More flexible than 1...d5: less committal to begin with, and keeps options open.

Rules out the King's Indian...

But Black could have played 3...Bb4 (Nimzo-Indian) or 3...c5 (Modern Benoni) each leading to very different sort of games.

That line has certain similarities to the Alapin Gambit in the Exchange Ruy Lopez.

What is the usual result of accepting the IQP? Black would obtain a fairly free sort of position with attacking potential, having control of the e4 and c4 squares. But the isolated central pawn on the open file is in itself a permanent strategic weakness.

Isolated QP's tend to be double-edged. The side without has a target, but the side with usually gets open lines and good piece play to make up for it. IQP's were considered strictly as a weakness until Botvinnik came along in the 1920's and 1930's and demonstrated the dynamic nature of IQP positions, for example check out the Panov-Botvinnik variation in the Caro-Kann. And that's today's lesson in Chess History   The relevance here is that IMO, 9...Qxe7 and accepting an IQP gives black better prospects for counterplay than the move actually played.

Bxh7+ isn't really a threat yet, as the black cavalry can still ride in to save the day at the last moment. Either 10...cxd4 or 10...h6 are more usual here.

I presume, then, that Black's actual choice is predicated upon the absence of a dark-square Bishop in White's army. But the g6-pawn now presents a possible target for a charging h-pawn, and the dark squares remain vulnerable to attack. I shouldn't be surprised if White castles long, here, despite his loose looking position on the Q-side.

I think I would have played 12.h4. This looks as though it's going to be one of those classic castling on opposite wings games, a race to see whose attack gets in first. Opening lines against your own King makes an odd impression...

... as does inducing the enemy rook to occupy said open file.

I'd play 15.Nc4 here too, but there's also something to be said for 15.Nxd7, eliminating a potential defender of the black K, then following up 16.h4 etc., trying to conjure up a K-side attack. That wouldn't be a bad plan here.

Having figured you weren't that interested in the K-side pawn storm, 15.Nc4 was also my 'guess'.

I don't like 16...e5, which creates a gaping hole on d5. It's easier for white to exploit the hole on d5 than one on e5, as black has a pretty good grip on the central dark squares. This pawn needs to stay on e6 for the time being. Better is 16...Ne5, challenging the N on c4, or 16...Nf6, looking to take up residence on d5.

I agree with hityerclock. So good is the Knight's spot on c4, I thought it would be in Black's interest to challenge it there. There seemed to be no reason not to do that at once.

It was tempting to take with the rook in order to double on the d-file. What dissuaded me was the possibility of ....Bb7. It might still be a playable line, though...

The bishop is wonderful there, and otherwise e4 block it!

I think 20.f4 works, too. Then 20....Bb7 fxe5 simply wins for white, and if for example, 20...exf4 21.exf4 Qxf4+ 22.Kb1 followed by 23.Rhe1 black's position collapses (just as it did after 20.Rd2 etc.) as the rook on the open e-file is creating too many threats for black to handle.

I was wondering about 20.f4, myself. In fact that was my 'guess' (my first 'wrong' one since I stopped guessing h2-h4!), although I found the 21...Qxf4ch line a bit disconcerting. Even so, White seemed to have a lot of open lines to play with.