Sometimes engines are stupid. Sometimes they are brilliant.
If you want my take on engines - for what it's worth - I have recently installed an engine to Arena Chess which has a bunch of parameters you can change. A few of those are piece activity, king safety and pawn structure. By increasing their values, I find that the moves that the engine plays are much more sensical than moves by Stockfish. It's been a great assistance ever since, even though I only have it for a short time it does wonders (I used to do analysis with another engine that comes with Arena when you download it with an optimistic outlook setting, but the new one is way more personalized to what I'm looking for)
I really have to wonder about the engine's opinions about moves. Here's and example where Nf6 forking both rooks and the queen was called a mistake. The reason involved a long and convoluted line that did lead to a somewhat better position for white (me), but what human in his right mind would take such a roundabout path when the simple direct approach gets the job done? As it turned out my opponent noticed the rook fork, but failed to notice that the queen was also forked, and lost it.
The engine said Bxc6 was best.
So my question is how concerned should we be about moves that the engine calls mistakes? I've made a lot of moves I consider good in light of the outcome, but which the engine called mistakes because there was some other convoluted line that resulted in a marginally better outcome. I guess I just get a little annoyed when the engine calls my winning move a mistake.