The algorithmic choreography of the impressionable subject

Introduction Advertising has become the dominant business model of the internet. As one of the early pioneers, Zuckerman (2014), suggests, it is “the entire economic foundation of our industry, because it was the easiest model for a web start-up to implement, and the easiest to market to investors.” The fact that the business model of the internet is advertising is historically contingent. It is neither inevitable, nor the only possibility. Nonetheless, once the advertising business model became established as the default model, a certain logic flows from it. One might say that such a logic requires, or necessitates, that all the actors (in the world of the internet) become positioned in particular ways-be they users, technology developers, advertisers, digital entrepreneurs, etc. Specifically for us, advertising needs an audience. Not just any old group of people, rather, the right person, at the right time, to see the right advertisement. That is, it requires impressionable subjects. Subjects that are so impressed-pressed into or imprinted on-that they are highly likely to convert. That is, do something of value for the company whose advertisement it is-such as click on the advertisement, register on the site, buy a product/service, and so forth. Thus, in the business model of advertising, the users of the internet need to become produced or positioned as impressionable subjects, specifically-and such positioning requires a particular regime of knowledge (or truth), as Foucault (1991) would suggest. The human subject is not an impressionable subject from the start, as it were-impressionability is neither necessary nor originally founded. Such subjects need to be produced, or perhaps, more precisely, enacted (Foucault 1980). To produce these impressionable subjects, the ones that will convert, a complex choreography is needed-a choreography in which algorithmic agency is playing an increasingly sophisticated part, as we hope to show below. For Foucault there is an intimate connection between power, knowledge, and subjectivity (Hall 1997). He suggests that power is relationally enacted and productive. Power is not an origin, but rather the outcome of the ongoing relational positioning of subject/objects within material discursive practice (Barad 2007; Foucault 1978, 94). Such positioning becomes constituted through regimes of knowledge. Knowledge is understood here as that which can be produced

through a series of methods, techniques, and technologies. These methods, etc., include mechanisms for inscription, recording, and calculation-that is, diverse ways of observing, and of encoding, subject/object positions. Through these domains of knowledge, subjects can become amenable to intervention and regulation-they can become positioned, or governed, in specific ways (Foucault 1991). Thus, Foucault (1980, 52) concludes that “the exercise of power perpetually creates knowledge and, conversely, knowledge constantly induces effects of power. [. . .] It is not possible for power to be exercised without knowledge, it is impossible for knowledge not to engender power.” Moreover, power “produces reality; it produces domains of objects and rituals of truth. The individual and the knowledge that may be gained of him belong to this production” (Foucault 1977, 194). In this ongoing circulation of power and knowledge the subject becomes enacted or positioned, and governed, in particular ways, in particular material discursive practices. For example, using specific methods and techniques-such as IQ tests, progress tests, classroom observations, etc.—some students become positioned as ‘good’ students, and others as ‘struggling’ students, in the material discursive practice of education. Over time, such positioning becomes the taken-for-granted-one might say, the valid and necessary-material discursive frame, relative to which subjects negotiate their own positioning, or position themselves. That is to say, as the way they take themselves to be, within such a regime of knowledge-‘I am a good student’ or ‘I am a struggling student.’ In our discussions below, we want to show how algorithmic actors emerge as producers of particular domains of knowledge, using very specific-and historically contingent-mechanisms of inscription, recording, and calculation, which position internet users in specific ways, in order to enact them as particular impressionable subjects. Specifically, as algorithms produce knowledge of us (indirectly through our online behavior as journeys) we become positionedalso by ourselves-as this or that type of subject-for example, one that is active, likes sport, listens to particular music, etc. Indeed, what makes online advertising different to other media is the diversity of methods, techniques, and technologies (mostly algorithmic) for the production of a particular domain of knowledge-that in turn function to choreograph certain subject positions, meticulously. Based on this knowledge, we are shown advertisements, or not, by the algorithms. Through these advertisements, we also get to ‘know,’ and position, ourselves. Hence, over time, as we become positioned, and start to position ourselves, in particular ways-we find ourselves, and taken by other, to be subjects that need, want, or desire those products shown to us in these advertisements. It is of course not the case that these algorithmic actors make us become these subjects, it is rather that the regimes of knowledge-based on historically contingent mechanisms of inscription, recording, and calculation-produce the very conditions under which our subjectivity becomes negotiated, and, freely taken up by us, as being this or that type of person. Thus, rather than taking the subject as an individual with some reducible and internal core of meaning (beliefs, needs, desires, etc.), Foucault’s work on power/knowledge suggests that

the subject is produced historically and contingently-in and through regimes of knowledge. That is, the subject is constituted through being positioned in correlative elements of power and knowledge (algorithmically produced knowledge, in our case). In such positioning: “[c]ertain bodies, certain gestures, certain discourses, certain desires come to be constituted as individuals. The individual . . . is I believe one of [power’s] prime effects” (Foucault 1980, 98). In considering the production of the impressionable subject, in online display advertising, we will be interested in the production of power/knowledge through the flow and circulation of agency in and through the sociomaterial whole, of the internet. Agency does not just flow through humans, it also flows through nonhumans, as suggested by Latour (1988, 2005)—specifically, in our case, algorithms. In the sociomaterial whole of the internet agency is always borrowed and translated from elsewhere; and is never at the singular bidding of any human or non-human actor per se (Latour 2005, 46). The impressionable subject is produced but there is no producer, as such. Thus, tracing the flow of agency through a heterogeneous array of actors in the sociomaterial assemblage (or the ‘agencement’ as Çalışkan and Callon [2010] would call it) of the internet is difficult, if not impossible-even more so for digital actors (Introna 2006; Introna and Nissenbaum 2000; Introna and Wood 2004). Digital agency is often subsumed, inscrutable, and opaque, even if one can read the code (Barocas et al. 2013). Still, tracing the flow of agency through socio-digital assemblages (including algorithms) is important because these actors do not only act they also simultaneously enact-to be more precise, they are performative (Barad 2007; Pickering 1995; Butler 1990). In other words, it does not just trace the subject, it produces the subject, as was suggested above. Or, as Whitehead (1978, 23) suggested: “[H]ow an actual entity becomes constitutes what that actual entity is. [. . .] Its ‘being’ is constituted by its ‘becoming.’ This is the principle of process.” Thus, in tracing the algorithmic choreography of the impressionable subject, below, we will attend to the ‘technical details,’ though not as mere technical details, but as performative material discursive practices (Orlikowski and Scott 2015). That is, as historical and contingent mechanisms of power/knowledge that progressively enact the impressionable subject in increasingly complex, and very specific ways. In short: as circuits in which power, knowledge, and impressionable subjects co-constitutively circulate-in and through the flow and circulation of agency (Ingold 2011). The flow of the narrative of the enactment of the impressionable subject, presented below, is in a sense historical-but not as a linear story that somehow adds up to the history of online display advertising. We rather highlight what seems to us to be important constitutive enactments, which can shed light on the ongoing production of the impressionable subject. In a way, we attempt to do a sort of genealogy in the Foucauldian sense (Foucault 1984). That is, we take careful note of the seemingly insignificant and historically contingent ‘technical’ practices, which are in fact constitutive of the becoming of the impressionable subject. This is not the only narrative possible, of course, and by no means the authoritative narrative. In articulating this narrative, we will focus on, or locate, four significant

enacting moments that seem to us to be constitutive of the impressionable subject. We have named these enactments as, the production of: (1) the gazing subject, (2) the animated subject, (3) the individuated subject, and finally, (4) the branded subject. In what follows we want to show, in as exact detail as possible, how these subject positions are produced in the becoming of the sociomaterial whole, of the internet, in order to produce the impressionable subject.