The photoelectrochemical reduction of water or CO2 is a promising route to sustainable solar fuels but hinges on the identification of a stable photoanode for water oxidation. Semiconductor oxides like Fe2O3 and BiVO4 have been gaining significant attention as promising materials. However, they exhibit a major drawback of a large required overpotential for solar water oxidation. In this Perspective, recent efforts to characterize and reduce the overpotential are critically examined. The accumulation of photogenerated holes at the semiconductor-liquid interface, recently observed with multiple techniques, is rationalized with surface state models. Transient absorption spectroscopy and electrochemical impedance spectroscopy suggest that surface treatments designed to either passivate surface traps or increase reaction rates (as catalysts) actually perform identically. This calls into question the definition of a catalyst when coupled to a semiconductor photoelectrode. In contrast, results from transient photocurrent spectroscopy suggest that two separate loss mechanisms are indeed occurring and can be addressed separately.