I already saw one puzzle thread about a bad puzzle, so I decided to make a more general one instead of one for each puzzle. Puzzle 5928 looks wrong. It looks like it takes an inferior line while forced mate in a few moves is possible with a different second move
You're right, that doesn't make any sense. Qxf2+ Kh1 Qf1+ Rxf1 Rxf1# should be the correct continuation. If it was you who downvoted that puzzle, good, downvoting means that other players are unlikely to see it.
There's still to be a moderator feature that allows us to remove bad puzzles, unfortunately this is currently done via manually editing database entries.
For the past week, ~1000 puzzles have been generated each day, unfortunately this means some bad ones may get through. This problem stems from stockfish considering positions such as 5928 to be material advantage positions, and not mate in N positions, thereby ignoring mate in N solutions.
There's still to be a moderator feature that allows us to remove bad puzzles, unfortunately this is currently done via manually editing database entries.
For the past week, ~1000 puzzles have been generated each day, unfortunately this means some bad ones may get through. This problem stems from stockfish considering positions such as 5928 to be material advantage positions, and not mate in N positions, thereby ignoring mate in N solutions.
It's understandable that there will be some problems with new features, right now I am just wondering should I continue to post bad puzzles in this thread (along with downvoting)?
I found another one where there is a forced mate line, http://en.lichess.org/training/1918 Qe6+ on the second move (black's reply to white's first move is also suboptimal, though I don't know what's the position on that)
I found another one where there is a forced mate line, http://en.lichess.org/training/1918 Qe6+ on the second move (black's reply to white's first move is also suboptimal, though I don't know what's the position on that)
http://en.lichess.org/training/2197
The continuation of the lin in the puzzle will lead too a first mate, but there is a quicker one with 2.Qxf2+ Kh1 3.Qf1+ Rxf1 4.Rxf1#
The continuation of the lin in the puzzle will lead too a first mate, but there is a quicker one with 2.Qxf2+ Kh1 3.Qf1+ Rxf1 4.Rxf1#
You're right, neither of these puzzles are correct. I'll have to work out a bug fix. It's a curious issue.
If you come across any bugged puzzles, you can send a short description directly to me. I think I know what's causing these bugged puzzles, and I'll be getting on it shortly.
As for now, downvoting is good. Even -1 means that it's almost impossible for other people to play that problem compared to the rest of the puzzles that are 0 or +X.
As for now, downvoting is good. Even -1 means that it's almost impossible for other people to play that problem compared to the rest of the puzzles that are 0 or +X.
After all and beside the problems in new training feature I wold like to thanks to all Lichess team for this great site, its simply fantastic, full of nice surprises every time I log in. Thanks!
Now to puzzles:
Looks like puzzles are taken from game where an computer analysis has been requested. Those analysis are with short time per move and not always are best play.
So probably that is the reason some problems aren't good.
A possible solution can be to do a re-verification on the proposed solution line before including into puzzles.
Also I see many puzzles with too much material difference, to a point making the problem uninteresting.
For example:
http://es.lichess.org/training/1175
http://es.lichess.org/training/841
In those cases you have mate in X moves, or if you miss the line you also win! So a possible solution is to look in lines where the forced continuation wins and in case you fail the score is close to 0, say [-1<score<1] , no advantage.
Regards!
Now to puzzles:
Looks like puzzles are taken from game where an computer analysis has been requested. Those analysis are with short time per move and not always are best play.
So probably that is the reason some problems aren't good.
A possible solution can be to do a re-verification on the proposed solution line before including into puzzles.
Also I see many puzzles with too much material difference, to a point making the problem uninteresting.
For example:
http://es.lichess.org/training/1175
http://es.lichess.org/training/841
In those cases you have mate in X moves, or if you miss the line you also win! So a possible solution is to look in lines where the forced continuation wins and in case you fail the score is close to 0, say [-1<score<1] , no advantage.
Regards!
http://en.lichess.org/training/2553
Strange puzzle. It's all good as such, but there's more than one way to skin a cat!
The puzzle demands: Rook H2, Queen D1, Queen E2, Queen B5
But rejects: Rook H2, Queen D1, Rook H4 (which would leave the only defence Knight G4), Rook X G4
I'd suggest that the puzzles are great, but that it's perhaps a bit dodgy to have a personal score associated with them, if there is even ONE dodgy puzzle or puzzle with a choice of endings where only one is accepted.
Of course, I realise that I may have missed something totally, in which case I will do 100 lines and recite the Lord's Prayer often etc.
Strange puzzle. It's all good as such, but there's more than one way to skin a cat!
The puzzle demands: Rook H2, Queen D1, Queen E2, Queen B5
But rejects: Rook H2, Queen D1, Rook H4 (which would leave the only defence Knight G4), Rook X G4
I'd suggest that the puzzles are great, but that it's perhaps a bit dodgy to have a personal score associated with them, if there is even ONE dodgy puzzle or puzzle with a choice of endings where only one is accepted.
Of course, I realise that I may have missed something totally, in which case I will do 100 lines and recite the Lord's Prayer often etc.
Rh4+, Kg4, Rxg4+, Qf4, Rxf4#
is longer than
Qe2+, Kb3, Qb5#
It should've given you a retry, it's probably because it's an earlier puzzle.
is longer than
Qe2+, Kb3, Qb5#
It should've given you a retry, it's probably because it's an earlier puzzle.
Here is a puzzle http://es.lichess.org/training/2553
Black has a Mate in 4, but an human player will go for other line: forcing queen exchange and playing a KRRB+5P vs KN+ 3P ending ...
This problem was selected from a position with eval:
(-27.09 → Mate in 4) Checkmate is now unavoidable. The best move was Qb3+.
I do think this problems can be avoided they are selected from positions (~0 → +/- BIG SCORE)
Regards.
Black has a Mate in 4, but an human player will go for other line: forcing queen exchange and playing a KRRB+5P vs KN+ 3P ending ...
This problem was selected from a position with eval:
(-27.09 → Mate in 4) Checkmate is now unavoidable. The best move was Qb3+.
I do think this problems can be avoided they are selected from positions (~0 → +/- BIG SCORE)
Regards.
This topic has been archived and can no longer be replied to.