Where in the World is the Internet? Locating Political Power in Internet Infrastructure

With the rise of global telecommunications networks, and especially with the worldwide spread of the Internet, the world is considered to be becoming an information society: a society in which social relations are patterned by information, transcending time and space through the use of new information and communications technologies. Much of the popular press and academic literature on the information society focuses on the dichotomy between the technologically-enabled virtual space of information, and the physical space of the material world. However, to understand the nature of virtual space, and of the information society, it is critical to understand the politics of the technological infrastructure through which they are constructed. In this dissertation, I study the infrastructure of the Internet to examine the processes through which the dichotomy between virtual space and physical space is produced. It is through Internet infrastructure that the seemingly placeless appearance of virtual space is produced over disparate physical telecommunications infrastructures which are fixed in space, such as copper telephone wires, satellites, and optical fiber cables.

My examination of Internet infrastructure focuses on the system of interconnections amongst the networks which make up the Internet, which is called the inter-domain routing system. For all that the Internet is spoken of as a singular entity, it is in fact a complex distributed system of over 47,000 interconnected networks spanning the world. It is these interconnections - in the shape of the inter-domain routing system - which allow the Internet to appear to be a single entity, and provide the means through which the apparent placelessness of virtual space is produced. I approach the problem of understanding the production of virtual space by examining the mechanisms involved in the maintenance of order within the Internet's inter-domain routing system.
    
To examine the mechanisms involved in maintaining order in the inter-domain routing system, I study the technology and practices involved in the interconnection of networks. The technology which enables network interconnection is the Border Gateway Protocol (BGP), which allows the establishment of network interconnections without any need for centralized oversight. In the absence of centralized oversight, I found that the practices involved in operating BGP rely on coordination and collaboration amongst the technical personnel responsible for managing the interconnection of networks. Coordination and collaboration are enabled amongst the Internet's technical personnel through social relationships of trust, running across corporate and state boundaries. Even though the inter-domain routing system operates without centralized oversight, it does rely on centralized institutional structures for specialized functions, such as standards-setting activity, and the allocation of unique numbering resources required to identify networks in the inter-domain routing system, and to identify computers within networks. The personnel involved in managing network interconnections, and in administering centralized institutions do so - to varying degrees - with an ideal of serving the common good, acting "for the good of the Internet".
    
I argue that order is maintained within the inter-domain routing system through a distributed system of trust relationships, which are anchored by centralized institutional structures. As an arrangement of mechanisms for maintaining order, I consider this to be a governance arrangement, which I term "distributed governance".
    
Distributed governance is an unusual, and possibly unique, model of governance. It has three distinguishing features which mark it off from hierarchical and market-based models of governance. First, in its reliance on a distributed system of trust relationships. These are produced and reproduced in the practice of interconnecting networks, and through professional communities of the technical personnel responsible for managing network interconnections. Second, in its centralized institutional structures, which are uniquely organized amongst global governance institutions. None of these centralized institutional structures are formed by international treaty, and all of them are strongly committed to openness and participation. Third, in its operation over the particular technological form of BGP which emphasizes coordination and collaboration. To change the technology of inter-domain routing would be to change the range of governance possibilities for inter-domain routing, modifying the nature of distributed governance itself.
    
These distinguishing features are sites of contestation. Although technical personnel do owe allegiance to their professional communities, and to one another through trust relationships, they are also employees of corporations which invest in Internet infrastructure, and citizens of nation states which regulate (and sometimes invest in) Internet infrastructure in their territories. Distributed governance is accordingly complicated by market relationships, the interests of nation states, and international relations amongst nation states, just as markets, nation states and international relations are complicated by distributed governance.
    
To make sense of distributed governance as a global system, I study its instantiation in professional communities of the Internet's technical personnel, centralized institutional arrangements, and state and market interests across two different regions: North America, which is relatively central to the global Internet, and South Asia, which is relatively peripheral. This provides the opportunity to perform a comparison between these two cases, to understand at once how distributed governance varies under different conditions, and how different articulations of distributed governance are linked into a single global system of governance.
    
The range of social possibilities within a society are shaped by the model of governance which provides it with order. To understand the nature of the information society, it is essential to understand the mechanisms of distributed governance. Indeed, I argue that the social values of "freedom" and "democracy" which are often ascribed to the Internet are only made possible through distributed governance. Equally, these social values cannot be maintained through technology alone. They must be served by an appropriate combination of technology and social arrangements of technical personnel who act with these social values in mind, "for the good of the Internet".