For understanding I went through relevant games in Kevin Goh Wei Ming's Chess Developments: Sicilian Najdorf 6. Bg5. For lines I used ICCF archive, since especially in Najdorf every book is a bit outdated the moment it comes out .

In archive (which I last updated sometime this year, so some fresh games are probably missing) both moves are played by similiar frequency. If I limit myself to games since 2016:
14. ... Qb7: 0 white wins, 4 draws, 1 black win. 
14. ... Rab8: 2 white wins (one thanks to me :/), 2 draws, 1 black win.

I think that Qb7 may indeed be more accurate and I will probably use it in future. Especialy since it seems that there is no clear mainline yet - white players are still trying many different moves at this early point. 

However I think that Rab8 should also be sufficient for equality.

Today chat reminded me of one of my first ICCF games, which I had finished last year (game moves will be in bold, while variations will be in regular font face). The game began with:

1. e4 c5 2. Nf3 d6 3. d4 cxd4 4. Nxd4 Nf6 5. Nc3 a6 6. Bg5 Nbd7 7. Bc4 Qb6 8. Bb3 e6 9. Qd2 Be7 10. O-O-O Nc5 11. f3 Qc7 12. Kb1 b5 13. g4 Bd7 14. a3 Rb8 15. Rhe1 h6 16. Bh4 Nxb3 17. cxb3

This position was already reached by two other ICCF games, which both continued with 17. … O-O and ended with a draw. Both me and Stockfish (after some convincing) found 17. … b4!? N quite interesting. This move is a pawn sacrifice, due to 18. axb4 Rxb4?? 19.Bxf6 Bxf6 20.Nd5 exd5 21. exd5+ Kf8 22. Qxb4 +--. However white ruined queenside pawn structure looked like compensation enough. The game continued with: 

18. axb4 O-O 19. g5 Nh7 20. Bg3 Bxg5 21. Qg2 Rb6 22. Nc2 Bf6 23. Qd2 Be7 24. Rg1 Kh8 25. Bf4 Qd8 26. Na4 Rc6 27. e5 d5 28. Rg2

At this point Stockfish realized that white has quite some advantage. Beforehand it thought that 28. … Rc7 is sufficient, but here it realized that 29. Rdg1 Bg5 30. Nc5 gives white probably decisive advantage. So I went for 28. … Rg8 in hopes of stabilizing the g-file situation.

I think that position after 24. Rg1 is already quite dire.  24. … g6 results in 25. h4 h5 (practically forced, since white attack is decisive otherwise) 26. e5 d5 with quite some advantage for white.

After 29. Rdg1 h5 30. Be3 g6 31. f4 I prematurely resigned. Few evenings of analysis have convinced me that white has decisive advantage, but I should have had played a few more moves to make my opponent prove that. Sadly I don’t have analysis from back then saved and it will take me a few days to reconstruct it, since I am currently at seaside without access to strong engine.

I am pretty sure moves before 17. … b4 are not to blame for my defeat, so the only question is whether white attack after it is already decisive or was the novelty successful and I lost due to a later mistake?

I will continue with analysis during next few days, but in the meanwhile what do you all think?