kernel_sees_what_posting_cannot

Gap-2 Room B is a necessary-reasons census for insertion asymmetry: recognition structure is asked to force the carrier-enlarging rate law sizeBlindBirthPerLabelDeath (one creation opportunity per tick, one deletion choice per existing label), or a counting-equivalent law $\mu(n+1)=(n+1)\lambda n$ that is not baked from a weight. The target is a search directive, never a proof premise.

D12 asserts existence of a selector on posting-rated worlds that respects present dynamics and separates those counting rates from the equal-per-slot decoy. Module status records D12 as refuted: bare posting reachability is blind to the attached rate law, so no such selector exists. That scopes D10 (existence of counting rates is derived; selection by bare dynamics is not).

D13 sharpens the close: the carrier-enlarging birth-death kernel is rate-sensitive, and the observation that the up-step weight out of size one equals one factors through the kernel, selects the counting rates, and rejects equal-per-slot. This theorem packages that wall/close contrast on the same pair of worlds.

Term-mode triple pairing. The proof is the product inhabitant ⟨D12_refuted, kernelUpObs_selects_counting, kernelUpObs_rejects_equalPerSlot⟩. First conjunct discharges $\neg$D12 from the already-banked refutation that no posting-reachability-respecting selector separates the two rated worlds. Second and third conjuncts apply the D13 kernel-observation lemmas: the up-step observation holds on the counting-rate world and fails on the equal-per-slot decoy. No further rewriting; the packaging is the content.

Feeds insertionAsymmetryReasons_certified, which packages the D10/D11 closures, scored decoys, the D12 wall, the D13 close, the D14 ledger-counted schedule, D15 canonical pinning, and D16 freshness forcing into the Gap-2 reason table. Without this contrast, the census would only show that counting rates exist (D10) and that bare posting cannot select them (D12); the kernel observation is what actually separates the asymmetric law from the equal-per-slot decoy under rate-sensitive dynamics.

In the Recognition framework this sits inside the gravity seven-gaps program: insertion asymmetry on the carrier is forced by move multiplicities tied to the tick (one creation opportunity per tick) rather than by a weight-baked schedule. It does not yet move the measure flag, and it leaves D14's ledger provenance and later freshness rows as the remaining forcing steps toward a full selection story.