Hi.

In the endgame line mentioned:
1.e4 c5 2.Nf3 d6 3.d4 cxd4 4.Nxd4 Nf6 5.Nc3 a6 6.Bg5 e6 7.f4 Be7 8.Qf3 h6 9.Bh4 g5 10.fxg5 hxg5 11.Bxg5 Nbd7 12.O-O-O Ne5 13.Qf4 Nd5 14.Bxe7 Nxf4 15.Bxd8 Kxd8 16.g3
I would also not go so far as to say good compensation. More like white may have problems making progress.

Leela also gives the line:
1.e4 c5 2.Nf3 d6 3.d4 cxd4 4.Nxd4 Nf6 5.Nc3 a6 6.Bg5 e6 7.f4 Be7 8.Qf3 h6 9.Bh4 g5 10.fxg5 hxg5 11.Bxg5 Nbd7 12.O-O-O Ne5 13.Qf4 b5 14.Nf3!? (alternatives exist) 14...Ng6 15.Qg3 Nh5 16.Bxe7 Nxg3 17.Bxd8 Kxd8 18.Rg1 Nxf1 19.Rgxf1 Ke7
With a different endgame.

Other than that I was just going to say that when running the moves Leela likes 10...hxg5 more compared to 10...Nfd7 so you may be on to something.

Have a nice day.

The Goteborg 
1. e4 c5 2. Nf3 d6 3. d4 cxd4 4. Nxd4 Nf6 5. Nc3 a6 6. Bg5 e6 7. f4 Be7 8. Qf3 h6 9. Bh4 g5

is often said to be refuted, I believe (nowadays) mainly due to 10. fxg5 Sfd7 11. Qh5.

However, there is also the interesting idea 10. ...hxg5!?, which has only been played a couple of times (interestingly there are some old games by Pilnik).
After 11. Bxg5 Nbd7 things seem to be rather unclear.
12. 0-0-0 Ne5 13. Qe2 (13. Qf4 Nd5!? 14. Bxe7 Nxe4 15. Bxd8 Kxd8 leads to a rather typical endgame for this line. Black seems to have decent compensation?!) seems critical.
Now Black wins the pawn back by 13. ...Nxe4 14. Nxe4 Bxg5+ 15. Nxg5 Qxg5+ 16. Kb1 Bd7. White might be slightly better objectively, but with black's central pawns, things could quickly change, especially if the queens come off. 

Another try for White is some quiet move like 12. Be2, and after 12. ...Ne5 13. Qf4 Nh5 14. Bxe7 Nxf4 15. Bxd8 Kxd8 16. 0-0 Nxe2+ 17. Ncxe2 b5 the same type of endgame as above appears on the board. I think, black has decent compensation here.

How do you assess these positions and what do you think about this line in general? I would be curious to hear your opinions on this