We study the issue of "spectrum zoning:" the problem of setting stable rules and bandplans -which may or may not give flexibility in use, allow markets, a primary/secondary division, etc. The approach (partially inspired by Rawls) might fairly be called "zoning as robust optimization." The goal is in part to create a framework within which new policy and technical problems can be formulated and attacked quantitatively. For researchers on the technology side, the idea is in part to change our perceived "customer." Traditionally, we have implicitly focused on the needs (perceived and as yet unperceived) of private wireless system builders (e.g. wireless carriers or their suppliers like Qualcomm). To understand the critical issues in zoning, the focus needs to switch to the needs (both perceived and as yet unperceived) of the government regulators. In particular, regulators need a principled way of deciding amongst alternatives for zoning to yield the greatest social good. This can be accomplished in a quantitative optimization framework. In this paper, one such framework is set forth and issues caused by moving away from a 'command and control' regime towards a more modern approach to spectrum management involving white-space channels and spectrum markets are explored. As an initial fruit of this framework, it can be seen that flexible band plans perform well in response to uncertain use preferences, even while not being optimal in a Pareto-efficiency sense, provided the overheads are not too large.