The Gravitational-Constant Discrepancy as a Parameter-Free Channel-Selective Planck-Unit Bifurcation

The determination of the Newtonian gravitational constant G presents a persistent metrological crisis. While the CODATA-2022 least-squares adjustment recommends G = 6.67430(15) × 10⁻¹¹ m³ kg⁻¹ s⁻², modern macroscopic torsion-balance experiments, notably the 2026 NIST/BIPM redetermination, report highly isolated but incompatible values near 6.67387(38) × 10⁻¹¹ m³ kg⁻¹ s⁻². We propose that this approximately 64 ppm discrepancy is not an intractable systematic error, but the signature of a fundamental quantum-gravitational bifurcation. Using the parameter-free finite-cell formulation of Mittermeier Attractor Theory (MAT), we reconstruct the Planck length from the Rydberg spectroscopy readout and test the hypothesis that gravitational coupling is channel-selective. Quantum-coherent states (N = 1) couple via the electromagnetically healed atomic branch, precisely enveloping the CODATA adjustment: G_atomic(R∞) = 6.674300 × 10⁻¹¹ m³ kg⁻¹ s⁻². Conversely, decohered macroscopic bulk matter (N → ∞) decouples from this branch and is subject to a calculable topological trace-anomaly penalty ε_cl. Evaluating this exact pairwise boundary projection yields a macroscopic mechanical source projection of G_source(MAT) = 6.673871 × 10⁻¹¹ m³ kg⁻¹ s⁻². This coincides with the macroscopic NIST/BIPM 2026 result without adding an experiment-level fit parameter and identifies the measurement, within MAT, as a candidate observation of macroscopic quantum-gravity source response.