The oxygen evolution reaction (OER) is central to several sustainable energy technologies. Catalyst development has largely focused on lowering the overpotential and eliminating reliance on precious metals, revealing stark differences in alkaline and acidic OER. In alkaline electrolyte, precious metal-free catalysts have approached the limiting overpotential from established free energy scaling relationships, and our survey of complex metal oxides shows that this limit can be approached with a broad range of catalysts. In acidic electrolyte, electrochemical instabilities create a dual challenge of a dearth of nonprecious metal OER catalysts with overpotential below 0.5 V and a high dissolved metals concentration for most precious metal-free catalysts. On device-relevant time scales, the high dissolved metals concentrations compromise device stability, for example, through a decrease of performance and due to metal exchange between anode and cathode catalysts due to finite permeability of ion exchange membranes. These considerations motivate a substantial increase in monitoring and reporting of dissolved metals concentrations in OER experiments. To facilitate durability-based screening in continued catalyst discovery campaigns, we introduce a durability descriptor based on the d-electron count of each metal element compared to that of its Pourbaix-stable oxidation state, which enables rapid down-selection of candidate metal oxide catalysts. We discuss the importance of a codesign approach to catalyst development, where a device architecture can set specific requirements for dissolved metals concentrations and/or cathode and anode catalysts can be designed to tolerate cross-contamination. This device-level guidance of basic science will facilitate deployment of new catalysts to meet the societal needs for accelerated sustainable technology development.