Perception · W4 — Critical-3 Resolved

W4, extended: 2026-04-19 to 2026-04-26.
Canonical bump: v2.1 → v2.3. Critical open problems: 3 → 0.
The short version: a problem that had looked like a missing mechanism turned out to be a category error between static energy, dynamic endpoints, and the layer where object-count is even defined.

Start here#

This was the week SCC stopped being blocked by the old F-1 / M-1 / MO-1 knot.

• F-1, M-1, and MO-1 all closed on 2026-04-24. Their final statuses are SPLIT-RESOLVED, LAYER-CLARIFIED, and SIDESTEPPED.
• Four Cat A entries joined canonical. T-PreObj-1, T-PreObj-1G, Lemma 4, and T-V5b-T.
• The theory gained two organizing commitments. CN15 Static / Dynamic Separation, and the σ-framework for Hessian / irrep / nodal-count signatures.
• The release path reopened. The v2.0 blocker had been asking why a static K=1 minimum and a dynamic K greater than 1 endpoint disagree. W4 made explicit that they were never the same object.

The week is best read as a resolution story, not as a theorem dump.

What changed#

The one-line takeaway: SCC's pre-objective character is now theorem-backed, not just an ontological preference.

The week as a narrative#

1. The old question broke#

The week began with the same question that had stalled the project for months:

If pure boundary morphology prefers K=1, how can SCC honestly explain K greater than 1?

W4's first move was not to answer that question. It was to notice that the question had been splicing together incompatible layers. The theory's ontology is continuous: the primitive is the graded cohesion field. But the operational language had repeatedly jumped to discrete counters: hard K, thresholded formations, axiom branches.

That mismatch became N-1: Soft-Hard Switching Asymmetry.

2. The search space got exhausted#

Day 2 evaluated eight candidate framings against two matrices: which open problems they resolved, and which Cat A theorems survived. The Pareto frontier looked respectable: B, B+C, E, and C+E.

It was still not the answer.

That was useful. The matrices made the hidden assumption visible: every candidate was still trying to reconcile a static energy statement with a dynamic protocol observation. W4's real route, later named Option D, was outside that candidate space.

3. The layer split became explicit#

The K-soft and thermal-free-energy work on Day 3 moved the Critical-3 cluster. F-1, M-1, and MO-1 stopped looking like missing theorems and started looking like poorly separated layers:

• pure boundary morphology,
• full SCC energy,
• dynamic protocol endpoints,
• single-formation smooth manifolds,
• multi-formation cornered manifolds.

Once these were separated, the old contradiction weakened.

4. The theorem arrived#

Day 6 was the pivot. The Pre-Objective Mechanism cluster established that the F=1 single-disk candidate is not a critical point under full SCC. In plain terms: once closure and separation are both active, the theory does not sit at the single-object analogue. It leaves toward a multi-peak configuration.

This is the mathematical form of the pre-objective claim.

5. Verification changed the final theorem#

The extended day, 2026-04-26, was supposed to be routine V5b confirmation. It was not. A mode-indexing artifact appeared in the Goldstone analysis: near-degenerate modes could silently swap labels.

The result was not thrown out. It was re-verified mode-agnostically.

That split V5b into:

• T-V5b-T, Cat A: translation-invariant graphs, sub/super-lattice spectral dichotomy, 2D Goldstone doublet, 1D Goldstone branch, universal nodal count 2.
• V5b-F, Cat C: partial Goldstone behavior on boundary-modified graphs, carried forward as NQ-173.

Compact daily log#

The new Cat A entries#

Pre-Objective Mechanism cluster#

Lemma 4 is the technical hinge. At the F=1 disk, the pure boundary gradient vanishes. The closure and separation gradients do not. Their positive-definite inner-product matrix gives a strictly positive destabilization direction for every nonzero coupling vector.

Pre-Objective Goldstone#

T-PreObj-1 says F≥2 is the attractor. T-V5b-T asks what soft modes appear once a multi-peak formation breaks translation symmetry.

The answer:

• On a 2D torus, there is a 2-fold Goldstone doublet with commensurability splitting.
• On a 1D cycle, there is a 1-fold Goldstone branch.
• Across tested translation-invariant graph classes, the Goldstone nodal count is 2 universal.

How the Critical-3 closed#

F-1: split-resolved#

Original worry. K=2 looked vacuous because, without externally fixing each formation's mass, energy minimization squeezes one formation to zero.

Resolution. F-1 splits into two correct statements:

• On pure $\mathcal{E}_{\mathrm{bd}}$, K=1 really is cheaper. This is T-Merge (b), already Cat A.
• Under full SCC, the F=1 single-disk is not a critical point. This is T-PreObj-1, new Cat A.

The two statements live on different energies. They do not contradict each other.

M-1: layer-clarified#

Original worry. K=1 is energetically preferred over K=2 on the fixed-mass boundary landscape.

Resolution. That statement is true, but it belongs to the static pure-boundary layer. It does not determine the full-SCC dynamic endpoint.

This is CN15 Static / Dynamic Separation:

• Static: global minimum on pure $\mathcal{E}_{\mathrm{bd}}$.
• Dynamic: protocol endpoint under full SCC gradient flow.

The apparent paradox came from demanding that these two objects agree.

MO-1: sidestepped#

Original worry. Multi-formation manifolds have corners, so smooth Morse theory does not apply directly.

Resolution for W4. The W4 results operate on the smooth single-formation manifold $\Sigma_m$, where the new σ-framework is well-posed. The current theorem path does not need Morse theory on $\Sigma_M^K$.

This is not a permanent solution. Multi-formation σ work will re-open the stratified-Morse issue.

What did not work#

• The first V5b variants were not stable enough. V4 had a sign issue; V5a had the mode-indexing vulnerability later caught by NQ-172.
• The candidate matrix did not contain the final answer. It helped expose the hidden premise, but Option D was not in the candidate list.
• P-Unified-1 stayed falsified. $\Lambda_{\mathrm{coupling}}$ is useful as a structural classifier, but not as a monotone predictor of Persist degradation.

These failures matter because they show what was not silently promoted.

Verification note: NQ-172#

The V5b result was almost accepted too early.

The artifact: mode labels were assigned by eigenvalue sort order. Near degeneracy, two modes could swap labels while the raw spectrum still looked correct. That creates a dangerous error: the numerical output is right, but the interpretation is wrong.

The repair was to stop identifying the Goldstone mode by label. V5b'' measures the low-frequency subspace and nodal invariants in a mode-agnostic way. This is what made T-V5b-T canonical-ready.

The lesson is straightforward: verification is a discovery channel. W4 spent two days on verification, and those two days changed the theorem.

Carry to W5+#

Canonical impact#

Canonical pages touched:

• Canonical Spec — Part 5 · Results Registry
• Canonical Spec — Part 4 · Interpretation
• SCC Theorem Registry
• SCC Status (April 2026)
• SCC Glossary

Pointers#

• Full theorem registry: Canonical Spec v2.3 · Part 5
• Resolution details: Canonical Spec Part 4 · W4 Resolution Banner
• Status overview: SCC Status (April 2026)
• Curated theorem guide: SCC Hero Theorems

Aligned with Perception_theory canonical CV-1.4 (2026-04-26). The v2.0 release path is unblocked after a year.