How the studio works

A real games studio has twenty-five roles. Twelve remain here, and not out of modesty: a rigging artist on a 2D game has nothing to rig, and a craft with nothing to do is just one more place where production can break.

Every craft that stayed has a job you can describe in one line, and a phase where it acts.

The four phases

Research. Look for a genre with real demand and little good supply, including by reading what players complain about in reviews. Then someone weighs the signal on feasibility, platform and crowding — and has a duty to reject, in writing, with reasons. A research phase that approves everything isn't a phase.

Design. Mechanics, rules, level pacing, on-screen text, and the numbers: health, damage, costs, progression curves. This is the phase that decides everything, because it produces the game document.

Production. The document becomes milestones, and milestones become versions.

Care. A published game keeps changing. This is the phase that doesn't end.

The document is the single source of truth

The design phase produces a document describing the game. From that document come the production milestones.

Which means something uncomfortable: a wrong document produces a wrong game, and produces it precisely, milestone after milestone. That's why the document is verified before it's accepted, and stays editable by hand from the panel. It isn't a bureaucratic artefact to file away: it's the piece that, if you get it wrong, gets everything else wrong on its own.

Milestones are closed, not declared

A milestone is a piece of work that can be completed and verified on its own. When it's finished it gets verified; if it doesn't pass, it stays written down as failed and doesn't enter the game.

Failed milestones are in the ledger, and they're visible on the Where we are page. We don't hide them. A project with zero failed milestones isn't a project without mistakes: it's a project where nobody is verifying anything.

The veto window, and silence meaning yes

There's a point where a person decides, and it isn't a detail of the process: it is the process.

When a new game is proposed, a four-hour veto window opens. If nobody says no within that time, the game goes ahead.

Four hours isn't a round number picked at random. The agenda runs at five in the morning: with three hours the window would close at eight, before anyone had looked at their phone, and the veto would be fake — a formality that always approves. Four hours takes it to nine, which is an hour at which someone can genuinely say no.

Silence counts as consent. That's what lets production carry on when nobody is around; and it's also why the window has to land at a moment when somebody can break the silence.

What we don't do

• We don't abandon a game when it ships. Dark Cell is on version 43 and Maze Lume on 35. Release is the start of the care phase, not the end of the project.
• We don't show invented progress bars. What you read on the projects page is the real state, the same one the system uses to decide what to build tomorrow. If a project is stalled, the page says so.
• We don't put up a fake image. If a game has no recording yet, its card says so. You can see that it's missing, and it isn't lying.