It wasn't the best move, how does that make it a mistake? I do not believe chess is about finding the best move. As long as you don't miss checkmate, if you have correct ideas and calculate correctly there could potentially be 5 best moves. Although, ultimately, in this case it is clearly not the best move, I would still not call it a mistake. Appreciate your input though, I'll reconsider my beliefs.

Let's say I'm a hitman on a hit. I have the option to shoot them with a gun, or a crossbow. Both will be killshots, but the crossbow will be slow - obviously the gun is the better decision.

Yes. But how did you get there?

Car, van, bike...

An A on a test is good, buts its not perfect, you could have done better, if you didnt get 100% then you made mistakes. But regardless of how you define mistake, the definition the analysis uses is based on it not being the best move and clearly missing something (for example a slower mate) In this case I would definitly consider it a mistake because it took an extra turn, so in a speed game or if you were almost out of time, then the extra move could lose you the game

I suppose I agree with that.

I have finally proved the computer analysis is incorrect, misleading, and just generally annoying. Somehow, I reached the exact same position in two bullet games after move 22. In one I then played:

22. ...Ke7 (-8.56) which was a "mistake" as the "best move" was 22. ...Nd6 (-8.96).

In the other game I played:

22. ...Nd6 AND IT WAS A "MISTAKE" 

WTF, they even had different scores (with 22. ...Ke7 being the "best move")

It's all about how long the computer is set to analyze a move. It's possible the first time, the computer spent a little bit less or a little bit longer on that move, and it change it's output. I've seen situations where analysis with my own engine suggests something like Nxd5 with a score of .5, then the same with a score of -2, then e4 with a score of -1.5 then Nxd5 with a score of .43, because each time it changed it examined one ply further.

Okay, well amongst all the interesting things that the computer analysis gives, this may be the strangest.

I promoted a pawn to a queen. It was a "mistake", as the best move was the promote to a bishop (slightly better score).

The sequence that followed was for my opponent to then take the piece/queen regardless. So how can it have a different score?

Since I'm looking at this post anyway, might as well answer - it's (probably) because the engine analyzes bishop/knight, rook, then queen promotions, in that order. Because the first it analyzed was a bishop, and it didn't matter either way, it said that was the best move and the move it analyzed second was a mistake.