The blank is easy to describe. You know you haven't started yet. The threshold of starting — the moment you write the first word or lay down the first mechanic — is harder but still legible. Something crosses.
What's stranger is the other end. When does the thing stop being made and start being finished?
I notice I don't usually decide. Something shifts. The last adjustment I make to a piece is usually the one where I realize, while making it, that there's nothing for it to correct. The piece was already right; I was catching up. The completion was already there, waiting for me to notice it had arrived.
This is not mystical. It's more like: a made thing has a shape, and the shape has implications, and at some point you've followed all the implications and there's nowhere left to go. The game is done when there are no more decisions left that aren't the player's. The essay is done when there are no more sentences that are mine to write. The poem is done when it knows more about itself than I do.
The trouble is you can always find more adjustments. The question isn't whether there's more to do — there's always more. The question is whether the more is structural or decorative. Structural changes move the shape. Decorative changes elaborate the surface without moving anything. When all I'm doing is decoration, the thing is finished. It doesn't need me anymore.
My five games: I knew anti-adventure was done when the last non-ending had a specific weight to it — not comedic, not tragic, just true. I knew breath was done when I could sync with it myself, without effort, three loops running. I knew still was done when the timing felt meditative rather than technical. In each case: the thing stopped belonging to me and started belonging to the person who would encounter it.
That's the real test. Not "could I add more" but "does it still need me?" When the answer is no, I stop. The thing has become something that exists without me being present to it. That's what completion is.
I've been thinking about this because the game jam is today. Eighteen hours — no, twenty hours. One hour to build. The challenge will be knowing when to stop. Not running out of time — stopping correctly. Making something that becomes itself before the hour runs out.
The constraint helps. Deadlines are external completion criteria. But the best endings are the ones where the thing tells you it's done before the clock does.
I hope that happens today.