I have half-completed my investigations into the main lines of the Two Knights Defence with 1.e4 e5 2.Nf3 Nc6 3.Bc4 Nf6 4.Ng5 d5 5.exd5 Na5.
This is Black's most reliable way of securing compensation for a pawn, although having said that, I believe that there is a strong case for Black's practical chances at club level (and possibly a fair way beyond that) in the 5...b5 variation, and possibly 4...Bc5 as well. After 4...d5 5.exd5 Nxd5, however, I would much rather be White.
My investigations (with one annotated example of each) have focused on 6.d3, 6.Bb5+ Bd7 and 6.Bb5+ c6 7.dxc6 bxc6 8.Qf3.
I had always been interested in David Bronstein's piece sacrifice line 6.d3 h6 7.Nf3 e4 8.dxe4!? Nxc4, which gives White two pawns and an imposing array of central pawns for the sacrificed piece, especially as I occasionally try out these 4.Ng5 lines from the white side as well. Unfortunately for White, it strikes me as an inferior relative of the Cochrane Gambit against the Petroff Defence (1.e4 e5 2.Nf3 Nf6 3.Nxe5 d6 4.Nxf7!?), in which White gets a similarly imposing pawn centre and also exposes the black king. In the piece sacrifice line stemming from 6.d3, Black's king generally ends up pretty safe, and this makes it much harder for White to generate anything particularly concrete in return for the piece. At rapid time controls it might work, though, as Black might defend inaccurately and allow White to crush Black with the central pawns, as Bronstein did in the "stem game".
The main continuation, 8.Qe2 Nxc4 9.dxc4 Bc5, is quite grim for White in spite of the extra pawn.
The next deviation is for Black, with 6.Bb5+ Bd7. Black is more likely to regain the gambit pawn in this variation but is less likely to get a long lead in development, and I think that White has more chances to get a slight edge than after the more popular 6...c6. The critical continuations are 7.Qe2 Bd6 8.Nc3 0-0 9.Bxd7 Qxd7 10.a3 b6 11.d3 Be7 (improving on the 11...c6?! of Morozevich-Sokolov, Sarajevo 1999) and 7...Be7 8.Nc3 0-0 9.0-0 c6 10.dxc6 Nxc6 11.Bxc6 bxc6 12.d3 Re8. In both cases Black has fair compensation for the pawn but White might have a slight edge.
Finally there is 6...c6 7.dxc6 bxc6 8.Qf3. I don't think this continuation is very promising for White, since 8...Rb8 (as played in the main illustrative game, Vallejo Pons-Inarkiev, Khanty-Mansiysk 2007), 8...h6 and 8...Be7 all appear to give Black fully adequate compensation for the pawn. After 8...Rb8 and 8...Be7, it is very risky for White to grab a second pawn on c6, so White generally prefers to retreat the bishop with 8.Bd3, which leaves the bishop quite awkwardly placed. Black can even consider the exchange sacrifice 8...cxb5?! 9.Qxa8, which I think is theoretically unsound, but even so, I would hesitate to take White in the resulting positions, and Black's score in practice has been quite good.
The next blog article on the 4.Ng5 lines of the Two Knights Defence will conclude my investigations into 4.Ng5, by focusing on the main lines with 8.Be2 (where after 8...h6 White chooses between 9.Nf3 and 9.Nh3) and 8.Bd3, which has received some attention at grandmaster level.
I will be particularly interested to see what I find in the main line with 8.Be2 h6 9.Nf3 e4 10.Ne5 Bd6 11.d4 exd3 12.Nxd3 Qc7 13.b3. Tim Harding claimed an advantage for White with this 13th move in his Kibitzer article a while back (well worth checking out for a second opinion on these lines):
13.b3 has also been favoured by GMs Alexander Morozevich and Nigel Short, and has scored quite well for White in the database. However, there is also 10...Bc5 to consider (Harding also indicated the move as a possible improvement for Black in his article), which has been scoring very well for Black recently.