Recent debates over Internet policy have focused on the network’s architecture. For example, the central justification for the Federal Communications Commission’s Open Internet proceeding is that the Internet’s architecture was critical to its past success and must be preserved. Unfortunately, policymakers and commentators have failed to provide any theoretical basis for determining whether any network configuration is optimal or when changed circumstances might justify a change to the architecture. The result is a regulatory approach that is unnecessarily static and reflexively accepting of the status quo, while failing to provide a basis for distinguishing between architectural changes that are part of the network’s natural evolution and those that are potentially anticompetitive. This Article fills the void by providing an analytical framework for assessing network architecture based on modularity theory. It synthesizes modularity theory into five key concepts: near decomposition, interdependencies, abstraction/information hiding, requisite variety, and testing/integration. It then surfaces the tradeoffs inherent in the architectural decision by identifying the benefits and costs associated with modular architectures and analyzing the dynamics of architectural change. It concludes by using the framework to evaluate a series of recent policy applications, including the Telecommunications Act of 1996, network neutrality, the transition from IPv4 to IPv6, calls for opening the Apple iPhone’s application programming interfaces (“API”s), and the evolution of future Internet architectures. An architectural approach based on modularity theory yields a number of important insights. It surfaces the inevitability of the tradeoff between generality and cost minimization and how generality can obstruct certain types of innovation. It underscores how parallel experimentation can accelerate innovation at the same time that the lack of coordination can stunt it. Moreover, like any design hierarchy, modular systems can resist technological change. Finally, it underscores the contingent nature of network architecture and provides heuristics for identifying when architectural change can be beneficial.
The full text of this Article is available to download as a PDF.