Why small games end well

Endings are the part of a game most likely to be described as a disappointment, and the complaint clusters heavily at the long end. Two-hour games are routinely praised for finishing well. Forty-hour games are routinely forgiven for not. It is tempting to put this down to expectation, but the mechanism is more concrete than that.

A short game can still be about the thing it was about at the start. That is the whole of it, and almost every other advantage follows.

The drift problem

Any game that runs long has to keep introducing material, and introduced material accumulates. New verbs arrive, new systems arrive, new places arrive, and each one has a claim on the ending. By hour thirty the ending has to satisfy a set of promises that were made by six different design decisions across six different months, several of which contradict each other.

The short game never accrues that debt. Its ending answers one question because only one question was ever asked. When people say a small game felt “complete”, they usually mean its ending addressed the same subject as its opening, which is a much lower bar than it sounds and one that scale makes surprisingly hard to clear.

Endings are a last use of a verb

The best endings in games are not cutscenes; they are a final use of an action the player has been performing all along, arranged so that it means something different. The door you have opened four hundred times, opened once more. The resource you have hoarded, spent. This requires that the action still be legible, which requires that it has not been buried under nine other actions introduced since.

A small game keeps its verbs close to the surface, so the final use lands. A long game often has to invent a new mechanic for the finale, which is precisely why finales feel detachable — they are.

The pacing of a single sitting

A game finished in one sitting also gets to use the player's own physical arc: attention rising, then tiring. Designers of short work can rely on the last ten minutes arriving while the player is still holding everything in mind. Nothing has to be re-established. There is no need for the character who reminds you what you were doing, and no need for the recap that flattens the material into a summary of itself.

Long games have to plan for interruption, and planning for interruption means repetition, and repetition dulls the very material the ending needs sharp.

What larger work can take from this

The lesson is not “make everything short”. It is that an ending needs a subject that survived the middle. Three things help.

First, keep a written statement of the question the game asks, and check new material against it rather than against a feature list. Material that does not touch the question is a candidate for cutting even when it is good.

Second, protect one verb from the very beginning to the very end, and let nothing crowd it out. It does not have to be the flashiest; it has to be the one the ending will use.

Third, treat the last act as a compression rather than an expansion. The instinct at the end of a long project is to add — a larger space, a bigger encounter, a new system. The endings people remember tend to subtract instead: fewer options, a narrower space, a single remaining action.

A caveat about scale

None of this means long games cannot end well, and it would be a poor argument that pretended the successes do not exist. But the successes usually show the same discipline described above, applied under harder conditions. They keep one question in view for a very long time. That is an achievement of editing, not of length, and it is available to anyone willing to cut.