Big G as a Quantum-Coherence Readout: Parameter-Free Resolution of the Gravitational-Constant Discrepancy

The Newtonian gravitational constant G remains the least precisely known fundamental constant: its CODATA-2022 recommendation, G = 6.67430(15) × 10⁻¹¹ m³ kg⁻¹ s⁻², carries a relative uncertainty enlarged by an expansion factor of 3.9 to absorb a Birge ratio of order five among mutually incompatible direct determinations. The 2026 NIST/BIPM torsion-balance replication belongs to the low-G branch; NIST's public news coordinate is 6.67387 × 10⁻¹¹ m³ kg⁻¹ s⁻², while the NIST publication metadata currently lists 6.67366(20) × 10⁻¹¹ m³ kg⁻¹ s⁻². We show that an offset of this sign and magnitude arises internally, without any experiment-level fit parameter, from a channel-selective boundary-response bifurcation of the parameter-free finite-cell formulation of Mittermeier Attractor Theory (MAT), and we offer this as a falsifiable resolution candidate. The Rydberg identity is used only once: it fixes an atomic SI anchor, R∞,SI ℓP,A^MAT = (αA²/4π) μe, where the dimensionless electron-to-Planck coordinate μe is fixed by a finite-support boundary action. Direct macroscopic measurements of G are then modeled not by re-evaluating the Rydberg identity with a different electromagnetic constant, but by multiplying this atomic anchor by a finite boundary-response kernel, GXY^meas = GA KXY. Here X,Y ∈ {A, ∂} are labels of the two active source boundaries, not tensor indices: A denotes the coherent atomic anchor and ∂ denotes a finite macroscopic boundary source. Coherent atomic readout fixes GA = 6.674300 × 10⁻¹¹ m³ kg⁻¹ s⁻², on the CODATA central coordinate at 17.1 ppb (+0.0008σ). Decohered bulk-source pairs carry the MAT response K∂∂ = [αM(1 − 2εcl/3)/αA]⁴, yielding Gsrc = 6.673871 × 10⁻¹¹ m³ kg⁻¹ s⁻²: 0.166 ppm from the NIST public low coordinate and 1.06σ from the current NIST publication-metadata coordinate. This formulation leaves R∞,SI, me, αA, and the atomic Planck-length anchor unique; only the finite gravitational source response changes with the measurement channel. Every rational factor entering the construction (π/4, 5/2, 9/8, 2/3, 14, 28) is traced to a closed identity of the MAT support spine, and the entire 64 ppm split is shown to decompose exactly as 52.98 ppm (atomic anchor → bare support response) plus 11.30 ppm (bare support → scar-dressed finite-source response). The single residue δαchart,π⋆ that drives the bifurcation is determined by the finite-support geometry descended from the algebraic seed ρ; it is independently confirmed by the flow-side closure condition δ₃(Λ₀⋆) = 0, which fixes Λ₀⋆ = 2.8169987329 × 10⁻¹²². The atomic fine-structure constant, the local atomic G anchor, the finite-source value of G, the electron mass, the Planck/DESI matter fractions, and the vacuum energy are therefore locked readouts of one upstream residue descended from ρ. The bifurcation is sharply falsifiable: finite macroscopic source–source experiments must converge to Gsrc, while a genuine coherent source–source quantum-gravity measurement must approach GA; ordinary free-fall and atom-gravimeter tests are protected because the response kernel is an active finite-source channel, not a universal passive-mass renormalization.