It might be worth noting that in Raymond Keene, Aron Nimzowitsch: A Reappraisal, he offers an interview with Bent Larsen who believed that Nimzowitsch did not really have a system, despite offering considerable value to chess players.
This is something I learned from a deep dive into Tigran Petrosian's games - Nimzowitsch did have a system, but he didn't follow his own rules very rigorously. Petrosian follows the script of My System with far greater dedication to the rules than Nimzowitsch himself did. When Nimzowitsch lost games, it was almost always due to a move or set of moves which ran counter to one or more rules he gives in his book. Petrosian's moves often look wild to the classically-trains chess player, but always perfectly natural in the context of My System. This helped me solidify the concepts Nimzowitsch was writing about.
Of the twenty legal first moves with White, Explorer tells me there are only two that I have never played: Na3 and Nh3.
However, I often play Nh6 several moves in playing the French against the advance variation.
Yes. And Nh6 happens in the Gurgenidze variation of the Caro-Kann.
That's why after e4 c6 d4 d5 ... white is considered to be better defending his pawn with Nd2 than Nc3. And does.
Because after 3) Nc3 black would have g6 ... with a g7 bishop being aided and enhanced by that knight blocking c2 pawn configuration.
Of course that doesn't mean Nc3 there 'loses'.
But here's the Irony.
Say white does play that Nc3 move (including hoping and wanting black to play g6) - so black goes ahead and plays dxe Anyway !! Not g6.
'No! We're doing main line Caro! Lets see if you want to waste 'knight-time' going after my c-bishop.'