The New Cartographers: Trust and Social Order within the Internet Infrastructure
21 Pages Posted: 19 Jan 2012
Date Written: August 15, 2010
Internet governance is often studied in terms of the overlapping interests of nation states and those of formal organizations, such as the Internet Corporation for Assigned Names and Numbers (ICANN) and the Regional Internet Registries, largely in the context of allocation and administration of resources such as Domain Name Server (DNS) designations and Internet Protocol (IP) address space. In addition to recording and allocating resources, governance implies the maintenance of order among those who use these resources. Social order may be maintained through a variety of ways, including the use of legitimate force within a nation state (Weber 1958), market mechanisms or hierarchical command-and-control structures within organizations below the level of the state (see Powell 1990 for a survey). However, the maintenance of order in the infrastructure of the Internet presents a special challenge because the Internet spans all nation states, economies, and organizations. In this paper, we describe a history of Internet governance rooted in social norms of trust. We argue that this informal system has evolved amongst network administrators to maintain stability, and therefore one type of order, in routing flows of data amongst the many individual networks which collectively form the Internet. To understand the fundamental problem of order in the Internet infrastructure, we must begin with how information gets from one place to another. Routing flows of data between networks is termed inter-domain routing. If we think of the postal service in the United States as an analogy to the Internet's inter-domain routing infrastructure, post boxes would be starting and ending points for information that travels though many different nodes (e.g., postal stations) before reaching an intended recipient. The problem of routing information, whether it is a postal letter or electronic data, involves two questions. First, how does one know what is available on the network? Second, how does one know what path(s) to take to get the information to the recipient when there are no direct paths? The reason that a letter dropped into a mailbox in North Carolina can arrive at the intended recipient’s address in San Francisco is largely accomplished because both routing problems have been solved. The postal service analogy is helpful for thinking about the structure of routing in a large network. However, the analogy fails once we consider the fact that addressing and message delivery are centrally managed and maintained by the postal system. On the Internet, there is no centralized repository of addressing and routing information. A helpful analogy in this case would be learning a new environment by exchanging hand-made maps with others. If you arrive in a new location and someone gives you a copy of a map with routes through the region, how do you know that you can rely on the information? By extension, can you trust the person who gave you the map? What interest does the person have in giving you a reliable map in the first place? The problem can be infinitely extended if the map is passed from one person to the next over time, extending the range of the mapped territory, and taking into account changes in routes over time. Relying on the routes, geography, topography, etc requires one to depend on every entity that has ever added to the map. The internal logic of any system is perhaps clearest when it fails, as the Internet's inter-domain routing infrastructure did for YouTube (the popular online video sharing site) on February 24th 2008. For about two hours on that day, the Internet believed YouTube's servers to be located in Pakistan. This kind of confusion seems absurd, but is in fact not an uncommon occurrence (Ju et al. 2010). In this instance, the Government of Pakistan issued an order to censor a particular video on YouTube, and the Pakistan Internet Exchange (PIE) chose to implement this by issuing a claim to all networks connecting to the larger Internet through them that they were aware of a more specific route to YouTube, which PIE directed to a dead end, effectively blocking YouTube in Pakistan. Unfortunately, this claim also spread outside Pakistan, through PCCW, an Internet service provider in Hong Kong, which provides international connectivity for PIE. In response, PCCW disconnected PIE until the rest of the Internet could recover from this routing fault. The reason for this failure is that the technology which supports the Internet's inter-domain routing infrastructure – the Border Gateway Protocol (BGP) – provides no mechanisms to evaluate the veracity of claims that networks make to one another about routes which they can carry. This is intrinsic to the Internet's success in developing so rapidly as a global infrastructure, as the administrative burden for a new network to become part of the Internet is relatively low. Just like the analogy of sharing maps in previously unknown territories, the Internet largely works because all networks freely share routing information with one another and the default assumption is always that this information is accurate. The problem of inaccurate routing information is, in part, a legacy of the trusting nature of the relatively small research community which originally developed the Border Gateway Protocol. There are risks and uncertainties related to routing information on the Internet, and we argue that trust is an important way to alleviate such concerns without formal governance. In our examples, uncertainty deals with the reliability of information flows (e.g. will the information be accessible to others?) The risk is what is actually at stake in the transfer of information across the Internet. Risk is equivalent to the importance of information that would not be accessible or properly transferred to the intended target in the event of a failure. Using a series of qualitative interviews and historical research, we argue that social relationships and the maintenance of trust are essential for safeguarding the stability of the Internet. We show how social relationships structured as exchanges amongst network administrators, along with a social structure following the structure of the Internet itself contribute to the maintenance of order despite the lack of formal governance. Furthermore, we discuss how the specific form of technology and associated technological practices contribute to the spread of trust transitively across the Internet, in turn driving the formation and maintenance of community and reputation to maintain stability. This is a story of entwined social and technical systems which together contribute to the maintenance of order in the absence of centralized authority.
Suggested Citation: Suggested Citation