I got some positive feedback from a puzzler about today's 12x12.

Part of my reply:

"Given that the puzzles are constructed by my software, so I have only indirect
influence of the end result, i.e. nice "logic sequences" are not in
there because I put them there,

I'm thinking it could be a good idea to have people rate a puzzle: so after
you solve it, you give it 1-5 stars for example.

Then after a while I'd know more about what type of puzzle is more entertaining,
and maybe push the software in that direction.."


Of course in the end I'd train some AI software so it knows what's a good puzzle,
and let it loose on the world

That sounds good! I would gladly rate puzzles.

-Ville

Ok, I added a simple "like" function to the solved box

(you can click a "+" if you especially liked a puzzle,
for 7x7 and up only, and the 6x6 difficult)

So have you learnt anything about solvers' likes and dislikes, perhaps by reviewing the likes, or by use of AI to scan the like marks?
For myself, I mark that I liked a puzzle when it made me work hard, or it was somehow "different," or had a new (to me) feature.

Thanks for the reminder, I'll take a look at the data

I'm running some stats now, a first result is the "most liked puzzle", with 94 likes: the 8x8 medium of January 25th this year:



https://www.calcudoku.org/en/2023-01-25/8/2

edit: ^^^ I changed it such that this puzzle is always solvable online (also for non-subscribers / people who are not logged in).

I also posted it on reddit, so please vote up if you have a reddit account:

https://www.reddit.com/r/puzzles/commen ... ku_puzzle/

I am not sure if this helps.
I am not one for rating puzzles.

I am biased and want to say we all like easier puzzles because they are easier. There are puzzles where I have brain fart I and mess up an obvious assumption (boy do I feel stupid when I figure that out), and there are puzzles that I sigh at because while the pattern comes every week its just not one I like.

Puzzles that I tend to like are puzzles that have a cage or two that are unusual. For example, a corner cage for 7+ that is (2, 2, 3) or (1, 3, 3) and not (1, 2, 4). Something that allows for more possibilities. Larger cages allow for more possibilities but there was (at least) one user submitted puzzle that had very large cages and would have been unfair to complete in 24 hrs. It took time to solve.