Idea behind e-pawn sacrifice in Open Sicilian: Najdorf, Freak Attack (6. Rg1)?

Question

There are 10 games in my database with this variation played on a high level and the White score is +6-2=2, almost always White is the elo-wise weaker player. The highest rated White player to employ this line is GM Tigran L. Petrosjan and he won his two games on a 2400+ level, against higher rated opponents.

Since it is played at such high level, repeatedly and with success, I suspect there is an idea to this sacrifice, but reviewing games and consulting the engine did not give full clarity. In fact the engine gives lines different to what is played, mostly going for Bf4 and trading queens.

The sacrifice in question is 8. g5!? (instead of say 8. Bg2), which compels Black to take on e4.

Only one Black player supported his Bishop with 10...d5, the rest retreated with 10...Bb7. My questions are: Why does White not only give, but provoke Black to take the pawn and why do most players not play ...d5 (after ...d5, f3 Bg6 and the position would seem very solid for Black)? Is it because they do not know the position and it looks as if the Bishop could be harassed and restricted after closing the diagonal?

[White "Petrosian, Tigran L (2568)"]
[Black "Smirnov, Pavel (2645)"]
[FEN ""]


1. e4 c5 2. Nf3 d6 3. d4 cxd4 4. Nxd4 Nf6 5. Nc3 a6 { B90 Sicilian Defense: Najdorf Variation } 6. Rg1 b5 7. g4 Bb7 8. g5 Nxe4 9. Nxe4 Bxe4 10. Qg4 Bb7 11. Bg2 Qc7 12. Be3 g6 13. O-O-O Bg7 14. f4 O-O 15. h4 Nd7 16. Bxb7 Qxb7 17. f5 Ne5 18. Qh3 Qe4 19. h5 Nc4 20. Rge1 Rfc8 21. hxg6 hxg6 22. fxg6 fxg6 23. Kb1 Qd5 24. Bc1 Rf8 25. Ne6 Qf5 26. Qxf5 Rxf5 27. Nxg7 Kxg7 28. Rxe7+ Kf8 29. Rh7 Kg8 30. Rh6 Kg7 31. b3 Ne5 32. Rxd6 Rf1 33. Re6 Ng4 34. Rexg6+ Kf7 35. Rxa6 Rd8 36. Rhd6 Re8 37. Rd4 Rg1 38. Kb2 Ree1 39. Rd7+ Ke8 40. Rh7 1-0

Here's a game where Black played ...d5:

[White "Nadj Hedjesi, B. (2311)"]
[Black "Milosevic, A. (2269)"]
[FEN ""]

1. e4 c5 2. Ne2 Nf6 3. Nbc3 d6 4. d4 cxd4 5. Nxd4 a6 { B90 Sicilian Defense: Najdorf Variation } 6. Rg1 b5 7. g4 Bb7 8. g5 Nxe4 9. Nxe4 Bxe4 10. Qg4 d5 11. f3 Bg6 12. f4 e6 13. Be3 Qa5+ 14. Bd2 Qb6 15. f5 Bc5 16. fxg6 Bxd4 17. gxf7+ Kxf7 18. Rg2 Nd7 19. g6+ hxg6 20. Qxg6+ Ke7 21. O-O-O Rae8 22. Rg3 Kd8 23. Ra3 Kc8 24. Ba5 Qa7 25. Bc3 Bc5 26. b4 Be3+ 27. Kb1 Qb6 28. Bxg7 Rhg8 29. Qf7 Rd8 30. Bh3 Rge8 31. Rxd5 exd5 32. Bxd7+ Rxd7 33. Qxe8+ Kc7 34. Rc3+  1-0

Answer 1

I think that a lot of the answers to your various questions about "why this or that" come down to the fact that it is just very unexplored theory, and the black players are just using their best judgment, probably mostly on-the-fly since Rg1 is so rare. I have just 18 games in my database after 9...Bxe4, and 7 of those games were played by white players below master level. That is not a lot of practical practice.

As far as white having a lesser Elo-rating, I assume that I have most of the same games as you (I use Mega 2020), and for the most part, the rating differences are not that great, so I do not consider it a big deal. 5 of the 11, white out-rated black, and the other 6, black out-rated white. Only one had a huge disparity with a 2500 white player vs. a 2100 black player.

Obviously, the main idea is to get a lead in development where black has barely even started developing the kingside. The latter part of that sentence is the key. In addition, it seems to me that sometimes, it can also be hard to get that development going. For example, I think that instead of trading the Bf1, that a quick Bd2 and 0-0-0 may be even more dangerous, highlighting that inability to develop.

As to why the bishop goes back to b7, I think that it is much more natural. The bishop going to g6, while solid, could still come under attack by h4-h5. It is too early to tell, and most strong players would probably be weary of that. With the bishop on b7, it controls a lot of squares, and it almost forces white to trade the light-squared bishops since on d3, the white light-squared bishop will surely be blunted by the natural g6. Trading off a potential attacker in this position is, clearly, desirable.

There is a lot of room for analysis in this opening.

 [FEN ""]

 1. e4 c5 2. Nf3 d6 3. d4 cxd4 4. Nxd4 Nf6 5. Nc3 a6 6. Rg1 b5 7. g4 Bb7 8. g5 Nxe4 9. Nxe4 Bxe4 10. Qg4 Bb7 (10... d5 11. Bf4 {This developing move, stopping e5, is probably best.} (11. f3 Bg6 {And while solid, the kingside is still far from being developed, and the Bg6 may still become a target.})) 11. Bg2 (11. Bd2 g6 12. Bc3 e5 {If this is the computer's best suggestion, then I really like this as white.} 13. O-O-O {May even be more dangerous.} exd4 $4 14. Bxd4 Rg8 15. Bg2 Nc6 16. Bf6 Be7 17. Rge1 $18) 11... Qc7 {Trying to keep control of the h1-a8 diagonal.} 12. Be3 g6 {This is where I would rather have this on c3, really giving the black king nowhere to go.} 13. O-O-O Bg7 {And black has a solid pawn, but the position is very open, and tactical in nature.}