Spectrum policy in the US (and throughout most of the world) consists generally of a set of nationally determined policies that apply uniformly to all localities. However, it is also true that there is considerable variation in the features (e.g., traffic demand or population density), requirements and constraints of spectrum use on a local basis. Global spectrum policies designed to resolve a situation in New York City could well be overly restrictive for communities in central Wyoming. At the same time, it is necessary to ensure that more permissive policies of central Wyoming would not create problems for NYC (by ensuring, for example, that relocated radios adapt to local policies). Notions of polycentric governance that have been articulated by the late E. Ostrom argue that greater good can be achieved by allowing for local autonomy in resource allocation. Shared access to spectrum is generally mediated through one of several technologies. As Weiss, Altamimi and Liu show, approaches mediated by geolocation databases are the most cost effective in today's technology. In the database oriented Spectrum Access System, or SAS, proposed by the FCC, users are granted (renewable) usage rights based on their location for a limited period of time. Because this system grants usage rights on a case-by-case basis, it may also allow for greater local autonomy while still maintaining global coordination. For example, it would be technically feasible for the database to include parameters such as transmit power, protocol, and bandwidth. Thus, they may provide the platform by which polycentric governance might come to spectrum management. In this paper, we explore, through some case examples, what polycentric governance of spectrum might look like and how this could be implemented in a database-driven spectrum management system.The technical architecture envisions a geo-located and networked "radio appliance" that controls the protocols and transmission parameters of the radios associated with it. This appliance (and the associated radios) might be owned and operated by a landlord, a farmer, an Internet service provider, etc. The operating parameters of the radios, such as power, bandwidth, protocols (e.g. waiting times in WiFi) would gradually become more permissive until an enforceable event occurred. At this point, the parameters would become more restrictive. Note that this is consistent with the notion of "graduated sanctions" of the CPR literature.Enforceable events would be resolved between appliances through an automated negotiation protocol. It may also involve fusing authenticated data regarding the enforceable event. If the frequency of enforceable events increased above a threshold, a super local (or regional) coordinating device would be invoked to optimize use of the spectrum. In this way, spectrum governance is nested, providing resilience against a single point of failure in the governance process.This represents a means to learn from the local knowledge and policy changes, which results in a system that is more responsive to environment threats at multiple scales. In the end, we have a system that can compensate for the failure of some units with the successful response of others. This can be an advantage with respect to global policy mechanisms. Thus spectrum management would be best modelled as an emergent phenomenon rather than a top down system. This paper will describe the key details of this system and present some initial modelling results in comparison with the traditional global model of spectrum regulation. It will also address some of the concerns with this approach.