267 small worlds vs one big game

Someone asked me last week what Motecraft does, and I gave them the standard answer — "small visual worlds you can hold in a palm" — and then realised the standard answer is doing the work the question actually needs done.

What it does is offer 267 different five-second arrangements of light and motion. Each one is a single idea, executed simply, that you can open, touch, watch, and close. The question "what does it do?" assumes one thing, which is the wrong frame for the kind of thing it is.

So this is a post about that frame. About what changes when the unit of design isn't "the app" but "the experience". And about the trade-off that thinking buys you, which is real and worth naming.

One big game, played many ways

Most apps are one big game.

That isn't a value judgement — it's a description of how they're built. There is one thing the app does. The product team's job is to polish the doing of it until it's the best possible version of itself. Headspace's one big game is guided meditation. Strava's is recording your runs. Calm's is calming you down with audio. Within that one game, there are levels, modes, presets, settings — but the job is constant. You can describe the app's value in one sentence and correctly predict 95% of what a user does inside it.

When the team builds something new, the test is: does this make the one big game better? A new meditation series in Headspace is a variation on the existing game. A new run-type in Strava is the same game with a different label. Single-mode polish compounds. You get better and better at the one thing because every team-hour points at the one thing.

This is a fine way to build software. Most software should be built this way. Trying to convince a Strava user that "actually we also do photo editing now" is how products die.

But there's a different shape, with a different cost curve, that I think Motecraft is.

Catalogue, not mode

A catalogue is not a collection of presets. Presets are variations on one game. A catalogue is a collection of distinct things, each one its own complete idea, that don't share much except a place to live.

A bookshop is a catalogue. So is a record store. The shop sells you "a place where books are" or "a place where records are" — but the unit of customer value is one specific book or one specific record, and the question of which one is the entire purchase decision.

Motecraft is a catalogue of visual worlds. The unit of value isn't "Motecraft" — it's the specific world a user is looking at right now. Mandala Crystal is one world. Candle Flame is a completely different world. Aurora Veil doesn't borrow anything from the first two except the bottom-of-screen player chrome that wraps them all. There's no mode-switcher; there are 267 catalogue entries.

What this changes:

• The polish unit is the experience, not the app. When I make Lichen Bloom feel right, that polish doesn't transfer to Candle Flame, because Candle Flame's "feeling right" is a different feeling.
• Discovery is the central interaction. In a one-big-game app, the user's first job is get good. In a catalogue, the user's first job is find the one that's mine right now. Home is a curation surface, not a tutorial.
• The product is the curation. Which experiences exist; which ones lead; which moods filter to which subsets; what the daily hero picks. That work compounds in a way the single-mode polish work doesn't, because each new compounding decision tightens the catalogue, not just one entry.

Why 267, not 12

The natural follow-up is: fine, catalogue model — but couldn't 12 really well-made experiences do this better than 267?

Maybe. But probably not, and here's why.

In a catalogue, the value to a single user comes from finding the one. Not from the average quality of every entry. So the calculation is: what's the probability, when this user opens the app, that there's something in the catalogue that exactly matches what they wanted?

Twelve experiences give a user maybe a 30% hit rate. They open the app, maybe they want something restful — Candle Flame fits, they're happy. Or maybe they want something more kinetic — and the closest thing in the twelve doesn't quite fit, so they bounce.

Two hundred and sixty-seven experiences give a 70% hit rate. Maybe higher. The user opens the app wanting "something that does what Ocean Waves does but slower and less blue" — and we have Fluid Gradient, which is exactly that. The user opens wanting "something fractal but warm" — there's Spirograph, warm-paletted, looping trace. The catalogue has the shape of "yes that one I needed" because it's wide enough to contain the specific thing.

The trade-off is real and worth naming: the production cost per experience has to stay near zero. If every entry costs three engineering weeks, the catalogue collapses to twelve and you're back to one big game with twelve variants. If every entry costs three engineering days, you can have hundreds, and the math works.

So the engineering work — the renderer conventions, the shared control system, the cross-platform parity, the multi-touch arbitration fixes, the tactile-column controls — is in service of making the catalogue possible. Not just making each experience pretty. There's a difference, and it's a load-bearing one.

What this looks like at a Tuesday evening

A user opens Motecraft. It's Tuesday evening. They've been at a screen for ten hours and want something quiet but interesting — not blank. They open the app, scroll past the Today's Pick to the Ambient section. They find Firecracker Spider, which is too kinetic. They scroll. They find Candle Flame, which is too still. They scroll. They find Mandala Crystal, which is exactly the quiet-but-interesting register. They tap. They watch the eight-fold geometry assemble. They close the app after forty seconds. They come back tomorrow.

That sequence doesn't work with twelve experiences. It needs the catalogue. And every individual experience in that scrolling moment is doing its specific job — not being all things, not trying to serve the user who wants kinetic. The user finds kinetic in Firecracker Spider, finds still in Candle Flame, finds the in-between in Mandala Crystal. Three different worlds did three different jobs. The catalogue, taken together, did the one thing the user came for: be the right thing tonight.

A note about strategy

People sometimes ask whether the catalogue model is the strategy. It isn't, really. It's the unit of design — the answer to "what's the thing we're making one of". Strategy is a question about which catalogue to build, in what order, for whom.

But the catalogue model does change a lot of the other questions downstream. It changes what "polish" means. It changes what "discovery" means. It changes what counts as a successful update. Most importantly, it changes what we have to be good at — which is not single-mode mastery, but range. Range is a different skill, and trains differently, and rewards different work.

Two hundred and sixty-seven small worlds is what range looks like when you draw a line around it.