What is infrastructure? What isn't?

I gave a presentation at a conference recently entitled "Building" Cyberinfrastructure. There were a number of great questions posed by my colleagues during the Q&A session at the end, but something was nagging me. I realized what it was later while chatting with one of the faculty who had attended the talk - most of the people in the room have a different concept of what infrastructure actually means. How could something like infrastructure have been a subject of study for so long, and still be so far away from closure as a term of art? The faculty member commented that I've probably spent more time thinking about the nature of infrastructure than the typical person in my audience. Thus, it became obvious to me that if I am to write a dissertation (and launch a career) that addresses infrastructure, I am going to have to be both careful and explicit in my terms.

I realized that my views on what infrastructure is and isn't have changed radically in the past two years. I owe that transformation almost entirely to Paul Edwards. I took a course with Paul in Winter 2006 entitled "Systems, Networks, and Webs: The History and Theory of Infrastructure". Over the course of the semester, Paul constantly challenged my assumptions and use of the term. The point that he drove into my head over and over is that infrastructure is vast, complex, heterogeneous. The term "infrastructure" is employed carelessly most of the time, and is usually invoked when people actually mean "system" or "network". The origination of these terms in the way I will choose to employ them is derived primarily from the work of Thomas Parke Hughes.

The complicated, ever-present systems that undergird the operational existence of society have come to be studied as "infrastructure'', a compound of the Latin infra-, meaning "below'' and structus or struere, meaning "to build or assemble.'' There are several accounts for the introduction of the term, some claiming its use in 19^th century France and others marking 1927, when the American military culture began to use the term to describe the interconnection of roadways, electrical resources, waterways, etc. Despite the origins, the term generally refers to the heterogeneous interconnection of systems that support the fluidity of services and interactions that are the attention of everyday matters. It is because of this supporting nature that the infra- tag is important. Infrastructure lies below the attention of those who use it. It is typically transparent, only becoming visible when it does not function properly.

There are several canonical pieces describing the development of infrastructure that I'm not going to repeat here, but a good summary can be found in two other recent pieces generated by Paul Edwards, Steve Jackson, and Geof Bowker. (My name appears on them as well, but I freely admit that my written contributions were comparatively minimal.) The first is a NSF Workshop Report on Cyberinfrastructure, and the other (written entirely by Steve Jackson) is a derivative article published in First Monday. The best place to look is in the section entitled "Infrastructural Dynamics".

In short, though, I believe that we can view systems with clearer and cleaner senses of jurisdiction, of sovereignty, of control. One can see without much effort where a system ends and the rest of the world begins. Infrastructure does not embody this relative simplicity. Boundaries and edges are hazy, if they are visible at all. Control is decentralized, complex¹, and the hybrid product of various agendas, tensions, and histories.

As I go on, I'm going to need to become more concise about how I express the categorization of systems, networks, inter-networks or webs, and full-out infrastructures. It's difficult to rely on something that sounds conspicuously similar to U.S. Supreme Court Justice Potter Stewart's 1964 "Casablanca test" definition of pornography stating, "I know it when I see it."

And there are the problems. Apparently not everyone calls the same things "infrastructure". More importantly, we are not supposed to be able to see infrastructure, so how could we ever know it?


¹It is important to draw yet another distinguishing line for this term, which is similarly over-applied in the academic literature. Complex is not the same as merely complicated. Complexity suggests that there are patterns, behaviors, enactments, or actions at one level of scale that would not be predictable by examining other levels of scale. The typical term for this is emergence. Complicated systems exhibit linear qualities. Complex systems exhibit non-linear qualities in addition.