Algorithmic Life: Calculative Devices in the Age of Big Data, edited by Louise Amoore and Vohla Poitukh

Reading Algorithmic Life makes one wonder if the social sciences’ excitement about ‘big data’ is somehow misdirected. The book reminds us that big data simply cannot mean anything without algorithms to make sense of them; in a ‘world characterised by a vast proliferation of structured and unstructured data’, as the editors so nicely put it (p. 2), what really matter are the devices we use to make sense of those data. The nine contributions in this volume, therefore, seek to illuminate the algorithms, or calculative devices, themselves: no small task in view of the secrecy that surrounds many of these devices, whether due to commercial value or national security. Such an endeavour is likely to be an ongoing concern of social scientists, and we should be grateful to Amoore and Poitukh, as well as the contributors, for their preliminary mapping of routes into the algorithmic unknown.

Amoore and Poitukh use the introduction to set out four key aspects of ‘algorithmic life’, which serve as parameters for this and future studies. First of all, algorithms filter what can be seen, creating ‘novel ways of perceiving the world and new visibilities and invisibilities’ (p. 5). Second, these calculative devices have the power to reorder space and sovereignty, just as they (third) can reorder time; and fourth, they ‘transform the nature of human subjectivity, pushing the limits of what can be read, analysed and thought about’ (p. 9). Crucial here is the visibility, or lack of it, of these devices: they ‘shape our capacity to decide and to act in the world in ways that cannot be fully excavated or known to us’ (p. 9).

The contributions echo Tarleton Gillespie’s (2014) argument that algorithms are arbiters of ‘public relevance’, helping us to determine what we should know and how to know it. Algorithms shadow us online, tracking our consumption and collecting our likes and dislikes. So much we know, but we often fail to recognise that those same shadows fall in front of us, selecting the content we see in accordance with our existing preferences, the so-called ‘filter bubble’. Widmer’s chapter, focusing on the personalised restaurant reviews offered by Foursquare, captures this process and the ambiguities it sets up. Foursquare users are not dupes. They knowingly accept the narrowing of their world to a risk-free and comfortable domain of recommendations from people like themselves: any new place instantly homogenised to a digital landscape of hipster coffee bars, with Starbucks algorithmically excluded. But invisibility is a political matter, even if we are aware of it. Deville and van der Velden, who examine the tracking devices employed by ‘payday’ lenders, position their piece within Susan Leigh Star’s (1991) call for a sociology of the invisible – the task of the social scientist being to illuminate the erased and silent work, here not of humans but of calculative devices. Deville and van der Velden use a tracker of website trackers (based on the tracker detector Ghostery) to produce novel visualisations of the tracking activities of these sites. The rich, dense traces they discover suggest that sub-prime lenders take far more notice of customers’ online activities than they do traditional credit scores.

Like Deville and van der Velden’s tracker, visualising threads and densities, much in this volume feels experimental, the chapters offering authors opportunities to play with ideas, metaphors, and methods. So van Otterlo’s chapter settles on the metaphor of librarian as it explores the ‘libraryness’ of algorithms. His chapter is as unruly as the universal digital library he imagines, illustrated with Lego-figure librarians surrounded by order or chaos, or – his real point – stacks of order arising from chaos as user behaviour shows what might be worth knowing. But the digital librarian takes note of the knowledge we need and increasingly knows what we should know until it ‘does not let us in the library, but answers all our questions at the door’ (p. 51). Algorithms qua-librarians shape our subjectivities by filtering and curating the knowledge available to us.

Other metaphors see calculative devices as mediators, public officials, or even judges; how algorithms mediate and judge emerges as a central concern of the volume. Birbak and Carlsen set out to render the organising principles behind the online algorithms we encounter every day in more familiar language – to caricature them, as they put it – and at the same time to establish the order of worth (Boltanski & Thevenot 2006) by which such principles might be justified. So Google is ordered by authority, running a continuous vote among websites as a just and useful measure of peer esteem. Facebook, on one hand, uses a principle of sociality based on a world that is open and connected, its web ranking driven by personal connections and privileged sharing. Twitter, on the other, understands sociality differently and considers events worthy if they cause a temporary affinity in people who are not previously acquainted. Birbak and Carlsen conclude that ‘algorithms offer a plurality of ways for us to orient ourselves in messy environments, and that these orderings come with justifications that, if taken seriously, explicate the world and the just vision of the public enacted by each algorithm’ (p. 31).

Such algorithmic orderings assemble ‘calculated publics’, in Gillespie’s phrase (quoted by Nisa, p. 117) to be treated not only as willing customers, but also as risks: to be detained or even bombed. In the chapters by Nisa and Belcher, we see two different, equally shocking, accounts of calculative devices at war. Nisa explores how biometric technology employed by the US military in Iraq and Afghanistan has sought to produce populations parsed into friends and enemies, changing the encounter between soldiers and civilian (or terrorist) and the moment of capture (or release) into a meeting mediated by a remote technology. Belcher gives an account of the attempts during the Vietnam War to use cutting edge computer technology to produce map of the country at a previously impossible level of detail, each hamlet coded by level of political threat. The system, fuelled by rough guesses and overoptimistic estimates, failed to deliver any meaningful advantage on the ground. But Belcher locates it within the tradition of data-driven American colonialism (as opposed to the deep cultural knowledge prized by European colonials) and argues that it served to usher in a new era of objective combat knowledge based on data science.

These calculative engines change the role of human agents, replacing discernment or expertise with data-driven objectivity, often calculated at a great distance: the role of the biometric-wielding soldier in Afghanistan is to gather and circulate evidence, while the expertise of the American general and their sensitivity to the ‘art’ of war is eroded by mountains of data and unimaginably detailed maps. Closer to home, I witnessed such changes in my study of transplantation, with surgeons’ power to determine who receives an organ greatly diminished by the implementation of national allocation algorithms (Roscoe 2015). Again, echoing the ambiguities elsewhere in this volume, there are shifting orders of worth at play. Surgeons were often relieved by the passing of such dreadful responsibility and one wonders if similar relief is felt by troops on the ground or generals in the war room – the consequences of algorithmic calculation are often nuanced and complex.

As the introduction makes clear, calculative devices compress time as well as space. O’Grady’s chapter sees the role of Fire and Rescue Services (FRS) metamorphose from responsive to anticipatory and entrepreneurial, with a responsibility to stop fires before they happen. The algorithmic prescience of the FRS depends upon the bricolage of consumer credit rating software with its forward-facing inferences, and historical fire records in a process O’Grady calls ‘over-representation’ (p. 81). Such over-representation does not coax insights from correlations in two related sets of data, but ‘play[s] on the capacity and value of different data to be inhabited by each other’, moulding together two separate temporal registers to instantiate the risk of fire in the present. Coleman identifies a similar temporal unravelling in her study of the UK government’s change4life programme, a social marketing campaign designed to combat future obesity. While she explores how the social marketing campaign has rendered obesity subject to the norms of neoliberalism – an individual problem, a site of personal responsibility, a public health issue to be understood in terms of cost-benefit analysis – Coleman’s concern is with the calculation of future obesity and the ability to make the future present in the now, in terms of shortened lives and higher risks of disease. Such a move is, like all the contributions in this book, a political one and takes us circling back to our original source of unease – knowledges of us, often collected without our knowing, repurposed into something else altogether. Knowledge of our past habits and satisfactions slides into present-day manifestations of our future selves at risk and in need of rescue, or ourselves as future risks, in need of special measures in the now.

In summary, calculative devices are performative of particular social worlds. Perhaps Mackinnon misses this crucial point in her study of ‘love’s algorithm’ and her analysis of the Gale Shapley ‘stable marriage’ algorithm: if the outputs of a dating site resemble the outputs of the algorithm, it is likely that they do so because the algorithm has been used in the dating engine itself (Roscoe & Chillas 2014). Moreover, though the universe of matches achieved by the algorithm is stable, it may not be optimal for everyone concerned; there is a hidden politics of gender at work within these calculating engines which Mackinnon’s voyage through Agamben and Lacan overlooks. On the other hand, Mackinnon is astute in her recognition that online dating is powered by ‘the prophylactic of instantaneous novelty’ and that the quantification of love’s pure randomness (though perhaps this is an idealism of another kind) is an essential step in rendering it of service to capitalism.

Certainly, this is a lively volume. At times insightful, at times confusing and obscure, often experimental, at times still in a process of becoming, it seems to mirror the world that it begins to open up. Linking all the contributions, however, are the editors’ and contributors’ shared concerns about security, privacy, agency, and freedom, and we should take them seriously: the book is a fitting manifesto for a sociology of the unseen agents that increasingly shape our ‘algorithmic life’.