Abstract
<jats:p>The lithosphere-asthenosphere effective friction (0.01&ndash;0.03 from geodynamic inversions) is vastly lower than Byerlee&rsquo;s laboratory rock friction (0.6&ndash;0.85), an unsolved long-standing contradiction. This work states the Interface Selection Principle: the system picks the minimum &mu;N vertical interface as the sliding plane, making macroscopic friction a thickness-dependent emergent property rather than fixed rock material constant.Interface selection acts as a minor trigger (1&ndash;10% weakening), while distributed creep shear dominates reduction (80&ndash;90%). We derive the core creep friction relation: \(\mu_{\text{eff}} =\frac{C}{\Delta z},\quad C=\frac{1}{N}\cdot v\cdot \frac{d^2}{A}\exp\left(\frac{Q}{RT}\right) \tag{1}\) \(\mu_{\text{eff}}\) inversely scales with creep thickness \(\Delta z\); standard mantle parameters yield \(\mu_{\text{eff}}&asymp;0.0067\) at \(\Delta z&asymp;366\ \text{m}\). A secondary trace-melt weakening formula accounts for the remaining 10&ndash;20% reduction via melt-film contact replacement: \(\mu_{\text{eff}} =\frac{\mu_0}{1+\beta_{\text{eff}}\Phi},\quad \Phi=\phi^{2/3}D^{1/3} \tag{2}\) This dual framework revises prior melt-only interpretations: asthenospheric friction was overestimated (0.01&ndash;0.03 vs 0.0067), while trench megathrust friction (0.3&ndash;0.5) was severely underestimated due to absent creep layers. This study transforms friction into a first-principles emergent variable and redefines Byerlee&rsquo;s law as the melt-free baseline (\(\Phi=0\)) of our thickness-controlled model.</jats:p>