Why was this counted blunder?

Having now looked at the position for ten minutes, after ...f6 I don't like white's position at all, compared with Bd3. I think that black equalises, whereas if white goes in for a sacrificial line by playing Bd3 after f6, and then castling after fe, then white has to consider the two lines ... where black pushes the pawn to e4 with a fork and where black plays a queen sacrifice by playing ed. The queen sacrifice line looks difficult to assess. Black gets a rook, a knight and a pawn for the queen and may be alright ... and maybe not. But that's unimportant since the pawn push line looks as though it equalises for black.

Contrast this position with the one after white plays the engine's recommended move of Bd3. There, if black castles, white seems to have immediate attacking chances. Alternatively, if black plays f6, white can transpose into the lines discussed above or can play f4. White has the further option of Ne2 after f6, which may be more interesting than Be2 after Nf3, although I haven't tried to analyse it. There's no need, because black would need to analyse it and white needn't, which is time saved for white. Consequently, Nf3 has to be regarded as a poor move, simply because it reduces white's options by so much and makes black's game so much easier. If the attack after Bd3 ... 0-0 is strong for white however, then Nf3 is definitely a blunder, because it throws away strong, winning chances for white.

I haven't used the engine and have just done this by looking at the diagram, so I may be wrong. In any case, I tend not to trust the engine regarding anything other than simple tactics. But I would think that Nf3 is a blunder and not just a rather weak move.

I think Nf3 f6 is strong. If Bd3 first then f6 Qh5+ g6 B:g6+ it's so obvious guys

All people know here this. But who will play f6?? After Bd3. Also it is Bxg6. 

I just didn’t get at the start why was it a blunder. After discussing, I saw many things comes after Bd3 as a position. There is no tactic after that . But I still think it should be labeled as a mistake.

This is very true and very important.

To put it into perspective: the chess engine (in this case Stockfish), is a result of more than 60 years of hard work by many-many chess players and programmers.

The software that assigns the labels was probably written in a couple of hours by some chess.com software developer guy.

I actually thought before that guy wrote this that chess.com uses stockfish engine when analyzing . I guess  it is good that I know this now.

Well, there's no difference between a mistake and a blunder. "Blunder" just puts more emphasis on it. But if it changes the probable outcome, as it might, maybe "blunder" is appropriate. After all, it might be that it changes white's chances from "excellent" to "about equal or even worse than equal".

You got a  point

I’m shocked though. This thread gone 54 comments in 3 hours. (Half by me)

That's the reason. No further comment needed

That’s not the main reason actually. One of the main reason is that the engine wants to play f4. Nf3 basically block it.

Because the engine works by counting up controlled squares and f4 controls dark squares especially in the absence of white's dsb. But conversely, f4 is a "positional" move and the engine is useless at positional assessment.

I can't imagine this being a blunder, but the engine here can be pretty harsh in dealing these out, where actual blunders are often overlooked.  I've seen a player hang their queen and it consider this a "good" move.  There are some bugs in it for sure.  You're going to want to move your knight soon even if you move the bishop out first as it suggests.  This helps take away a couple of key squares from the black queen and I'm not even sure why this is a mistake  I'd want to be clicking on what the engine thinks is so much better and at least get a sense of why.

Don't forget that the analysis is done on the server of chess.com

When I analyze a game with chess.com it take 5 to 10 second. When I analyze a game with Fritz on my laptop (good enough for VR gaming ) it take several minutes :-) . 

Analysis of chess game can be VERY resource intensive, on my old laptop (also VR gaming capable) the Peshka software of chessking was using a ton of resource when analyzing my position, as much as for a triple A game or VR one. 

When using the server do not expect to have so much calculating power available, calculating power = money. 

If you want serious analysis, you need a desktop or laptop I guess.  

I have the same question lol

https://www.chess.com/analysis/game/live/42661191925

on move 4, hanging black's queen is an inaccuracy??

Don’t know what is going on

Well in fairness to the engine it did count 2...Ba3?? as a blunder.

Chess.com wants you to upgrade to the highest paid membership to get the brilliant advice from Mr. Chess Coach.   If you do this you will have no more questions on move grades.