At the Cascadia subduction zone, the Juan de Fuca plate (JdF) subducts obliquely under, and deforms, the North American plate (NA). Subduction deformation dominates deformation on land, as observed by geodetic measurements such as the Global Navigation Satellite System (GNSS, including GPS), limiting the resolution of deformation due to surface faults on the overriding North American plate. A kinematic block model applied to oblique JdF-NA convergence interprets deformation in terms of microplate rotations and elastic deformation on locked and partially locked faults, revealing partitioning between surface faults in the Pacific Northwest region and the subduction zone interface. Beginning with an initially dense block model, an L1 regularization known as Total Variation Regularization (TVR) algorithmically selects the most active faults. We find a tradeoff between megathrust activity and surface fault activity: A megathrust constrained to dip-slip-only requires more active surface faults along with greater fault slip rates than a freely slipping (oblique) megathrust. However, even with a dip-slip-only megathrust, trench-parallel strike-slip faults near the trench are neither active nor form a simple forearc sliver. Furthermore, estimated slip rates in models with a freely slipping subduction zone are more consistent with geologic slip rates than those models with a constrained dip-slip subduction zone. In other words, oblique subduction is sufficient to explain onshore GNSS observations or geologic slip rates, and strain partitioning between a dip-slip megathrust and strike-slip hanging-wall faults is not required