How do I counter the Smith-Morra Gambit?

You've missed the point, I think. 

If the Smith-Morra were as horrible as LT says, Anand would've taken out an hour to prepare 3...dc3 against Esserman. The fact that he went for the pragmatic approach means that Black isn't obviously better, nor winning after 3...dc3, otherwise he'd have just done his homework and taken an easy win.

     You more or less say that any patzer with an engine can claim he understands a line more than Anand. Of course you do forget that memorising a few Stockfish moves hardly means "understand".

   Even if I agree with you (I don't) , I can't accept that a guy  can see just one line in Stockfish and that immediately qualifies him to claim that he understands Smith Morra more than Anand . In my opinion this is the definition of ignorance and stupidity(EDIT: This is for Tsvetkov).

   Do you know how much Anand has analysed Smith Morra? No you don't.

Even if he has a second(I know he had) , it's your theory that "he does his dirty work" and I don't even know what that means. Does it mean that Anand is sleeping all day and the second works 24/7? No, of course not. It means the second finds something and then both work hard on it analysing and examining the positions.

     In my opinion , and I might be wrong , you simply have failed to understand what it means to be Anand. I'm sure he has spend a lot of time in Smith Morra long before he even had the need to have a second. 

Understood.  Is there a way to play e7-e5 at once?  I understand blunting the bishop, but it seems like a waste of tempo to have to play e5 so soon after playing e6.

On my club most Sicilian players and I think in general play 3...Nf6 to decline the Smith Morra. But I heard very good GMs just grab the pawn and never returns it and white sees it back never again. This first play of Nc6 d6 and e6 and then after you develope your pieces with Nf6 and Be7 this is why this is not a wast of a tempo. because after you developed your pieces the attack runs out of steam and when the treath of e5 becomes actual only then play e5 with black. I heard this is a good aproach to the Smith Morra to just keep the pawn and keep it.  

Correct:  Stockfish is able to find moves that I'm unable to find OTB.  You rarely find anything of substantive value (Your Qe4 suggestion, for example, which was added to the database in 1995!).

105% of your statistics are completely made up, but going down what you are insinuating:

The first 10 moves of virtually every opening line have remained static for most of the last 80 years (note that roughly 50 of those years were prior to strong computers engines!).  The "changes" to opening theory have largely been with regards to extending it.  For example, in the Najdorf, there are lines that go out to 30+ moves (well into the middle game!).  The first 10 moves of a given line in the Smith-Morra, Alapin, etc., have been fairly well-known for quite some time.  If you think you are suddenly going to "throw out" many of those lines simply because of your misuse of an engine, you are more delusional than I had previously thought.

For example, here is the position from one of the Alpha Zero vs Stockfish games:

7. d5 is the 3rd most popular move (behind Nc3 and Re1) according to the Chess.com database (with Nc3 being a clear first with over 5300 games).  [Granted, Stockfish was playing without an opening book and not optimally configured - that is outside the scope of this point]  The first game I can find on record with this line is from 1957!  That is, 60 years ago (well before strong computer engines), master-level players had determined this move was a good one.  To dismiss a line because it is "old" is simply ignorant and arrogant.  To assert that your silicon friend is going to "revolutionize" opening theory (when it is designed to rely on an opening book in its optimal configuration) is laughable, at best.

 what is the best first chess move?

There are 300 openings and 5000 subopenings.

he can not prepare for everything.

     You more or less say that any patzer with an engine can claim he understands a line more than Anand. Of course you do forget that memorising a few Stockfish moves hardly means "understand".

   Even if I agree with you (I don't) , I can't accept that a guy  can see just one line in Stockfish and that immediately qualifies him to claim that he understands Smith Morra more than Anand . In my opinion this is the definition of ignorance and stupidity(EDIT: This is for Tsvetkov).

   Do you know how much Anand has analysed Smith Morra? No you don't.

Even if he has a second(I know he had) , it's your theory that "he does his dirty work" and I don't even know what that means. Does it mean that Anand is sleeping all day and the second works 24/7? No, of course not. It means the second finds something and then both work hard on it analysing and examining the positions.

     In my opinion , and I might be wrong , you simply have failed to understand what it means to be Anand. I'm sure he has spend a lot of time in Smith Morra long before he even had the need to have a second. 

I don't know.

I looked into it very carefully, and it is a very complicated tactical play in almost all variations, but black gets on top almost always.

One possible line after Na4 is this one, Nh4, to displace the bishop, followed by Bd8, guarding the b6 square.

SF reaches 50-60cps black edge.

A pawn more is a pawn more.

I don't know if that is sufficient for a win though.

As said, very complicated tactically, you need days to analyse it, but black will always retain a clear edge.

I think there’s something called the Siberian trap

Imho, in the current state of art, there are 3 clear candidates: 1. d4, 1. e4 and 1. Nf3. 

I don't know why I have to explain everything to you. Marc Esserman is a well-known specialist in the Smith-Morra. Even if Anand had no idea who Esserman was, he could check a database and see that Esserman plays the Smith-Morra. The Smith-Morra was then played in the game.

300 openings and 5000 "subopenings" are irrelevant.

You are implying that because something is played more it means people evaluate it to be "the best".  That is an assumption on a massive scale.  Theory has not changed.  In fact, you can even argue that (given how Alpha Zero "taught" itself) that the decision it made was not even the "best", but simply allowed it to get into a situation where it had a better understanding of the position than a brute force engine (which is the same problem many engines have with many opening lines, by the way)

The order for moves in most opening lines has not changed much in the last 100 years.  They were playing QGD, Italian, Ruy Lopez, and many lines in the Sicilian 100+ years ago.  Sure, Steinitz played a few openings that we do not see played much these days (the Vienna, for example), but that is not due to them being unsound (at least not all of them), but rather the preferences of modern players are different. 

These statements are so freaking idiotic that I could write an entire post just on them.  One of the big complaints about the Alpha Zero vs Stockfish results was that (wait for it ...) Stockfish was running without an Opening Book (and without an endgame tablebase and odd time controls - which you like to put to shame with your even more absurd time controls)!  Why was that a complaint?  (drum roll ...)

To state that "no one uses opening books" with engines anymore is asinine.  From the Komodo 10 release notes:

Komodo can use Polyglot opening books. When playing with a book, Komodo will move instantly if it finds a legal move leading to a position in the book. We have a free book with lots of variety on our site on this page

Stockfish also uses Polyglot opening books, FYI.

You are confusing opening lines with chess being solved.  The latter is not the case.  There are no "best first moves" in chess.  e4, d4, Nf3, and c4 are all quite good (probably slightly better than the rest), but you are not likely to lose a game based on your first move alone.  Hell, even a3 can be a decent first move when used appropriately:

Your obsession with centipawns at the opening is amusing, but that is about it.  Now, please go back to complaining about how you don't have time to play OTB while spending hours upon hours trolling forums online with half-baked ideas.

Thanks for the response. This looks interesting, but couldn't White also try to force a perpetual with 21.Bf4? 21...e5 looks a bit risky to play on.