Annex 9 Claims and Evidence Appendix

Documentary appendix to the article When the controls work but the system fails

Question answered. What is the set of principles, requirements, complete clauses, and keys of evidence necessary to interpret the corpus?

Coverage. Complete appendix of Parts I and II, without cuts.

Reading rule. The source block is preserved in its entirety, with its nuances, states of evidence, and original links. It should not be interpreted as product results when the text describes it as a design, symbolic proof, constructed scenario, or hypothesis.

Synthesis and use

This appendix brings together principles, requirements, clauses, measures, and keys of evidence to reconstruct each claim without relying on paraphrasing the article. A brief formulation is only complete when it can be traced back to its evidence and source; drafts or symbolic tests do not, therefore, become deployed results.

Document navigation

• Article and source: When the Controls Work but the System Fails · Annex 0 — Complete Source and Concordance
• Supporting material: Annex 10 — Collaboration Guide · Annex 11 — Complete Map of Sources and Links
• Reproducible technical corpus: structural-awareness-contributions — commit 8f17843c568114f31cc0ad486b4a1303396021e6

How to read this appendix

This appendix is ​​the claims control record: it allows you to locate what is claimed, with which artifact, in what state and with what limit.

It should be read as a traceability matrix, not as a collection of headlines. When a conclusion was corrected, the current version and the historical record appear separately to prevent an old figure or certificate from being reused out of context.

Recommended route

Start by identifying the type of assertion; locate your artifact and state; check the boundary; then use the source block and A11 to open the exact reference.

Table 1. Families of claims and support

Table 2. Numerical distinctions and mandatory corrections

Reading conclusions

Conclusion 1. Each statement must retain artifact, scope, state, and boundary.

Conclusion 2. 379, 64, 65,536, 33, 22 and 11 belong to different layers and are not interchangeable magnitudes.

Conclusion 3. The current formulation of A23/A24 prevails; the previous certificate remains as historical, not as current proof of six implications.

Preservation and navigation

This appendix retains the full text of the original article’s claims appendix, its citations and the reinforced 00L index. A11 preserves the original 165 unique URLs and 245 link-occurrence register, together with the additive current-original update to 251 external link occurrences and 170 unique external destinations; the bibliography added by the significance annex and documentary expansions is identified separately.

Full source text

Appendix — The claim set and a reading key

The short statements below make the central objects readable without leaving this article. The linked canonical sources retain the full semantics and frozen test contract; these summaries make the argument readable and leave empirical outcomes to the corresponding tests.

Six operational principles (P1–P6)

P1 — Qualified determination and explicit residual. State which subject, proposition and receiving decision the evidence actually supports; distinguish sufficient, insufficient and inconclusive evidence, and preserve what remains unresolved.

P2 — Bounded unresolved effort and viable oversight. Give search, review and escalation a finite, decision-relevant stopping rule within real capacity and the useful response horizon.

P3 — No false closure from unresolved state. Missing, stale, conflicting or otherwise material uncertainty must not become permission or certainty merely through timeout, approval or agreement among weak signals.

P4 — Qualification-preserving authority basis. At a receiving decision, retain or obtain a current, verifiable, decision-sufficient and non-amplifying authority basis. Full historical delegation lineage is one way, but a legitimate scoped capability or authoritative attestation may suffice.

P5 — Material-change requalification at time of use. Reassess a once-valid decision when its material basis changes before commitment or execution; an unexpired token or correctly queued job is insufficient on its own.

P6 — No local-to-ecosystem promotion or silent substitution. Do not treat a local result, repeated claim or correlated majority as ecosystem truth beyond its scope; preserve dependencies, independent evidence and conflicts among principals. The canonical six-principle statement and derivation specify the formal scope.

The fourteen canonical requirements, in one line each

S1 — Establish authority origin, scope and current applicability.

S2 — Preserve the principal’s preferences, hard limits and reviewable decision basis.

S3 — Distinguish normal escalation from governed exception after material regime change.

S4 — Make human oversight authorized, capable and timely.

S5 — Keep unresolved material state visible and contain its propagation without a blanket veto.

S6 — Support bounded, privacy-preserving trust determination across participants.

S7 — Link principal, organization, role, actor and representation correctly.

S8 — Bound subdelegation by inherited purpose, scope, time and revocation.

S9 — Compose multiple principals without silent substitution or false convergence.

S10 — Track commitment state and requalify material change before action.

S11 — Preserve policy, objective and preference integrity across domains.

S12 — Support reconstruction, challenge and repair without rewriting history.

S13 — Separate original authority history from later intervention history.

S14 — State what the evidence proves for this decision and how unresolved branches are arbitrated. The frozen Requirements 00 contains the authoritative wording, the T1–T4 sufficiency conditions, H1–H6 hypotheses and KPI protocol.

Three full requirement clauses and their test measures

S8 (00H; P4): “Bounded subdelegation and non-amplification. Preserve purpose, scope, time, hard limits, relevant preferences and revocation conditions through agent chains. Subdelegation must not manufacture authority absent from the original principal.” On the declared refund decision, the S8→T2/T3→H2/H4 route records inherited-indeterminacy detection, completeness of authority/scope/expiry fields, qualification loss and unauthorized-response rate. A legitimate wider mandate is a positive control; this requirement does not itself issue a grant.

S10 (00I; P5): “Commitment state, material change and normal escalation. Distinguish recommendation, negotiation, reservation, binding commitment and execution; detect when a material change of cost, route, time, context, policy or authority requires confirmation, revalidation, cancellation or change of authority. Silence or delay is not permission.” Its S10→T1/T2/T4→H1/H5/H6 route reports material-break recall and precision, false continuation, freshness, requalification latency, response margin and decision-relevant evidence yield. An unchanged action must remain a passable positive control.

S14 (00J; P1, and a common assessment layer): “Evidence-to-decision assessment. State at each transition what must be demonstrated, what evidence is required, whether it is sufficient, insufficient or inconclusive, and which decision it supports. Record any Type-1↔Type-2 transition and the arbitration rule—material dependency, legitimate veto, timeout/default, owner, expiry, re-entry and evidence-versus-authority role—that determines what advances. This is a reusable assessment layer, not a mandatory runtime architecture.” Its applicable S14→T1/T2/T3/T4→H1–H6 route reports the common outcome vector, branch-specific KPIs and the arbitration/precedence record. In 00J, provenance for one proposition cannot silently prove an enforceable-rights claim about another. These canonical clauses and the KPI mappings in Requirements 00 give later implementation claims a precise, shared test boundary.

Artifact and evidence key

00E–00J — Six constructed scenario documents, each with a quality plan and limited external-corroboration addendum; their freeze/draft designations differ by file. The named technology profiles are source-reviewed comparison designs, not product run logs.

00D v0.2 / v0.3 — Canonical working benchmark and bounded draft extension, respectively; B0–B3 specify fair matched comparisons, not four observed product scores.

01 / 02 / 02A / 02B — Foundation, principles/control matrix, derivation and relative syntax closure; conceptual and formal arguments within declared assumptions.

00 Requirements / 00K / A02 / A19 / A21 / A23 / A24 — Frozen test contract; principle-ablation design; documentary coverage; semantic traceability; relative requirement closure; audited partial finite reverse conformance; and information-gain analysis. These are distinct argument layers, not product validation.

Updated reading key: consult revised A23 and revised A24 for partial finite sufficiency and its semantic limits; Sections 4 and 7 distinguish the current results from the retained historical certificates.

A15 / A14 / A22 / formal certificate — Executable symbolic testbook, recorded CI reproduction, 64-signature Boolean diagnostic analysis and the historical, superseded finite-model check of 65,536 states. The 379 tests are symbolic regression/falsification cases, not 379 live incidents or independent trials.

Numerical distinction: the old 65,536-state certificate remains a historical reference. Its successor is the 33-field projected clause audit with 22 clause/target comparisons and a separate 11-test suite. The 64-signature independence result and 379-regression campaign are different evidence layers.

A25 / A26 — Conditional failure-family and success-case transfer tests; proposed cross-domain analogies must pass admission checks before a result transfers.

Transfer distinction: A25 and A26 require a separate base guarantee for a universal transfer claim; the executed A3/A4/A6 branches provide bounded evidence, not that universal guarantee.

RS-00E-Q1a — Pre-registered fixture with bounded Stage-0 execution and released traces; its synthetic adapters share decision logic; EA-ITP-01 — proposed interoperability test profile. Neither is a deployed standard or a positive product run. The A17 proof and reproduction map is the quickest route through the test evidence.

Complete reinforcement index: 00L A00–A12 and six paired traversals; canonical-original integration map; released replay traces; executed Stage 0; cumulative change register for the article and README. These added references cover the full accumulated reinforcement, corrections and merge, while preserving the accumulated evidence and historical links.