The BAO standard ruler is only as good as its calibration. DESI fits r_d near 147 +/- 1 Mpc and Planck's ΛCDM chain infers 147.09 +/- 0.26; the agreement looks like closure but is model-loaded, because the CMB-side number is not a measurement of the plasma, it is an inference through the full ΛCDM parameter chain. That chain-calibrated ruler is what holds BAO to the low H₀ and quietly powers the Hubble tension, and every early-time patch proposed to shorten it imports tuned new physics and pays in the damping tail.
Planck does not measure r_d; it infers the ruler through the full ΛCDM parameter chain, damping tail and matter density and peak structure all weighed by the model at once. The assumption is that the chain is the only way to compute the number: remove ΛCDM and the calibration loses its floor, so any strain between tracers or between H₀ anchors has nowhere to go except systematics or tuned early-time new physics.
SCT leaves the computed step alone. By the Plasma Equivalence Theorem the collision-thermalized plasma reaches recombination in the thermodynamic state the standard equations describe, so the sound speed keeps its textbook form, c_s² = c²/[3(1+R_b)], and the drag horizon follows from ω_b and ω_m alone, independent of H₀ (P29, P30). Evaluated that way the ruler lands at r_d = 146.8 +/- 5 Mpc, within 1σ of DESI's measured 147 +/- 1 and consistent with Planck's chain-inferred 147.09 +/- 0.26. The point is not the agreement but its price: Planck's number is bought with the entire ΛCDM chain, while SCT reaches the same ruler from ω_b and ω_m with standard acoustics and no chain, so the ruler stops being hostage to the model that carries it.
An earlier SCT reading ran this tension through a modified acoustic medium, inserting the coherent (CAR) sound speed c_s² = c²(1+R_b)/3 at recombination; the registered ledger now retires that step as a category error. Coherence physics belongs to the late-time sector, where the same cascade-derived baryon loading R_b = 0.2545 +/- 0.032 (0.17σ from observation) sets the coherence floor 1 + R_b/3 and with it S₈ = 0.783 with zero free parameters. The collision signature lives downstream of recombination, not inside the ruler.
PROVISIONAL: r_d = 146.8 +/- 5 Mpc awaits independent third-party verification of the Boltzmann pipeline. The earlier CAR-at-recombination outputs (149.1 Mpc and the 161 to 178 Mpc analytic integrals) are superseded, and the 28 Mpc gap the old ledger carried is resolved as the category error itself. The number is a published prediction, not a settled result.
The registered kill: DESI Year 5 measuring r_d above 150.5 or below 145.0 Mpc at 3σ falsifies the standard-horizon value; within the band, the decisive test is the verification itself, since an independent Boltzmann run failing to reproduce 146.8 from the full solver would retire the provisional number. The companion derivation shares jeopardy: observed R_b outside 0.158 to 0.350 breaks the cascade-geometry chain behind the late-time sector.