Doubt it. Try Qf6 instead of Nxf3 in that g6 line say, and I imagine other things. By the way, g6 really is an absurdly computerish move! So is Qb3 I guess.

What this line definitely is is rather unpractical - white can force black into several endings with nil chances for black, and slight niggles for white.

The main line after dxe5 here for instance.

Not really what you want when you're playing a moderately risky line.

Or play 11...Qc7, maybe? But I agree, I like the move in the game. It is very thematic.

I have recently played a game as Black that went: 



In the game I immediately snapped off the Knight with 14...Rxf3, and I had a comfortable game after 15. Bxe4?

Checking the Mega Database 2016 after the game 14...Rxf3 is the overwhelming choice of Black players who reach this position, with exactly 75% of people playing it in 288 games. It certainly looks like a very natural move, and I remember playing it without too much thought. Yet it's not even in the top 10 of Stockfish 6 with a depth of 25 ply.

It seems like the computers like 14...Ng5 or 14...Bd7 here, but the MD2016 mentions Ng5 was last played in 2008 and Bd7 has been played only 8 times.

I'm a mere 1800 USCF player, so I'm not sure what's happening here. Is this a case of computers not understanding the positional elements of a position, or is there something else going on? After something like 14...Rxf3 15.gxf3 Ng5 16.Kh1 e5 17.dxe5 Nxe5 everybody has responded with protecting the pawn with 18.Ng1 but the engine points out an interesting alternative with 18.Bc2!?, letting Black take on f3 and instead focusing on Black's weak (and pinned) d5 pawn.

As I'm improving in strength I find that I'm playing this variation of French Tarrasch more and more, and it's always fun to find these interesting and weird bits. I'd like to know what you guys think about it.