Is the analysis wrong sometimes ?

Hello, so i'm a really low elo chess player and i'm confused about the analysis of my game. It says i'm gonna lose a bishop and on the analysis it doesn't take the enemy bishop with the king and i would be happy to know why ?

Actually i lose my second bishop on his pawn D6 but i still don't know why the computer doesn't take the enemy bishop ? Sorry my chess language kinda suck too lol.

Here's the analysis 

I think the reason computer doesn't take the bishop on e3 is because the king would still be pinned by the rook. If the king takes the bishop, black plays pawn to e5, and white end up losing the other bishop as well. And because white is down on material, this would be good for black.

I don't have a very high rating myself and sometimes the analysis can be very confusing because the computer plays at such a high level and nothing like the opponents I play. I think the lesson you need to learn here is that pinning your own pieces as defenders can be very dangerous.

Not true.

My most recent job, by the way, was as a computer systems analyst for the military. Just sayin'.

But the computer IS correct in this case. Kxe3 would not help White at all, because then his King, his Bishop and Black's Rook would all be lined up together, and Black could just play his Pawn up to d5... winning the Bishop anyway since it cannot legally move (since that would leave White's own King in check).

higher elo than me

Yeah, one game it analyzed if I promoted my pawn to a rook, I would get mate in 5. To a queen, I got mate in 7, which didn’t really make sense. The mistake was corrected when I reanalyzed.

You are right. I think this has to do with the way a move is evaluated. Stockfish doesn't evaluate the move in itself, but instead looks at the evaluation of the position it leads to and compares this to the positions the other positions lead to.

I don't know the exact numbers that are currently used but as a rough guideline (these numbers might be a bit off):

Between +/- 0.25 - 0.75 is a slight advantage
Between +/- 0.75 to 1.50 is a clear advantage
Above +/- 1.50 is a decisive or winning advantage.

With these numbers as guidelines Stockfish can't give useful feedback in extremely winning or losing positions. For an engine that wins any position with a 1.50 rating or higher, there is no meaningful difference between + 15 and + 19. They are in the same ballpark. 

At least that's what I think.

White is lost either way. -14 and -17 is pretty much the same practically speaking. The time for real blunders is behind us at this point. That is when the engine might show a certain bad move as "good".

"Good" move (in engine terms) means that it is not the best, nor it is among the best moves, but it doesn't really change things too much. In this case, things are so bad that one piece, more or less doesn't change things too much. 

On the other hand, if it was equal material, those moves would get double question marks from the engine... they would be considered as blunders.

It accelerated the lost and saved time? Hence, more efficient. After these two moves the game went from -14 to -30. Just kidding, you probably analyzed it when too many people were on the server.