http://en.lichess.org/training/33307After ...nxr, kg1 nf3, kg2 I played a move that is difficult to see but also outright wins Qh2, kxn qh1, ke3 qe1 and black either gets mated in the same fashion that the puzzle intended or loses his queen. All I'm saying is can we get a redo on qh2 on this particular puzzle?
Seems justified!
I don't get these Stockfish analyses when mate in x is on the cards. What's going on on moves 40-42? Mate in 9 after 40... Qxa7, then suddenly it's mate in 11 after Stockfish's 41. a4. Couldn't the programm see one move ahead here? And the same thing but even more puzzling on moves 41-41, when the computer opts for being mated in 6 rather than the mate in 10 which it'd calculated half a move ago.
I've noticed the same thing in many computer analyses of my own games here and it puzzles me.
It's likely because it is not a single computer processing the analysis. It's distributed across multiple servers at limited depth per move in order to generate fast results. Therefore sometimes one server may generate a different result for the next move in a sequence depending on that server's load. So in essence, yes, that computer couldn't see one move ahead because maybe on that instance the server was overloaded and got stuck looking at a slower mating sequence. Some accuracy in analysis is sacrificed for speed of results and conservation of resources. Therefore, take all Stockfish results with a grain of salt and work it out on your own with the engine's guidance.
Today's daily puzzle is another example of this.
http://en.lichess.org/HEnssVA6/blackHi inkspot,
the point is that the best move is requested. Qh2 is wrong because it will not force a mate within the best sequence. I agree that tons of puzzles require improvement especially in middlegame positions with tactical elements.