There are half dozen high profile games with 11... bxa4, one early example being Karjakin - Bu Xiangzhi 2005 won nicely by Black .
Looking more closely at the database stats, we can see that 10... b5 is still being played by strong players, and almost all White players go for 11. 0-0-0 or 11. g5, not 11. a4. It was also the case between 1993 and 2005.
Conclusion : contrary to Nunn's claim, many people did not take 11. a4 as a refutation of 10... b5 seriously, at least from the practical point of view.
Sicilian Najdorf (B90), English Attack: 6.Be3 e5 7.Nb3 Be6 8.f3 Nbd7 9.g4 Be7 10.Qd2 b5 11.a4 bxa4!?
Sort:
@Nerwal
@Nerwal: Yes, this is what I was saying. What I don't understand is how such a strong GM as Nunn didn't think of this line (even though this often drops the a pawn, White's King is in a very bad state), and why b4 is still played by many...
I was going over the Nunn - Sadler game of the Lloyd's Bank op 17th, London, 1993, in Nunn's book "Grandmaster Chess move by move". In it, he talks about 11.a4 being a refutation to 10. ...b5, and that several other GMs stated the same. Curiously enough, the most intuitive reply to a4, in my mind, is simply to take it, force the Rook to recapture (or misplace the Knight), to both destroy the King's pawn protection, and keep it in the center. Just intuitively to me, the resulting position looks very comfortable for Black, and logically, the imbalances on the Kingside essentially allow Black to make significant compromises to neutralize White's attack, and he is perfectly fine... I suppose it is counter-intuitive to just abandon the Kingside attack as Black, however at some point an attack must be converted into a positional imbalance (unless, of course, you mate or promote), and what Black stands to gain seems so blatantly beneficial for him, that I really don't understand why 11. ...bxa4 is not mentioned by Nunn, nor has anyone in the Big Database 2019 played it (bar one master).
I understand these variations are relatively out of date now, but still, it is very perplexing to me how such a continuation was not even considered. To me, it makes sense, and running it through Komodo 14 BMI2 x64 MCTS to depth 31, it finds bxa4 to be the overwhelmingly better reply, leading to complete equality (I think the final evaluation was +0.08, compared to +0.50). It considered 11.a4 to be very plausible, but also prefers 11.g5 slightly when beyond depth 30.
Without an engine refutation, it is really difficult to understand why this hasn't been considered. I suppose the contempt setting is not high enough on my engine, and simply, against humans, taking a more attacking line and creating complications is likely stronger than simply trying to find the most objective strategy.
Or is it simply that players that play b4 aren't aware of just how little this promises Black, and players who would play bxa4, and those who prepare against such players, simply wouldn't go for these lines?
Can someone shed some light on this? I want to learn the Najdorf, but it's hard to decipher some of these essential lines when I can't quite analyze the depth of these ideas...