The deciding measurements, registered in advance of their data

The claim - verbatim

ARC-Beta-k, the programme's registered measurement arm, files its verdicts before the answers. The correction-leverage exponent is measured directly from running systems, and the falsification number that separates the framework from a no-framework null is the cross-class to same-class correction-exponent ratio: the null pins that ratio at exactly 1.00, because architecture-blind averaging has no free parameter for composition class; the framework says materially above 1.00. Two predicted numbers for one measurable quantity, registered in advance of either dataset existing; the ratio itself has not been measured on real data. The earlier reciprocity test that asked for one number to be measured two independent ways and required agreement is withdrawn as vacuous before filing. If the surviving measurements later contradict the framework, the framework is withdrawn with the same prominence it was published; both component measurements stand on their own registrations. Details appear at registration.

Anchored evidence

• research/registered-programme/index.html (the registered programme: what runs next, registered before its data)
• research/data/registration-diligence-table.json (fifty-six-row diligence table; zero submitted; the withdrawn reciprocity row named)
• research/data/arc-bound-framing-question.json section twoWaysWithdrawal2026_08_12 (the withdrawal record for the reciprocity test)
• research/data/claims-ledger.json claim CEILING-DERIVATION-LAW (the cross-class ratio named as the unification carrier)

Limits and citation guidance - what this claim does not support, and how to cite it

Limits. None of these registrations has been submitted or accepted: the programme's rule is that the AI prepares drafts and only a human files. Any public sentence implying N registrations are filed is false until submissions begin, and this ledger says so first. The cross-class to same-class correction-exponent ratio has not been measured on real data; never describe it as measured. Preregistrations are drafted and awaiting human submission; never say 'filed on OSF'.