Reading “The Ethnography of Infrastructure”

Scholarly texts are not always easy to read. An academic book or journal article is often written in a way that’s challenging to those who aren’t fluent in its disciplinary language. Beyond the language and complexity, these works are often long. In my (very) periodic “Reading” series, I’ll be offering digested interpretations of works I find important. These are my interpretations, so they’re not authoritative readings (if such a thing is even possible), but I hope they offer an entry into texts that would otherwise be challenging to even begin. It’s a bit like Cliff Notes, but also very much not. My goal isn’t to help you pass the test or do the assignment, but to get you interested enough to, just possibly, give the original a try. Start with my reading, and maybe discover that you’d like to do your own.

In her 1999 article “The Ethnography of Infrastructure,” Susan Leigh Star (who you may have seen mentioned by me previously) lays out rationale and some methodology for studying infrastructure. Importantly, Star also recaps and fleshes out a definition of infrastructure. It’s a recap because the definition comes from an article published previously (Star & Ruhleder, 1994), in which the traits of infrastructure are laid out alongside an ethnographic case study. “The Ethnography of Infrastructure” gives infrastructure the spotlight on its own, and offers further insights on why it’s important to study infrastructure, and how that might be achieved. When I use this article myself (which I do – a lot), the part I normally focus on is the start. It’s a convenient source for the traits of infrastructure, and this is how a lot of people use it. But the second part of the article, focusing on methods, deserves love too. So that’s how I’ll be dividing this text: first, I’ll focus on the description of infrastructure and how Star supports its importance; and then I’ll look at the way Star lays out methods for studying infrastructures.

What is infrastructure, and why study it?

So let’s start with the most obvious question: what is Star writing about when she writes about infrastructure? The term “infrastructure” is used casually in a lot of contexts, and can mean things like roads, electricity, or fibre optic cables. The definition Star provides is simultaneously narrower and broader than the common sense view. She does not restrict infrastructure to being a certain class of thing, but instead provides a set of nine traits that are present in infrastructure. These are: embeddedness; transparency; reach or scope; learned as part of membership; links with conventions of practice; embodiment of standards; built on an installed base; becomes visible upon breakdown; and is fixed in modular increments, not all at once or globally. For each of these traits, Star provides a definition.

All told, the traits of infrastructure take up two pages in the article itself, so I won’t faithfully recap them here. Instead, the speed read of these is as follows: infrastructure is inside the things people use (embeddedness), “invisibly supports [...] tasks” (transparency), is not limited to a one-time or one-site use (reach or scope), becomes obvious to people who are in the know (learned as part of membership), “shapes and is shaped by” the people who use it (links with conventions of practice), makes standardized use of other infrastructures (embodiment of standards), come out of what has come before (built on an installed base), just works, until it doesn’t (becomes visible upon breakdown), and, being made up of a lot of independent parts, “is never changed from above” (is fixed in modular increments, not all at once or globally). This is an extremely whistle-stop tour around the traits of infrastructure, and I’ll dig into some of them a little deeper as we go on.

Star also highlights something of crucial importance in understanding infrastructure: it is relational, in that “[o]ne person’s infrastructure is another’s topic, or difficulty” (380). An example provided in the article is that of railway track, which can be infrastructure for most people, and a topic for the railway engineer. But the concept of relationality in infrastructure applies across the board. The infrastructure-ness of a system is never totally and universally fixed, as it must be a subject for someone.

But now that a quick (Seriously! This was the condensed version!) definition of infrastructure is on the table, I want to roll back to why Star says it matters. She writes:

My teacher Anselm Strauss had a favorite aphorism, ‘study the unstudied.’ This led him and his students to research in understudied areas: chronic illness (Strauss, 1979), low-status workers such as janitors, death and dying, and the materials used in life sciences including experimental animals and taxidermy (Clarke & Fujimura, 1992). The aphorism was not a methodological perversion. Rather, it opened a more ecological understanding of workplaces, materiality, and interaction, and underpinned a social justice agenda by valorizing previously neglected people and things. (Star, 1999, 379)

Star is putting her heart on her sleeve here and explaining the importance of placing one’s focus on things that do not normally receive the focus they should by rights deserve. The “unstudied” things of Strauss’ work are important because they not only make the world go ‘round, but they open up new opportunities to see the work of people who are not otherwise sufficiently seen. She goes on to argue that her articulation, the study of boring things, has a similar capacity to the study of the unstudied: “[t]he ecology of the distributed high-tech workplace, home, or school is profoundly impacted by the relatively unstudied infrastructure that permeates all its functions” and, in neglecting infrastructure, "you miss equally essential aspects of aesthetics, justice, and change” (379).

In short, Star’s argument for the study of infrastructure is that it opens worlds, makes the unseen seen, and provides opportunities for justice that would not otherwise be available if the boring things remained a substrate instead of a subject.

We may take for granted today that it is important to study the way systems function, and especially the systems that underpin new or controversial technologies. Research on the environmental and labour impacts of tools like large language models and other generative AI tools is essential to the debate about the use of these new systems. But the assumption we take for granted – that the study of infrastructures is important – comes out of the groundwork of people like Star.

Star argued in 1999 that good studies existed on the uses of computers and networks to do novel or interesting social things like community creation or the development of individual identity, but less work had been done on “the effect of standardization or formal classification on group formation, the design of networks and their import for various communities, or on the fierce policy debates about domain names, exchange protocols, or languages” (378). All of these issues were part of the understanding of infrastructure for which she was advocating. That legacy is alive today, both in the trans-disciplinary field of infrastructure studies, and also in many other fields. When a sub-sea cable or a specification for a standard becomes a subject of study, Star’s definition of infrastructure is generally not far away.

How to study infrastructure

Star indicates that it is not surprising to see less work on digital infrastructure (though partially because of her work, there is much more of it now than there was in 1999), thanks to the locations in which such decisions are made and the manner of their implementation. “Theirs is not the usual sort of anthropological strangeness. Rather, it is an embedded strangeness, a second-order one, that of the forgotten, the background, the frozen in place” (378-9). Star argues that new methods are needed to study this different kind of strangeness, and the intertwined relationship between what she refers to as “human organization” and infrastructure.

Calling back to the 1994 article with Ruhleder, we see a story about the development of a computer-based system for biologists: “the difficulty was not in the interface or the representation of the work processes embedded in the system, but rather in infrastructure – incompatible platforms, recalcitrant local computing centers, and bottlenecked resources” (380). It is the features underlying the apparent subject that turn out to be at issue. And indeed, these are relational issues. As I highlighted above, Star takes great pains to point out the relationality of infrastructure. What appears to be infrastructure for most of us is inevitably a subject, an issue, or a concern for someone.

The relational approach to infrastructure has methodological implications – “sites to examine then include decisions about encoding and standardizing, tinkering and tailoring activities [...], and the observation and deconstruction of decisions carried into infrastructural forms” (382). Fieldwork includes “a combination of historical and literary analysis, traditional tools like interviews and observations, systems analysis, and usability studies” (382). In short, the study of infrastructure does not look like the study of many other subjects, and requires a multi-faceted and multi-method approach.

Star doesn’t leave us hanging. The bulk of the methodological portion of the paper comes under two headings: “Tricks of the trade” and “The thorny problem of indicators.” In these two lists, she gives us a small arsenal with which to begin finding cracks where the light shines through opaque systems.

Her tricks of the trade boil down to three things:

• “Identifying master narratives and ‘others’” (finding what has been characterized as “other” and using that to discover what is dominant);
• “Surfacing invisible work” (work gets embedded and encoded in information systems, and finding it means looking for the traces left behind); and
• “Paradoxes of infrastructure” (understanding why small inconveniences can turn out to be barriers, and recognizing that the story of why something doesn’t work is often much bigger than what’s visible on the surface).

Indicators bring a new set of traits around the fundamental problem that there are different ways (on different levels) of understanding a subject. Star lays out three levels on which we can read information infrastructure:

• "a material artifact constructed by people";
• "a trace or record of activities";
• "a veridical representation of the world."

Untangling different levels of these indicators can be difficult. The difficulty of “discovering the status of indicators” is “partly due to our own elisions as researchers, and partly due to sleights of hand undertaken by those creating them” (388), which is to say that we may often need to look differently, and occasionally that need comes from the way something is being presented by those who have created it. Understanding which level to read an infrastructure on allows us to cut through deliberate opacity or obfuscation more effectively. When should we see it as a true representation, when as a record, and when as a material thing?

Star ends the article by referencing a popular parable in the academic study of technology. It is a story about low bridges preventing access to a beach. (I won’t go into it here, but there’s an excellent documentary on the topic called Misleading Innocence.) The crux of the issue is that on a highway, a low bridge may prevent access to buses, and this could have been a deliberate choice to create exclusion. Star writes “[t]here are millions of tiny bridges built into large-scale information infrastructures, and millions of (literal and metaphoric) public buses that cannot pass through them” (389). She goes on: “[i]n information infrastructure, every conceivable form of variation in practice, culture, and norm is inscribed at the deepest levels of design” (389). In short, the design of systems is neither simple, nor straightforward, nor apolitical. It is imbued with heft, and it has importance for those who will use it, or be impacted by it. This, ultimately, is one of the reasons we must study infrastructure.

References

Star, S. L. (1999). The ethnography of infrastructure. American behavioral scientist, 43(3), 377-391.

Star, S. L., & Ruhleder, K. (1994, October). Steps towards an ecology of infrastructure: complex problems in design and access for large-scale collaborative systems. In Proceedings of the 1994 ACM conference on Computer supported cooperative work (pp. 253-264).