Reasoning. An Eddington limit requires luminosity L = eta dM/dt whose radiation force etaMdotsigma/(4 pi m r^2) cancels gravity at the sonic radius; the modified Bernoulli equation then shuts off accretion when the Eddington term dominates, giving the 4D limit Mdot ⇐ 4 pi m G M/(eta sigma) and its higher-dimensional generalization Mdot ⇐ f(Gamma,D) 8 pi m R_B c_s^2/(eta sigma). Whether this is ever reached depends on how much radiation is generated and whether it can propagate outward. (1) Generation: for spherical (Bondi) inflow onto a bare horizon there is no hard surface or accretion disk; free-free emissivity integrated over the supersonic region is small — the well-known inefficiency of spherical accretion (Shapiro-Teukolsky; Frank-King-Raine). (2) Pressure comparison: ambient photon pressure over fluid pressure is T^4/p ~ 10^-6 (young white dwarf, T ~ 10^8 K vs degenerate pressure), 4 x 10^-12 (Earth center), < 10^-11 (neutron star), and the ratio p_gamma/p scales as rho^(3Gamma-4), staying small inward. (3) Transport: in white dwarfs the ambient photon mean free path is l_0 ~ 5 x 10^-8 cm (opacity calculations, falling 100x by 1 Gyr age), so beyond tiny radii the medium is optically thick; the photon-trapping condition r^2 < R R_B^3/l_0^2 holds out to the sonic radius once R_B > l_0/c_s — outward diffusion loses to inward advection even for very small 4D holes. Earth is the least opaque case (l_0 ~ 0.5 cm, an optically-thin 4D window between R_C and l_0): here luminosity can escape, but attaining Eddington with efficiency eta forces the thermalization radius r_T ~ R/eta^2 to be large, which caps the last-scattering temperature and undercuts the required luminosity — a self-consistency failure matching the known difficulty of radiation-limited spherical accretion; and any melting merely reproduces the fluid treatment already assumed. (4) Observable cross-check: were white-dwarf holes Eddington-limited radiators at L_Edd ~ 8 pi m R_B c_s^2/sigma, the thousands of accumulated holes would exceed 10^-2 L_sun once N R_B/cm >~ 60, interfering with measured cooling rates (10^-1-10^-3 L_sun) — not observed. Hence Earth safety cannot be rescued from the fast-accretion corners by radiative feedback, and (conversely) the white-dwarf/neutron-star destruction times are not lengthened by it.
Step 6 verdict
Verdict: approved (checked). Load-bearing step: for Bondi inflow onto a bare horizon, radiation cannot both be generated efficiently and propagate outward against the flow - (1) spherical-accretion radiative inefficiency (standard result), (2) p_gamma/p ⇐ 1e-6 in all three bodies, (3) photon trapping/advection once optically thick; Earth’s optically-thin 4D window is closed by the eta-vs-thermalization-radius self-consistency argument. Plus an independent observational modus tollens: Eddington-limited WD holes would exceed 1e-2 L_sun, interfering with measured cooling - not observed. Each sub-step traced at the physics-argument level; no defeater found conditional on the no-disk spherical premise (reasonable for a hole at rest in pressure-supported matter). Note: A-7 (Plaga) contests the premise set for the metastable Hawking-radiating case; that is a premise dispute for steps 7-8, not an inference failure here.