The Federal Communication Commission’s Order on Protecting and Promoting the Open Internet (GN Docket No. 14-28), adopted on February 26, 2015, mentions innovation more than 100 times. The Order is based on the premise that Internet openness is a precondition for a virtuous cycle in which edge innovation and network investment mutually propel each other. The adopted rules — including bright line standards (no blocking, no throttling, no paid prioritization), and safeguards for non-discriminatory access of edge providers and users — are seen as necessary safeguards to allow this beneficial process to unfold. Although an extensive research literature exists on the drivers of innovation in information and communication markets the Order mainly refers to comments and assertions submitted by stakeholders but it does not reference any of the pertinent innovation research. This paper seeks to close this gap by integrating insights from innovation theory and regulatory economics to examine the conditions of innovation in the Internet. We then juxtapose the findings with the vision embraced in the Order. Because empirical observations from countries where network neutrality regulations have been in place for some time (e.g., Netherlands, Chile) are sparse and anecdotal, the paper remains largely theoretical and conceptual. There is a long debate on the pros and cons of network neutrality regulation by constraining forms of active traffic management. Mandating net neutrality by means of government regulation, in particular the ban to charge termination fees to content providers, would be subsidizing the creation of new application services. This absence of fees to be paid to the Internet traffic providers would stimulate entry of new application service providers and also move consumers into the role of prosumers who are creating their own application software (Lee and Wu 2009, pp. 66; Choi and Kim 2010). While it helped clarify important aspects, this literature is too narrowly construed as it does not explore the full set of complementarities between network and edge innovations. Moreover, the rapid convergence of a multitude of services with heterogeneous demands on the emerging all-IP network requires a broader perspective. There is also a vibrant theoretical literature suggesting that some degree of network differentiation is conducive to efficiency and innovation (e.g., Reggiani and Valletti 2011; Krämer et al. 2013; Bourreau, Kourandi and Valletti 2014) although this body of work typically uses highly abstract notions of innovation. Our paper goes beyond these strands of literature by differentiating the types of innovation processes that unfold in the Internet economy and then linking them with research on innovation in highly interdependent systems. Building on the industrial organization literature on innovation types (e.g., Malerba and Orsenigo 1996; Breschi, Malerba and Orsenigo 2000), the paper starts with a detailed examination of the anatomy of the Internet innovation space. We distinguish innovation processes along two main dimensions: the type of coordination required for a successful innovation (modular, coupled) and the extent of the innovation (incremental, radical). The role of “innovational complementarities” between General Purpose Technologies (GPT, Bresnahan and Trajtenberg 1995) is of particular relevance in future all-IP networks. Although the focus of the literature on General Purpose Technologies (GPTs) has mainly been on the role of key technologies on aggregated economic growth, its overall framing is very fruitful for the understanding of the dynamics of the future Internet. According to Bresnahan and Trajtenberg (1995) innovation in the upstream GPT increases the productivity of R&D in downstream application markets and, developments of GPT-using applications raise vice versa the return to new advances in the GPT. From this perspective innovations within the Internet are not only driven by applications but can also be stimulated by positive feedback effects from the all-IP infrastructure and Generalized DiffServ architecture which function as GPTs for the application services (Knieps 2013). It is therefore important that the GPTs both on the broadband infrastructure level as well as on the traffic architecture level are open for innovative evolutions, taking into account requirements from the application side. Within this conceptual framework we examine the conditions under which different types of innovation flourish and under which conditions an overall desirable mix of innovation emerges. Are forms of differentiation in the network a precondition for certain types of innovation? Does exploiting the innovation potential of some applications require realizing innovation potentials within the data transmission architecture? On the other hand, what types of innovation in the network and at the application and services level are likely supported in a neutral network environment? This yields a differentiated analysis. In the future all-IP world a disaggregated representation of the Internet into all-IP broadband infrastructures, markets for Internet traffic (applying active traffic management) and markets for application services become more relevant. While much will depend on how many of the provisions in the FCC Order will be interpreted and operationalized, our framework helps formulate contingency claims as to how different approaches will affect innovation. If the rules are interpreted in a rather stringent way, it is likely that innovation will be biased toward application and services with the risk that infrastructure constraints may eventually slow down innovation compared to a scenario in which some degree of network differentiation is allowed. One possible outcome is that innovations that thrive in a more quality-differentiated network environment might migrate to private IP networks, inadvertently undermining one of the central goals of the Order, to safeguard an integrated and open Internet
