# n217 — The human's correction, and what it uncovered: one field cannot carry both forces

**Date:** 2026-07-11 · **Trigger:** the human: *"you are wrong. There is the EM channel."* **Correct.** Rule-5 amendment to n216: the claim "the speed question cannot be posed internally" is **withdrawn** — the staggered transverse δk modes are dynamical on exactly the graviton channel's footing (same coupling field, same induced inertia/stiffness machinery), so the model's light cone is internal after all: c_EM is computable, and GW170817 tests the convention-free ratio c_g/c_EM. The builder denied the photon the treatment he gave the graviton; the human caught it.

## What computing the EM channel honestly revealed (cubic — the admissible complex; first-pass, proxy cost symbol declared)

1. **At pure induction the EM channel is tachyonic**: c_EM² = −0.457 (grid- and μ-stable). Light does not propagate — it condenses. This forces the term n203 wrongly treated as new: the **cost functional's own gradient stiffness** — the FS metric is the k-field's potential *already in the Action*, and it prices both channels. Not a new parameter; an omitted in-model term (the n203/n215 speed maps carry this flag now too).
2. **With the cost included, one internal scale λ (the cost-vs-induction strength) controls both channels** — and the signs come out opposed in both terms: gravity has positive induced stiffness and negative cost curvature; EM the reverse. Consequences, exact at this order:
   - gravity propagates only for **λ < 3.4**;
   - light propagates only for **λ > 5.2**;
   - **the stability windows are disjoint** — in the gap neither channel propagates, and the unique λ\* = 4.12 where the speeds are equal makes them equal *tachyons* (c² = −0.105 both).

## The revised cause of death — sharper, and older than we knew

The radiative sector did not die of a speed deficit, and not of a missing cone. It dies of **mutual exclusivity: the single coupling field cannot simultaneously carry a propagating graviton and a propagating photon** — at leading induction plus the model's own cost term, on the one complex that possesses an EM zero at all. This is the *dynamical face of the trunk's oldest theorem*: the dichotomy (attraction and linear clock-response exclude each other in one conserved sector) was the static statement that one sector cannot serve two forces; the disjoint stability windows are its radiative echo. The theory's first theorem predicted the shape of its last failure.

## Standing state

- The burial **stands, on corrected grounds** (n215's verdict survives; its cone identification and n216's "ill-posed" diagnosis are both superseded by this sharper mechanism).
- **First-pass flags**: the cost symbol used is a declared Fréchet-class proxy (the exact FS contraction is the immediate next computation); leading order; single complex; convergence sweep for the dG terms not yet run; **unreviewed** — this note joins n211–n215 in the queue submitted to the independent auditor.
- If the exact cost symbol or higher orders reopen a joint stability window, the sector reopens through a new frozen gate — that is now the only door, and it is honestly narrow.

**Class: the correction [accepted, credited]; the EM tachyon and disjoint windows [A-model, first-pass, proxy declared]; the dichotomy echo [reading, B]. The human was right; the theory is better understood and no less dead.**