Ethical Ontology, Volume III: Gigantomachia peri tes ousias - Controlling Plan V15.4, Governance and Deposit Protocol V2, Deep Research Prompt Register V3, Universal Loss-Condition Register V2.2 (unsealed)

Record A, controlling architecture V15.4 - restricted deposit. The controlling architecture of Ethical Ontology, Volume III, with its execution protocol, research register, and the unsealed master loss-condition register, deposited before the research programme runs. Files are closed; metadata, timestamp and hashes are public. This version carries the Book Plan from V15.1 to V15.4 through three repair releases, none of which changes the architecture. The nine Parts, thirty-nine chapters, both Stations, three theorem legs, eleven audits, four ratification statuses, twenty-eight tailored opposition ledgers, Threshold distribution and eleven-line self-test are carried unchanged throughout. V15.2 restored the Chapter 3 referee-bout perimeter from temporal-deictic access through Dasein alone to the persisting unifying methodological office across Heidegger's whole path - a widening of what the doctrine must defeat, and therefore ratchet-permitted. It repaired a WHO-LOCUS defect in the Chapter 38 self-audit, where a locus had been made to answer for a vacancy that only a party, institution or practice can answer for. It received the Heidegger exposure event: the opened works are E1 or E2, may defeat or restrict, and confer no confirming power; four falsification tests returned non-defeat and, under the no-pooling rule, aggregate to nothing; one prediction was defeated and is permanently retired. V15.3 propagated the restored perimeter through the registries, fixed Chapter 27 as an unscored deployment of the one scored Heidegger bout rather than a second verdict, repaired the Actus Station's mode-of-containment field so that discovery-side material performs no argumentative office, and installed a vacancy-clause selector barring any Conclusion form from preselecting the seat stays empty before Report 26 has earned its criterion. V15.4 harmonises the self-audit with that selector. V15.3 convicted the volume wherever kept-empty answerability could not be shown, while the selector separately provided for no criterion earned - a result the loss-condition register states does not automatically defeat exhibition or every theorem leg. That row was unreachable, because the audit convicted first. The audit now asks the ordinary archic question, and the two-vacancies criterion becomes a further typing of non-occupation rather than a precondition of ordinary acquittal. This removes one route to self-conviction, and the ratchet argument for admitting it is printed in the release itself rather than left implicit: the removed route arose from a collision of two rules rather than by design, and the volume is now barred from describing its threshold at all unless the criterion is earned - a bar the predecessor did not carry. No other loss condition is relaxed. V15.4 also adds a non-bout risk crosswalk, so that six load-bearing theses which are not adversarial bouts cannot escape risk merely by being architectural. The paired Universal Loss-Condition Register V2.2 is included and is unsealed and non-operative. It is the master from which the per-Part B.n instruments will be cut after Report 16 D0; its deposit here is a timestamp on commitments made before the evidence, not a seal. It adds a target-integrity rule distinguishing a genuine opponent victory from the discovery that a planned opponent was never source-grounded at the claimed width - a distinction that prevents a framing failure from being printed as a victory over an opponent who was never there. Nothing in this deposit is evidence. Everything in it is architecture and printed risk. The credibility of the volume is the demonstrable order of its timestamps.