It took me a second to realize "has the answer E" meant "as the answer" not "as an option," but I think that's a glitch in my brain, not the puzzle.

I did accidentally pull down too hard on my screen when I was trying to scroll up, which accidentally refreshed and deleted all my answers on all difficulties, not even just the one I was working on/hadn't solved yet, so that's less than ideal. I almost had level 4 done, but no way I wanted to go back and redo it all!

I'd also love an "answer lock" button, and/or floating undo/redo buttons, because sometimes to take the next step, I need to put in an answer and game out how the rest of the puzzle comes together with that answer, and it would help to be able to undo back to that point, not just try to count my steps and keep scrolling down to the undo button for more of them. I'm not likely to get out a piece of paper or pop open a notepad app to game it out that way.

Sorry about the pull to refresh glitch! Completely forgot to disable that. Will fix ASAP.

Thanks for the other feedback!

Came here to say the same thing. I think the wording could be a little better.

Thanks for the feedback! I'd appreciate hearing how you (or the parent) think the wording could be improved. English isn't my native language, so suggestions are very helpful.

Maybe "Which question is solved by E?" or "How many questions are solved by E?" - another way would be "Which question has E as its correct answer? // Which question is answered by E?" or "How many questions are answered by E?"

I personally like the solved wording.

Also another vote for clearer colors about possibility state - for example, the cells are marked green if that specific answer is internally consistent, but it can still be an incorrect solution overall, which means you did not get the correct answer whatsoever.

I think people (myself included) have been trained to think green = good, meaning that answer is correct and shouldn't be moved, you found that one, now work on the others - so an intermediary color as mentioned, like yellow, meaning 'yep that's plausible but not necessarily right' would be a good fit there.

I would go with, "Which question has E as its answer?" or "the answer" instead of "an answer" would fix it for me.

Similarly, "How many questions have E for the answer?"

I think it also threw me because we have a trick question in English, "How many months have 28 days?"

All the months have 28 days, but people usually answer 1, so for those of us who are used to approaching questions, especially logic questions, as literally a as possible, all the questions have the answer E. It's just not the right answer. "How many questions have E as the answer?" would clear that up for me.

These are really fun, thanks for posting this. Only weirdness I've seen is that I got a red bar for selecting a correct answer on one question because I had not yet selected the correct answer for the question it was about, which threw me off briefly. Working on level 3 now.

Hey, thank you for playing and for the feedback. I'm not sure how to solve that, though.

I've tried to make the game behave similarly to the Zebra Puzzles [1], where each clue is validated against whatever is on the grid right now, not against the expected answer.

[1]: https://www.zebrapuzzles.com/ (I run this one as well)

I feel like there might be a green/yellow/red situation here, where red indicates inconsistency with marked information, while yellow indicates "not satisfied YET, but could be with marked information"

If Q1 is "the answer to Q2" and I mark A:A, Q1 should be red if A is crossed off for Q2, yellow if A is still an option for Q2, and green if A is the chosen option for Q2.

Thanks, that seems like the way to solve this. If this game gains some traction I'll implement this. Right now the crossed off alternatives are local only and the server side checker wasn't made thinking of that.

Both Gemini and Claude can effortlessly write a python program to solve the first problem, using z3

https://claude.ai/share/86224755-5245-4635-ba6d-66fad5573118

https://g.co/gemini/share/d94e7e3d675e

Not really surprising, but was still nice to see

That's interesting. Will try later to see if the reasoning models can solve the harder puzzles without z3.

I probably shouldn't be trying this so late at night, but that was a lot of fun! Thanks for sharing!

The undo/redo buttons were very helpful, I was certain I had figured a test out but ended up being wrong. Being able to go back to the assumption I made early on instead of having to restart was much appreciated

That's great, I'm very glad you enjoyed it!

“what is the only answer with ${q}?” seems like a bug.

Indeed, thanks for spotting. Deploying the fix right now.

edit: deployed, sorry about that!

Oh I thought I was supposed to fix the bug myself— that would be VERY self-referential.