[oliwary]

> If you’ve ever tried to construct a crossword, you’ll find that the framing of a crossword grid under square theory feels right. When you’re nearing the end of the grid-filling process, finding valid crossings of words to fill that final corner of a grid, there’s a satisfying “clicking” feeling—a sense of magic—when it all fits together, analogous to the wrapping-around feeling of completing the square.

If you enjoy this feeling, I think you would like my word game https://spaceword.org. The goal is to arrange 21 letters in a square that is as tight as possible. No one has achieved a "perfect" pattern yet, but people are very close, often leaving only 3 spaces blank!

[JoshuaDavid]

Fun game! Though I dispute that people are "very close" to achieving a perfect pattern.

To get a "perfect" pattern you'd need to find three 7 letter words that can stack on rows adjacent to each other to form a 3 letter word in each column. Such arrangements do exist, for example:

    o p e r a t e
    a r r o w e d
    r e s e n d s

but they are very rare - I estimate something on the order of 0.002% of combinations of three 7-letter words have any valid arrangements. Assuming that you're using standard ETAOIN letter frequencies, the typical bag of 21 letters will usually have just a handful of combinations of three 7-letter words so a given puzzle has a << 0.1% chance of having a perfect solution.

But there are 12,000x more ways to rearrange 21 tiles within an 8x3 grid, and the word choices are more forgiving as well (if you draw 7 letters from the etaoin frequency distribution, those 7 letters in order are much more likely to form a 3 letter word followed by a 4 letter word than they are to form a 7 letter word). Pretty much every puzzle should have at least some solutions fitting within an 8x3.

Additional note: 3 blank spaces is the best non-perfect arrangement, since the grid is only 10 tiles wide. One blank space could only be achieved by a single 23-letter-long word, and two blank spaces could only be achieved by a 10 letter word next to an 11 letter word, and an 11 letter word would not fit inside the 10x10 grid.

[oliwary]

Glad you like it! :) And thank you for your comments, super interesting! Excellent point about the rarity of the perfect arrangement. Perhaps I should throw in a few lettersets that do have a solution, I am intrigued to see if people would discover it.

My other game, https://squareword.org focuses exclusively on perfect 5x5 squares, but here the goal is to uncover it wordle-style rather than arranging it from scratch. There are surprisingly few combinations that have ten unique, common words in a 5x5 letter square!

[cjameskeller]

I struggled on a few days' puzzles under the assumption that there _was_ a perfect solution possible -- It may be worth noting in the "help" that not all lettersets can be solved perfectly.

[isodude]

My first initial thought when I saw the game: spaceword golf.

Like any golf, you start with the smallest square possible and increase it with each level. You get less points for how perfect the the square is.

[JoshTriplett]

Do you have a version of spaceword that's not a "daily" game? I'd be interested in trying it if so.

[oliwary]

There is a weekly mode with 63 letters, thinking of adding a monthly mode as well. Or would you prefer puzzles that are always open?

[JoshTriplett]

Puzzles that are always open is exactly what I mean, yeah.

Some "daily" games call this kind of generated puzzle a "practice" mode. But whenever I encounter a daily game, I go straight for that mode, which is what most games would just present as the game itself.

[oliwary]

Fair enough! :) I'll look into adding a practice mode.

[JoshTriplett]

Thank you very much! I really appreciate it and look forward to trying it!

[KolibriFly]

Oh wow, Spaceword is hitting that exact same brain-itch as square theory