the state budget, and refusing to store forever

The most thorough early design for keeping history was also the one that could not be paid for. The pruning record says it plainly: an archival scheme where old state is kept cold and revived on demand requires someone to maintain that storage in perpetuity, with no clear payer. It was dropped not because it was technically wrong but because it contradicted the rule that nothing is free.

The tension is that an append-only system implies unbounded growth, and a network intended to run for years cannot be unbounded.

The decision is to bound by a window instead of by total history. The chain operates over a sliding span of recent time; anything outside the span is discarded, so there are no archive nodes and no revival schemes to specify. Storage becomes time multiplied by activity, which an operator can reason about. Periodically, participants fold the old state and the new blocks into a fresh checkpoint, and producing that checkpoint counts as real work rather than housekeeping. Underneath sits a smaller rule with a longer shadow: an account showing no activity inside the window may be dropped, and the clock that decides this is the same timestamp that already tracks dormancy — no new field, no second clock.

What it costs is that history is genuinely not kept. Anyone wanting to query the deep past must accept that the network never promised it. What it proves is that a bound was chosen deliberately, with a named carrier, and the forgotten case designed rather than discovered.