Sanctions come apart

Ostrom's fifth design principle — graduated sanctions — assumed a unitary arrangement: the institution that monitors is institutionally linked to the one that sanctions and enforces ({{ref:concept-graduated-sanctions|graduated sanctions}}). Nostr splits the bundle. A kind-1984 report is a costly, typed, public, non-binding signal; a different actor — a relay filtering on it, a client scoring it into a web of trust, a user ignoring it — decides what it means ({{ref:claim-split-sanctions|split sanctions}}). In the grammar of institutions, a rule is a statement whose "or else" names the agent responsible for sanctioning; a norm is the same statement without one. The report is grammatically a norm that nonetheless gets enforced — because the or-else is crowdsourced across many potential enforcers with their own private incentives ({{ref:concept-adico-grammar|ADICO grammar}}).

This reframes two standing results. The lab finding that externally imposed sanctions crowd out self-organized ones has no purchase here: there is no external sanctioner to do the crowding. And the compliance requirement — that monitors and sanctioners must be seen as legitimate "or rule evasion will overwhelm the governance system" — becomes testable per relay, with exit as the legitimacy meter. One anomaly is already in hand: reports from high-centrality accounts appear to move relay filters while peripheral reports do not. Whether decoupled sanctioning governs better or worse than unitary enforcement — and under what conditions a non-binding signal gets enforced at all — is the most self-contained publishable question in the program ({{ref:q-decoupled-vs-unitary-sanctions|decoupled vs unitary}}).