. In addition to 19.Bd1, I like 19.Qc2 or 19.Qd1. Then 19...Nxe2 20.Qxd2 and white's position seems to be holding up pretty well.

@hityerclock: why would Black give up his strong knight for White's e2 bishop after 19.Qc2 of 19.Qd1?

Next black's move would be g4, and white must defend...

I disagree with the GK engine (as I often do), The h-pawn is meaningless. Opening up lines and getting the light-square bishop into the action is more important.

Actually, I don't agree with the engine either, figuring that the pawn was worth the gain in time.

After this move I started getting paranoid that it was a big mistake. perhaps because of the Bd3 threat (Though not a real threat) I dont know, maybe I just dont like trapping my other rook the way I did.

Here as white I'd take a good, long look at 13.a4!?, a positional exchange sacrifice aimed at consolidating white's control of the light squares. If black takes the R (giving up his LSB), I don't think he has winning chances any more.

It seemed cunning and different at the time. I thought it was a strong spot for the knight and if black wished to remove it he would probably have to give up the bishop.

I'd suggest 25.Nf3, getting the other N back into the game.