We will be following the game Lecic(2149)-Nikolic(2145), played in 2002.
1 e4 c5 2 Nf3 d6 3 d4 cd 4 Nxd4 Nf6 5 Nc3 g6 6 Be3 Bg7 7 f3 0-0 8 Qd2 Nc6 9 g4
This variation is played rather often by higher-rated players in the Pacific Northwest. The two most popular lines are 9 0-0-0 and 9 Bc4, played about equally often overall. 9 g4 is the third-most popular, played about a fifth as often as each of the two main lines. All three lines yield about a 60-40 advantage for white. However, I consider 9 g4 to be inferior to the main lines. It "threatens" 10 g5 N moves 11 Nd5, but other than that all it does is hand over the initiative to black.
9...Be6 is played slightly more often than this move, but I don't like allowing white to double my e-pawns after the capture on e6. Besides, the line usually transposes if white does not take on e6, and black then captures on d4.
10 Bxd4 Be6 11 0-0-0 Qa5 12 a3
The fact that white has to stop and play this move demonstrates his defensive posture.
12...Rab8 has some independent significance, as black can then get in the ...b2-b4-b5 push a move earlier than in the main line.
William Schill, a Master from Washington state, played 13 Kb1 against me, but his move is almost a blunder. 13 h4 is far better.
13...Rab8 14 h5
14 Nd5 leads to massive material trades and leaves white with no advantage. An example is Schreufer(2205)-Fleck(2400), played in 1984, which continued 14. Nd5 Qxd2+ 15. Rxd2 Nxd5 16. Bxg7 Kxg7 17. exd5 Bd7 18. Rd4 Rc7 19. Bd3 b5 20. c3 Be8 21. g5 f5 22. h5 Bf7 23. h6+ Kf8 24. Bc2 Rc5 25. Bb3 Ke8 26. Kd2 Kd8 27. Ke3 a5 28. Rhd1 Kc7 29. Kd2 1/2-1/2
b5 15 hg fg
15...hg is often played as well, but taking with the f-pawn works out better for black.
16 g5 Nh5 17 Bxg7 Kxg7
17...Nxg7 is more often played, but I prefer taking with the King.
18 Bh3 Bxh3 19 Rxh3 b4
Black proceeds thematically. 19...Rc4 was played in Sieber(2145)-Schneider(2248), but it turned out badly for black: 19...Rc4 20. Qd5 Rbc8 21. f4 Qb6 22. Qe6 Kf8 23. Rxh5 Qe3+ 24. Kb1 gxh5 25. Qh6+ Ke8 26. Qxh5+ Kd8 27. Qxh7 Qxf4 28. g6 Qg4 29. Rf1 R4c5 30. Nd5 Qg5 31. Rf8+ Kd7 32. Nf6+ Ke6 33. Qf7+ Ke5 34. Qxe7+ Kd4 35. Qxd6+ Ke3 36. Qd3+ Kf4 37. Nd5+ Ke5 38. Rf5+ Qxf5 39. exf5 Rxd5 40. Qf3 Rd6 41. Qg3+ Kd5 42. g7 Rdd8 43. f6 Ke6 44. Qg6 1-0
20 Nd5 Qa4 21 f4
21 Nxb4?? Rxb4! and white has lost his knight, since both 22 ab Qa1# and 22 Qxb4 Qxc2# are out of the question, as is 21 Nxe7 ba 22 b3 Rxb3 23 Nxc8 Rb1+ 24 Kxb1 Qb5+ 25 Qb4 Qxb4+ 26 Kc1 a2 27 Nxa7 a1(Q)#. After 21 f4, White's rook on h3 guards the vulnerable a3-pawn; however, computer analysis now puts black ahead in the game.
21...Rc4 22 Qe2 Rbc8 23 Rd2 b3 24 c3 Rxe4 25 Re3 Rcc4 26 Rd4 Rcxd4
If 26...Rexd4, ostensibly winning a rook, white draws with 27 Rxe7+ Kf8 28 Rf7+! Kxf7 29 Qe7+ and white has a perpetual check with the queen.
27 cd Rxe3 28 Qxe3 Qc4+ 29 Nc3 Qf1+ 30 Kd2 Qxf4 31 Ne2 Qf7(?)
I would be inclined to trade queens and go into the pawn-up endgame. Black has a weak pawn on b3, but white has one himself on g5. If black doesn't want to take the time to win the pawn on g5, he can simply play h6 and create a dangerous passed pawn on the g-file.
32 Nc1 Nf4 33 Nxb3 Nd5 34 Qg3 Qf1 35 Nc1 Nb6 36 b3 Qf5 37 Ne2 Qa5+ 38 b4 Qd5 39 Kc3 Qc4+ 40 Kd2 Qa2+ 0-1
White must have run out of time, because the position is not yet resignable. Regardless of that, an interesting game indeed.