Puzzles Don't Choose Challenging Responses

Thanks for replying @Martin_Stahl! I agree that it's impossible to determine which line is always harder/easier than another, but in puzzles like this, I feel like it's not about the difficulty of seeing the line, but rather how important the line is to actually being able to make the correct move confidently. You're never going to be able to see everything all the way through, so being able to sort of eliminate moves that tactically feel wrong through the process of elimination is an important skill. When I think about Nxc3, I can quickly see at a glance that Bxh5 does not work for white. (It's a 2-mover, after all). Seeing that sacrificing my queen does not lose on the spot and/or results in a completely winning position is a bit more abstract, and during a real game, these are the variations that take up much of your time because you HAVE to make sure.

I think revising the current "report the problem" feature would be great. Currently, you're only supposed to report the puzzle if the solution is wrong, or if there are multiple answers. That is not the case here. The solution is right, and there is only one answer in the line that the program chose to respond with. I would like to be able to report problems like this and add a little message that explains why I think it is not wrong, but could be improved.

Thanks for summarizing my entire post into 3 sentences for me, I totally agree

Puzzles rating is extremely inflated, but it's not really a problem because no one decides to judge a player's playing strength by their puzzles rating on an online site lol

There should be multiple correct options, when I was on chesskid, I got a puzzle wrong and showed it to my coach. He said I was correct and chesskid was wrong, and told me to use chess.com.. But now it seems that both need to be better.

oh okay bro !

Yeah, there will always be puzzles that are not the best, but I'm just hoping we can mitigate the problem if you know what I mean

I am not so sure that 23. ... Nxc3  24. Nxd7 Nxd7! is hard to find on the 2900 level. You already have 2 pieces for the queen while attacking 3 white units. A quick scan ought to reveal you remain at least a piece up.

Clearly 24. Bxh5 is not all that challenging. I doubt though that the puzzle engine has a clue about what "challenging" means. It simply plays a random losing move after it finds it loses, probably from the top 3 or top 5 continuations. The fun thing is that the engine treats you with the courtesy it should only extend to other engines like: Iallmightyengine trust you to find the 25 move one way road to checkmate and I therefore rather give up a piece which extends my life with at least 5 moves . 

This is in store for us with quantum computing. After you open 1. e4  your steel eyed opponent surprisingly offers you a draw on the logic that its evaluation function reveals it has a 0.001% chance of losing the game and no chance of winning. Why underestimate you? Btw, I advise to take the offer!

@Arisktotle - it's true that after Nxc3 Nxd7 Nxd7 black will probably win another piece, which means he'll get 3 pieces for a queen, but it's not obvious whether or not that's actually good for black in the first place, or if he can hold onto all of the pieces. Just because something seems initially good doesn't mean it works out in the end - just like how Bxh5 felt wrong but is not losing until we find the forcing continuation that refutes it

That's a fair comment from the human viewpoint. It will not change the way the puzzle engine looks at the variations as they are all in the piece of cake range for engines. What's probably missing is an AI module which dumbs down the engine power such that it is capable of appreciating our human difficulties and take them into account. It's not a simple thing to resolve.

Understood, I was hoping that chess.com could revise the "report a problem" feature to allow for puzzles to be manually updated. It might be too tedious though

I agree, some puzzles seem to give a non-challenging response. If I spend 3 minutes looking at a critical line and the computer cops out with, say, material loss a person would never, I'm bummed. Of course I can just note the rating of the puzzle at the end of my puzzle rush (survival) but that's one more step. Also, puzzles ending too early or with weak lines leave the temptation to skate.

Now, I'm very okay with analyzing analyzing the puzzles I didn't totally understand but got right anyway. That's the way to learn. And the AI for deciding what's the critical line for a human is nontrivial. But sometimes it's a bit frustrating for a puzzle to end too soon or to give a response that just leaves the computer down material. I feel a bit cheated. First world problems.

I have a goal for Puzzle Rush survival, as well as ratings for daily puzzles, but once I get there, I may move on.

I do agree chesstempo is better if only for the interesting notes in the comments section including "what about move X?" so I don't have to jump through hoops and consult an engine. And also so I feel slightly less silly when I miscalculate.

Get the ct art dvd. It's a lot more challenging.

My puzzle issue

I'd rather do a 5 move combo that either ends in mate at best or me up a minor piece at worst, than do a simple 2 move trade queen for queen and pawn.

Going for a possible win is tough on your puzzle rating, if all it wants is  to go for an exchange that favours me slightly.

@December_TwentyNine for that problem, Qc7 would be mate if it weren't for Be7! blocking your pieces. After winning the piece the position isn't the clearest thing but it's definitely winning

you never maked a post lol

I almost never post. Never have something worth saying.

Puzzle- gives an  opportunity to put opponent in check, then take a knight after their king moves or  their queen interposes.

They use Queen to interpose, I take knight I win puzzle as knight for nothing. - NOPE.

I'm supposed to take their queen, they take my queen using the knight that was on back rank; so it goes forward 2 ranks. Then I'm supposed to take their original knight with a pawn.

This makes Queen for Queen plus knight, but allows opposition to better their position with their capturing knight.

My way was I get a knight for nothing; but I'm wrong.

Puzzles can be hard to understand

HELLO EVERYONE

wait ur an nm... how are you not like 3.3k or smth

anyways, yes, i agree. but how would chess.com know which is the hardest line?