This Keres Variation is nowadays the "official" DDG bust, recommended by Watson and Schiller among others.
9.Bb5+ Bd7 10.O-O O-O 11.Bg5 Bxb5
Not good. Better alternatives are h6, a6 or Re8.
12.Nxb5 Qb6 13.Bxf6 Qxb5
Black thought it was best to resign without facing any of these:
A) 14...gxf6 15.Nxh7
B) 14...h6 15.Bxg7! hxg5 (15...Kxg7 16.Qg4!) 16.Bf6 Nd7 17.Qh5 Nxf6 18.Qxg5+ +-.
C) 14...g6 15.Nxh7 Kxg7 (15...Re8 16.Qd2) 16.Rf3 Qd7 17.Qd3 +-.
D) 14...Qc4 15.Qh5 Qd3 16.Bxg7 Qg6 (16...Kxg7 17.Rxf7+ Rxf7 18.Qxf7+ Kh6 19.Rf1) 17.Qxg6
E) 14...Qb4 15.Qh5 h6 16.Bxg7 +-.
F) 14...Nd7 15.Nxh7 Kxh7 16.Bxg7 +-.
The way I found 14.Ng5 was a funny incident. Having analyzed the position after the move 13...Qxb5 for maybe half an hour, I had decided to reply 14.Nh4.
When I was in bed, the position suddenly came to my mind, and so did the aggressive move 14.Ng5. Without a chessboard, I was able to picture the main variation 14...gxf6 15.Qh5 fxg5 16.Qxg5+ Kh8 17.Qf6+ Kg8 18.Rf5, which looked like a win for White.
In the following day, I checked the variation on the chessboard. First I found that 15...Qd3 seem to spoil it, but then came another bomb: 15.Nxh7! instead of 15.Qh5 keeps the attack going on.
Later on it turned out that my first option was probably as strong as 14.Ng5. I had thought that 14.Nh4 gxf6 15.Nf5 Bxh2+ 16.Kxh2 Kh8 is better for Black, but two of my chess friends convinced me that 17.Qh5 decides the game. Due to deep combination trees, Fritz first believed that Black has better game, but with a little help, it finally agreed that White is winning, e.g., 17...Qxb2 18.Rae1 Nd7 19.Re2! Qb6 20.Rf3 Rg8 21.Rh3 Nf8 22.Qxf7 Qd8 23.Re8 +-.