Mishurd x Patchwords

Two word games in one Reddit app, built on the same bet: the community is both the puzzle-maker and the audience. Which game a subreddit gets depends on where the app is installed, Mishurd, where the crowd disguises a phrase so it only sounds like the answer and everyone else tries to crack it; and Patchwords, where the crowd clips a shared pile of word cut-outs into the funniest one-line note.

Inspiration

Party games share a shape: one person sets up a payoff, a phrase that snaps into focus, a sentence so wrong it's perfect, and the one to get the room react to react wins. It either works or it doesn't.

We wanted to rebuild that feeling for Reddit, as daily games where the community can be both the maker of the game and the audience, and I did it twice. The same bet pointed at two different payoffs, the oh right of a mishearing and the laughter from an absurd sentence, so it became two games sharing one engine: Mishurd and Patchwords.

What it does

Mishurd is "write one, crack many." You open the day's post and land straight in the cracking flow. A line of real words sits in front of you, like jury sick park. You say it out loud and immediately hear Jurassic Park, and type what you heard. I give five puzzles to solved in a 60-second run, so figuring them out fast earns a speed bonus, and you score more the fewer people who cracked the same one. Then you can switch sides: you're handed a hidden phrase and write a disguise of it, built only from real English words. It goes into the databank for another round.

Patchwords is "build one, judge many." Everyone gets the same pile of word tiles and the same one-line prompt, your villain origin story, in one line. You snip tiles into a note and submit; identical notes merge into a single card, and the earliest author is the one credited. Then you switch sides and judge: two notes at a time, which is funnier? The day's notes rank by an Elo-style score from those head-to-heads. The prompts come from the crowd too: the top-voted reply to each post's pinned comment becomes the next round's prompt.

Around both loops sit the things that make a daily worth keeping: streaks to protect, leaderboards, and content the players feed themselves. Mishurd's submit-vet-graduate disguise library and its small hint allowance and Patchwords' community prompts and a voter "taste" score that rewards picking the note the crowd ends up loving.

How we built it

Both games are one app and each lives inside a single interactive post a quick inline version you can play right in the feed, and an optional expanded full-screen version. A small dispatch layer routes by subreddit to the right game; the two share the scheduler, storage, and Reddit wiring, but keep their own rules, screens, and look.

Mishurd's heart is a pair of checks every disguise has to pass. The first asks does this actually sound like the answer? The second asks is it built from real English words?.That second rule is what makes it a game and not a typing trick: joorassik park and jury sick park sound identical, but only the second is made of real words whose meaning sends you the wrong way, that misdirection is what makes it a tricky game. On top of that sits a self-balancing scoring system and the submit-vet-graduate library that keeps it feeding itself.

Patchwords' hardest piece was judging without native votes. Ideally I'd use the reddit native upvote system to judge the top sentences but since the app can't upvote, I built the comparison itself, duels saved as in-app votes, and rank notes with an Elo-style system: a minimum number of "head-to-heads" before anything can take the crown, plus a bounded bonus for originality so novelty helps but can't beat a genuinely funnier line. Each post carries its own tile pile and prompt so posts never bleed into one another, and the next round's prompt is mined from the pinned comment's top reply, auto-checked for length, links, and profanity. Both games post on an automated scheduler, Mishurd once a day, Patchwords twice.

Challenges we ran into

The biggest challenge was a question, not a bug: the joy of these games is something you normally hear or laugh at out loud, with people in a room. Does the aha, and the laugh, still land in silent text, played alone and a day apart? Everything else is built on that bet.

For Mishurd, the phonetic check was the hardest piece to get right. Telling a clever-but-fair disguise from a lazy respelling is subtle balance and can only be fine-tuned with real play.

For Patchwords, the hard part was fairness without upvotes: making pairwise comparisons converge on a real winner without endless voting, and keeping a comedy game funny rather than mean, so its moderation had to be a design problem, not an afterthought.

And both faced the classic community-game catch: they need content on day one, before there's a community. We seeded a starter set for each, so the very first players always have something to crack or a pile worth cutting up.

Accomplishments that we're proud of

Both full loops work end to end and are genuinely playable. Mishurd: crack-first onboarding with no tutorial wall, the two checks, the timed run, the scoring, scorecard comments, leaderboards, streaks, hints, and the community content pipeline. Patchwords: the shared-tray compose, the merge-identical-notes step, pairwise judging with Elo ranking, community-sourced prompts, and the voter taste score.

We're proud that two complete games live in one app without stepping on each other, and that in both the community grows and vets the content, which turns moderation into part of the game instead of a chore. In Mishurd, the scoring even makes both sides want the same thing: rare cracks are worth more and writers earn from every crack, so puzzle-makers and solvers are both pulled toward "hard but fair" and the difficulty tunes itself. And it's great to see that each game has a real point of view and look: Mishurd's one expressive character and "spoken word made visible," and Patchwords' magazine-cutout, ransom-note collage.

What we learned

Building two games on one engine also taught us what's genuinely shared, the crowd-as-content loop and the Reddit plumbing, and what has to stay bespoke: each game's validation, its scoring, and its feel. And in both, teaching by doing, letting someone crack one easy puzzle, or pick the funnier of two notes, before anything is explained beats any rules screen.

What's next

From here I will grow each community library and sharpen the pipelines, keep refining the feel and the two distinct looks.

Built With

• devvit
• react