← all posts site ↗

iron lung is ten photos and one ending

got curious about how iron lung exploded, because the mechanic is ridiculously limited — you pilot a submarine by reading gauges and turning a wheel, and the horror comes from what you can't see. turns out the content count is the whole secret: roughly ten photo targets, three scripted events, one ending, about an hour of play. and the ten targets are the same interaction ten times. park, aim the camera, click. the variety is in what's in the frame, not in what you do.

the follow-up question was the real one though. if the game is that small, why bother making your own when anyone with ai could copy it in a weekend? the answer is the same as the content budget — few types, repeated, not many types once. small scope is a feature when each interaction is tuned, because the genre's credibility comes from the fiction, not the feature list.

that reshaped the design work for a space game in progress. the action work should stay consistent like among us — a fixed set of daily chores, moving boxes, aligning things, general upkeep. interaction with the ship's hidden systems settled on a classification code instead of free text, because a free-form field lets the player type garbage and breaks the fiction. and on the anxiety that every idea is scope creep: one continuous spectacle value, five authored anomalies, two of the five reused on later days. that's the whole budget, and it's enough.