Theoretical descriptions of ice-bed friction beneath glaciers and ice sheets are key to predict changes in sea level. Applicability of these theories at the natural landscape scale and over long periods has however not been tested. Here we test hard bed friction laws by analyzing a unique data set of in-situ basal sliding measurements collected over three decades under an Alpine glacier enduring large changes in geometry. We report many observational features that are in striking agreement with theoretical predictions. However, we also observe an undocumented behavior where the basal stress state stabilizes near Iken's limit under meltwater input, which suggests the basal effective pressure is primarily set by bed shear stress rather than by water input and drainage specifics as commonly thought. As a result, long-term changes in year-averaged sliding velocities follow a simple power law scaling with bed shear stress as opposed to more complex pressure-dependent relationships.