“…The primary result of the modeling exercise is to demonstrate that a simple vertically homogeneous TVS with kinematic boundary conditions determined from relative plate velocities, body forces set by the distribution of GPE, an approximate representation of crustal faults using lateral variations in effective viscosity, and realistic coupling of the Pamir and Tibet fits the full GPS surface velocity field within uncertainties for the entire Indo-Asian collision from Afghanistan to SE Asia. Quantitative comparisons of the best TVS solution to fully 3D forward models that incorporate additional structural and dynamic complexity based on a priori additional knowledge of the region, such as stresses applied by a coherent lithospheric slab in the Pamir, the development of very weak lower crust in Tibet (Royden et al, 2008), sublithospheric tractions (Ghosh et al, 2006), or the introduction of 3D architecture (Lechmann et al, 2014;Yang & Liu, 2013), demonstrate that increased complexity does little better than our minimally complex formulation in reproducing the observed surface velocities, and therefore violates an information criterion of model efficiency. On the other hand, simpler models than the TVS, such as a three-block elastic formulation (Avouac & Tapponnier, 1993) or a homogeneous Newtonian sheet (England & McKenzie, 1982), perform so poorly at matching the high spatial resolution of available surface velocities and topography that they also violate the information criterion.…”