The Evisceration of the Human Under Digital Capitalism 

This is a pre-print of Carrigan, M. (2018). The evisceration of the human under digital capitalism. In Realist Responses to Post-Human Society: Ex Machina (pp. 165-181). Routledge. Please see the final version if you want to cite this.

Introduction

In the summer of 2008, editor-in-chief of Wired magazine and techno-evangelist Chris Anderson wrote a much circulated article claiming that the ‘data deluge’ entailed the ‘end of theory’ and had rendered the ‘scientific method obsolete’. Anderson’s (2008) point was one which was made frequently over subsequent years, namely that the volume of data now available, advances in the computational sciences and the interaction between the two took us past a threshold where theorising was a necessary or useful endeavour. This polemic marked a coming out of ‘big data’ into popular consciousness, as well as being cited over a thousand times as the exemplar of computational evangelism. While much critical work on this topic has been undertaken within the academy, recent years have also seen the ascendancy of data science, widely hailed as an occupation which will be of ever greater prestige and influence as digital capitalism develops. I explore the sociology of this evangelism, drawing on the account of digitalisation developed in the previous Centre for Social Ontology book series[1], as something which links changes underway within the academy to many others taking place beyond it (Archer 2015, Lazega 2015, Carrigan 2017).

By placing the emergence of ‘big data’ in historical context, we can account for the ascendency of its practitioners and techniques in terms of a prior series of social, cultural and technical factors. ‘Big data’ as a concept has served as a discursive shield through which a range of problematic assumptions about the social world and our knowledge of it have been advanced. It is necessary for this reason to disentangle the epistemology of transactional data from the social and intellectual advocacy surrounding it. My argument is that transactional data is inherently generative of social asymmetries, as its institutionalisation divides the engineers from the engineered, but that this is only one mechanism at work within an open system. There are urgent questions about the political implications of digitalisation, such as those raised by Lazega (2016), which the social sciences must address. However the rise of data science threatens the capacity of the social sciences  to perform this role, as we confront the possibility that digital capitalism may be generating a digital social science which is unable to reflect back critically upon the conditions of its own emergence.

In this Chapter, I unpack these political questions through the prism of human agency, identifying how an epistemological stance and social outcome converge in what I term the ‘evisceration of the human’: the reduction of human agency to its behavioural traces. This leaves us with a circumscribed understanding of ‘online order’, naturalising the horizons of privately owned platforms and obscuring the political economy logically and empirically prior to them. There are methodological and philosophical reasons for seeking to resist this move, leading as it does to inadequate accounts of digitalisation. However there are also political imperatives, with the reintroduction of the human being the most potent means through which to analyse the emerging political economy of digital capitalism. The points made parallel those set out in other chapters regarding the issues raised by transhumanism.  

The Emergence of ‘Big Data’

What does it mean for data to be ‘big’? If we pose the question so bluntly, it can seem almost asinine. Data lack extension, such that the adjectives ‘big’ or ‘small’ can only be figurative devices intended to signify a contrast. This is often presented in a straight-forward way, as new technology leading to a volume of data which has outgrown our capacity to process it, necessitating innovation at the level of storage and analysis. However the claim of capacity is always relative to the analyst: the census is an obvious example of ‘big data’ in an analogue mode, in the sense that it encompasses the entire population rather than a sample of it, yet it has been within the capacities of state agencies to conduct it. This lack of clarity concerning volume, as a definitive characteristic of ‘big data’, leads to other characteristics being mooted: velocity, variety, variability, veracity, visualisation and value etc. These attempts at definition are tied up in an emerging political economy of what Kitchin (2015) calls ‘data-boosterism’, reflecting vested interests at work in the invocation of ‘big data’ which ought to be the object of sociological analysis. As Kitchin (2015: loc 225) writes, “the terms big data and open data have become powerful memes not just a way of describing data but symbolic of a wider rhetoric and imaginary that is used to garner support and spread their roll-out and development”.

Following Beer (2016), I suggest we can usefully distinguish between ‘big data’ as a material phenomenon and ‘big data’ as an ideational phenomenon, even if the two are interconnected in practice. Analysis of the former can address its social, technical and historical elements while analysis of the latter considers how ‘big data’ has been constructed in technical, philosophical and cultural terms. Through doing so, it becomes easier to identify how advocates of ‘big data’ infrastructures, techniques and procedures are finding material sponsorship within organisational contexts with specific characteristics, facilitated by the cultural valence attached to ‘big data’ as a concept while also obscured by its discursive sheen. In other words, the breathless rhetoric surrounding ‘big data’ can lead us to pay insufficient attention to the social context in which these claims are made, with the epochal terms in which these innovations are framed (data is, after all, the ‘new oil’) obscuring the many continuities which obtain with earlier stages of capitalist development (Srnicek 2016). ‘Big data’ is something which needs to be placed in an historical context.

The notion of ‘big data’ is rather misleading, as if the technological innovations to which it refers are radically discontinuous from that which has preceded it. The large-scale collection and analysis of data, as well as the reliance on digital databases and their analogue predecessors, has a long history across government, commerce and science. For this reason, Williamson (2017: 30–31) distinguishes between first wave big data and second wave big data. The former began with the “nineteenth century avalanche of analogue numbers” orientated towards the analysis and control of populations under conditions of rapid social and cultural change. The latter began with the expansion of digital infrastructure and digital devices, producing transactional data through the mediating role of the aforementioned elements, as well as the technological innovations which have facilitated the storage, retrieval and analysis of this data.

We can offer a material history of ‘big data’ as a form of social data, placing it at the intersection between technological change and the history of social statistics (Beer 2016). However we must also account for the concept itself, one that “has achieved a profile and vitality that very few concepts attain” to an extent which surely outstrips the instrumental value which can be found in its actually existing practical applications: 

That is to say that we have little understanding of the concept itself, where it came from, how it is used, what it is used for, how it lends authority, validates, justifies, and makes promises. In other words, we now need to work through a detailed account of what might be thought of as the birth of Big Data. (Beer 2016: 1–2)

Once placed within the history of social statistics, the apparent novelty of ‘big data’ comes to seem more complex. We can identify historical precedents to the “sense that there is a sudden and unstoppable wash of flowing data about people, a pooling of data that is on a scale that was not previously imagined” (Beer 2016: 2). This earlier ‘avalanche of numbers’ was entangled with industrialisation and the transformation of governance in response to emerging social problems. It would be a mistake to reduce innovations at the level of method to these features of the social context in which they emerged, but it would be equally mistaken to deny the ‘social life’ of methods if we wish to understand them as emergent social forms (Savage 2013, Law et al 2012). If we wish to understand how methods of generating social knowledge have emerged, it is necessary to understand their context of emergence, the motivations of their developers and the uses to which they have been put.

While there is much of value in Beer’s (2016: 2) exploration of “the discourse, terminology and rhetoric that surrounds it and which ushers and affords its incorporation into the social world” to this end, it is an account curiously lacking in people. As is often the case with Foucauldian approaches, it proves adept at tracing out the discursive powers latent within the concept of ‘big data’ at the expense of gaining traction upon how those powers might be enacted within actual social situations. We can approach the former in a purely ideational register, in terms of how ‘big data’ allows the social world to be imagined and carved-up in novel ways. However the latter is a much more complex matter, involving the interplay between structure, culture, agency and technology (Archer 2012). Effective analysis of it requires that we consider how concepts are deployed in actual meso-social contexts, the strategies and tactics they facilitate and the vested, as well as ideational, interests served by this activity. How are specific claims about social method and social data deployed in this way? How are inferences and explanations licensed by invoking the concept of “big data”?

Some of the challenge posed by ‘big data’ for social analysis stems from the technical character of these elements. This challenge reflects a disciplinary order in which the technological and the social have been objects of concern in different arenas of the academy which rarely, if ever, meet. Under these conditions, it is inevitable that a tendency to ‘black box’ the technological exists when it emerges as an irreducible element during inquiry into the social (or indeed vice versa), reducing it to an assumed relationship between ‘inputs’ and ‘outputs’ without concern for the mechanisms at work or the conditions of their operation. So we talk about ‘big data’ or ‘algorithms’ or ‘machine learning’ as if these were stable objects, with internal relations which could be precisely articulated by someone with the appropriate technical expertise. Whereas in reality each remains contested at the level of conceptualisation, as Kitchin (2015) for example unpicks in the case of ‘big data’. This conceptual ambiguity reflects in part the rapidity with which the technology itself is changing, as Williamson (2017: 33–35) summarises in the case of ‘machine learning’. It is not simply a matter of technical specificity, but rather of social embedding e.g. it matters how autonomously machine learning systems can operate, how structured their training data must be and the conditions in which this training data is produced.

If we remain at the level of the black box, we inevitably work with a technical ontology which is hypostasised, occluding the variation which is so integral to unpicking the dynamics of socio-technical systems. In so far as our chosen terminology retains an air of technicity, it further occludes the cultural ramifications of our mode of analysis. For instance, as Beer (2017) observes, the notion of the ‘algorithm’ has become ubiquitous in a way overloaded with emergent anxieties about autonomy and control within digital capitalism. Our use of an avowedly technical term can obscure how our analysis contributes to these increasingly prominent narratives about socio-technical futures. It can also obscure the continuities in social thought, hindering reflection upon whether our approach to social ontology is adequate to changing social and technical realities. As one prominent research within the field observed, the term ‘algorithm’ is becoming a master category of social thought analogous to ‘discourse’, without necessarily furthering our understanding of how social and cultural structures operate in relation to human agency and vice versa (Archer 1995).

This is why it is necessary to isolate the epistemological characteristics of ‘big data’, as I do in the next section, without losing track of the social and cultural context in which it is being institutionalised into socio-technical systems. In taking such an approach, we are forced to confront media infrastructure in a way which is still relatively rare within social theory. What Couldry and Hepp (2017) identify as a rupture between social theory and media theory creates problems because it either ignores the role of media or social infrastructure or operates with an empirically inadequate and often outdated conception of what that infrastructure entails. This poses a challenge to the disciplinary order of the academy, itself being transformed in part through the processes that cannot be adequately studied unless it is transcended (Halford 2015). The danger is that the digital social sciences emerging from this transformation (particularly data science and computational social science) come to be constituted in a way in which this broader horizon is lost, leaving the social sciences that remain unable to address the conditions of their existence, as they come to be replaced by a narrow empiricism with a descriptive and explanatory horizon no broader than what registers on digital platforms. If we cannot offer a diachronic account of what Pasquale (2015b) describes as the ‘online order’ then the political economy of digital capitalism is hypostasised as condition for, rather than object of, social scientific inquiry. The emancipatory potential of social science is lost if this orientation becomes hegemonic, leaving (digital) social scientific knowledge as little more than a mechanism through which present conditions can be reproduced by private actors into a future which is greeted with a disempowered shrug. 

The Epistemology of Transactional Data

We can read the emerge of such data is historical terms, dependent as it is upon digitalisation having reached the degree of ubiquity which means that significant swathes of social activity is mediated by digital data infrastructures. This new epistemological dimension to social life, in which all manner of everyday activities now generate digital data as byproducts which are stored within and cross data infrastructures both public and private, invites a reconsideration of established social scientific methodologies and their applicability to the everyday realities of digital capitalism. As Burrows and Savage (2007: 892) put it, “in the current situation, where data on whole populations are routinely gathered as a by-product of institutional transactions, the sample survey seems a very poor instrument”. This inevitable moment of comparison, seen by Burrows and Savage (2007) to herald a crisis for empirical sociology, poses important questions concerning what Halford (2015) calls ‘the politics of discipline’:

How we define Big Data matters because it shapes our understanding of the expertise that is required to engage with it – to extract the value and deliver the promise. Is this the job for mathematicians and statisticians? Computer scientists? Or ‘domain experts’ – economists, sociologists or geographers – as appropriate to the real-world problems at hand? As the Big Data field forms we see the processes of occupational closure at play: who does this field belong to, who has the expertise, the right to practice? This is of observational interest for those of us who research professions, knowledge and the labour market, as we see how claims to expert knowledge are made by competing disciplines. But it is also of broader interest for those of us concerned with the future of Big Data: the outcome will shape the epistemological foundations of the field.

The analytical challenge lies in unpacking the epistemological, technical and institutional factors at work here. The former is embroiled within the latter, in so far as transactional data is understood both as a new possibility and as an epistemic gain with institutional consequences. In its bluntest form, the epistemic gain is explicitly tied to an institutional claim for priority: this is better data, more cheaply and easily obtained than expensive sampling methods, which ought to be recognised as a crucial advance in the horizons of the social sciences. But in what sense is it understood to be better? Claims about the epistemological characteristics of transactional data are entangled with disciplinary politics because of the institutional context within which these debates are taking place. This is why how ‘big data’ comes to be framed has existential ramifications for the social sciences, rather than simply being an intellectual dispute between equal partners.

Despite being untenable in terms of the philosophy and sociology of science, ‘big data’ evangelism does important institutional work which even those who rejects its excesses can capitalise upon. There are important historical continuities which the overly sharp dichotomies of the evangelists obscure, as if we have entered a new age of social science in which we must turn our back on all that has gone before. For instance the unobtrusive measurement facilitated by transactional data (often framed as what people really do rather than what they say they do) is hailed as epistemological gain with revolutionary consequences for social scientific practice. However as Marres (2017) notes, unobtrusive methods have always been part of the repertoire of the social sciences, even if there are new possibilities for it in a context of ubiquitous digitalisation. The difficulty arises because the ontology of data (produced as a by-product of a transactional mediated by a digital system) and its epistemological merit (an unprecedented capacity to reveal the truth of social systems) are being routinely conflated in ways that drive methodological polarisation.

The concept of transactional data entails questions about the ontology of data which too rarely receive scrutiny. Though scrutinising the production of data is associated by some with constructionism, it is nonetheless the case that a focus upon the production of ‘facts’ has been an integral facet of critical realism from the outset (Bhaskar 1990). The conceptual resources of critical realism allow us to distinguish between the constitution of data and the construction of data: how data is stored and how data is created. What is novel about transactional data is not a matter of its constitution: storing data in a digital form is now standard practice, even if the precise point in the research process when that digitalisation takes places varies because of any number of factors. Nor is what is new about transactional data that it is constructed as digital from the outset. This can of course be the case for even qualitative research, in so far as that interview data might be produced in real time through the causal capacity of a voice recorder.

The novelty of transactional data lies in the combination of digital constitution, digital construction and a lack of elicitation. This is a genuinely new phenomenon, with potential implications which justify talk of a ‘data revolution’, even if we remain critical of the way that affirmation of this novelty is deployed by those committed to a project of restructuring the social sciences. Transactional data is digital from the outset, reflecting its status as a byproduct of data infrastructure through which social activity is transacted. It shares the digital constitution which eventually characterises much of the data with which social scientists work, but this is not an outcome brought about for the convenience of the researcher but rather an intrinsic feature of the data itself. Furthermore, it being produced as a by-product of existing activity, rather than as a consequence of a deliberate intervention in the world gives prima facie grounds for conferring a different epistemic status to that of the qualitative and quantitative data produced through the established methods of the social sciences. This natively digital character, constructed and constituted digitally, generates emergent properties which orthodox accounts of transactional data as ‘behavioural traces’ fail to capture.

In a much circulated article heralding the ‘coming age of computational social science’, Lazer et al (2009: 721) illustrate the daily activities common for those who “live life in the network”:

When we wake up in the morning, we check our e-mail, make a quick phone call, walk outside (our movements captured by a high definition video camera), get on the bus (swiping our RFID mass transit cards) or drive (using a transponder to zip through the tolls). We arrive at the airport, making sure to purchase a sandwich with a credit card before boarding the plane, and check our BlackBerries shortly before takeoff. Or we visit the doctor or the car mechanic, generating digital records of what our medical or automative problems are. We post Blog entries confiding to the world our thoughts and feelings, or maintain personal social network profiles revealing our friends and tastes.

Through pulling together the “digital breadcrumbs” which each of these transactions leaves, it is possible to overcome the limits of “one-shot self-reported data on relationships” in order to achieve “increasingly comprehensive pictures of both individuals and groups” (Lazer et al 2009: 721). Each of these ‘breadcrumbs’ has been constructed through a digital transaction and has generated data which is constituted within a digital infrastructure. However the actions themselves likely vary in terms of their degree of reflexivity, ranging from the maximally reflexive (e.g. expressing our thoughts and feelings through social media) through to the minimally reflexive (e.g. choosing a sandwich shop which allows us to accumulate reward points through our use of the credit card) and the entirely habitual (e.g. walking the same route we do each day, unaware our movements are captured by video camera). It is particularly significant when individuals act in a way orientated towards the infrastructure through which the transactional data is constituted, producing ‘breadcrumbs’ which are shaped by these systems while remaining unable to represent this influence in the ensuing data.

If such variation is systematically excluded then the human has in an important sense been emptied of properties and powers. Human agency is reduced to human action, as known through the transactional data it produces. This epistemological blindspot is most pronounced when we consider the user-generated content which is produced, circulated and consumed through social media platforms. There are undoubtedly important issues which can be raised about the ‘addictive’ characteristics of such platforms, particularly in terms of how they are continually (re)engineered to maximise the length and frequency of user engagement (Carrigan 2017, Eyal 2014, Gilroy-Ware 2017, Schüll 2012). But much of the activity which takes place upon these platforms is obviously social action, concerned with and orientated to the (re)actions of others. This sociality is mediated by the technical architecture of the platform, as expressed through constructs such as follower counts, ‘likes’, ‘shares’ and ‘retweets’ (Van Dijck 2013). Social reflexivity in this domain often entails some element of technical reflexivity: evaluating oneself in relation to others, when mediated through social media platforms, draws upon categories coded into the infrastructure itself. Users calibrate their behaviour in relation to feedback from others and/or the system itself, contributing to a reshaping of the context which other connected users inhabit (Pasquale 2015a). Furthermore, the many commercial imperatives which platforms are subject to inclines them to continually refine the architecture of these platforms in pursuit of greater user engagement. These platforms are characterised by a profound recursivity, arising from the reflexive capacities of users/designers and the digital constitution/construction of the data they draw upon in their activity. To reduce the relational complexity of this environment to human behaviour, registering in behavioural traces left by digital transactions, ignores much of great urgency about how the ubiquity of these platforms is bound up in a transformation of social life and those participating in it.

We need to recover the specificity of transactional data as something which is not easily reduced to ‘behavioural traces’ (Marres 2017). The case of social media data illustrates why this reduction is problematic but the point is a broader one: the orthodox framing of transactional data reduces action to behaviour, obscuring swathes of meaningful human activity and failing to account for the role of infrastructure in shaping it. Resisting the dissolution of the former into the latter does not entail a denial of behaviour as a relevant factor in understanding (digitalised) social life. Not all action is social action, despite a growing theoretical tendency to assume this is the case (Archer 2000, Campbell 1998). Furthermore, some of the most sociologically interesting and politically pertinent features of digital capitalism involve the deliberate attempt to intervene and manipulate at the level of behaviour (Davies 2015). But this dominant framing of ‘big data’ obscures the role of the human in social digitalisation, forming a crucial part of a powerful cultural movement towards the ‘evisceration of the human’: treating individuals as if they lacked reflexivity, being passively susceptible to behavioural interventions, rather than as active contributors to a profound transformation of social life (Couldry and Hepp 2016).

For all the excitement attached to the ‘flood of data’ and the possibilities it offers for a resurgent naturalism in the social science, there is nothing inherent to transactional data which demands a behaviouristic interpretation (Conte 2012). There is an ideological supplement at work here, a cultural project to overcome the irreducible ambiguities of the hermeneutic dimension to social life, driven by the belief that digital data enables us to see ‘who we are when we think no one’s looking’ and a commitment to a ‘utopia of total social legibility’ (Couldry 2015, Little 2015, Rudder 2015). The intellectual commitments we can find amongst designers, engineers and scientists across academia, government and commerce exercise a real influence over human action: the categories of their thought are coded into now ubiquitous systems for mediating social life, proving real in their consequences if not accurate in their descriptions. ‘Big data’ evangelism should be treated as a cultural form, reflecting a belief that the ‘book of the social’ can be read in the same way as the ‘book of nature’ has been, with all this entails for the status and reality of the now mathematicised human being (Barnes and Wilson 2014).

The Politics of Transactional Data

The scope of digitalisation inevitably poses questions which are best described as political, in the sense of struggles over the recognition of interests and distribution of resources. Much as with the other dimensions of transactional data considered in this chapter, its politics is manifest across a dizzying array of fields that easily lends itself to a focus on individual cases without attempting to draw out their commonalities. Undertaking ontological analysis is helpful in relation to such a multifaceted phenomenon because it allows us to (fallibly) identify mechanisms which are operative across a wide range of empirical cases. This is important because the functionality we can empirically discern could always manifest itself in different ways, with its mode of implementation reflecting features of the prior context rather than something intrinsic to the technology itself. In fact, the notion of ‘big data’ can serve as a smoke screen, providing a discursive capacity to make an epochal cut (“this is how we do things now”) which forecloses the other social possibilities inherent in the affordances of the technology, as well as the interests being served by such a foreclosure.

Identifying the role of contingent factors in shaping the institutionalisation of transactional data does not imply that its future is radically open, as if all that would be needed for it to serve progressive ends is to put the same technology in different hands. However it does foreground the centrality of agency to the establishment of socio-technical systems and the reproduction or transformation they undergo after their emergence: the role played by individuals, networks and organisations, their concerns and vested interests, producing what are later ossified as technical systems inexorably unfolding in a way governed by nothing more than their own technical logic. It makes it easier to ask who are driving these changes, what motivates them and how are their interests served through this labour.

To leave the analysis at this stage would offer an unsatisfying political ontology, presenting a digital capitalism bifurcated between the engineers and the engineered. This offers a dystopian mirror image to the cyber-utopian outlook which presents social change as emerging from the creative actions of a pioneering elite (Turner 2005). Each accords a primacy to technological mediation, seeing it as the engine of either social control or positive transformation, which inclines analysis to an over-estimation of the causal powers of technology and an under-estimation of the causal powers of social structure. This leaves us with an inadequate account of capitalist dynamics, failing to recognise the engineers as operating within firms embedded within a politico-economic context that profoundly shapes their operation (Srnicek 2016). It demarcates technology firms from capitalism as such, seeing the technology sector as sui generis in a way which acts discursively to make an epochal cut (“we are not like other firms”) that obscures genuine novelties at the level of economic firm by implicitly casting it all as novel. It also leaves us with an inadequate account of everyday life under digital capitalism, not merely at the level of phenomenology (i.e. the richness with which digital devices and digital activities, generative of transactional data, come to be experienced by ‘users’) but also the concern which individual users come to have in these data-generating affairs and how this mattering shapes their social action in ways which have important aggregative and collective consequences.

These framings obscure the collective dimension to agency, an oversight which Archer’s (1995) social realism can help us correct. The engineers are themselves primary agents, possessing vested interests through shared social position(s) and the distribution of resources contingently attached to them. Their imagined primacy within digital capitalism is not borne out by the available data, with product managers and designers consistently earning more, even if monthly salaries for engineering interns nonetheless receive twice the media wage for the rest of the United States. While data scientists, infamously described in the Harvard Business Review as the ‘Sexiest Job of the 21st Century’, routinely enjoy higher starting salaries, their growing oversupply seems likely to suppress salary growth (Coren 2016). Academic ethnography, technology journalism and popular culture converge in depicting life in Silicon Valley as presided over by a series of towering figures, enjoying celebrity status in virtue of their business accomplishments, while many thousands of aspirants imagine one day joining their gilded ranks (Marwick 2014, Martinez 2016). Furthermore, the privilege apparently enjoyed by many of these aspirants when considered in national or international terms is complicated when the rapidly spiralling costs of life in the Bay Area and rapidly declining quality of life are factored in.

The political economy of Silicon Valley is complex and simplistic invocations of the engineers as a new political class fail to represent it adequately, even if there are nonetheless important questions we can ask about the emergence of a (variegated) digital elite. This raises issues concerning the collective agency emerging within Silicon Valley, ranging from the sustained growth in the depth and complexity of corporate lobbying activities through to the unusual distribution of political beliefs across these populations (defying familiar distinctions of left and right) and how these are coming to exercise an influence over national and international politics (Ferenstein 2015). Epistemology, ontology and political economy are linked here because it is precisely these people whose agency is occluded in conventional accounts of transactional data which deny their role in its generation, circulation and interpretation (Couldry and Hepp 2017: 4978). ‘Raw data’ is and always will be an oxymoron, with the concerns expressed through its denial being susceptible to empirical analysis (Gitelman 2013). While the vested interests underlying this might be more complex than the self-enrichment of a postulated ‘engineer class’, it would nonetheless be difficult to analyse the converging project represented by this evisceration of the human if interests are expunged from the analysis at the outset. These interests can be served by the projects of engineers and architects even if these interests might not be their immediate concern, if indeed they even recognise them as vested interests they share with similarly situated others.

Much as it is simplistic to invoke a nascent ‘engineering’ class as the direct beneficiaries of digital capitalism, it would be a mistake to construe the ‘engineered’ solely as an atomised and distracted mass. This is one strata upon which transactional data is operating as a social mechanism: facilitating analysis and intervention at the level of the individual in a manner which is intrinsically prone to opacity. The assembly of people under false pretences in a way intended to create misleading impressions of this mobilisation occurred prior to digitalisation. However digital media provide a powerful array of tools which can be used to this end, as well as insulating the instigators from identification or scrutiny (Tufekci  2014). Nagle (2017: 118) argues that those organising against the current American president are now at risk from an “ability to send thousands of the most obsessed, unhinged and angry people on the Internet after someone if they dare to speak against the president or his prominent alt-light and alt-right fan”. This might overstate the case somewhat, reading back ‘an ability to send’, which produces collectivity from the observable fact of converging action, however it accurately recognises the central role played by digital media, particularly Brietbart, during Trump’s ultimately successful campaign for the presidency (Green 2017).

The radical possibilities for mobilising distracted people in this way raises questions about the long-term future of democratic political forms (Carrigan 2016). However there remains the possibility that the reconfiguration of primary agency currently underway might generate collective agents who are not being influenced in such a way. Interventions which bring about a change in the life chances of an aggregate, even one that might not have existed as such in a prior state, concurrently raise issues which those within that group face individually and/or collectively. The inherent opacity of transactional data makes it difficult but by no means impossible for groups to organise collectively around the shared fact of predictive privacy harms or digital enfranchisement. The difficulties which union organisers are now confronting when seeking to organise workers within the gig economy, with labour relations rendered opaque and tenuous, pale into comparison compared with the practical questions of political mobilisation entailed by these subtle forms of harm, disenfranchisement and marginalisation (Woodcock 2015).

However these empirical difficulties do not negate the possibility that these nascent primary agents can be organised into corporate agents, as well as their capacity to organise themselves in this way. The identification of these aggregates has important consequences for how we conceptualise the agency of collectives, as well as that of individuals. Identifying, representing and acting upon populations through ‘big data’ carries consequences for the public who are ‘carved up’ in new ways (Williamson 2017: 62–63). This ‘carving up’ is not new, nor are its effects. While there may have been a gap between the rhetoric and reality of the mid-century advertising business, its ambitions “to call group identities into existence where before there had been nothing but inchoate feelings and common responses to pollsters’ questions” should be taken seriously (Frank 1998: loc 640–657). The market segmentation which became the focus of advertising in the latter half of the twentieth century was partly identification of latent differences within the population and partly an artefact of the methods used to identify those purported differences. What has changed are the means available to this end, offering an unprecedented degree of granularity in the analysis of consumer behaviour and opacity in interventions made on this basis. This might be a matter of primary agency, in Archer’s (1995) sense, with an act categorisation leading to a shift in the life chances facing members of an grouping. The opacity of such data-driven interventions means the patterning of this effect is liable to be unclear, something which can at best be inferred from individual cases in which what Crawford and Schultz (2014) call predictive privacy harms have been established through legal challenge, journalistic expose or activist campaigns. What we don’t know is the potential scale of these harms, as well as how their distribution intersects with existing socio-demographic categories. Such opacity is a characteristic feature of platforms, resulting from the personalisation of user experience and the recourse to ‘corporate confidentiality’ in the face of demands to audit propriety algorithms. It obstructs any move from primary agency to corporate agency, as those harmed by data-driven interventions will often be unable to identify others who are similarly harmed with whom to organise collectively, if indeed they are even cognisant of the harms they themselves have been subject to.

What Tufekci (2014) calls the computational politics emerging from these developments upsets many of our traditional assumptions concerning the public sphere and how grievances are aired within it. The manner in which life chances are liable to be configured by opaque processes poses profound legal and political challenges which academics, activists and policymakers are only beginning to scratch the surface of through initiatives driven by notions such as algorithmic accountability (Pasquale 2015a). These unseen and unheard harms, unlikely to be conceptualised as such by those leveraging the ontological asymmetry of transactional data for their own competitive advantage, find reflection in more easily identifiable forms of inequality which are no less pernicious for being familiar in their operation. As Couldry (2012: loc 1534) points out, the ubiquity of the Internet across sectors increases the salience of digital skills and digital strategies, individual capacities likely to be correlated with socio-economic status and education. While the existing data on the digital divide makes for sobering reading, reminding us that … the tendency within this debate to conceptualise access in zero-sum terms obscures more subtle forms of digital disenfranchisement that seem likely to persist, even as access to the Internet trends ever upwards. The increasing assumption of digital-by-default in the provision of public services risks further excluding those who have fallen behind, as well as symbolically erasing them if transactional data becomes the basis for service-planning as well as service-delivery (Dunleavy 2014).

Andrejavic (2013) goes further and suggests we are liable to see a big data divide emerging, in which the data-rich and the data-poor are forced to orientate themselves towards the social world in diametrically opposed ways. These asymmetries are embedded into the expanding infrastructure of digital capitalism: an infrastructural divide “shaped by ownership and control of the material resources for data storage and mining” and an epistemological one manifested in “a difference in the forms of practical knowledge available to those with access to the database, in the way in which they think about and use information” (Andrejavic 2013: loc 464). The political economy of such a divide eludes the analytical resources of the conventional social sciences, requiring a reconfiguration of disciplinary boundaries to facilitate analysis all the way down to algorithms and all the way up to social structures (Davies 2017). However the digital social science being generated by digital capitalism is not the digital social science we need, if we are to resist the inequities endemic within it.

Conclusion

In this chapter I have argued that the asymmetries we can see produced in so many domains are a necessary rather than contingent feature of transactional data, necessitating that claims made at the level of epistemology should also be analysed at the level of political economy. It is possible to recognise a common intellectual project, emerging in a manner which is tied up with the reconfiguration of vested interests as we move into what Sernicek (2016) calls platform capitalism, without reducing that project to these vested interests. This is what I have described in this analysis as the evisceration of the human: the commitment to the reduction of human agency to the behavioural traces of human action. It is common in the sense of converging rather than coordinated, to be analysed in terms of shared pre-conditions and shared outcomes rather than prior organisation.

It can be found across a range of social domains, in each case animated by different proximate concerns, despite the overlaps which license talk of a singular (albeit diffuse) project. The closest manifestations of it for those working within the academy are data science and computational social science, reflecting a radical empiricism committed to the aforementioned dissolution of human agency. My argument has been that we need to see their emergence as proximate manifestations of a broader trend, itself susceptible to (and urgently necessitating) sociological analysis. To treat the emergence of ‘big data’ within the academy in terms of its epistemological radicalism or methodological novelty obscures a broader transformation taking place, in which positivistic dreams of reading the ‘book of nature’ expand to incorporate a similarly mathematicised ‘book of society’ (Barnes and Wilson 2014). The epistemic hubris which has come to surround what Little (2015) calls “the utopia of social legibility”, as well as the projects which respond to it while also seeking to bring it into being, can be framed in terms of the longue durée: a new form of social science emerging concurrently with a new phase of capitalism (Bratton 2016; Srnicek 2016). It is one which takes the ‘online order’, profoundly shaped by corporate actors, as the given foundation for a social science conducted within the boundaries of these established platforms and constrained by their ever-shifting conditions. Analysis of these issues is plagued by the problem of self-reference: it is itself part of the reconfiguration of knowledge production which it is purporting to analyse. In this chapter I have attempted to lay some of the conceptual groundwork for addressing these issue in a systematic way, though it is undoubtedly a project which exceeds the scope of what is possible in a single text. It is nonetheless crucial if we hope to resist the emergence of a digital social science which is unable to ask systematic questions about the digital capitalism which has produced it. Unless this trend is abated, the possibility of resisting the evisceration of the human will decline precipitously.

References

Andrejevic, M. (2013). Infoglut: How too much information is changing the way we think and know. Routledge. [Kindle DX version]

Archer, M. S. (1995). Realist social theory: The morphogenetic approach. Cambridge university press.

Archer, M.S. (2000) Being Human. Cambridge: Cambridge University Press.

Archer, M.S. (2012) The Reflexive Imperative. Cambridge: Cambridge University Press. 

Archer, M.S (2015) “How agency is transformed in the course of social transformation: Don’t forget the double morphogenesis” In Generative mechanisms transforming the social order, pp. 135-157. Springer International Publishing. 

Barnes, T. J., & Wilson, M. W. (2014). Big data, social physics, and spatial analysis: The early years. Big Data & Society, 1(1)..

Beer, D. (2016) ‘How should we do the history of big data?’, Big Data & Society, 3(1): 1-10. 

Beer, D. (2017) Why is everyone talking about algorithms? Discover Society (40). 

Bhaskar, R. (2010). Reclaiming reality: A critical introduction to contemporary philosophy. Taylor & Francis.

Bratton, B. H. (2016). The stack: On software and sovereignty. MIT press. [Kindle DX version]

Savage, M., & Burrows, R. (2007). The coming crisis of empirical sociology. Sociology, 41(5), 885-899.

Campbell, C. (1998). The myth of social action. Cambridge University Press.

Carrigan, M. (2016). The fragile movements of late modernity. In Morphogenesis and the crisis of normativity (pp. 191-215). Springer International Publishing.

Carrigan, M. (2017). Flourishing or Fragmenting Amidst Variety: And the Digitalization of the Archive. In Morphogenesis and Human Flourishing (pp. 163-183). Springer International Publishing.

Conte, R., Gilbert, N., Bonelli, G., Cioffi-Revilla, C., Deffuant, G., Kertesz, J., … & Nowak, A. (2012). Manifesto of computational social science. European Physical Journal-Special Topics, 214, p-325.

Coren, M. (2016) The highest paid workers in Silicon Valley are not software engineers. Quartz. https://qz.com/766658/the-highest-paid-workers-in-silicon-valley-are-not-software-engineers/

Couldry, N. (2012) Media, society, world. Cambridge: Polity. [Kindle DX version]

Couldry, N. and Hepp, A. (2017) The Mediated Construction of Reality. Cambridge: Polity. [Kindle DX version]

Crawford, K., & Schultz, J. (2014). Big data and due process: Toward a framework to redress predictive privacy harms. BCL Rev., 55, 93.

Davies, W. (2015). The happiness industry: How the government and big business sold us well-being. Verso Books. 

Eyal, N. (2014). Hooked: How to build habit-forming products. Penguin.

Ferenstein, G. (2015) Silicon Valley Represents An Entirely New Political Category. Tech Crunch. https://techcrunch.com/2015/11/08/silicon-valley-represents-an-entirely-new-political-category/

Frank, T. (1998). The conquest of cool: Business culture, counterculture, and the rise of hip consumerism. University of Chicago Press. [Kindle DX version]

Gilroy-Ware, M. (2017). Filling the Void: Social Media and The Continuation of Capitalism.. Repeater. [Kindle DX version]

Gitelman, L. (2013) (Ed.) “Raw Data” is an Oxymoron. MIT Press. [Kindle DX version]

Green, J. (2017) Devil’s Bargain: Steve Bannon, Donald Trump and the Storming of the Presidency. Penguin Press. 

Halford, S. (2015). Big Data and the Politics of Discipline. Discover Society (23). 

Kitchin, R. (2015). The data revolution: Big data, open data, data infrastructures and their consequences. London: Sage. [Kindle DX version]

Law, J., Ruppert, E., & Savage, M. (2011). The double social life of methods. CRESC Working Paper Series: 95. 

Lazega, E. (2015). Body captors and network profiles: A neo-structural note on digitalized social control and morphogenesis. In Generative mechanisms transforming the social order (pp. 113-133). Springer International Publishing.

Lazer, D., Pentland, A. S., Adamic, L., Aral, S., Barabasi, A. L., Brewer, D., … & Jebara, T. (2009). Life in the network: the coming age of computational social science. Science (New York, NY), 323(5915), 721.

Little, D. (2016). ANT-style critique of ABM. Understanding Society. Available at http://understandingsociety.blogspot.co.uk/2015/12/ant-style-critique-of-abm.html. Accessed May 10th, 2017.

Marres, N. (2017) Digital Sociology. Cambridge: Polity Press. 

Martinez, A. G. (2016). Chaos Monkeys: Obscene Fortune and Random Failure in Silicon Valley. Harper Business.

Marwick, A. (2013) Status Update. Yale University Press. [Kindle DX version]

Nagle, A. (2017) Kill All Normies: Online culture wars from 4chan and Tumblr to Trump and the alt-right. Zero Books. [Kindle DX version]

Pasquale, F. (2015a). The black box society: The secret algorithms that control money and information. Harvard University Press.

Pasquale, F. (2015b). The Algorithmic Self. The Hedgehog Review. 17(1). 

Rudder, C. (2014). Dataclysm. New York: Crown. [Kindle DX version]

Savage, M. (2013). The ‘social life of methods’: A critical introduction. Theory, Culture & Society, 30(4), 3-21.

Schüll, N. D. (2012). Addiction by design: Machine gambling in Las Vegas. Princeton University Press. [Kindle DX version]

Srnicek, N. (2016). Platform capitalism. Cambridge: Polity. [Kindle DX version]

Tufekci, Z. (2014). Engineering the public: Big data, surveillance and computational politics. First Monday, 19(7).

Tufekci, Z. (2017) Twitter and Teargas. Yale University Press. [Kindle DX version]

Turner, F. (2005). From counterculture to cyberculture: Stewart Brand, the Whole Earth Network, and the rise of digital utopianism. University of Chicago Press. [Kindle DX version]

Williamson, B. (2017) Big Data in Education. London: Sage. 

Woodcock, J. (2015) Working the phones. London: Pluto Press.  [Kindle DX version]

Van Dijck, J. (2013) The Culture of Connectivity. Oxford: Oxford University Press. [Kindle DX version]