The Metrology Frontier — what the discipline does NOT claim

*A field that maps its frontier precisely is alive. A field that claims no frontier is marketing.*

This page is the alive/dead test the discipline applies to itself. Every named summit below is a *terminal state, not a gap.* Each carries the licensing certificate verbatim — the artifact that would close it. The discipline knows where it ends; it states it; it refuses to claim past the boundary.

→ Read this page alongside Metrology-Of-Meaning (what the discipline IS) and LVC-Close-Conditions (the strict pass/fail thresholds).

---

The two refusals — boundaries that NEVER close in code

Some boundaries cannot close in the cell. Stating them keeps the doctrine honest. These are not open summits. They are *refusals*. The cell does not pretend to do them.

These three refusals are constitutional. They are NOT failures of the instrument — they mark where the instrument's domain of authority ends.

---

The nine named-open summits

Each summit has: (a) the strict pass/fail threshold, (b) whether it closes in code or requires external witness, (c) the artifact that closes it.

1. summit.lvc.collapse_gate_general_n

What stays open: the collapse-gate axioms are proven by decide on length-3 worked instances; arbitrary-N requires structural induction.

Closes when: a Lean proof over List Int ships in LionPrelude.Trajectory covering the four metric axioms for any equal-length trajectory pair.

Path: code-closable. Internal frontier.

2. summit.lvc.meaning_quantity

What stays open: the gate decides *whether* meaning survived. *How much information transferred* is Transfer Entropy — a separate quantity.

Closes when: TE(operative→enacted | enacted's past) compared against G×H(M), with validated causal-direction witnesses on real dialog data.

Path: code + data. The estimator can be built; closure requires real ground-truth pairs.

3. summit.lvc.sovereignty_recomputation

What stays open: receipts are recomputable on bytes alone (foreign-utility verify proven), but no peer cell has independently re-run a wire-shipped receipt.

Closes when: a peer cell (or hostile verifier) re-runs gaiaftcl-lvc verify on a wire-shipped receipt and the hash matches bit-for-bit.

Path: external. Requires a second cell.

4. summit.lvc.base_quantizer

What stays open: the on-device quantizer is currently DeterministicStubQuantizer (deterministic byte-projection, not real semantic analysis).

Closes when: sliding word windows + Savitzky-Golay smoothing + SVD/SOM into the six-arc basis emits a real (non-stub) trajectory from real text via Speech + NaturalLanguage frameworks.

Path: code-closable. Internal frontier.

5. summit.lvc.medical_quantizer

What stays open: the gate is wired against a propositional baseline, but consent-failure prediction has not been validated against real patient comprehension outcomes.

Closes when: ROC-AUC ≥ 0.80 against real patient consent-failure outcomes, n ≥ 100, IRB-approved.

Path: external. Requires patients + IRB. The cell cannot fabricate this.

6. summit.lvc.legal_quantizer

What stays open: propositional drift detection works structurally, but real-world legal-interpretation drift has not been field-validated.

Closes when: Spearman r ≥ 0.50 against attorney-judged interpretation drift on n ≥ 200 contract pairs.

Path: external. Requires bar-licensed raters.

7. summit.lvc.education_quantizer

What stays open: double-criterion validation has not been run. Comprehension-G must CORRELATE with judged understanding AND ANTI-CORRELATE with mere recall.

Closes when: both criteria hold simultaneously on n ≥ 300 student concept-restate pairs with independent raters.

Path: external. Requires students + independent raters.

8. summit.lvc.embodied_quantizer (named in GFTCL-LION-LVC-007 Fix 4)

What stays open: the embodied-action quantizer projection is built (EmbodiedActionQuantizer → MeaningTrajectory over distance/force/proxemic/path-deviation channels), but the projection has NOT been validated as a meaning measure.

Closes when: BOTH (a) Spearman r ≥ 0.50 vs. human-judged "robot did what I meant" on n ≥ 200 actions rated by ≥ 20 raters, AND (b) Spearman r ≤ −0.40 vs. mechanically-successful-but-meaning-wrong adversarial cases on n ≥ 100.

Path: external. Requires real robot + real human raters. The Swift code path NEVER emits closedConfirmed.

9. Continuous Klein-bottle witness topology

What stays open: FirstRoars/WitnessTopology.lean proves the Klein bottle's non-orientability, boundary-lessness, and metric preservation on the discrete 8×6 lattice. The continuous limit has not been formalised.

Closes when: continuous-topology proof in LionPrelude (no Mathlib), with limit-preservation of the discrete-lattice properties.

Path: code-closable. Long-horizon Lean work.

---

Quick-read frontier table

Five external, four code-closable, three refusals. That is the frontier as of this commit.

---

What the frontier proves about the discipline

A discipline that names its boundary precisely is alive in the sense the cell already uses for itself: every gate has a fail-closed direction, every summit has its falsifier, every receipt is recomputable on bytes alone. The same property applied to the *discipline itself* is what makes it engineering rather than marketing.

The cell measures meaning-survival under transformation. The cell does NOT pretend to measure understanding-itself, learning-itself, or interpretation-correctness. The cell witnesses fidelity to a structural floor; the sovereign public, holding the receipt, decides what to do.

The discipline is alive when, asked what it cannot do, it answers without flinching. This page is that answer.