Lichess tactics suggestions

A lot of tactics are underrated because in lieu of knowing the correct answer, people tend to just play moves that look the most forcing. I have a couple of solutions for this.

1. Instead of having the computer play the objectively best move in response, have it play the move that is hardest to refute.

2. Have the player counter all possible refutations to the tactic, instead of just one

3. Don't tell the player when they've selected the wrong move, only check if they have the necessary advantage / equal position after the target number of moves.

Also: When a player makes a wrong move, show the computer refutation to that move so they can see why it's wrong, then allow them to continue

And maybe: automatically show the analysis after the tactic is solved

I completely agree with you on point 2 and 4.
But all of the tactics on lichess.org are based on decisive advantage. so "3. Don't tell the player when they've selected the wrong move, only check if they have the necessary advantage / equal position after the target number of moves."

wont work because unless that particular line is played, the position is either drawish or losing. You can analyse a few tactics to check that out

5. Don't show the link to the game. Show it after the puzzle is solved.

@vedantkarandikar I'm not sure what you mean by "if that particular line is played." Aren't tactics forcing a very particular line? It would be checking to make sure the player followed that particular line, and it would also allow tactics with a couple correct moves, which would mix things up a little bit and force people to memorize new patterns. (For example this tactic, which favors one winning move (+4) over another (+3.5) The suggestion seems a bit weird but if a player gets the first move of a tactic right without seeing the whole thing, I want him to get discouraged and think he messed up.

@Aks_Oks I didn't even realize they did that on flash, I usually use mobile. That's actually a hilarious exploit, but good point. It makes it way too easy to cheat without being detected.

1. Yes, although that is a subjective measure. It's impractical to have humans go through the 1000s and 1000s of puzzles and decide what move for the computer to play.

2. Yes, but again some practical problems with this idea. Such as, many puzzles aren't forcing, and sometimes there are 50 moves the opponent could play and they are all evaluated as equally losing. Is this again going to require humans to go through every puzzle to decide which lines the computer should test you on?

Also, even if the puzzle is pretty forced, if there are 3 different options, and the puzzle is 3 moves deep, you're already at 9 variations you'd have to solve. I can see this getting tedious, especially if many of the variations have a similar solution.

3. Good puzzles shouldn't have multiple solutions.

4. Yes, that would be a nice feature. Would require Stockfish to run any time someone made the wrong move in a puzzle, but I bet that would be doable.

Many good ideas. Unfortunately some of them might be too hard to implement. Maybe Lichess could try to build a new tactic trainer using these ideas, and start with a smaller puzzle database so that it's manageable for humans to make these additions. I do think the Lichess puzzle trainer could use some improvements for sure.

@Es-tactic as you said after "target number of moves" multiple lines are possible. so if some "other" line is played, is what i meant.

As I have already written in another topic, I think the only problem with lichess puzzles is that the score often does not match the real difficulty of the solution

@jposthuma You bring up some good points.
I didn't realize that a computer program picked out the tactics. Perhaps there could be a way for lichess to crowdsource the creation and judging of tactics, such that only ones deemed sufficient become rated?
Also, the computer wouldn't have to play every variation, just every reasonable one. I know the CT-Art apps do this and it works really well.
Good tactics problems don't have multiple solutions, but realistic ones sure can.
But you're right, it's probably very difficult to implement changes. These are just some ideas that try to reduce the problem @LanciaFulvia_HF said, which is that tactics problems are often poorly rated.
Thanks for your input, as a master level player, you definitely add a perspective that I wouldn't see on my own.

@Aks_Oks Agreed!

The main problem is related to puzzles that have been played a few times.
When a score of 2400-2450 points is reached, puzzles often require long reflections,
And so far everything is right and all normal.
But every correct solution is worth a few points + 2, +4 and so on.
It's enough a mistake with a complicated puzzle to lose suddenly -25 -30 points.
And often this happens because a very complicated puzzle (but with few attempts) is evaluated - erroneously - 2000 points