Abstract: Replying to Matthew C. Stephenson, Legislative Allocation of Delegated Power: Uncertainty, Risk and the Choice Between Agencies and Courts, 119 Harv. L. Rev. 1035 (2006). Matthew Stephenson models "the decision calculus faced by a rational, risk-averse legislator who must choose between delegation to an agency and delegation to a court." On the assumption that delegation to agencies tends to produce interpretive consistency across issues while delegation to courts tends to produce interpretive consistency over time, a risk-averse legislator interested in reducing variance along either dimension will face a tradeoff between intertemporal risk diversification and interissue risk diversification. From this basic tradeoff, Stephenson derives comparative statics about the rational, risk-averse legislator's choice of delegates under various conditions. I suggest that Stephenson's legislative-delegation model rests on excessively artificial assumptions and is unable to yield significant predictions - in either the political or statistical sense. In particular, it is unsatisfying to model legislators as entering a "policy lottery" by enacting ambiguous delegating statutes, but then also to picture them as clearly specifying the identity of the delegate. The same institutional and political factors that tend to produce a first-order policy lottery over statutory substance also tend to produce a second-order "delegation lottery" over the question whether agencies or courts have ultimate interpretive authority. Moreover, the factors the model includes are, at best, second-decimal considerations relative to the factors it excludes.