============================================================================== MVPS ADVERSARIAL-AUDIT METHODOLOGY -- INVARIANTS RECORD (The discipline that makes every MVPS draft falsifiable, traceable, and audit-survivable. Companion to draft-melegassi-irtf-mvps-methodology-00.) "A framework that cannot be broken is not mathematics. A framework that has been broken and rebuilt -- eight times -- and still stands: that is the MVPS proof." -- after the Ten Commandments ============================================================================== Mathematical assembly: Leonardo Melegassi (Catellix) Produced: 2026-05-28 Companion: docs/draft-melegassi-irtf-mvps-methodology-00.txt Law: docs/MVPS_TEN_COMMANDMENTS.txt Authority: docs/MVPS_MATHEMATICAL_EXISTENCE_PROOF_V4.txt (v4.0) Registry: docs/MVPS_IETF_FOUNDATIONS.txt Numerical receipt: scripts/validate_methodology_discipline.py (exit 0) evidence/methodology_discipline_receipt.json ============================================================================== 0. WHAT THIS DOCUMENT IS ============================================================================== This document states the MVPS methodology as a set of MACHINE-CHECKABLE META-INVARIANTS (M-1..M-9). It does NOT prove any detection theorem. It proves that the family enforces a discipline -- the same discipline that allowed v4.0 to defend 44 adversarial findings across eight audit rounds. These invariants are not aspirational ethics; they are engineering constraints checked by scripts/validate_methodology_discipline.py and recorded in a receipt, exactly like every other MVPS draft. The methodology is itself subjected to the discipline it describes: its invariants are stated, classified, and validated. ============================================================================== 0.1 ELEMENTARY DEFINITIONS ============================================================================== F-M1. CLAIM. Any assertion uttered in the name of MVPS. F-M2. CLAIM CLASS. Exactly one of: [T] Theorem -- proved from imported results + axioms, [D] Design -- explicit architectural choice (not falsifiable), [C] Conjecture -- stated with a written falsification protocol. F-M3. IMPORTED THEOREM. One of I1..I12 (v4.0 Section 0.2): the only external results MVPS is permitted to stand on. F-M4. TRACE. A finite chain of substitutions from a Theorem to one or more imported theorems I1..I12. F-M5. FALSIFICATION PROTOCOL. The 4-tuple (observable, data source, statistical test + significance, current blocker) attached to every Conjecture. F-M6. TRIO. The (proof companion, validator, numerical receipt) triple that backs a submittable draft. F-M7. AUDIT ROUND. A labelled adversarial pass that produces a set of findings; each finding is either defended (Theorem corrected) or reclassified (demoted to Design/Conjecture). ============================================================================== 1. THE META-INVARIANTS (M-1 .. M-9) ============================================================================== M-1 (EXHAUSTIVE CLASSIFICATION). Every claim maps to exactly one class in {Theorem, Design, Conjecture}. Any other label (e.g. marketing language) is rejected. CHECK: M-CLASS-1. M-2 (IMPORT COMPLETENESS). The imported-theorem registry is exactly I1..I12, each named to a published result. CHECK: M-TRACE-1. M-3 (THEOREM TRACEABILITY). Every Theorem names a trace into I1..I12. A claim that cannot trace is not a theorem. CHECK: M-TRACE-2. M-4 (NO INFLATION). A claim with no valid trace MUST NOT be admitted as a Theorem. This is the negative control that guards against Conjecture-to-Theorem inflation (the cardinal sin, Commandment IV). CHECK: M-NOINFLATE-1. M-5 (LEDGER COMPLETENESS). The adversarial-audit ledger is complete: eight rounds (K, G, H, W, S, B, L, N); 35 findings against v1.0-v3.0 plus the L-series (3) and N-series (6) = 44 total, each defended or reclassified by v4.0. CHECK: M-AUDIT-1, M-AUDIT-2. M-6 (FALSIFIABILITY). Every Conjecture carries the 4-field falsification protocol of F-M5. CHECK: M-FALSIFY-1. M-7 (TRIO PRESENCE). Every draft that declares a validator has that validator on disk; the receipt corpus is commensurate with the validator corpus. CHECK: M-TRIO-1, M-TRIO-2. M-8 (COMPANION PRESENCE). The methodology's companion documents (the Ten Commandments, this proof, the v4.0 catalogue, the foundations registry) are present. CHECK: M-DOC-1. M-9 (HONEST MATURITY). The maturity ladder Exist -> Prove -> Calibrate -> Harden -> Optimize is declared and MVPS is positioned honestly at step 3 (Calibrate); optimality language is forbidden until a step-5 dominance proof exists (Commandment X). CHECK: M-PROGRESS-1. ============================================================================== 2. THE AUDIT LEDGER (8 rounds, 44 findings) ============================================================================== ROUND SERIES FINDINGS DISPOSITION ---------------------------------------------------------------------- K,G,H, v1.0-v3.0 adversarial 35 all defended by v4.0 W,S,B (six rounds) (explicit definitions, bounded domains, distributional fixes) L cross-draft reconciliation 3 L1 -> L_DL (D-15); L2 -> L_LT.A/B; L3 -> Bonferroni demote (R-NVD) N two new-draft rounds 6 N1-N3 -> orbital L_ORB.* N4-N6 -> leadtime L_ZD.* ---------------------------------------------------------------------- TOTAL 44 validate_v4_against_ all_attacks.py = 44/44 Each finding is recorded with (i) the original attack and (ii) the mechanism in v4.0 that defends it. A defense passes when v4.0 either avoids the claim entirely (reclassifies to Design/Conjecture) or implements the algebraically corrected mathematics. ============================================================================== 3. WHY THIS IS A RESEARCH CONTRIBUTION (IRTF framing) ============================================================================== The contribution is not a new detector; it is a reproducible DISCIPLINE for measurement-security drafts: (a) a three-class claim taxonomy with a machine-checked partition, (b) a closed import basis (I1..I12) with mandatory traceability, (c) an adversarial-audit ledger with defend-or-reclassify dispositions, (d) a falsification-protocol requirement for every conjecture, (e) a proof/validator/receipt trio per draft, regenerable in CI. This is offered to the research community (IRTF) as a template that other measurement frameworks MAY adopt, independent of MVPS's specific mathematics. ============================================================================== 4. CONJECTURES (none of mathematical type) ============================================================================== The methodology introduces no empirical conjecture. Every invariant M-1..M-9 is a definitional/constructive check, validated by exit-0. ============================================================================== 5. CLOSING STATEMENT ============================================================================== What this document proves: (i) the family enforces a three-class taxonomy (M-1), (ii) every theorem traces to a closed import basis (M-2, M-3, M-4), (iii) the 44-finding audit ledger is complete and defended (M-5), (iv) every conjecture is falsifiable (M-6), (v) every draft carries its trio (M-7, M-8), (vi) maturity is stated honestly (M-9). What it does NOT do: (a) prove any MVPS detection theorem (those live in their own companions), (b) assert optimality (MVPS is at maturity step 3). The methodology is the reason the other drafts survive review: not because they are never attacked, but because every attack has a recorded disposition and every claim has a class. ============================================================================== 6. REFERENCES ============================================================================== [I-D.melegassi-irtf-mvps-methodology] Melegassi, L., "The MVPS Adversarial-Audit Methodology", draft-melegassi-irtf-mvps-methodology-00, May 2026. [MVPS-V4] Melegassi, L., "MVPS Mathematical Existence Proof v4.0", May 2026. [TEN-COMMANDMENTS] Melegassi, L., "The Ten Commandments of MVPS", docs/MVPS_TEN_COMMANDMENTS.txt, May 2026. [MVPS-FOUNDATIONS] Melegassi, L., "MVPS IETF Foundations: Theorem Traceability Table", May 2026. ============================================================================== END OF MVPS METHODOLOGY INVARIANTS RECORD ==============================================================================