Social Relations and Routing Architectures (SIGCIS / Command Lines 2017)

This is a talk I will be giving at the 2017 SIGCIS meeting, Command Lines:

In my paper I will reassess the historical significance of the End to End Principle in light of an approach to core Internet technologies that I call the social relations of routing architecture. My objective is to provide an alternate historiographical approach to routing that incorporates an analysis of social relations, and of labor in particular. I choose routing because it is a common technical and architectural feature of all “switched” communication infrastructures: that is, communication in which messages are passed from node to node in an network (e.g. telegraph, telephone, Internet) before reaching their destination. Routing, and routing architecture more broadly, refer to the technologies (and, in my argument, techniques) that control how nodes work together to transmit messages from origin to destination. I use the End to End Principle as my case study because of its centrality in histories of the Internet and its significant consequences for a range of technological challenges faced today. Put simply, the End-to-End Principle is a design philosophy arguing that the maximum amount of intelligence (e.g. error correction) should be placed at the “ends” of a network, in the connected devices. (Adherents argue, convincingly, that placing services within the network itself will lead to complexity in interconnecting networks, and that these services will ultimately need to be duplicated in the connected devices all the same; Saltzer et al 1984) I argue that this principle also reflects a specific set of social relations in the information architectures to which it applies.

Elsewhere I argue that each form of message switching mandates, as a condition of its functioning, certain forms of labor, as well as power relations necessitated by the organization of, and discipline over, that labor. This argument is most intuitive in the case of the manual switchboard, with clearly gendered labor relations at work in a sociotechnical system in which humans play a central role in routing. Where this argument is less obvious, however, is in the hardware and software of modern routers, digital and algorithmic switching machines that route packets across and between the component networks of the Internet. To be sure, important efforts are underway to demonstrate the significance of labor in general in modern communications infrastructure (Russell & Vinsel 2016); I aim to expand this analysis by providing a new way to understand the specific role of labor in switched communications over time. Here I focus on the debates associated with the End-to-End Principle, as it was here that many of the boundaries around the functions of modern routers were settled. While existing STS approaches to these debates demonstrate their ideological character (Gillespie 2006), these debates are not yet understood as related to the social relations presupposed by the infrastructures for which they argued. In addition to revisiting the end-to-end debates, I will also draw on alternative models for routing architecture considered during the 1970s and 1980s.