Audience
Who is actually going to load this, and who are they going to make load it with them.
Gribble's is a two-player game with no account system, which means it has an unusual acquisition property: the first player recruits the second one in person, or over a link, within about a minute. Almost every design decision below exists to protect that minute.
The four people who matter
1. The Sender
Late twenties to forties, works a job with a queue in it, is the person in their group chat who posts the weird thing. They will not finish a 40-hour game. They will absolutely spend six minutes making a friend cook a Beverage Loaf.
What they need: a link that works instantly on the device already in their hand, and something screenshot-worthy inside the first two minutes.
2. The Recruited
Did not ask for this. Has been sent a link with no context. Their tolerance for setup is close to zero — a download, an account, or a 200 MB asset load loses them permanently.
What they need: to be in the kitchen before they have decided whether they are interested.
3. The Streamer / clipper
Small-audience creators need games that generate a reaction from someone else on camera. Asymmetric information games are structurally good at this: the audience knows both halves and neither player does.
What they need: the two screens to be legible side by side, and audio that is funny on a phone speaker.
4. The Kid at the Library
Playing on a locked-down machine with no install rights, no email address they will hand over, and a 15-minute session limit. This player is why the game is static files and synthesised audio.
What they need: no install, no account, no assumption of a fast connection or a good GPU.
What the prototype has already told us
The playable build has been running as a two-window and two-device test with unstructured hallway playtests. Consistent observations:
- The laugh is at the send, not the cook. The funniest moment is the diner committing to something indefensible. The kitchen's reaction is the punchline, but the send is the joke. Production should protect the send: fewer confirmation steps, more irreversibility.
- Nobody reads the tutorial. They drag a spider onto a plate and understand the game. The hairdryer needs no explanation at all — people point it at food instinctively and are delighted that it works.
- The cook wants to be crueller than the game allows. Players immediately looked for the bin. Incinerate needs to be as satisfying as serve.
- Both players ask for a way to talk, then are funnier without it. This is the single most important finding: the constraint is the game. It is not going to be relaxed.
- The specials board gets read aloud. Every session. Procedural naming is punching well above its cost.
What we do not know yet
Honest gaps, and how the funded work closes them:
| Unknown | How the grant period answers it |
|---|---|
| Does it survive round three? | Structured playtests at 3, 6 and 10 orders, measuring where the deadpan curdles |
| Does the cook role hold up solo-lonely? | A single-player mode where the "diner" is the seeded generator, tested against a live human baseline |
| Do remote strangers work, or only friends? | Public-room testing on the always-open default room |
| Does it read on a phone? | Device-matrix playtesting, including a deliberately slow Android and an old iPad |
What we are not chasing
Not a service game, not a live-ops game, not a retention product. Micro-scale means finite, complete, and rewarding to abandon. Fifteen good minutes and a screenshot is a successful session, and the design should not punish anyone for stopping there.