I see lots of puzzles being downvoted

I think that's because some people somehow don't know a bishop and a knight is stronger than a rook, or a minor piece is stronger than the exchange,

or they don't analyze puzzles afterwards. They should do that before downvoting.

Please analyze puzzles before downvoting. there might be a reason why the opponent gave up an exchange.

Thanks.

I never vote pro or contra, but there are some strange puzzles, for example, where you should just take hanging figure and opponent doesn't have any countergame. So not all of puzzles are ideal.

I downvote quite a few puzzles because the algorithm to pick them is very weak compared to other sites.

A lot of them require you to learn how to understand Lichess' mechanics more than they require you to actually understand chess.

Two minor pieces being superior to a rook is not Lichess' mechanics.

lichess.org/training/74535

One of the puzzles I tend to complain about.

In order to succeed you need to find that fxe5 winning a piece because of Nxd5 followed by cxd5.

Here's where the puzzle ends.

Why wasn't Nd7 an option for black?

Playing vs Stockengine lvl8 as black and Nd7 instead of nxd5 led to me being ahead in material until 9 turns later, when white's positional advantage was too much to bear, forcing me to give material back.

That is eight turns on _perfect_ play until Nd7 turns out to be the weaker move.

In a puzzle that ends after two moves.

This is frustrating for two reasons. One: The puzzle following this one was this: lichess.org/training/74538
No threats. Push the pawn. Win a piece.
Two: The puzzle ending after two moves means you succeed without having proven that you understood the solution.

Doesn't make sense that both of them are on the same difficulty level either.

I agree. I pretty much downvote every puzzle that doesn't force you to play the most testing line. I know it plays the "best" move from the computer's perspective, so it makes sense why it won't test you on the interesting line. But I believe then that the puzzle was not that interesting.

Figuring out what the "most human move" would be seems to require a human to look over the puzzles, so it's probably inevitable that lichess won't include that feature. But it would be nice. Perhaps, one day, we can have Leela Chess Zero play against you in the puzzles instead of Stockfish?

To claim there is such thing a "most human move" is to perpetuate an embarrassing myth. People who argue this are simply unhappy that their answer was wrong, period.

There is not one move that is the "human" move, because humans are all different and will not all assess the situation in the same way.

In any case, demanding that your puzzles force you to respond to inferior moves rather than the best move on the board for your opposition is a good way to undermine the benefit of the puzzles, if such a benefit even exists.

I think the puzzles on this site are great. They seem to come from real games (played on this site) and are at a level that are appropriate for me (i.e., hopeless patzer). I'd like to thank the people who have created and maintain this site and all its many awesome features: studies, practices, puzzles, live games, tournaments, etc.

Agreed! @BoLoSi

@Chessty_McBiggins You miss the fact that what the computer plays is not always the strongest move. There's a very specific issue with computers that people are talking about.

Imagine we have a position where there's only one tactical idea. But it leads to an incredibly complex series of moves that culminates in an amazing mate in 15. Well you make the first move of the combination, really not sure if you've got it completely solved - but you at least see the general ideas and themes.

And the computer's response? It immediately gives you its queen, rook, and knight - for 100% free.

Why? The computer sees that if it doesn't give all of its material, it loses to the amazing mate in 15, and so it plays to prolong the game. But in the process of this, it plays vastly worse. There's no player that will ever lose after being given queen, rook, and knight odds but from the computer's perspective it's better than allowing the mate in 15.

You should of course never hope your opponent will play something less than the best move, but given the choice of a position where you lose 100% of the time and another where you lose something less than 100% - you should always go with the latter. Unfortunately, computers are not yet able to do this and will always choose the former if it prolongs the game.