Planck 2018 measures the primordial tilt at n_s = 0.9649 +/- 0.0042; ACT DR6 measures 0.9743 +/- 0.0034, just over 2σ higher, and joint fits with DESI BAO push toward 0.974 to 0.977. The two flagship CMB experiments disagree on the cornerstone parameter of early-universe physics, and each shift reorders the inflationary model landscape, from Starobinsky-favored to attractor-disfavored to renewed talk of n_s = 1.
In ΛCDM the spectral index is a free input, set by the shape of an undetermined inflaton potential. The model carries no expectation for where n_s should land, so it cannot say which measurement is wrong, cannot flag either value as surprising, and silently re-ranks an entire literature of potentials every time the central value moves.
SCT replaces the hot-dense-center with a superluminal collision cascade, and with it the spectral tilt stops being a free input. Successive collision stages spanning a finite range of scales produce a power spectrum with index n_s = 1 - 1/L, where L is the number of hierarchical nesting levels the cascade traverses (P36, P37, P38). The red tilt is a boundary correction from the finite dynamic range of a truncated geometric series: a counting of levels, not the shape of a field potential.
Read through this formula, the two measurements are statements about cascade depth: Planck's 0.965 corresponds to L of 29, ACT's 0.974 to L of 38, and the all-data 0.977 to L of 43. The independently estimated plausible band is L of 20 to 40, so both flagship values fall inside or at the edge of the same physical window, and the dispute shifts from which potential is correct to how deep the hierarchy runs. SCT requires only that n_s sit strictly below 1 by an amount 1/L for a physically reasonable level count; it does not need the two experiments to agree before the systematic floor is found.
The same cascade that sets the tilt sets the rest of the primordial sector: near-zero non-Gaussianity from central-limit averaging over thousands of collision events, isocurvature suppressed by complete thermalization, and no inflationary gravitational-wave background (r below 10⁻⁵). One mechanism, several testable corollaries, no inflaton.
The registered kill: a precise measurement of n_s inconsistent with n_s = 1 - 1/L for any physically plausible L of 20 to 40, that is, n_s below 0.950 or above 0.980 at 5σ, falsifies the cascade-tilt derivation. A confirmed exact Harrison-Zeldovich spectrum (n_s = 1) would do the same, since it requires infinite cascade depth. CMB-S4 reaches the needed precision.