Today marks a full year of Jumblie puzzles! I made the theme “appreciative” for a reason: I am so grateful for all the love the game has gotten over the past year.

Without burying the lede too much: I’m going to stop making Jumblie puzzles now.

The game will still work, daily, it will just loop from now on, where it starts from Puzzle #1 after Puzzle #365.

How the game was received

I have loved seeing how much people like the game! It made my day, consistently, to see people playing it and enjoying it.

It’s been particularly cool seeing the international community embrace it, with fans of the show The Devil’s Plan playing it, and folks making their own versions in their own languages.

I also had the pleasure of showing it to some large publications that you know of, without revealing who they are, and popular puzzle creators like Josh Wardle (of Wordle). It actually was looking like some folks might acquire the game a few times, but none of those conversations came to fruition (which is okay!).

The game averages between 800-1200 players a day, which has been really fun to see consistently happen, and again, I’m so grateful for that!

How it came together

I’ve written before about how I built Jumblie and how I upgraded it over time, and quietly added smaller features like dark mode, different accessibility modes, animations, an archive, user-submitted puzzles, and more.

Building the actual game was a labor of love, for sure. There are definitely features that I wanted to add and didn’t have time to (like different hint systems and difficulties), but overall it felt nice to have a fairly polished, legitimately used product.

How it fell apart

Y’all… being a puzzle editor is incredibly hard, time-consuming work. If you don’t have a puzzle that can “generate” itself like a sudoku or maze game or math game… you need a human in the loop.

Before you jump into solutions-mode and say something like, “BUT HAVE YOU TRIED AI???” Yes, yes I have!

Ultimately, making puzzles every day for a year for a free game (that actually costs me time and money) revealed to me that it was too much to maintain. I tried AI/LLMs to generate them, but they’re not particularly good at sticking to letter-counts in words, and would kind of fall apart after a certain number of themes (or get repetitive). I tried using human suggestions, which were awesome, but there simply weren’t enough of them. I tried using the Datamuse API and other various thesaurus tools, but they only got me so far. I tried so many more things. I even tried talking through through how to make it easier/faster/less overhead to generate puzzles with professional puzzle editors and puzzle enjoyers alike, and they all ended up saying the same thing: Jumblie is the type of puzzle that needs a human editor.

Yes, it’s only one puzzle a day. But again… I’ve been doing this for a year. It’s a lot of brain-power to try and come up with yet another theme that is accessible to enough people to keep it going. It’s time to admit that unless I actually hire a human editor (which adds more to the costs), or get the game acquired (which is still possible, and I’m still open to it), I can’t keep this up.

So… now what?

That’s about it. I’m not as bummed as I thought I would be! In a way, saying goodbye to maintaining Jumblie feels like a failure, but in a larger, louder way, it just feels like closing a chapter in this book of side projects.

The little sigh of relief I did when I chose this direction was probably the biggest indicator that it was the right move for me.

Anyway, all this being said: thank you again for playing, for reading, and for checking out my little game.

• You can still play Jumblie here
• You can find the source code here
• You can ping me to re-up acquisition conversations on socials!