I, too, would like to know how black can survive the endgame. 14...Qxg5 15. hxg5 Nd7 16. Bxg7 Kxg7 17. Rh4! looks very strong for White (who is planning f3-f4-f5) and even if white isn't winning here, he certainly has all the winning chances in my opinion.

Unlucky, as far as 14...b5!?, there are so many deviations from the main line (most of which are bad) so I will give what moves have been determined as "best" to a point. 15. Qxb5 Qc7 16. Qa4 Rab8 17. Kb1 is pretty much best for both sides. Here, Ward gives 17...Qb7!? as offering reasonable chances for black and his analysis can be found in Winning with the Sicilian Dragon 2. Also of interest might be 17...Bd7!? 18. Qxa7 (18. Qa3 e5 is fine for black) 18...Rb7 19. Qa3 e5 with adequate compensation. Certainly not a monograph, but enough to let you know where things stand.

Also, some new material. In the aforementioned main line: 16. Bd3 Bc4 17. Rh3 b5 18. Bxc4 (18...b4 cannot be allowed) bxc4 19. Ka2 Rb6 20. g4 Ra6 was Szuk-Schneider, A. 1996 which looks fine for black. In addition, 17. g4 b5 18. Bxc4 bxc4 19. Ka2 Rb7 20. Rh3 Rcb8 "with mutual chances" is given by A. Schneider on his "Dragon for Experts" CD.

I find the lines more difficult to play against is when black tries to skip qc7 and when he does play qa5 in one go. Unless there is some reason that is looked down upon...

I completely agree  with Dragon expert. The only other line in this variaition I found that Black was doing poorly (statistically) in was the 17. g4 b5 line which Tiviakov gives as "main line" inh is 1995 monograph. But even from here, 18. Bxc4 bxc4 19. Ka2 Rc6 (with the idea of Ra6) looks fine for Black. Seriously, guys, this is only critical because of how precisely both sides must play. Black is facing more trouble in other lines the 12. h4 13. h5 14. h6 endgame still looks depressing for Black while 2. h4 13. g4 14. a3 Rab8 15. h5 b5 16. h6 still looks like trouble for Black unless there is something new to be had in the 16...b4 lines (I am currently investigating 17. Nd5! Bxd5), then 16...Bh8 17. Nd5 Qxd2 18. Rxd2 definitely looks at least somewhat better for white.

In the variation you cited first, 18.Nxd5 Bxd5 19.exd5 instead of 18.exd5 is the "official" way to a white advantage, if I remember correctly. I have not examined the second line you proposed on the board but have the impression that the inclusion of a6 and Rh4 might favour white even more...

I have to admit that there are other lines except the 14.hxg6 line which I don't like so much either, but I think none of them comes as close to being a refutation as that line.

After 15.a3 the main line is Rab8 16.Bd3. Now black has a choice whether to play 16...b5, 16...Bc4 or even 16...a6.

a) 16...b5
This is one of the lines which at least made it into WWTD2, if even only as a small comment. After 17.Qg5! many often cited lines are known to be bad for black. The less known idea 17...Qd8, as given in WWTD2 seems to fail after 18.e5! according my analysis with similar lines as 17...Qc7 18.e5. If black wants to switch back to the 16...a6 line with 17...a6 and the idea of 18.Rh2?! Rc5! with a lot of counterplay, he has to reckon with 18.f4!

b) 16...a6
This move has the idea of playing 17...b5 without letting white play Qg5 immediately. 17.Rh4 looks like a better move than the once played 17.Rh2 because after the sequence 17...b5 18.Qg5 Rc5 19.Bxc5 dxc5 20.Qxc5 Nd7 21.Qb4 Qc7 the rook is not attacked and white wins a crucial tempo.

c) 16...Bc4 I once came to the conclusion that the lines starting with 17.Rh3 lead to a white advantage but the NIC yearbook I mentioned before shows that things are not so easy. In another source 17.Rh2 is given as clearly better for white but I have not looked at it extensively yet.  In a recent game 17.Be3 was played, and the conclusion was that after 17...Nd7 18.Bxc4 Rxc4 19.Nd5 Qxd2 20.Rxd2 Re8 21.Bxa7 b6 things were not so easy for black as thought before.

Apart from that black has the option 15...Bc4 which is often criticized because of 16.Rh3 but perhaps it is not so easy too refute after all. At least in my database I have not analyzed it to a clear advantage but I think there should be one.

Hoping for a lot of comments...

Not sure what you mean. This line looks fine for Black to me (will have some moves by this weekend). IMHO, the most critical lines are the 12. h4 13. h5 14. h6 and 12. h4 13. g4 14. Qg5 and 14. a3 (after h4 and g4) Rab8 15. h5 b5 16. h6 lines. This one looks like it should be a minor concern.

In my opinion this line is just better for white. I played this line from both sides for many years and have come to the conclusion that black is just worse, whatever he plays. I could provide some moves if there is interest, but I remember that one of the last NIC yearbooks (67 or 68) came to exactly the same conclusions as I did. So you might find enough in there to get convinced of this assessment.
By the way, don't get me wrong. I would love to hear of an improvement for black which makes this line playable
Patrick