Why is an infrastructure like an ontology? Why is an infrastructure like a shaggy dog? These are two fairly simple questions, but the immediate response might be to ask for a punchline that makes sublime sense of the incongruity (see Koestler 1964).

Let’s start with shaggy dogs, because tales about these critters precisely have no punchline. The joy is in the constant embellishment of detail, with a final groaner at the end that justifies the endless, rambling excursus. Infrastructures, for their part, do not have plotlines or heroic figures, the kinds of temporality that we associate with much historical storytelling. There are neither Napoleons nor Alexanders (though there is the occasional Baron von Haussmann). Theirs are messy stories of making things hook up together, of hopefully continual maintenance—America, take note—and no real endpoint. You never complete an infrastructure in the way you complete a novel; it is always and ever in the making. An infrastructure never even has the grace to die. It just hooks into an emergent infrastructural configuration—land lines into the cell phone network, and then the Internet, more broadly—until it persists in afterlife as a thoroughly metasomatized entity. As it transitions, it frequently brings with it a ghostly, persistent image of itself. The default terminal screen is eighty characters wide, just like IBM punchcards.

If you want to track an infrastructure, you need to come up with new historiographical skills. Our generic historical model is one of human mortality: people are born, flourish, and die; empires rise and fall; social movements succeed and fail. It is difficult to study things that do not have a singular identity at any one moment, that do not have clear life cycles.

This brings me to my first question. I offer a plonking response, worthy of a shaggy-dog story, before I map out further issues. Infrastructure is ontology. The kinds of social entities that make up our world are niches available within our infrastructural environment, an assertion somewhat overdeveloped by Ian Hacking (1998) in his work on the traveler’s fugue. The death of the extended family in Great Britain and North America can be connected to the rise of the steamship, the train, and the telephone. It is ironic that, during the 1990s, the telecommunications company Sprint was advertising the role of the telephone for keeping families together: you could be a thousand miles away, but still connected. The infrastructure of the telephone had insinuated itself into the extended family, much as Facebook is doing today. The very technologies that permit dispersion and hyperindividuality promote themselves in terms of closeness and sociality. The possibilities for what it means to be a person, what it means to be a democracy, and so forth emerge from infrastructure. Thus Richard John (1995) argues for the impossibility of conceiving the United States as a nation without subsidized postal service for newspapers allowing the development of a national discourse. Our sacred cow, democracy, is fundamentally different from an agora or the thing once assorted infrastructures have inserted themselves into the mix. I do not have the space here to develop the related assertion that the objects we see making up the world (species, genes, quarks) are themselves infrastructurally determined. However, in brief, I see no general need for a divide between the social and the natural. Infrastructure is ontology.

Given that infrastructure is variable just like ontology (see Latour 1993), it would be good to have new sets of tools to explore its historicity. One intriguing idea—I cannot track down the original reference—is to index terms by superscripting them with dates. So, if I use the term mass in my writing, then I might write mass¹⁹⁰⁵ to indicate that one author was using the Einsteinian concept of mass, while I could use mass¹⁶⁸⁷ to indicate that a different author was still working from a Newtonian ontology. This is not a very useful practice within the traditional forms of knowledge expression; there are so many dates for so many ontological and infrastructural variations that no one reader would be able to keep them all in her head. However, a visualization tool that would show, for an infrastructure, what its shape and nature was at that moment would be invaluable. As you click, you get an infrastructural landscape that displays an infrastructure along with its associated ontologies.

Mapping the temporalities of infrastructure in this way would give ways of escaping the dead weight of progressivist historiography. The discipline of history is only slowly being weaned from a vision, whether Hegelian or Marxist, that accompanied the very rise of progress as core to human development. Infrastructures expand and retreat, support more or fewer people; see the push to end net neutrality or Richard Beeching’s railway cuts in Britain in the 1960s. Acknowledging this would mean that we could start to play with complex temporalities in theorizing infrastructure. One might be Louis Althusser’s (1969) concept of combined and uneven development: instead of looking for the measurable rise of a society through a series of stages, one would look at punctual developments that permit the leapfrogging of a putatively natural sequence. The way in which smartphones in Africa have leapfrogged wired access to the Internet is a case in point.

Infrastructures are ontologies and shaggy dogs. Their temporalities remain core sites for theoretical and empirical elaboration: I have proposed one infrastructure to assist in this process. Infrastructures do not inhabit human lifetimes.