ABSTRACT
To make sense of the increasingly complex information systems that now undergird so many social enterprises, some social scientists have turned their attention to the ‘algorithms’ that animate them. This “critical sociology of algorithms” (see Gillespie and Seaver 2015 for an evolving catalog of this work) has revived longstanding concerns about the automation and rationalization of human sociality, the potential for discrimination inside of bureaucratic and formulaic procedures, and the implications of sociotechnical systems for the practices that depend on them. Algorithms offer a powerful focal point for this line of inquiry: a hidden core inside these complex systems that appears to hold the secret, embedded values within. They are instructions, after all; the mechanic ghost in the machine? Tempting (Gillespie 2016; Ziewitz 2015). But, in our enthusiasm to install the algorithm as our new object of study, we (myself included) may have fallen into the most obvious of intellectual traps: the tendency to reify the very phenomenon we hope to explain. Much of this work positions ‘the algorithm’ as the thing to be explained, as the force acting on the world. This is hardly a new misstep; rather, it is one that has plagued the sociology of technology (Bimber 1994; Gillespie et al. 2014; Smith and Marx 1994; Sterne 2014; Wyatt 2008). Invited to consider “algorithmic cultures,” as we are in this volume, we might be tempted into the same trap: how has the introduction of algorithms changed the dynamics of culture? There are some interesting avenues to explore there, but they all run the same risk: of rehearsing a cause-and-effect story that treats ‘the algorithm’ as a single, bounded entity, presumes a stable and unsullied ‘culture’ that preceded this perturbation, and then looking for the effects of these algorithms on cultural practices and meanings-usually troubling ones. But we will certainly come up short if we tell simple cautionary tales about the mechanisms of production and distribution and their effects, or reassuring fables about how they merely answer to the genuine wants of audiences. These are the intellectual missteps that plague the study of culture. Culture is the product of both of these corresponding, but not isomorphic, forces (Bourdieu 1993, 230). Cultural objects are designed in anticipation of the value people may find in them and the means by which they may circulate; once circulated, we encounter cultural objects amidst a corpus of others, and attend to their place
amidst them (Mukerji and Schudson 1991). Moreover, culture is aware of this correspondence, self-aware and reflexive about its own construction. As we consume cultural objects, we sometimes wonder what it says about us that we consume them; and some cultural works are interested in culture itself, reading the popular as a clue to the society it is produced for and that finds meaning in it. Culture thinks about itself. The mechanisms by which culture is produced and circulated are sometimes drawn up into those debates, and the signals of valuation (Helgesson and Muniesa 2013) they generate-of what is significant, or popular, or newsworthy, or interesting-themselves become points of cultural interest, telling us something about the ‘us’ to which it is served. We not only debate the news item that made the front page, we sometimes debate the fact that it made the front page, the claim of importance made by the newspaper in putting it there, the logic by which newspapers choose and prioritize news stories, the institutional forces that drive modern news production. Evidence that we want a particular cultural artifact, or claims that we should, provoke us to ask why: why is this particular cultural object popular, how did it become so, are the artists and industries that helped generate it feeding us well, should culture be popular or should it be enlightening, are other kinds of culture being displaced in the process? Today, these questions have algorithms in their sights, particularly those algorithms that help select and deliver the cultural works we encounter. Algorithms, particularly those involved in the movement of culture, are mechanisms of both distribution and valuation, part of the process by which knowledge institutions circulate and evaluate information, the process by which new media industries provide and sort culture. In particular, assertions of cultural value, always based on prediction, recipes, and measurements of what makes something culturally valuable, are incorporating algorithmic techniques for doing so. Algorithms, then, are not invisible. While they may be black boxes in terms of their code, at the same time they are often objects of public scrutiny and debate. Not only should we ask, then, what it means for modern culture industries to adopt algorithmic techniques for producing, organizing, and selecting culture, and for knowing, tracking, and parsing audiences in order to make those choices (Beer and Burrows 2013); these are deeply important questions. But we should also examine the way these algorithmic techniques themselves become cultural objects, get taken up in our thinking about culture and the public to which it is addressed, and get contested both for what they do and what they reveal (Striphas 2015). We should ask not just how algorithms shape culture, but how they become culture.
