I've entered a thematic tournament on the French defence with the starting moves 1.e4 e6 2.d4 hosted by Lorem.
I have almost always played this defence : it's safe, solid with good counter-attacking chances. It's also rather easy to play and you don't need to know theory to start with it.
Unlike some others e4-defences, you won't face wild gambits which can be very scary if you are not prepared. I have a couple of white hairs due to the Morra gambit when I tried out the Sicilian.
No, the french defence requires from the white player to be patient and dynamic on the same time.
Patient because he can't rush quickly and has to build his attack.
Dynamic because he doesn't have time to fooling around with his pieces or playing purposeless little pawn moves otherwise it's Black who strikes first.

OK I'm happy with the French, but my real challenge in this tournament is to play with white against it.
I was ranked 2nd at the Elo so I expected to win and couldn't be contempt with draws.
I play very rarely 1.e4 and then I meet the French with the King's indian attack (1.e4 e6 2.d3 d5 3.Nd2 with g3,Bg2...) ; It is an easy way to tackle the problem, although theorically Black has different ways to deal with this systematic attack.
Here I was condemned to play the classical 2.d4

Which variation to choose ?

I exclude exotic variations such 3.Be3 or 3.Bd3, they have their points but don't promise much. I've also enough experience with the French to stick with a mainstream line, except the classical 3.Nc3 Nf6.

I thought a good way to start the search was to know which variations promise a theorical advantage.

And who makes theory ? Grandmasters.
These guys are professionals. They don't play for fun but for make their live so they take chess very seriously.So I asked them what they play on 1.e4 e6 2.d4 d5 and get the answer by searching in my large database (mainly Chessbase megadatabase with some others picked up here and there)  for players rated over 2500

3.Nc3    54%
3.Nd2    34%
3.e5       8%
3.exd     4%

So from a theorical point of view, for an opening advantage White should play Nc3 or at least Nd2.
And if I play correspondence chess with a pure "scientific" approach with books,databases & engines (when allowed) where psychology or memory have no importance, my choice would be already made.

But I don't. I want to play on OTB style using no help. And most of my oppponents seem to play the same way.
In any case, we aren't at GM's level but at club player's.
It's nice to get an opening advantage but even more important is the knowledge of the position : What is going on there ? What should each player try to do ?

I divide this knowledge in two parts : the practical experience that you get by playing games and the theory/study that you get by analyzing positions at home.

My own knowledge is my concern : my games, my books, my analyses...

But we're two on the board and the knowledge of my opponent plays an even role.
The best position you can get is the one where you know what's going on while your opponent don't.
Of course, I can discover the experience of an opponent by checking his games.
But I'm looking for a variation that I can use for all of my opponents.
So this time I asked the club players rated between 1200-1900 what they play on 1.e4 e6 2.d4 d5 (put in another way: what a French defense player meets at club level)

3.Nc3    36%
3.e5      24%
3.exd    21%
3.Nd2    19%

I can point out that more the player are weak more they prefer e5 while Nd2 disapear.

What about the theorical knowledge of a club player ? Is it possible to have an idea about it ?
Well, I took my own experience and picked up my reference book : "Play the French" 3rd ed by J.Watson which is a very serious and very appreciated work. I came with the idea that the time spent on studying a variation should logically linked to the number of pages it is dedicated.

3.Nc3    134 pages     58%
3.e5        47   "         20%
3.Nd2      39   "         17%
3.exd      10   "          5%

Logically, we find Nc3 at the top : GM plays it a lot and it leads to many intricates positions so there's a lot of theory.
At the bottom, the exchange don't get much. Because of his symetrical nature, it has a dull & drawish reputation : White leaves the idea to get any advantage out of the opening and reports the fight on the middlegame.


CONCLUSION : If I want to play a variation which can lead to an opening advantage and that my opponent doesn't have a great knowledge, then the logical choice is the Tarrasch 3.Nd2
It has the best rate between GM (34%) / Club level (18%)