LINE TWO
After failing to prove any advantage in the above line I switched lines, playing 18. g4 (Of course 15. g4 is somewhat in vogue, see Radjabov-Carlsen, Baku 2008). This line hasn't been as popular of late and I think I now know why. I have some new analysis, though, to support its non-existence since Negi-Papin, Istanbul 2005 (Note: In the Negi game, this move was played on move 17, without the in-between Bg5 and Rc5, though the position is of course the same).
Anyway, the line played out 18... hxg4 19. h5 Rh8 20. hg6 fg6 21. f4 and here my friend played Dearing's recommendation 21... e5!?
Position after 21... e5!?
Indeed, this line requires white to play very carefully, lest black get a winning endgame! The first time he played it I tested it head-on. However, this came to disaster. After. 22. fe5 de5 23. Ne6+ Bxe6 24. Qxe8 Rxe8 25. Rxe8 Nxe4 26. Nxe4 Rxe4 black is clearly better. Those pawns become very strong and the rooks are rather uncoordinated.
The only testing line I have found is 22. Nde2! There are still so many intriguing variations here. I'm wondering if there is any analysis on this position, or if I can get some help from those of you with more time and a stronger engine!