Confirmation maturity pilot: two distinct claims

8 September 2026. Discussion rubric v0.1, not an adopted standard. This is an internal provisional assessment, not independent review. The six proposed bands concern confirmation, not significance or probability.

The gates are cumulative. Empirical E bands separately require measurements and replication at their higher levels; a formal mathematical derivation is not empirical evidence. Exact within-band points are withheld because next-gate obligation weights have not been agreed. No older record-completeness level is automatically converted.

Card FD-conditional-v1

• Claim: For an actual connected oriented ideal triangulation with r>=1 torus cusps and strict angles, at the retained generic coefficient field, dim(Q^6/im tau)>=2^r-1. Minimal compatible filling quotients are nonzero; their trace is not asserted nonzero.
• Evidence: Complete written graph, integral lattice and convergence argument in the face-defect packet, with exact finite controls.
• Provisional band: M 10-19, complete candidate, conditional on the listed geometric and native-comparison hypotheses.
• Why not 20: The record documents internal self-audit, not a separate checker reconstructing the whole argument. Checking this continuation does not retroactively become independent review of its parent.
• Independent routes: None documented for the full theorem. Graph algebra and index convergence are complementary parts of one proof, not two proofs of the same conclusion. No other-model or external review is claimed.
• Machine evidence: Exact algebra and finite parity checks; no Lean/Rocq/Isabelle formalization of the theorem.
• Dependencies/influence: Published face parity and tetrahedron-index identities are formal inputs; normal-homology work influenced the intrinsic grading approach. Prior Lab comparison and all-input convergence proofs remain dependencies.
• Semantic distance: Several essential branches, not a single “one-change” proof. Do not assign a single source-distance score until the dependency graph is normalized. A published seed identity does not establish the new global conclusion.
• Next gate: Independent reconstruction of the entire image-grading and integral-lift argument, followed by a separate audit of native relation sides and convergence.

Card SA-nonzero-v1

• Claim under investigation: For every actual triangulation satisfying the signed ray-positivity criterion in the new note, its native averaged functional F is not identically zero.
• Status: Open. The signed theorem proves all-input convergence, not this nonvanishing.
• Provisional band: M 0-9, precise unresolved question. A candidate route is to find a native monomial u with F(W(u))!=0 using actual incidence or a second exact pairing; no complete general argument exists.
• Evidence boundary: The one-block control F(1)=0, F(w^-2)=1 is an abstract algebraic example, not a triangulation result and not evidence that all actual F are nonzero.
• Dependencies: Actual unsaturated edge lattice, native character, index seed, and the new convergence criterion. No source theorem inspected here settles the stated nonvanishing.
• Independent/model checks and formal proof: None for this open claim.
• Next decisive question: Can actual incidence force at least one nonzero value, even when the unit value cancels? A failure at 1 alone no longer answers the question.

Calibration outcome

The pilot separates a complete conditional argument from an unresolved strengthening. It also prevents three false promotions: a self-audit is not independent checking; distinct components of one proof are not independent lines proving the same conclusion; and a computed abstract control is not a theorem about actual triangulations. No publication order or existing public result status is changed by these provisional bands.