Optimization models for California’s hydropower system are designed to be decision-support tools and aids for climate adaptation decision-making. In practice, they fall short of this goal. One potential explanation is that optimization models are not more successful because they are built on, and depend on, a misrepresentation of law and politics. The legal reality of California’s hydropower system is a web of networked jurisdictions of multiple federal and state agencies, with varying levels of coordination, long periods of legally obligated stability with rigid rules, and prone to conflict, but with multiple procedures for conflict resolution. Barriers to climate adaptation from that mix vary according to where a given dam is located. The virtual institutional arrangements represented in optimization models are not a simplification of existing arrangements. Instead, they are a dramatic replacement. That replacement is deliberate and reasoned. As seen in two optimization models supported by the state of California, CALVIN and INFORM, the operation of the optimization function of computer models depends on a virtual system of rules that are centrally controlled, coordinated, nimble, and without the possibility of conflict (let alone conflict resolution). But that smooth virtual system comes with a real cost. Institutional economics suggests that this mismatch between existing formal law and represented law may upend the results of models, since value is determined from institutional context.