Templare2 wrote on 09/26/09 at 06:10:28:
If Nb5 is so strong, why not 15.., a6 preventing Nb5?
I am not sure whether you mean 14...a6 or 15...a6, so I will address both options.
a) 14...a6 does not seem at all convincing to me after 15.a4! (with the idea of playing a4-a5 before cd6 to control the key b6-square) 15...Rf7 (15...h5 16.a5 g4 17.cd6 cd6 18.Na4 and in Clavijo-Nunez, 2000, Black's attack never got off the ground) 16.a5 Bf8 17.cd6 cd6 18.Na4 g4 19.Nb6 Rb8 and in Schorr-Quelle, 1981, White should have played the simple 20.Nc8 Rc8 21.Qb3 with a clear advantage. Black's attack is not achieving anything.
b) 15...a6 is an important option, though, since it essentially forces White to sacrifice a piece. However, White achieves tremendous compensation: 16.cd6 (16.Na3 g4 17.Be1 g3! was already promising for Black in Pachman-Padevsky, 1957) 16...ab5 17.dc7 Qe8!? (17...Qd7 is more common but inferior after 18.Qb3 Ne8 19.Nc5 Qd6 20.Ne6! Qb6 21.Kh1 Be6 22.de6 and White had a tremendous advantage in Biriukov-Solovjov, soon agreeing a draw in a winning position) 18.Qb3 g4 (played in Novik-Ezat, blitz 2003) 19.Nc5 Nh4 and now instead of 20.Kh1, 20.d6 Kh8 21.fg4 Qg6 22.Qh3 would have been clearly better and probably winning for White.
In conclusion, White achieves clearly better chances in both the 14...a6 and 15...a6 variations.
In that last line 20. fg (as in Fissore-Petrillo, corr. 1984 and Berkes-Pavlovic, played in the Serbian team championship three weeks ago) is reputed to be clearly better for White. Panczyk and Ilczuk think 18...Qd7 is better than 18...Qe8, and that Black should deviate from Biriukov-Soloviov with 20...Bxe6. They give a couple of lines (one of them attributed to Soloviov) which are said to be unclear.