MNb wrote on 07/16/09 at 01:01:48:
Isn't 12.Qh5+ Ng6 13.Nf3 Qd6! 14.Be3 Bd7 15.Rd1 Qf8 16.Qg5 Qe7 17.Nd5 playable? The queens are exchanged, but the other white pieces are quite active.
Apparently 17...Qxg5 18.Nxg5 0-0 (18...Rc8 19.Bc5 Bb5 20.Ne6) 19.Nxc7 Rad8 is critical, and in the further course of events White is hampered by the unfortunate situation of his king: 20.Bxa7 Bc6 21.Bd4 Rd6 22.Nce6 Re8 23.f3 h6 24.Nxg7 Rxd4 25.Rxd4 Kxg7 26.Nh3 Re2 27.Rc4 Bd5 28.Rc7+ Kf6 29.a3 (29.Kf1 Rd2 30.Nf2 Bc6) 29...Bc6 30.b4 Re1+ 31.Kf2 Rc1 32.Ke3 Ra1 33.Kd4 (the "forced activity" of White's king, in contrast to his foregoing restriction on the first rank, is funny) 33...Ra2. Maybe 33.Kd2 offers better chances for a draw: 33...Rxa3 34.Rh7 Ra2 35.Rxh6 Kg7 36.Rh5 etc.
I'll look at the exchange sacrifice again, when I have more time. In the 12.Bf4 line I also have the impression that the "worst case" scenario with active knights, but only a pawn for the exchange, isn't too bad for White - in 8 games out of ten he should get a draw. The problem with using chess software is that it lets too many things appear very easy. Fighting with rooks against these active knights can be an unpleasant experience in practice - so many forks... And in a game many players will prefer the "safe" 13...Kf8, instead of the correct 13...Qe7.
Edit: An even bigger problem with chess software is that sometimes I forget to look at the screen. After your 17.Nd5 Qxg5, the continuation 18.Nxc7+! is stronger and was probably your intention. 18...Kd8 19.Ne6+ Ke7 20.Nexg5 Rhc8 and here we have several interesting possibilities, like 21.Nxh7 Rxc2 22.Nhg5 Rxb2 23.h4 Bc6 24.Bc5+ Ke8 25.Ne6 Kf7 26.Nfg5+ Kg8 27.h5
27...Nh8 28.Bd4 Rb4 29.a3 Rc4 30.Bxg7 Rg4 31.h6 Rxg2+ 32.Kf1 Rh2, e.g. 33.h7+ or 33.Rd6!, about =. Chess is a fascinating game. No doubt that after your 17.Nd5 endless analyses are possible, hard to believe that it should not be a good weapon for OTB play!