Does anybody want to know how to destroy the French Defense?

I don't care what Brainfish says. What does your brain say ?

OP is trying to find the best way to win French, So I proved that there is no way to win French.

Well my brain is around 2000+ rating/skill/knowledge, you can see in my profile.

I dont play French,. I  play CaroKann, I follow almost exactly whatever brainfish say, sometimes I can follow up to 8 moves, sometimes, up to 12 moves or more. I remember more than 15+ moves in certain variation. So I will play exactly the same as brainfish/stockfish in my prepared line. 

I get that part. Doesn't every player have a defense or opening that is a nemesis ? Trying to find a way to refute the French defense is about as much of a waste of time as is trying to memorize any opening 53 plies deep. I think a better thing to focus on would how did Bobby Fischer beat the French ? It was a nemesis for him. How does Magnus Carlsen tear it down ? If you look, you will find games of Carlsen playing about every variation of it that is really good for white and beating other GM's, regularly.  I would focus on learning the ideas that gave Carlsen or Fischer an advantage in their games and try to apply them to your own games. There are some really good YouTube videos of some of them. I watched one of the Tarrasch and scored a nice win with it, gambitting a pawn in the process. I find a big key in the French is being willing to trade some material for a good position that allows for chances to attack on the king side. 

I don't like to give away my opening secrets, but since you seem willing to try to give good and credible information, all I will say is look into how the positions of the French and Karo-Cann can transpose and learn to utilize them. I picked up on some patterns and finally figured out how to use them. I just haven't had time to employ them in my games yet. 

I could not understand

This is the current position evaluation by three engines. 

Brainfish say- 5. cxd5 is best, 5. Nf3 is  alternative move.

Both stockfish and Leela prefer 5. Nf3

Stockfish evaluation on this position( 0 depth ) for  evaluation table( on right side) 

Scores- positive score  for white, mainly due to  thread 0.17 and mobility 0.11.

It's useless to tantalize your engine like that without consulting major master games first.

For example, after the tricky 5.dxc5 (a Mickey Adams speciality), the reply 5...Qxc5 (which prepares the queen retreat to its optimal c7 square) is known to be more exact than 5...Bxc5.

 Chess engines were weak in earlier days but these days stockfish is stronger and stronger. In a few weeks ago, stockfish passed the different color bad bishop detection patch.  Stockfish still has major issuse in fortress detection.

If someone find of a weak play by stockfish in any particular opening variation that lead to strategic mistake/major mistake .(e.g  E.99 King Indian Defence) please let me know I will review with my brainfish and I might forward it to fishcooking google group.

PS, I am an amateur programmer, I am now more keen in chess programming than playing chess.

Then i guess you should study secrets in chess by that infamous author.

Brainfish actually choose 5. Nf3!!  as well.  One of end position of brainfish, stockfish says 0.15+ for white, it is equal. 

I guess you mean 5.Ngf3.

They are a wee bit late:

Tarrasch played this 128 years ago, and several thousands more games played after that.

As far as SF, it gave AlphaZero a negative eval all the way until it lost in the French.

@pfren I know those line as I read ECO 500 and Several Chess Informants in my teenage ( may not be very deep like you). I recommand to have a look at those computer chess opening books. 

https://sites.google.com/site/computerschess/perfect2017books

It is free. It is a popular book tested by millions of games between 3200-3500 engines..

please note, the opening lines not recommanded by this book would likely be Red Flags lines. ( suboptimal lines)

I don't think I will bother using this book. Actually I do not use any engine book at serious correspondence chess- I'm doing fine with par openings which offer room for creativity. More so for OTB- what is OK for an engine might well be unplayable in practice.

Again, nobody cares what an engine says, and we will not mimic an engine.

That said, if YOU want to mimic an engine, computers have a life span of 2 to 3 years.  Computers don't poop.  Therefore, I recommend that if you think everyone should follow everything that a computer does, then while everyone else should play 5.Nf3, you should go 2 to 3 years without pooping!

 even as a big fan of engines for opening innovations even i admit it has limitation. in practice it is almost alway much better to play something slightly inferior but with a clear set of governing principles and an ideal set up or goal, then a move that's obscurely concrete where you need to respond in 5 different narrow ways to keep your hard to understand 0.15. in many openings, certain sidelines score worse than the suspected eval for this very reason. "natural play" can often be a big advantage.

The modern opening is always "Ladies and gents, start your engines!!"

As to how engines work , my favorite is a line in the Caro-Kann.  It gives me an immediate big inaccuracy so big brother has no clue its all engine at way high depths . One guy was +1.3 and watched it go bye bye with no inaccuracies (Even with his SF engine soaking up time) I did lose on time, but I picked up +12 from who knows who that apparently cheated me earlier....so I ended up ahead in online rating? So go the times. I dont know if I'll pick up the 4 I lost by playing, but ya never know?

https://www.chess.com/article/view/deas-on-how-to-prepare-a-good-opening-repertoire2

This is how GM prepare their opening. If you check interview of most top GM , eg Caruna, they will say, their opening prep is assisted/ mainly influenced by engines.

https://www.chess.com/article/view/how-fabiano-wins