In my childhood, a radio station used to be on all the time, constantly broadcasting hits from the 60s, 70s, and 80s. In the post-Soviet years, that was probably a matter of a previously scarce resource (Western songs from my parents’ teenage years and youth) suddenly becoming widespread. Among them, there was one that I had always found strange. Even today it sounds to me like the ultimate stalkers’ anthem. The song is, of course, Every Breath You Take by The Police. With lyrics like ‘Every breath you take / Every move you make […] Every step you take / I’ll be watching you’, it can certainly sound sinister. However, this mantra is also evocative to anyone concerned about privacy and datafication in today’s digital world. Indeed, by now, as Cowley (2019: 96) attests, ‘it seems uncontroversial – even banal – to suggest that we have become reflexively aware that our actions, from the moment we wake up, are digitally mediated’. However, they are much more than just mediated—they are permanently measured and quantified, thus giving rise to comparison and competition, prognostication, and surreptitious adaptation of the digital architecture of the everyday. Hence, the online platforms behind these processes must be seen as not only reflecting but, even more so, producing ‘the social structures we live in’ (van Dijck et al. 2018: 2), thereby disrupting traditional patterns of life and governance.
Algorithmic governance—the increasingly prevalent form of governance in this digital world—is characterised by its tackling of problems through ‘their effects rather than their causation’: Instead of disentangling the multiplicity of causal relationships and getting to the root of every matter, this form of governance is intent on collecting as much data as possible in order to establish robust correlations; in other words, instead of decoding underlying essences, this mode of governance works by way of establishing connections, patterns, and, no less crucially, predictions (Chandler 2019: 24–29). These can be subsequently worked on and turned into algorithmically devised courses of action, changes in the digital architecture of our everyday environment, or nudging strategies. Notably, though, this attitude that prides itself on replacing causes with trends also has the effect of altering the place of human persons, effectively objectifying and commodifying them, turning them into data generators where the data footprint is all that matters and is taken for the person.
The shift towards algorithmic governance has happened courtesy to five major recent developments: ‘data, algorithms, networks, the cloud, and exponentially improving hardware’ (McAffee and Brynjolfsson 2017: 95). The abundance of data from sensors embedded almost literally everywhere has allowed access to the world and the individuals inhabiting it in unprecedented detail; algorithms have enabled the analysis of and pattern recognition in this deluge of otherwise raw material; networks have enabled cheap, instant, and virtually ubiquitous transmission of both data and the results of their analysis; the cloud has enabled vast and flexible storage and computing space to both perform data-related tasks and develop new, more potent, algorithms; finally improvements in hardware have added sheer power to the capture and analysis of data. All of these developments have severely disrupted (and, indeed, transformed) not only private life and the economy but also governance processes.
However, it is not only the matter of data capture, analysis, and use that is at stake. In fact, a crucial theme and concern of this book is how new data-based techniques of governance are changing the very matter of being human. Agency suddenly becomes debatable: after all, as Mau (2019: 3–4) stresses, ‘[i]f everything we do and every step we take in life are tracked, registered and fed into evaluation systems, then we lose the freedom to act independently of the behavioural and performance expectations embodied in those systems’. It also must be kept in mind that whereas various forms of regulation and enforcement of expectations traditionally used to have a public nature—i.e. were typically adopted, promulgated, and enforced by a public authority of some sort, the algorithmic governance regimes of today have a distinctly private character, being part of what is increasingly referred to as the platform economy or platform capitalism (see e.g. Srnicek 2017) or ‘digital capitalism/Big Data capitalism’ (Fuchs and Chandler 2019). In a broad sense, it can be defined as ‘a system in which a small group of powerful technology firms have vertically integrated a vast range of services and functions that they then provide to others’ (Hill 2019: 3). But perhaps more than integration, this system is characterised by a specific and distinct logic: provision of services or connection of service providers and users in exchange for data and the use of such data to better target and more efficiently discipline both users and providers. In fact, the efficiency aspect is paramount: platform economy is about the monetization of efficiency, be it transportation, accommodation, advertising, commerce, or any other domain, and data are paramount in achieving maximum efficiency.
More broadly, this is a book about the structuration of life in what Andrejevic and Burdon (2015) call ‘sensor society’, van Dijck et al. (2018) call ‘platform society’‚ or Mau (2019) calls ‘metric society’. The premise behind these terms is, nevertheless, the same: the capacity to collect data through ‘[h]uman-non-human assemblages of sensors’ (Chandler 2019: 34), render the world measurable and predictable, and put this newly acquired intelligence to (economic) use (see, notably, Schwab 2017). Moreover, an additional impetus for this analysis is that very soon (if not already) there will simply be no outside of digital platforms as we will either have no alternative but to carry out our activities through them or will be constantly surveyed by them even when not using them (Susskind 2018: 153). Still, this totality should not be surprising: ours is a world which feeds on the ‘the technical integration of previously distinct content streams, blending commercially produced entertainment with personalized logistics and everyday speech’, which is constantly being synchronised across devices (Athique 2019: 6).
The ubiquity of devices means that our activities and life patterns can be picked up in every detail, and the ease and low cost of both capture and storage means that even the most mundane of bits and pieces are collected (and, likewise, we are encouraged to share them ourselves, e.g. through social media activity or self-tracking). As a result, the mundane has become ‘one of the key sites through which Big Data is generated’, elevating it from something boring and uninteresting to ‘a domain of creativity and improvisation as well as a site of these everyday routines, contingencies and accomplishments’ (Pink et al. 2017: 1), all of them being repurposed in a data form for commodification (Zuboff 2015: 79). In this environment of ambient connectivity and data capture, ‘we generate more than we participate’—again, precisely due to the recording of the most minute and mundane of details—and even on occasions when we do actually participate through voluntary sharing activities, this participation also generates data (or, rather, metadata) by itself, independently of what is being shared (Andrejevic and Burdon 2015: 20).
As a consequence, populations occupy a dual role in the ecosystem of digital economy: they are simultaneously ‘the sources from which data extraction proceeds and the ultimate targets of the utilities such data produce’ (Zuboff 2015: 79). In other words, there is an inherent loop, in which humans become double producers of value: of data, which must be treated as capital in itself, not least because it can be directly converted to monetary value through the sale of data to third parties (or through allowing them to use your pool of data), and of direct returns, which could be financial (as in purchasing products or services) or behavioural (such as voting for a candidate). The two kinds of value production are mediated through algorithms that are perhaps best understood as ‘sociotechnical systems’ that ‘link society, technology and nature in a mesh of relations’—in fact, their operation is often ...