Chess.com computer is blind

I submitted a game for Chess.com computer analysis, and I found some glaring inaccuracies in it. At several points, when it was mate-in-6 or less, the computer commented on moves that they were 'mistakes', and once going so far to write 'lucky you, your opponent blundered', when it really wouldn't have mattered.

Can the 2000 computer not even see mate-in-6?

It gave me an inaccuracy for playing the Slav Defence and every subsequent book move in the line.

1.h4 is the move which extends the mate to a Mate in 10. The best move in this position

1...Bg1 leads to a Mate in 12. And Bxd4 leads to a quicker mate. But it is no reason to put an '?' if they both lead to mate

2.Rxg1 is the best move and white will be mated in 12 moves. Kh3 leads to Mate in 7

3...Ke3 is the best move for black

The ?s were awarded by chess.com's computer. The comments and alternatives were inserted by myself, to show how absurd the computer's analysis is.

After 1.h4 it's mate-in-6, as I've demonstrated.

After 2.Rxg1 it's mate in far fewer moves than 12 (I think 5).

After 1.h4 is it Mate in 9 not 6. I think you mean it is mate in 6 after 1.d5

1...Bg1+! leads to Mate in 9 (sorry my fault )

Ba dum tssshhh.

Chess.com Computer 2500 is even more strange at end game analysis.

After 1.h4 it's mate-in-6 if black pushes the a-pawn then checks with Rh1#. If white tries to stop the pawn using his rook from behind, white plays Bg3+ followed by Rh1#.

Stupid computers. They don't calculate well, so sorry again 

You don't have to use these computers. If you have Rybka, Houdini, Stockfish, etc. You can enter the game, and have the computer analyze the game with tremendously better results

No, we don't have to use the Chess.com computer, however Premium members are paying for this service and as such, it should be better than what it is.  Playing the Sicilian Defense, the Chess.com computer informs me that 1...c5, is an inaccuracy!  lol

Did you turn on the webcam?

  lol The Sicilian Defense is the most potent response against 1.e4 

There have been a lot of threads showing the error of that computer lately.

I feel like it used to work better and they made some code changes in an attempt to update it and actually made it worse.

This reminds of the outrage I got in response to asking the question if my Chessmaster was saying that e5 was the strongest response to 1. e4... or suggesting maybe it is programmed to select an opening defense to give you a demonstration of what happens when it is used to calculate the best line of play within that opening/defense theory.

It does have an opening book, but I asked it to play itself to see what would happen. I was specifically looking for it's opinion of the best opening and line of play that gives white the best chances to win each time.

When I mentioned that black won the game, it seemed to further incite one individual to the point that he accused me of implying that 1. e4 e5 was a victory for black. In the computer's estimation, this may be so, but it isn't relevant to the way people are capable of playing, generally speaking. To give relevance to my last statement, I am not sure how many ply a GM can see at standard time controls, but I doubt it is 60 ply for each possible line, for every move.

It also cracks me up how engines tend to favor a tactical line that is a win in the same amount of moves as a positional line. You would think the programmers of the better engines, would have put some effort into making sure that a user had justification of their choices, if winning lines are very close or the same, in the amount of moves to victory, or if the calculated numerical advantage that changes per move, was made irrelevant, in the case of victory being imminent. Don't get me wrong, I realize the importance of the engine having a system for finding and selecting the best moves and lines in the first place, but it need not treat us as if we asked it to pick the best color.

To add to that Nameo, I hate when the chess.com computer says mistake when the mistake had me at +11.05 and the suggested had me at something like +11.35. Really? I think with that much of a lead and that slight of a difference change we could simply say "Alternative".

I can understand if it was something like the mistake move being 0.05 and the better move to be 0.35 because that is much different and could lead to further incremental increases that lead to a win vs leading to a draw.

That is is exactly what I am referring to when I say that at some point it shouldn't ding you anymore, when it is going to be mate regardless. Even though it needs to still distinguish a best move, it also should be aware of when to stop worrying about making it a priority.

When victory is imminent and one line has a .0025 average difference for the 4 moves required to checkmate, it should no longer consider your move choice lesser than in terms of a mistake. If it considers it a mathematical 2nd choice, so be it. In fact, if I were the programmer, I would make it a point to reward a player on their way to victory, by labelling the move a winning move, even if it is a mathematical second best.

 + 1!  Also, with a service we are paying good money for, we should be getting something in return that is a lot more competent. Otherwise, it's pretty worthless.

This position is rather trivial, so it seems a bit silly to have all these "mates in x" being bandied back and forth, especially when they are all doubtless the products of silicon.

In a tournament situation I might well have opted for 1... Bg3+ 2 Rxg3 Rh1+ 3 Kxh1 Kxg3, which gives a simple win with "no hair."

I have complained about that very thing in the tactics trainer, where a simple two move mate was the alternative correct move to a less obvious queen sac, two move mate.