In the 1997 essay “Protected Mode,” the late media theorist Friedrich Kittler, with nostalgia for “the good old times” when using computers meant interacting with them in a way that made it impossible to ignore the reality of their basic hardware, expressed his disapproval of the user-friendliness of commercial software. In contrast to the true underlying operations of digital machines themselves, he asserted, commercial software hides from view the reality of computers’ operations determined at the level of material technological frameworks. “The higher and more effortless the programming languages,” he wrote, “the more insurmountable the gap between those languages and a hardware that still continues to do all of the work” (157). The problem with software, for Kittler, is that it seems to put the user in control when, in fact, what it really does is reduce the user’s agency by obscuring the user interface’s basis in hardware. Put in different terms, it performs an illusory reversal of the relationship between infrastructural and superstructural elements. One can only imagine that Kittler would be dismayed by the current state of digital media technology’s development in general and by the trend among technology startup companies toward increased reliance on cloud computing in the form of “infrastructure as a service.” At the same, I think this gestures toward a certain problem in the anthropological study of digital technologies.
“Infrastructure as a service” (or IaaS) is a model in which businesses outsource certain necessary components of their systems, including hardware, servers, databases, and other sorts of equipment, often to large companies like Amazon, Google, or AT&T. The ability of early-stage digital entrepreneurs to conveniently rent infrastructure located and maintained elsewhere has important implications for the amounts of time and money required to establish some digital startup endeavors. It eliminates some of the costs associated with acquiring and updating equipment and the labor associated with keeping it in working order. This is the idea anyway, whether or not it always works like this is another question. But, without wading into a debate about how much innovative new technology digital startups actually build these days, I think it is safe to say that, with respect to certain kinds of companies, the relative roles of hardware and software are at least put into question, and it may be productive to consider what the consequences might be for the anthropological study of technological innovation, especially in light of the discipline’s recent surge of interest in infrastructure.
Of course, the scale of the shift illustrated above with respect to any given startup depends upon exactly what kind of technology the commercial venture in question hopes to develop. Even so, in my research on the development of new digital media products, I am finding that the infrastructural elements outsourced though cloud technology are sometimes regarded by startup founders as necessities more like office space or paying taxes than as the fundaments of a dynamic technological context into which they will intervene or the results of past generations of innovation upon which they will improve. Certainly there must be variation, depending upon whose perspective is taken into account. A programmer may be more likely to dispute this characterization than a designer, but what I am suggesting is that it is exactly this difference that is important: it is often the case that the design-oriented aspects of a new digital media product are more helpful in explaining its processes of conception and circulation than are its technical aspects. There is a risk of reification in this distinction, but it can heuristically mark a difference between, say, the Stuxnet computer worm and Angry Birds. When one considers that the huge commercial success of a product like, for instance, the image-sharing application Instagram relies on rented Amazon Web Services infrastructure, basically working as a piece of software composed from technologies that existed before it was conceived, then it may not be appropriate to attribute a great deal of causality to the outsourced infrastructure that was certainly necessary for its existence, but also probably not particularly revealing of its particular behavior.
In my view, this has implications for how, as anthropologists, we should think about both the ethnographic examination of digital technologies development and the question of how infrastructure should feature in our research. When I asked one founder of an early-stage social media startup whether he thought the attention many startups in his industry devote to the usability and user experience of their products meant turning away from the direct pursuit of technological breakthroughs, he responded by shifting the conceptual frame of the conversation: “It’s not interesting to write a story about someone who’s gotten more efficiency out of a shoe factory. That was the story of a hundred years ago. There’s nothing new in that.” What this suggests to me is that if we want to understand what is new and unique about the innovations of today’s technology startups, we should not try to do so by analytically drilling down into the machines underlying other layers of computer technology to ask how it is being manipulated or, rather, this should not be assumed to be explanatory before the fact. It is an approach that can lead toward a sort of Kittlerian demystification, which may very well help to create an account of infrastructure, but might just as easily fail to detect what is emergent in media technological development.