an ordinary chessplayer wrote on 03/15/22 at 04:32:22:
FMCharless wrote on 03/12/22 at 02:51:12:
the best move for black is 4...cxd4!
this gives a clear plan for black which is to play for the pawn on d4.
It also gives white a good square on c3 for the knight. But in the database white only scores 44% after 4...cxd4, so I will grant that it's not completely ridiculous, but it seems horribly mis-timed. I don't see why 4...cxd4 would improve on 4...Nc6 5.Nf3 Nge7!? when black can still play 6...cxd4 and 7...Nf5, but white can't play Nb1-c3.
]
Its actually much more reasonable than that

Basically the white knight isn't
actually that good on c3, so black can seemingly afford to give the option away early without damage.
The main tabia in in a way is that 4.. Nc6 5 Nf3 Nge7 6 Bd3 Nf5 7 BxN exf5 stuff. Well established theory of course, quite interesting. Absurd to call it a refutation of white's opening
Jones in Coffeehouse tries to avoid that black set up with various ideas vs black's moves orders. There - vs 5.. Nge7 but I think its the overlap here - he recommends that white go for Na3 -> c2 - while black has no Bxa3.
That's pretty testing, computers can seemingly draw the mad complications in one of his lines, who is doing better in practice I've no idea.
The a3 based ideas that you're referencing are of course OK, but both SF14 and Leela seem entirely content with black after 9.. Be7 10 h5 then Qb6 or h6.
Various Be3 based positions, again nothing bad there.
What black gains, I don't know. There's definitely an active benefit as a move order against the c3 Sicilian. Mostly I suppose you might make white think for a while, trying to find the way to take advantage?
... 5 .. Nge7 6 dxc5 is legal but not very threatening either.