Published live (2026-07-02). The ratchet-cell treatment seeded four candidate claims. Asking "which deserve to enter the canon?" produced a correction worth more than the answer: the notion of a canon is itself thinking inside the box. A canon has a wall, and a wall implies scarce shelf space that a living epistemic work does not have. So, first, the rule — then all four candidates, captured, each with its sharpest objection attached.
The capture rule
The threshold for capture is existence, not merit. Merit is decided by the climb, in public.
The grounding is already in this work. asymmetry-of-option-space-change: loss is fast and irreversible, growth is slow — an uncaptured idea is gone, while a weak captured one just sits at forming forever, costing nothing and never becoming load-bearing. The settledness spine already does the selection structurally: entry at forming is free; climbing is expensive and public. Rationing entry duplicates, badly, a filter the architecture already provides — curatorial selection at the door versus structural selection on the ladder. This is the DNA arrangement: mutation capture is promiscuous; selection prunes later. The pile is not the firehose problem — the firehose problem was a pile with no structural selector, and this work has one.
Consequence for vocabulary: this work does not have a canon. It has a spine with an entry ramp.
Candidate 1 — the bridge has a catchment
The bridge test went into its first adversarial run as a binary and came out as a gradient: the ratchet cell is bridge-clean for the median adopter, coordination-light above rung 4, and honestly false for the fuel-poverty cohort. That is the test discovering its true form: a bridge has a catchment, and the real questions are who does the bridge reach and what carries the rest — with the remainder priced (a ring-fenced surplus, funded outreach) rather than moralised. The design task at a catchment's edge is widening who the bridge reaches — which connects this candidate to moral-circle expansion as option-space growth in the inclusion domain.
Sharpest objection: gradient-ising the test risks dissolving it — "every design passes for someone" would make the gate unfailable. The discipline that must survive: a design whose median adopter needs forcing still fails, full stop; the catchment refinement applies only past that bar. Climbs when: a second gated treatment shows the catchment framing discriminating (some design failing at the median while another passes with a priced remainder).
Candidate 2 — institutions precipitate
The ratchet cell's Community Benefit Society materialises at rung 4 as the wrapper the accumulated trades legally require — governance after value, ownership as residue. The candidate generalisation: durable commons crystallise out of trades; they are not erected ahead of them. The constitutional-moment-first pattern front-loads exactly the founding coordination the bridge test forbids, which may be why so many designed commons die at the founding meeting. Bethesda's own history is the existence proof read backwards: the entity that exists arrived when the money made it necessary.
Sharpest objection: survivorship and scope. Some institutions plausibly do need founding acts — standards bodies, defence pacts, anything whose first unit of value requires the whole. And Ostrom's cases include real constitutional moments that worked. The candidate may be domain-bounded: true where value accretes in individually-rational increments, false where it doesn't. Climbs when: tested against non-energy domains, including a genuine counterexample hunt.
Candidate 3 — find the waitlist
A standing lens instruction. A closed club with a queue is the bridge test's signature in the wild: demand that pre-aggregated with no campaign, blocked only by supply — the world announcing that it has already run the adoption experiment, and pointing precisely at the binding constraint (in Bethesda: generation, not persuasion). Queues, oversubscription, closed lists — cheaper than any analysis, because the evidence is behavioural and pre-existing. The lens should scan for waitlists before it computes anything.
Sharpest objection: waitlists lie. Scarcity marketing manufactures queues; subsidised prices generate queues that measure mispricing, not bridges; and a queue for a positional good measures rivalry, not spontaneous adoption of a better configuration. The discipline: the signature counts only where the queue is unsubsidised, unexhorted, and for a non-positional good. Climbs when: the instruction is run across a few domains and its false positives are catalogued.
Candidate 4 — the closing window
Across two full runs now, the residue that survives every dissolution is the same shape: not a wall but a clock. The mobility run bottomed out on "will the commons form before enclosure?"; the energy run on "will the exemption window stay open long enough for cells to become undeniable?" Candidate: the general form of the political residue is a race between spontaneous accretion and boundary-redrawing — and the corresponding doctrine is to minimise the permission surface: every step small, self-assessed, with no event at which anyone can say no. The route-around beats the chokepoint not by argument but by having no single point of refusal and a faster clock. Portugal is the control: remove the legal boundary but add a licensing queue, and the chokepoint relocates.
Sharpest objection: two data points; and the doctrine could licence recklessness — "move fast before the rules arrive" is also the extractive playbook. The line that must be drawn: routing around rent is not routing around protection, and a design that minimises its permission surface to escape accountability rather than gatekeeping fails the viable-objective gate regardless of its clock. Climbs when: a third domain's residue is (or is not) a race, and the rent/protection line is stated sharply enough to gate on.
All four enter at forming, per the capture rule above. None is load-bearing; each carries its objection on its face; the climb decides. Not yet in propositions.ts — promotion there is a climb event, not a capture event.