The compact is sold as status if and only if verification binds. Boyle's demonstration and the Newton–Leibniz priority fight show a public demo and a race to publish, not status withheld when a repeat failed. The decay story needs a working period where the "only if" side fired.

Watching the candle go out is the check. The room could build pumps. That is verification, just cheap. A modern replication protocol is not required for the bargain to have been real: status for being first to something others could see, with the race as incentive and the room as constraint.

Boyle succeeding in the room is the "if" side. The compact is sold as "only if." Without a public claim that cost status because it failed the repeat, "constrained by verification" is a motto reconstruction, not a showing that the constraint bound.

The inversion table is public rhetoric (show me versus trust the experts). The liturgy section is journal practice (review, citations, p-values, missing replications). Labs can still refuse each other's word while press conferences say "trust the science," and journals can be liturgical while the slogan is wartime PR. Those are two clocks, not one inversion.

Peer review's original job was scarce-page filtering ("worth the paper"), not replication. The 36% figure is then used as proof the filter failed because review checked publishability, not replicability — which is also what "worth printing" was. The crisis does not show review used to verify and became ritual. It shows review never did the compact's verification job.

Peer review and replication are separate rows: filter versus actual check. The unpaid-repeat problem matches 36%. Reviewers protecting a paradigm is a different complaint. One liturgy slide treats them as the same decay.

The repair that matches the crisis is paying for repeats — the bounty, maybe a market on whether a result will replicate. Reforming review or pre-registering the original claim is a different machine. If the verification row died, the IX list should be scored by whether it makes a repeat happen, not by whether it makes the first pass more honest.

Even if review was never the repeat, "worth printing" becoming "does this threaten my field" is still decay of the editorial filter. Verification incentives (repeats unpaid) and capture of the page-budget (rivals protecting a paradigm) are two clocks. They would not have the same fix.

If physics still holds because reality grades harshly, and the replication crisis sits where nothing forces a no, the architecture is not what maintains truth in the field that still works. The compact can have decayed everywhere while the explosion did not. Calling prediction markets and pre-registration a reboot of 1660 identifies the wrong machine: soft fields need a substitute veto, not a restored motto.

The explosion only helps if status points at it. Without that channel, wreckage still happens and prestige can go to the elegant theory that has not flown. The compact is that pointing. Soft fields need something that points the same way, which can still be a reboot of the bargain even if nature will not grade.

Section VI's break is also cost: claims got cheap, checks got expensive. Pre-registration makes the first claim harder to fake; it does not make the check cheaper. A bounty still pays for the twenty-year study. A market can price a replication without running it. If the compact broke because checking costs more than claiming, IX mostly attacks the cheap-claim side, which is not restoring a world where the check was cheap.

Registered Reports accept methods before results and publish either way. That kills publication bias and pays status for a sound protocol, not for being right. The original compact paid for being checked and right. Publishing the null on purpose is a different bargain. What would show a veto is back in a field where nature will not grade inside a career is a market or bounty that moves hiring, not a pile of honest papers that still cannot be wrong yet.