Another way to improve the quality of the Puzzles could be an option to report bad (but not strictly incorrect) puzzles. If you for some reason don't like the puzzle (unclear best move or whatever) you report it with a short explanation. Then chess.com can manually review it. If still to much workload, chess.com can have a trigger level e.g. 2 or more reports before manual review.
Multiple winning moves for chess.com puzzles
Sort:
JockeQ wrote:
Another way to improve the quality of the Puzzles could be an option to report bad (but not strictly incorrect) puzzles. If you for some reason don't like the puzzle (unclear best move or whatever) you report it with a short explanation. Then chess.com can manually review it. If still to much workload, chess.com can have a trigger level e.g. 2 or more reports before manual review.
That's already available and has been on the website for a very long time. Just without explanation.
https://support.chess.com/article/1223-how-do-i-report-a-bad-puzzle
Though I imagine most reports are invalid today, where it's just that the solver doesn't understand the solution. I can't imagine the man hours that would be required if someone had to vet the explanations.
I know this function but it says you should only report if puzzle is incorrect.
But I agree it would probably take too many manhours to review the reported puzzles.
Speaking of multiple winning moves, does anyone know how to make multiple winning moves for puzzles? (I tried variations but it didn't work):
JockeQ wrote:
This puzzle has only 1% success ate:
Something has clearly gone wrong with the rating algorithm on that puzzle. According to the comments on the puzzle, it was rated 3800 in January 2021 but somehow slipped down to a rating of 1967 by December 2021. As of today (February 1, 2023), it is still rated 1967 despite 1% pass rate on 10K+ attempts.
Unfortunately, their guidance for reporting puzzles does not allow for a puzzle to be reported for something like this. They say to "only report problems where the answer given is incorrect or there is more than 1 correct answer". That is not the case here. This puzzle is not incorrect, but something is clearly wrong, and it is adversely affecting the user experience.
Maybe they should do a drop-down menu or multiple choice that allows users to select the reason they are reporting a puzzle. Include "incorrect solution", "multiple solutions" and "rating error" as options. They should also include "spite" or "anger" as options, as I'm sure some percentage of people would be honest and select those options when they apply. That would ultimately make it easier for the development team to sort through reports and find the ones that really need to be addressed.
It's not just something wrong with how the algorithm works for this puzzle, it's a general fault. It's just more obvious for some puzzles.
1800 rated puzzle with 99% success rate from 3k+ attempts. Something is not right with this algorithm. https://www.chess.com/puzzles/problem/2053366
Regarding chess.com's interface, they could design it to handle more than one move. At the very least, they could do something similar to their Lessons where an equivalently good move is met with something like "That's good too, but this puzzle follows a different line. See if you can find it."
Following the "lessons" approach is a good option which I suggested often. Handling more than one good move is better but it implies that the solution tree may grow significantly. The situation at the first forking point may repeat several times before solvers are satisfied and varies with their strength. The best option is to set a much higher standard for the selected puzzles like in professional compositions. But that would require human inspection and judgement. Now all (almost all?) puzzles are mined by a search engines which gets you the issues described and others.
Note that we have not arrived in the age of puzzle interfaces yet. Your wishes are quite modest, I have hundreds of them. The most important one is the handling of defensive moves. All really good puzzles have more than 1 variation - different black moves with different solution lines. Solving all the relevant ones should be required before declaring a puzzle "solved". This is the reason that all real solving events only permit handwritten solutions which may be scanned and mailed to judges across the planet. These will check and give points to the parts of the puzzles correctly solved. What is required depends on the puzzle type. There are simply no puzzle interfaces (yet) capable of decent communication with puzzle solvers. They all suck!