Seeing the World Like a Platform
Discussion
Peter Törnberg:
"Characterizing governance through data power: seeing like a platform
Scott (1998) famously characterized how the modernist state made the social world legible and amenable to state power through a top-down population-based epistemology, exerting power through hierarchical command-and-control. As Scott argued, any understanding of the world necessarily requires abstraction: a narrowing of vision to reduce the unwieldy complexity of reality into something manageable. Scott studied the way that the modern state made the social world legible, readable, and thus amenable to state power. The state needs maps to navigate and act – and to be useful, maps need to reduce and leave out.
In examining the emergence of the modern state, Scott found that the state’s particular way of drawing its maps was at the core of a range of societal phenomena. When wielded by the state, a map becomes more than just a map: the world is reshaped and remade to fit the description that the map provides. A state registry which designates taxable property-holders does not merely describe a system of land tenure but creates such a system by giving its categories the force of law. Scott traced a long range of phenomena expressive of the modern state’s particular way of rendering legible – from the standardization of weights and measures, the design of forests, the creation of permanent last names, cadastral surveys and population registers, to the grid design of cities. From these emerge a common pattern, in which the state fought against diversity, mobility, local traditions, and individuality, seeking to impose a world that matched the homogenizing rows and columns characterizing its ways of representing the world. These changes were attempts at making legible, taking complex and diverse local practices and slotting them into a standard grid whereby they could be centrally recorded and monitored.
As Bauman (2000) argued, the high modernist state’s way of seeing emerged from the epistemic structure of the Fordist factory. Industrial capitalism was marked by the specialized division of labour, with specific characteristics: mass production based on standardization, rationalization, and the interchangeability of parts; mass production based on large groups of workers concentrated in factory settings, operating with functional specialization in administrative hierarchies and under strict managerial authority. The image of the Fordist factory shaped a modernity which was similarly obsessed with bulk and size, and impenetrable boundaries, with a preference for matching forms of planning and social organization – large factories and farms, huge dams, and grid cities.
Centrally, the modernist epistemology characterized a certain form of data, which matches the structure of the factory. It was the data of exact accounting, measurement, and statistics, printed sheets of IBM machines that governed every movement of the factory floor. The data of the ‘average man’, monitored through rows and columns of data, steered through top-down command-and-control. Drawing on such data, the image of the Fordist factory was transposed to society at large, institutionalized in schools, hospitals, family life, and personality – as figures like Robert McNamara brought to statecraft and warfare what they had learned from managing – through IBM machines, statistics, detailed control, and strict hierarchies – the factory.
The high modernist abstraction permeated the lived experience of its era, built around the image of the heavy machinery of the factory, with its precise structures and control. Society became factory-like, shaped by institutions that mirrored industrial organization: schools, hospitals, and even family life. Its science was that of average man and homo economicus, statistics, and systems theory, based on variable-variance analysis of representative samples of survey data. Its quantitative social sciences have dominated up until today, obsessed with measuring, classifying, and categorizing, finding regularities through means and variances, through assumptions of homogeneity and linearity.
The rising use of data power for governance, which characterizes platformization, is driving a shift in this epistemic foundation of modernity. As capital is using code to rewrite laws and employing the medium of digital technology to supplant the role of democratic institutions, platformization is bringing the rise of a new form of regulation. The platform mode of regulation comes with a particular way of seeing those governed, as the digital is coming to replace the Fordist factory as the chief ‘epistemological building site’ (Bauman 2000, 82) for contemporary modernity. As the high modernist way of seeing before it, data power is emerging as an incipient paradigm, reflected everywhere in society. As the logic of heavy machinery permeated first modernity, so does the logic of computer code and data coming to permeate the society of today. The rise of data power implies a shift in two deeply intertwined dimensions of power: the way of rendering legible, and the way of exerting power.
In terms of how platforms render legible, the shift consists of a move from traditional data to Big Data. This shift is not merely a question of new quantities of data or new tools – rather, in the words of Boyd and Crawford (2012), digital data are associated to ‘a profound change at the levels of epistemology’ (p. 665). While survey data is constructed for processing through variable-based analysis, requiring pre-compartmentalized data designed to be palatable for a scientific perspective that sees the social world through a lens of averages and variances, Big Data tends to be structured by and for algorithmic processing, implying indexed data structures and traversable networks (Mackenzie, 2012; Marres, 2017). While traditional data slots reality into fixed categories, variables, and variances, concealing its interactional elements (Conte et al., 2012; Lazer et al., 2020), Big Data are relational, interactive, heterogeneous, interactional, and emergent (Törnberg and Uitermark, 2021a). The social ontology that digital technologies operationalize is not focused on the summing up of a population in fixed categories, but rather on the individuals and their dynamic connections and interactions (Uprichard, 2013; Castellani, 2014; Törnberg and Törnberg, 2018). Rather than focusing on populations – assumed to be the sum of their parts – Big Data sees the world through clusters and patterns, located within a larger data structure. While traditional data was collected periodically, giving a snapshot of a defined population, Big Data is continuously gathered – and continuously fed algorithms that redefine clusters and patterns and seek to modulate their behaviour. Data power is fueled by a continuous flow of surveillance and control, from sensors that are seamlessly integrated into the urban fabric. Big Data thus gives space for forms of diversity, mobility, and individuality that traditional data erased – tracing individuals through thousands of ever-shifting attributes. While traditional data sees order from above, digital data sees it from below: traditional data imposes grids and straight lines, while Big Data allows fractal structures and diversity. But the epistemic shift associated with Big Data representations does not imply that the world is more correctly represented: as new aspects are brought into focus, others become blurry (Andersson and Törnberg, 2018). Any way of rendering legible requires abstraction, erasing aspects of the phenomenon.
In terms of how platforms exert power, the shift consists of a move from top-down command-and-control to a form of control mobilized through the design of programmable social infrastructure. If ‘the medium is the message’, as McLuhan argued, then consequently the one who controls the medium controls the message. This is the foundation of platform power. Platforms operate by providing the social infrastructures that underlie actions, and thus exert control by designing these infrastructures so as to generate certain outcomes – drawing on massive behaviour data to engineer social systems through infrastructural design. Yeung (2017) refers to this mode of control as ‘hypernudging’, as digital platforms engage in a rigorous process of designing the architectures to alter behaviour in predictable ways. Platforms shape their users through a mix between soft and hard discipline, combining gamification and scores with detailed tracking, algorithmic control, and at times threats of fines and expulsion – all A/B-tested and designed to efficaciously shape user behaviour.
To design infrastructures is to define the rules and goals of the social games that people are playing as they engage in the world. As Thi Nguyen (2020) argues, such games operate in the medium of agency: they have the power to determine not only the mode of interaction, but the goals and motivations of players – that is, to shape their very subjectivity. To control a social infrastructure is to gain some level of control over the goals and rules governing social life. This is not to suggest that data power vacates the role of individual agency – but rather to say that it situates and sets the context of agency. As Marx famously noted, we make our own history – but not under the circumstances of our choosing. Platform power implies control not over our choices, but over their circumstances, by defining the material life so central to conditioning social life.
While regulatory power targeted individuals, platform power thus operates on the interhuman and relational level, seeking to algorithmically modify the social rules that govern social behaviour. Platform power thus implies a relational approach to control, reshaping the connections and relations between people, leveraging social behaviour to generate social pressure for change. While an individual may, of course, choose not to play or to disregard the imposed rules of the game, this will, as in any game, inevitably imply losing in the eyes of those who are playing.
Twitter provides an example of this form of power in action (Nguyen, 2021; Törnberg and Uitermark, 2021b). When we engage in public conversation and discourse, we engage in a complex social activity in which each individual pursues their own goals – implicit, and often rich, subtle, and conflicting. Twitter’s interface thus constitutes the most profitable answer to the question: what type of game is public discourse? Twitter not only defines how we interact and with whom, but centrally supplants this nuance and diversity with simple points-based scoring systems to measure our conversational success – retweets, likes, and followers. By defining measures of our success that are irresistible in their simplicity and clarity, Twitter re-engineers our communicative goals. The effects of this are not restricted to the confines of the platform itself, but as social media have become the chief engine of public discourse in our society, the aims and motivations seep out to redefine public discourse and even political life more broadly – in a process that Hepp (2020) refers to as ‘deep mediatization’.
As Twitter applies this form of power to public conversations, so labour platforms like Uber are employing similar strategies for worker control. While purporting to provide a ride-share market, Uber sets the base rates its drivers charge, and limits the ability of drivers to accept or reject these offers – even creating ‘phantom cabs’ to give an illusion of greater supply to push down prices (Rosenblat and Stark, 2016). The Uber reputation system works as a normative apparatus, nudging both drivers and passengers toward a specific behaviour through scores, nudges, detailed tracking, algorithmic control, and threats of fines and expulsion – all A/B-tested and designed with precision to shape worker behaviour.
At the same time, platforms shape subjectivities of workers by having them interact as competitors in a market rather than collaborators in a team, designing interfaces to prevent communication, and seeking to prevent emergence of a critical political subject needed for resisting the disembedding brought by the labour platform.
Törnberg and Uitermark (2020b) and Isin and Ruppert (2020) situate this novel digital form of control in Foucault’s history of power, arguing that it signifies a move from regulatory power’s top-down ‘average man’ data epistemology to a power shaped by the epistemic features of Big Data: cluster-based, relational, interactional, fluid, and ostensibly bottom-up. In the same way that Foucault (2008, 259) suggests that the modern disciplinary power was reshaped by the biopolitical power exerted by neoliberal rationalities, so is the biopolitical power of neoliberalism thus now being altered by the digital power made possible by digital technologies (Cheney-Lippold, 2011; Pfister and Yang, 2018). Platformization thus constitutes the rise of a new governing logic, coming to shift the fundamental market ideology, discipline, and rationality. As the neoliberal rationality came with an associated ideology and belief in the legitimacy of market rationality in regulating every aspect of human life, so does this complex control come with its associated ideology: what Malaby (2009) terms ‘technoliberalism’, defined by faith in the legitimacy of emergent effects – ‘the emergent properties of complex interactions enjoy a certain degree of rightness just by virtue of being emergent’ (Malaby, 2009, 56). That is, the trust in the invisible hand of the platform algorithm.
Isin and Ruppert (2020) use the notion ‘sensory power’ to refer to this novel regime of power – as it is characterized by data collected from sensors. As Isin and Ruppert (2020) note, these regimes of power should be understood as layered rather than consecutive: it is not that old forms of power fall into complete disuse and become replaced, but rather that new forms emerge alongside them, nestling and intertwining, varying in salience across periods and contexts. Törnberg and Uitermark (2020a) instead use the term ‘complex control’ to describe the emerging regime of power, suggesting that it should be understood through the epistemology of the digital. The epistemic nature of the digital can best be understood through the fundamental distinction between complex and complicated systems (Érdi, 2007; Andersson and Törnberg, 2018; Törnberg and Uitermark, 2021a). The epistemology described by Scott’s characterization of the modern state was founded on complicatedness.
Complicated systems are like sophisticated machineries: top-down, hierarchical and bureaucratic, each of their components designed to carry out an organized function that fits into a larger structure. Such systems can be made highly efficient and capable of executing large-scale tasks with extreme precision, but they are at the same time brittle: fragile to internal and external disturbances, and lacking in their capacity to adapt to shifting circumstances (Michod and Nedelcu, 2003). With the rise of digital power, we are seeing the shift to a complex regime of power. Complex systems tend to have less functionally differentiated components, and are instead organized through large sets of interacting components on the same organizational level (Andersson and Törnberg, 2018). In complex systems, the components ‘are to some degree independent, and thus autonomous in their behaviour, while undergoing various direct and indirect interactions’ (Heylighen et al., 2006, 125). The macrodynamics of complex systems emerge through ‘self-organization’, and by controlling the infrastructure, the outcome of self-organization can effectively be designed. Complexity is the epistemic structure of the digital, and as capitalism is becoming digital, complexity is coming to lie at the epistemic foundation of contemporary modernity."