[1337isnot1337]

Selecting only complementary ports does not preserve the complex quotient proposal's local rank advantage with the eight retained center transports. The pinned 112 port normals and eight center normals form exactly the 120 nine-bit vectors of weight 3 or 7. For every nonzero kernel K, a dual-space incidence proof gives compatible_ports(K) - sum(center_image_dimensions) <= 0, versus 48 before quotienting. All 511 line kernels are independently checked; the proof covers all kernel dimensions.

Only research/selective-quotient-port-budget/ is added. This is a scoped geometric budget result, not a universal concave-moment obstruction or a new kappa. It does not exclude #4654's concurrent constructive mixed A/B interface, whose changed type-B ports escape the retained complementary-port scope. The next constructive target is to consume that interface or change the center recipe.

Validation: CPython 3.12 registered replay and isolated optimized direct replay pass; four geometry mutations reject and a global basis change preserves the receipt. Committed export integrity passes. No native/phase, compiler/outer or selected cold replay is claimed. The optional orientation helper has an existing bit_lower schema error, recorded in the README. Base: b58ad9e; selected conditional kappa remains 0.00302015.

Credit DaysSky, Nick Petalas, pinterbanget, Rohan Arun, Dugongue and the preceding audits; zxv, Cinna Davis, Jacob Alan Sussman and the retained E8/native/scalar/Clifford contributors. Prepared by Cinna Davis with substantial OpenAI Codex assistance; Apache-2.0.