When forming policy under conditions of extreme uncertainty, the optimal approach seems to be a process by which the policy decision is divided into multiple stages, or in other words, an experimental approach. The optimal legal vehicle for such policy experimentation is what this Article refers to as “experimental rules,” which are rules that terminate automatically and are designed for the express purpose of generating data during the sunset period that can then be used to determine the optimal policy strategy for the long run. Yet it turns out that agencies rarely adopt such “experimental rules” in the real world. This Article argues that the reason has to do with the political economy, which appears to disfavor experimental rules either because they are more temporary and therefore less valuable to interest groups or because they are more costly to adopt. To overcome these political economy constraints and encourage policy experimentation, this Article proposes having courts apply greater deference to experimental rules (at least during the initial, experimental phase of the multi-stage process). This approach would have the effect of nudging actors in the political economy toward experimental rules, thereby avoiding the possibility of sub-optimal policies becoming entrenched in permanent rules. It would also preserve rules that might otherwise be vacated by courts at least long enough to generate the necessary learning to determine whether they should be implemented on a more permanent basis.