Ill-posed problems.

Friends, not for the first time in solving the problems faced with the fact that they are drawing up incorrectly. Since this can be tolerated, and I even used to. But the task №16000 - it is beyond my comprehension. It is easily calculated even novice players. Well, tell me: Is R a1 - this is not one of the three options the right decision?

And here's another problem: 25294. The solution that the system considers true mate in three moves. I put the mat in the same three moves the other way, but the system does not see it, he says that the problem is not solved.

I beg your pardon: the task №25294 I was wrong

I dont solve problems here any more. A broken problem, thats a no go. Same for Opening training. Broken shit. I use Chess Tempo, thats bug free :-p

If you think a puzzle is inaccurate, downvote it. It gets removed below a certain threshold. As of the time of writing, at least 16 people think Puzzle 16000 was a good puzzle.

Moreover, all of them have been tested with very strong engines. I take it you tested your solution with an engine?

For Hellball.
The task does not require 16,000 check engine solutions. All three of these are identical stroke. You can check the engine - I do not mind.

For awesomer.
I like the color on the board lichess))

Ah see your writing was unclear. You meant to say that Ra1 also leads to a forced solution. I agree, it should be added as an alternative for that specific puzzle.

But it does not invalidate the puzzle, unlike what #4 said. :P

Yep this is a common problem on lichess' puzzles. Often there are simply so many forcing wins that you dont know which one to choose. Lichess often accepts only the shortest solution or the one that's mate in something. This would be totally fine had the other winning moves been considered as alternatives. But you get used to it. I also see it as a challenge of finding the "engine way " of winning certain positions :P

#8 i had puzzles with equivalent solutions (eg move transpositions) when one was 'solved' and another was 'not solved'. Probably a problem with ponder depth/not all alternatives checked. But also second best moves should never be marked as 'not solved' as long as they win, but as 'try again'.

Using an engine-only solution and not a review mechanism will not solve the problem (and will not make users happy). It will just shift it to a deeper ponder level.

Downvoting is too unspecific. There is a difference between 'i dont like it' and 'this is wrong' (at least for smart people).

A 'report puzzle' button is needed: If clicked, a new page is opened where the user reports the puzzle to a forum (pre-filled with a link to/fen of the puzzle and links to earlier posts regarding this puzzle and an encouragement to read these first). Then strong experienced users with a rating above some level and some extended rights can re-check the problem and adjust it if the user is right. If so, it would also be good to mark the problem so that like once in a year all solvers of this problem do get their points back if they didnt solve it (referring to http://de.lichess.org/forum/lichess-feedback/win-tourn--get-points-back#6). If so, this is also relevant information to be put into the prefilled forum post of the user reporting it ('the problem has already been reported and found to be incorrect and all solvers of this problem will get their points back on the next yearly reset.')

I know that is a bunch of work to implement but it is the only solution to get a perfect puzzle collection. 1. deeply check with comp, 2. refine using peer review. Just want to point that out.

Let me also add that i believe that strong human player + engine + time = a clear answer to the question if a chess position is won, lost or in the draw region. This means it can clearly be said which moves get a 'not solved' and which moves get a 'you have a better move'. That is the difference between chess and reality and the reason why a peer review for chess puzzles should not be as hackable as eg a peer review for a wikipedia article.

Thanks.