Multiple winning moves for chess.com puzzles

This puzzle has several solutions, fastest mate is in 3 moves but there are several other forced mares in 4 moves or more. Is the rule that in order to solve a puzzle you must find the fastest check mate?

https://www.chess.com/puzzles/problem/2062872

The issue in that puzzle is that it's 3 moves before 50 moves without captures or pawn moves occurs, so anything other than M3 is actually a draw.

So faster mates are always preferred and if you see a puzzle where mate is possible, no matter what, it means that it's probably has that draw constraint.

OK but not really a good puzzle then when you have several other checkmates in 4 or 5 moves. Especially since it doesn't say you have done 47 moves without capture or pawn moves.

I sympathize with you. Chess.com's puzzle criterion is this: there is either 1 single move leading to a clear win, or there is 1 single move leading to the quickest checkmate - which takes precedence. The catch is that when there is 1 move to the quickest checkmate then there might still be 100 moves that easily win as well. It's weird thinking but it's not a contradiction. Note that the "quickest" mate is hardly ever longer than 3 moves from the point where there is more than 1 winning alternative!

Note: it's even weirder. When you discarded the 100 wins and found the checkmate sequence, to your surprise, the puzzle interface might suddenly abort the puzzle. It means that - though only 1 single move started towards the quickest checkmate - there is choice of moves thereafter before actually getting there! Chess.com will never ask you to choose between moves which are equally effective in reaching the target it secretly assigned to you! It simply aborts the puzzle and grants you solving success!

To forestall the 50-move dangers you must always check the move counters in the FEN. There is no way of knowing that even a common position (like after 1. e4 e5) is not close to the 50-move line. Practically it turns out that chess.com only applies this technique to lean endgame positions. So you'll need to install a bio-alarm to identify those occasions.

Is there any way to know this when you encounter the puzzle?

No. But  almost every time I've encountered a mate no matter what endgame puzzle, it's one of those.

I think these stupid puzzles that assume the 50 move rule is almost reached should be removed. Or add a label "Find fastest mate".

Are these puzzles automatically computer generated or is there a manual process involved when they are created?

Most come from actual games or old tactics collections, but new ones are automatically pulled from site games. I think there may be some additional human checks, but I'm 100% sure on the release process of candidate puzzles 

For this particular example, it would mean that White had been unable to checkmate the lone Black king with Rook and Bishop for the previous 47 moves.  If this puzzle was actually pulled from a game, then it is a game that surely deserves to be featured.

I have also encountered a few puzzles where the difference in evaluation between "correct" move and second best move is marginal, something like +2 vs +2.5. Those are also not very good puzzles IMO. And I guess the should be easy to find and filter out by making a script that goes through all puzzles and compares the evaluation for the moves in the winning sequence to the next best move.

But to be fair it is very rare, I find >99% of the puzzles to have a clear best move. However, a couple of times the the best move is a draw. This I also don't like because since it is so rare I find myself sitting forever to look for a winning move that does exist.

It's part of the "puzzle philosophy" of chess.com. It treats puzzles as games. In a game you don't know what the outcome will be with the "best play" by both sides. It may be a win, or a fast checkmate or just a draw. So you gotta figure out what you can achieve and then how. The world of "professional"chess composition is different. It won't ask you to juggle the whats with the hows. All its compositions specify what is expected of the solver, like "checkmate in 4 moves", draw or win. As a solver your job is only to find the how! Without specified duration your solution may be as long as you like!

Note that in both types of puzzles there is still the requirement that only 1 move at every turn leads to the goal moderated by its philosophy. Once there is a choice in solution moves the puzzle terminates and counts as solved.

I understand the basic idea not knowing if the best move is for a win or saving a draw. And it would make sense if it it wasn't for that it is so rare. I have solved 4400 puzzles and maybe 3 or 4 of them the solution is a draw. Something in the order of 0.1%. And since it's so rare you don't trust yourself even it you come to the conclusion that the best move is really to save a draw.

OK, since it happens so rarely it's not a big problem. But on the other hand, since there are so few of these puzzles I think you might as well remove them.

I have solved a lot of puzzle rush survival puzzles and I found there are quite some draws - commonly stalemates - in the higher rating regions. How many there are may very well depend on the profile of chess.com's mining program. It might have a preference for wins (most puzzle solvers do) and the ones that are there may only come from the "old" database. No way of knowing until chess.com tells us.

I only do my daily dose (25 puzzles with my Gold membership). My rating is around 2100-2200 and I can only remember facing three or maybe four puzzles where the solution was a draw.

Since Puzzle Rush Survival is not timed I get to puzzles rated 3000 and higher. Probably more draws there. Not that it matters a lot. You still have to consider the possibility when the going gets tough in any puzzle. And who says chess.com's engineers won't read these posts and start mining for draw puzzles today!?

Much of the time it comes down to the depth the analysis looks at when the system select them as puzzle candidates. At that depth it's the best move/line but a much deeper look might find essentially equal options. 

There's a tradeoff on resources and compute time where the additional compute time, in most cases, isn't going to change anything. That's one reason there's the ability to report puzzles.

As to draws, sometimes I wish there were more and I would like a training method in puzzles where there isn't any tactic and you just have to play a decent move

On both this site and another site, I've gotten to a level of puzzles where a decent fraction of the puzzles no longer make sense to me, even when going through the analysis.  For example, after a few moves, the puzzle will end without any change of material and the analysis will not show any captures in the immediate future.  The engine just thinks that the position is significantly better for some reason. 

Sometimes I can figure it out after the fact (e.g., it sets up a mating net from which you have Mate in 12 or something ridiculous like that), but usually I'm left scratching my head.  Then the very next puzzle will be a straightforward Mate in 2 that unbelievably has the same rating.  It's frustrating and often leads to me cursing at my device.  ... And then I keep playing.

I also like the idea of a "Find a Decent Move" version of Puzzles.  I don't know exactly how you would score it (everything HAS to have a rating associated with it these days), but they could probably use their Game Review evaluations (Brilliant, Excellent, Good, etc.) to form a metric. 

Many posters have answered that question. like Martin Stahl. I explained it at least thrice on these Forums in the past week.The answer is very simple and logical. Every puzzle stops as soon as you would have a choice between 2 about equally effective alternatives for your next move! The chess.com puzzle interface is simply not designed to manage more than 1 correct move at any move turn. Occasionally you have seen neither of the correct alternatives ahead and therefore do not know of your advantage in the position. Run a StockFish analysis to confirm the theory. in retrospect.

Note: the interface might not even care to distinguish between winning a rook and winning a queen when both effectively win you the game without much trouble. But if the puzzle continues, take the queen

@JockeQ: I forgot about mentioning another draw-puzzle type, the repetition puzzle. Like in your games, occasionally your only escape is to sacrifice some material and then repeat a position or give a perpetual check. I have been faced with a good number of those puzzles by some chess.com puzzle module. A while ago though. Possibly the repetition puzzles were common in the (small) old database but no longer mined for the current one.

I didn't ask a question.  I understand why puzzles stop early and I said in my original post that I go through the engine analysis.  I am merely sharing my experience, which nowadays can include going 5 moves into each of the top 3 Stockfish lines following the "end" of the puzzle and still failing to see what the supposed "tactic" was.  I consider such puzzles to be  "positional" rather than "tactical", and that can be frustrating when I spend a long time looking for a concrete tactic.

Regarding chess.com's interface, they could design it to handle more than one move.  At the very least, they could do something similar to their Lessons where an equivalently good move is met with something like "That's good too, but this puzzle follows a different line.  See if you can find it."  Then maybe there would be fewer of us venting in the comment section and you would be spared from explaining this simple and logical topic thrice more.