Again Sevenviolets, there's no need to make a comment like "This isn´t a good evidence of quality of your analysis. I´m sorry. " as far as I can see... we all make mistakes in analysing, that's why chess is still alive. The day we can all calculate everything accurately, the game is dead and we have to move on.
I've finally had a chance to look at the position on a board (which is rare for me!) - Someone beat me to suggesting 17...Nd7 (although I think black's serious improvements should come a lot earlier, but instead of 20...Be7 in this line, in view of white's threatened Qa5-Qa4, I thought "Well, how about 20...f5?"
21.Nd6+ Bxd6 22.Rxd6 Qe7 23.Qe2! Qg5+ [23...Rhe8 24.Rxc6+! forces a draw] 24.Kh1 Qh4 and again, every line seems to lead to a forced draw for one side or the other - I could be missing something of course, as I don't profess to be a GM nor a computer, one of which would be needed to have any chance to find your way through the tactical maze here, but certainly 20...f5 seems to hold in this line.
21.Nc5!, however, looks like a big improvement here. My original thought was that black would be okay after 21...Bxc5 22.Bxc5 Kc7, but white here has 23.Qh4!! Qxg6+ 24.Kh2 Qf6 25.Bd6+! Kc8 26.Qb4! winning. So perhaps black has to try and hold the unbalanced endgame after 22...Nxc5 23.Rxd8+ Qxd8 24.Rxd8+ Kxd8 25.Qh4. Perhaps black can hold this, and with the two central passed/connected pawns, if he can get these rolling then I'm sure he'll be doing okay at the least. However, this should surely be better for white with correct play?
I will continue to say what I've been saying - the only critical test of 8.g4 is to instantly grab the loot on d4. I've not looked at the 8...h6 lines but I can't believe such a move for black.
The line I'm currently investigating appears to be Leisebein's Main Line after 8...Qxd4, 9.Be3 Qd6 10.g5 Nfd7 11.Qf2 Ne5 12.Bg2 Nbd7 13.O-O (given as strongest by Junior 9, which incidentally up to this point considers
white, not black, better) h6! (Class's move against Leisebein last year) 14.Rad1 Qe6 15.Ne4. Class continued with 15...f6!? 16.g6 Nxg6 17.Nc5 Nxc5 18.Bxc5 Ne5 19.Bxa7 g6. The game finished as a draw, and can be found on the thread given by Scholar above. I'm not entirely sure what to make of it though, it's not clear to me that black didn't have improvements and that he shouldn't have carried on in the final position, with two rooks against a queen. But it's all very complicated.
Still, this is my hope at proving an advantage against 8.g4. Certainly 8...e6 looks a little dodgy and 10...Nb4?! doesn't quite cut it, if the above endgame is all black can play for he certainly doesn't have an edge.
Regards,
Craig