So Fritz much prefers 10. a3 (your move). The reason Fritz prefers this move is the coming ... b6 and/or Nb4 from black.

 

My other idea, 15. Bh6 also doesn't work. So I'm 0-2 today.

Loomis: ahh, nice setup move. Sneaky.

I think 15. Bh6 before sacing on g6. That way your line looks like: 15. Bh6 Rc8 16. Nxg6 fxg6 17. Bxg6 hxg6 17. Qxg6+ Kh8 18. Qg7#. Obviously black can't take both pieces on g6. So you have to work out what happens when black declines the second sac.

Bf4 is a great idea! I certainly wished I had played it later in the game.

 The sac idea on move 15 is a good idea that I'll have to have a look at. Perhaps after the Knight and B sac, getting my Q and other B into the act could get some fireworks. I'll think about it some more, but a superficial look suggests it might not work, at best ending with a perpetual check (e.g., 15. Nxg6 fxg6 16. Bxg6 hxg6 17. Qxg6+ Kh8. And then it's not clear how the Q and B can get something together with a succesful attack.

 

 

Is 10. Bf4 a possibility? You comment that 10. a3 was to support b4 in part to develop the bishop. My instinct is that the bishop is more useful on the other side of the pawns. There it keeps the queen off c7 and puts a claim on e5.

 

After 14. ... g6 did you consider an attack on the king? For example, Bh6 and sacing twice on g6? I haven't calculated it, but it's the kind of thing that works and is a potential game ender in an otherwise long fight.