1. Introduction
As the structure of the Internet is yet again under transformation, with the introduction of subscription-based tiers by Facebook and Twitter, a wide array of literature has been expended in examining the operation of labour and value on the Internet. This paper attempts to develop a general concept of datafication, through a critique of the model provided by ‘data colonialism’. Examining the cases of online food aggregators and the proliferation of fake news and disinformation online, the paper demonstrates how Accumulation by Digital Dispossession can explain a wide range of processes linked to datafication on the Internet.
The representation of datafication as a space upon which new colonial expansions can take place emphasises a spatial metaphor for its understanding. With such an emphasis, the role of accumulation of surplus value within the capitalist mode of production is relegated within the analysis to the background of a primarily extractive process. The paper however argues that such an approach is deficient as it fails to both adhere to decolonial theory’s more radical assertions, while at the same time decentering labour from its analysis of datafication. Instead, we can focus on two main questions. First, what role does digital labour play in the datafication and datafied reorganization of capital and value? Second, can accumulation by dispossession as theorised by David Harvey emerge as more suited to explaining the contemporary processes in which the collection and processing of data is emphasised and is a site of extensive capital investment?
To address these questions, we examine two distinct processes broadly relating to digital labour. In the first case, we look at food aggregator platforms, whose basic service is to provide a consolidated list of menus from various restaurants, facilitating users to discover, order, and often receive food delivery from a diverse range of establishments through a single digital interface. However, as the paper points out, the relatively high valuation of food aggregator platforms relies on a speculative understanding of the value of labour time of digital platforms. This means that these platforms seek investment, not because they provide the most efficient service and therefore better business fundamentals, but rather because the platform’s copious accumulation of digital labour allows (at least theoretically) for the companies to anticipate and induce demand. The platforms operate, not independently, but through a monopolistic drive to accumulate a greater amount of digital labour.
In the second instance, we examine the rise of networked disinformation as an outcome of digital dispossession. In particular, we draw upon the existing research on the relationship between big tech platforms and the shrinkage of revenue for traditional media. While the problem has been recognised to some extent, attempts to remedy it through regulations like Australia’s News Media Bargaining Code or Canada’s Online News Act have fundamentally failed to grasp the problem (Katz 2023, Bossio, et al. 2022). Rather than the lowered barriers of entry to the publication of news online, it is the monopolistic control exerted through the algorithmic logic of big tech platforms. Far from fragmenting the online news market, we find that the trend in online media has been growing consolidation through disinvestment. The intensified news cycle has meant that news media itself has become more reliant on the editorial subsidy by PR practitioners, inverting the power dynamic between PR agents and journalists. The paper argues that within such a media landscape, disinformation is more easily accessible and economically sustainable.
The paper proposes accumulation by digital dispossession as a framework within which a wider array of phenomena linked to digitisation and datafication may be analysed. In particular the role of datafication and digitisation in rationalising and advancing a speculative form of capital aimed at disrupting ‘traditional’ expectations of profitability.
2. The Inadequacy of Colonial Metaphors of Datafication
The discourse on data colonialism has emphasized the ‘colonization’ of the body and the extraction of ‘data’ from the body (Couldry and Mejias 2019, Thatcher, O’Sullivan and Mahmoudi 2016). More specifically the theorists of data colonialism claim that ‘social life’ in itself is the subject of the new colonial extraction. Couldry and Mejias (2019, 90) argue that “Life, extracted through data relations, acquires a devalued meaning and becomes a mere factor in capitalist production”, defining social life as a territory that is colonized for the extraction of data. Here a crucial distinction is to be made from processes of expanding neocolonial expansion through the process of datafication, which we may identify as ‘digital colonialism’ (Coleman 2019, Kwet 2019, Young 2019). While the latter share a few premises with the former, the distinction in the colonial relationship between the Global North and Global South at large is maintained. Here we shall look at the key arguments made by Couldry and Mejias in favour of the use of colonial metaphors in their analysis of datafication, and then look at the shortcomings of such a framework.
2.1. Rethinking the Extractivist Framework
The problem with the extractivist logic of treating user data as ‘raw material’ is raised by Christoph Raetzsch (2016). While Couldry and Mejias rhetorically distance themselves from this extractivist framework, they actually reintroduce it into their analysis through their own framework. I argue that in retaining and insisting upon the effectiveness of the ‘colonial’ metaphor, they dislocate the term from its meaning altogether arguing “it makes no sense to read data colonialism as exclusively a Western project” (Couldry and Mejias 2019, 17). It can be argued that their approach fundamentally ignores the evolving framework of colonialism, especially the role of finance capitalism, which in turn has played a key role in the emergence of big data. The logic of data colonialism is also extended to explain the expansion of tech monopolies into domains long considered the exclusive prerogative of the state (Magalhães and Couldry 2020). That being said, let us look into some of the core claims made by the proponents of the data colonialism framework.
The running theme of Couldry and Mejias’ central text is the distinction of the process of datafication from ‘regular’ capitalism, and its proximity to the actual historical process of colonialism. “This legacy involves a more extreme degree of alienation than usually recognized within a traditional Marxist perspective, because subjects are estranged not only from the products of their labour but from their own personhood, their basic realities as living beings.” (Couldry and Mejias 2019, 84). Thus while the principal actors in the previous epochs of colonialism were individual states, here the role is ascribed to the ‘social quantification sector’, which incorporates an array of companies that are linked to ‘big data’. The issues that they raise deserve much attention within the social sciences, especially with attempts made by the major players of the social quantification sector to reconstitute ‘social physics’. The fundamental misreading of the Marxist theories on primitive accumulation and the appropriation of surplus value of capitalism itself leads further to the misreading of colonialism as disjointed from capitalist accumulation.
The idea that value is generated through the extraction of data itself has found quite a great deal of support within the field of datafication. That is how the World Economic Forum (2012) framed the issue for example. The simplicity of the metaphor of an extractive form of colonialism has been a key contributor to its popularity as it is easy to envision big data corporations drilling for data in our minds in the manner of fracking companies. The authors identify correctly that colonialism has relied on the myth of terra nullis that dispossesses and renders invisible the ‘ownership’ of territory by existing communities. Here they demonstrate that the metaphorical representation of data as oil also renders invisible the labour of users on social media platforms. However the distinction between oil as a ‘raw’ material and oil as a commodity is crucial. The question of the value of a commodity cannot be resolved without taking the congealed labour contained within the commodity into account. Here we may draw upon the critiques of the land-labour binary in the studies of settler colonialism made by Shannon Speed (2017). Pointing to the context of labour regimes in Latin America (Abya Yala) like encomiendas, repartimientos, or haciendas, she argues that such organization of labour was the very basis of the dispossession of indigenous people from their lands.
Stefano Calzati
(2020) attempts to recontextualize the claims regarding data colonialism by
proposing that the process emerges out of a ‘complex networked ecology’. Firstly, he argues that the reading of
colonialism while ignoring violent displacement as an integral aspect of it,
perils the conceptual clarity of datafication or colonialism itself. Following
Segura and Waisbord (Segura and Waisbord
2019, 416) “colonialism is unthinkable without violence—the takeover of lands
and populations by sheer physical force”, Calzati
argues that colonialism cannot be read without historically situating it in its
particular context. Secondly, Calzati problematizes the attempt to frame the question of
data colonialism around two imperialist poles (between China and the US). He
borrows
Christian Fuchs (2021) also points to the fact that while certain aspects of datafication are imperialistic, it doesn’t adequately address the whole range of activities related to the process. Fuchs draws from Rosa Luxemburg and Lenin in order to underscore the specific relationship between colonialism and capitalism. The relationship between physical violence and extermination with colonialism is emphasized, in contrast to data colonialism’s equivalence in the “gradual elimination of social space” i.e. the substitution of previously ‘social’ forms of interactions through digitally mediated forms (Couldry and Mejias 2019, 107). He therefore objects to the colonial analogy being used in the case of the Cambridge Analytica scandal, especially when child slaves continue to be used in places like Congo for “extracting conflict minerals” for computer hardware and data-driven military technologies inflict violence on civilians in the Global South (Fuchs 2021, 54). On the other hand, he points to programs like Facebook’s Free Basics that involves Big Data practices that enable the Global North to accumulate capital and power as a form of big data imperialism.
Densua Mumford (2022) agrees with the basic impulses of studying the extractive rationalities of data colonialism but feels that the authors do not adequately address the ‘decolonial’ possibilities. First, she points to the failure of addressing Quijano’s problematic of the origins of coloniality as a Eurocentric epistemology, thus converting coloniality into a motif rather than the main theme of the argument. Second, they fail to account for the analyses of capitalism offered by decolonial theories themselves and thus are also unable to engage with theorists of neo-colonialism.
The question that naturally arises is whether the colonial metaphor could be substituted while maintaining the overall integrity of the conceptual framework. The answer is emphatically in the negative. The fundamental claim to the distinct nature of datafication is based on asserting its epochal character, which Couldry emphasizes in his interview with Chang Jiang (Chang and Hao 2020). While the role of colonialism has been fundamental to the continuing existence of capitalism itself, the main issue they seemingly have with data colonialism is a form of dystopian function creep. It must be clarified that the problems highlighted by Couldry and Mejias aren’t to be dismissed altogether. Indeed, the issues of privacy, surveillance, the undermining of social security, and labour rights are critical issues to be addressed. However, the single-minded drive to retrofit these questions into a theoretical model proves to be counterproductive.
2.2. Theorising Digital Labour in the Colonial Context
The purpose of this essay is not to consign colonialism into the bin of history but rather to read it as an existing, evolving and integral process linked to capitalism. I argue here that the process of datafication itself must be studied through the lens of labour and its organisation in the dominant mode of production i.e. capitalism.
Alessandro Gandini (2021) has pointed out that digital labour has remained largely a catch-all category that has been variously used to describe a wide range of paid and unpaid labour mediated by digital technology. Thus acknowledging the two broadly distinct readings of digital labour is critical here. The first is linked to platform capitalism and mainly what is identified as the gigification (or uberization) of the economy. Here the reclassification of existing categories of labour through digital platforms is used primarily to bypass legal classifications. The second is linked to the emergence of social media giants and the source of their super profits. While we will be dealing generally with the second, it needs to be mentioned that the two readings have significant linkages.
In further outlining the key debates that have taken place in the conceptualization of digital labour, we find two distinct trajectories taken by the arguments. The first argues that the values generated by the interaction between users originate outside of digital media platforms. The key argument made here is that ‘value’ is produced elsewhere and is appropriated through financialization on the platforms. The ‘value’ in question is the labour power of users and the digital media platforms simply performing an innovative form of targeted advertisement (Arvidsson and Colleoni 2012, Kangal 2016, Bolaño and Vieira 2015, Robinson 2014). The second draws upon Dallas Smythe’s conceptualization of the ‘audience commodity’. Smythe (1977) argues that the ‘consciousness industry’ (i.e. advertisement and marketing through television and other media) encroaches upon the ‘leisure time’ of the viewers. Drawing from a Marxist interpretation of the Labour Theory of Value, proponents argue that this represents an expansion of ‘labour time’ (Jhally and Livant 1986). Christian Fuchs (2014, 93-94) advances this position further by modifying the concept of ‘audience commodity’ into the ‘Internet prosumer commodity’.
Recommender Systems are algorithms that form the fundamental basis of the so-called social web and Web 2.0 in general. When a video streaming platform recommends videos, or a social media platform loads a newsfeed, or an e-commerce platform suggests purchases, it is usually done via recommender systems. “The task of recommender systems is to turn data on users and their preferences into predictions of users’ possible future likes and interests” (Lü, et al. 2012). They rely on ‘data’ i.e. interaction of users made available to the system, to provide relevant recommendations to users. We can broadly generalise therefore that recommender systems become increasingly ‘accurate’ or efficient in ensuring a user follows through with the recommended interaction. I have argued for closer attention to be paid to the process of ‘recommendation’ rather than the accumulated user data (Saha 2020). This approach leads us to the understanding of the Recommender Systems as a ‘commodity’ that is constructed through historically constituted user labour. Viewed through this perspective, labour is a historic subject that develops through a dialectical process, not beholden to the immediate demand and supply pressures of the market. Yet at the same time, we may also interpret Recommender Systems as ‘congealed alienated labour’, in essence performing the same role as capital. Following Marx (1867, 342) we recognise that “Capital is dead labour, that, vampire-like, only lives by sucking living labour, and lives the more, the more labour it sucks”. In this interpretation, we focus on the iterative interaction between the prosumer or user and the recommender system, where the living labour of the former sustains the latter.
The difficulty of quantifying the value of labour of prosumers in a geographically dispersed model of production, integrated with an algorithmic mode of management presents its own set of challenges. We may look at similar challenges faced by scholars who have examined similar problems in the studies of Business Process Outsourcing (BPOs). Literature on BPO call centres in India has demonstrated how globalization(s) produce both connections and disconnections (Aneesh 2015). In particular, he demonstrates that the ‘innovation’ lay not merely in diverting calls from the US (and other English-speaking countries in the Global North) but through the twin processes of ‘mimesis’ and ‘neutralization’. The term ‘algocracy’ has been deployed in this regard to explain how the task of managing labour has been assigned to specific algorithms and can be used to represent the merger between management and capital. Algocratic modes of governance can be explained as a system operating through the logic of programming code that provides the operational framework for outsourcing operations (Aneesh 2009). While Aneesh writes primarily in terms of outsourcing operations, the rise of cloud computing has made such algocratic governance ubiquitous. The Salesforce workflow for example is presented as “an automation tool in Salesforce that offers time-saving solutions for making different internal processes of a company switch from manual work and for optimizing various procedures. Workflow rules indicate specific criteria to be met in order to trigger automated actions” (Husar 2020).
It is to be noted here that far from being rendered irrelevant by the existence of a globalized regime of management, the North-South divide remains an integral feature of algocratic governance. The pruning of ‘unwanted’ cultural particulars, identified as ‘neutralization’, and simulation of ‘desirable’ cultural traits, identified as ‘mimesis’ provide a sense of directionality in terms of the flows of power (Aneesh 2015). It is the workers in the Global South, who must invariably perform the labour associated with this process by breaking down all possible outcomes of the interaction down to a script. These interactions do not constitute a new domain of colonialism but simply a reconfiguration of the neo-colonial/semi-colonial dependencies that transposed the direct forms of colonialism.
The question for us becomes to examine if similar labour flows can be attributed to digital labour performed by prosumers on social media platforms. Put alternatively, does the digital labour of social media prosumers in the global south generate use values, which are then realized as profit in the global north?.
3. The Spatio-Temporality of Digital Labour
The confusion about Internet users' roles and labour is primarily due to the variety of activities and ways in which Web 2.0 functions. Many critics of digital labour refuse to analyse it within a Marxist framework due to their inability to perform a differential diagnosis of the rate of profit (and its decline). The second problem is the perceived lack of ‘materiality’ of digital labour which is related to the lack of a spatially defined workplace. Indeed the most ardent supporters of the increasing platformisation of labour tend to argue that removing the structural barriers to better quality of employment for workers in the global south by giving them access to an integrated labour market. In order to understand the flow and realization of profit in the Global North from values generated in the Global South, we must endeavour to understand how the process of datafication draws upon existing categories of labour.
It is important to bear two distinct
critiques in mind before we move forward. Antonio Casilli
(2017) warns that “Using notions such as colonialism, imperialism, and slavery
by drawing broad parallels between present and past times risks trivializing
and dehistoricizing the experience of colonization,
neglecting the specificities of colonial past and geographies” resulting in
theories that rely on the shock value of their comparison to colonialism, but
remain content with abstract equivalences. He proposes ‘coloniality’ as a
serviceable framework. On the other hand, Priyamvada
Gopal (2021) notes that an uncoupling of decolonial framework from
anti-colonial theorization and action may in itself serve towards the
appropriation of decolonial rhetoric into the managerial agenda of university
boards.
3.1. Monopoly Capitalism and the Emergence of the Attention Economy as a Spatio-Temporal Fix
The qualitative transformation of the Internet from Web 1.0 to Web 2.0 is most often reflected in a transformation of aesthetics. While Web 1.0 was dominated by static banner ads, hyperlinks, and HTML 3.2 elements like frames and tables, these features were the outcome of an Internet where data flows were mainly staggered, linear, and temporary. The Internet, mainly dissociated transactions from the physical marketplace, thus enabling consumers to choose from a wider range of goods and services. These accelerated rates of profit were further invested into the rapidly expanding realm of the Internet, which resulted in the speculative Dot-com bubble of the early 2000s.
Sweezy and Baran’s examination of monopoly capitalism fundamentally enables us to understand (1) the entanglement of the Internet and the global waves of economic liberalization in the 1980s and (2) the need for increasingly targeted advertisements as an effective tool for market monopolies. Baran and Sweezy (1966, 112-141) pay specific attention to the growing role of the ‘sales effort’. They argue that in the pre-monopoly phase of capitalism, the role of advertisement was minimal and played a marginal role in the broader supply and demand of commodities. The growth of advertisements is the result of the decline in price competition with the rise of Monopoly Capitalism, which has resulted in the problem of insufficient demand. Advertisement, while often represented in accounting as a production cost, is really a way for the utilization of economic surplus through often speculative investments in the generation of future demand.
Bailey et al. (2022) have drawn upon the early works of Nicholas Kaldor, who argues that advertisements must be understood as a subsidized service, which in turn subsidizes other services. In particular, the disruption to the traditional modes of advertisement that had helped sustain mass media in the previous decades has been key to the growth of social media platforms. Thus rather than a break with the past logic of accumulation under monopoly capitalism, the emergence of new technologies enables a more accelerated form of accumulation. The key issue for social media platforms thus becomes garnering the greatest possible amount of user attention and engagement in the digital space, which would draw greater ad revenue, leading to the formation of what is called the ‘Attention Economy’.
Even before the emergence of Web 2.0, we find increasing interest in the ‘Attention Economy’. Michael H. Goldhaber (1997) provides a primordial understanding of the concept, whilst situated in the midst of the Dot-com bubble that had been an outcome of the proliferation of the Internet. While the technological basis for the Internet has been around for decades, its emergence as a commercialised entity in the era of neoliberalism is by no means mere coincidence. Neoliberalism provided the fundamental rationale that has enabled the governance of the Internet on a multi-stakeholder basis (Chenou 2014, Simpson 2004). If the growth of advertisements can be understood as the outcome of a decline in price competition within monopoly capitalism, the ‘Attention Economy’ is to be recognised as an intensification of this monopolist drive. The apparent ‘competition’ for attention in Web 2.0 is illusory given that ‘views’ are primarily determined algorithmically and platforms prioritize paid views. Ariel Ezrachi and Maurice E. Stucke (2019) demonstrate that contrary to the assumption that greater transparency in data would lead to growing price competition and thus lower prices for consumers, algorithmic oligopolies can emerge at times even without explicit intent for collusion.
David Harvey (2004) broadly theorises a confluence of Marx’s general theory of capitalist accumulation and the specific set of historical processes that underlie the description of primitive accumulation. In order to identify the underlying causes of the persistence of the capitalist mode of production despite the repeated onset of crises brought about through overaccumulation, Harvey points to the existence of periodic ‘spatio-temporal fixes’. It is here that he places the theory of accumulation by dispossession which has four distinct expressions (1) privatization, (2) financialization, (3) managed redistribution, and (4) state redistribution. He locates the ‘new’ imperialism as exemplified in the War on Drugs and the War on Terror of the 1990s and 2000s through a close alignment of the interests of Wall Street and Washington that is necessitated by overaccumulation. Harvey’s (2001) theorization of a ‘spatial’ fix for surpluses generated under conditions of monopoly capitalism is relevant here as he links it to the overall problem of overaccumulation in capitalist societies. First, he argues that capitalism needs to be expansionary in order to survive. Second, innovation in communication technologies plays a critical role in this expansion. Third, the specific cause of a crisis determines the ‘mode’ of geographical expansion. Here we can already identify how the role of the Internet as a form of communication technology can benefit from such a theory, but we can move even further with the analysis.
In examining the rise of neoliberalism, Harvey (2005) traces the crisis of ‘embedded liberalism’ in the late 1960s. The Bretton Woods system had delivered over two decades of relatively steady economic growth but had started to falter. The compact between capital and labour was forged in the post-war era with an interventionist state as the mediator ensuring a relatively reliable welfare state. The crisis of the 1970s resulted in growing unemployment, stagnating wages for the working class, and growing inflation. More than responding to the crisis of capital accumulation, David Harvey (2007) argues that neoliberalism responded to the political threat posed to the capitalist ruling classes by a militant labour movement. Uneven geographical development, which is explained as an aspect of the ‘spatial fix’ under capitalism, was organised by the Wall Street–IMF–Treasury complex with the strategic intent of enabling FDI, especially US investments, into developing economies. This meant that the targeted countries would suddenly find themselves flooded with ‘hot money’ in speculative investments, which could just as swiftly be withdrawn with devastating consequences for the economy (Stiglitz 2002, 17).
The emergence of the commercial Internet stabilised neoliberalism. In particular, the growth of ICTs enabled the rapid inflow and outflow of capital between regions (Neubauer 2011). The new media also transformed the availability of consumer goods and the patterns of consumption by the time of the widespread adoption of the Internet in the US. As an example, Dell’s business model relied on being able to use ‘virtual integration’ between computer component manufacturers in Asia and Europe, and its consumers in the US (Dell 2024). By 1998, Dell.com was set up as an online shop that allowed users to choose different components and build their own computers. This integration between offshoring manufacturing jobs and greater diversity in consumer goods made neoliberalism an acceptable settlement of the capitalist crisis in certain regions. Mike Berry (2019) explores the role played by mass media in promoting the notion of a ‘people’s capitalism’ as neoliberal practices were being implemented in the UK. The rise of online retail sites like eBay and Amazon.com leveraged interactivity between consumers on the website to accelerate consumption, while presenting the Internet as the elixir to the bottleneck of globalised consumption and production.
This brings us back to the Dot-com bubble of 1995-2000 and the high valuations of the Internet-focused companies that were considered unprofitable. Priceline, in the way of example, had an IPO of $16 but opened the first day of trading at $81 a share. These valuations were largely speculative given that Priceline, a company that primarily auctioned excess inventory on flights with a revenue of only $35 million, could not justify such valuation on the basis of conventional metrics (Cassidy 2002, 216). Dot-coms in turn expended generously on advertisement leading to a media bubble, which further pumped the Dot-com bubble. Websites like Yahoo.com relied on this explosion of ad revenue to develop their own businesses and encourage news publishers to branch out online to earn a share of online revenues. The burst of the Dot-com bubble meant that online publishers now had to compete for relatively lower revenue sources and therefore demonstrate better return on investment. This meant greater market segmentation and the rise of social media platforms that could feasibly provide such services.
The general tendency is to view the Internet and social media platforms as a digital space on which social and commercial interactions take place. However, the Internet is equally a temporal ‘fix’ where speculative investments are made in expectation of future profits. The prosumer labour, discussed earlier, interfaces with the speculative investments that allow labour to be congealed into recommendation algorithms to generate use value in the form of consumer demand. Instead of the general accumulation of capital, we have an intensification of the recommender algorithm that leads to the dispossession of existing spaces for ad revenue.
3.2. The Materiality of Digital Labour
Given that the outcomes of most processes of digital labour are often intangible, the materiality of digital labour is often called into question. Hardt and Negri (2004) develop the concept of ‘immaterial labour’ as labour “that creates immaterial products, such as knowledge, information, communication, a relationship, or an emotional response” (108). Given that the labour of users on the Internet is also largely based on the circulation of affects online, digital labour has broadly been categorized as affective (Dean 2014, Jarrett 2016, Pybus 2013). In his examination of the global and transnational dimensions of digital labour, Christian Fuchs (2013) expands upon its definition. Drawing upon Jairus Banaji’s (2010) argument that the capitalist mode of production is compatible with a wide variety of forms of labour. Fuchs thus negates the classification of digital labour as ‘immaterial’ demonstrating the materiality intimately linked to digital labour.
In this reading of the International Division of Digital Labour (IDDL), Fuchs points out that different processes in this New International Division of Labour correspond to different modes of labour organization. Marx’s own conception of the value of labour power, which often has been central to these debates on materiality, notes its equivalence to the value of necessaries required to ‘produce, develop, maintain, and perpetuate labour’ (Marx 1865). Contrary to scholars like Leopoldina Fortunati (2007), who argues that the distinction between ‘material’ and ‘immaterial’ labour corresponds largely to the division between ‘simple’ and ‘complex’ labour, it should be pointed out that a closer approximation of our understanding of material labour is simple reproduction (or caeteris paribus) i.e. labour whose outcome is replicable in measurable quantities.
Utopian fantasies of a ‘post-labour’ and ‘post-socialist’ revolution have themselves been reliant on the racialized organization of labour, often corresponding to the kind of labour that the political vision of automation seeks to do away with (Atanasoski and Vora 2015). Casilli (2021) notes that despite the ongoing rhetoric that proclaims that complete automation is right around the corner, most automated tasks rely on a variety of digital and platform labour. This includes on-demand labour, micro work, and social-networked labour, often reliant on workers located in the Global South. On the other hand, the emergence of ‘digital platforms’ responded to the need for an accelerated process of capital accumulation after the saturation of European and Japanese markets from the post-war recovery in the 1970s. The process of datafication allows platforms with control over data to “maintain a chokehold on key points for accumulation and using the threat of exclusion to extract monopoly rents” (Törnberg 2023). Capitalist ‘fixes’ in the aforementioned forms enable markets to stabilize and ensure future consumer demands, which can be satisfied by exacerbating the precarious conditions of labour in the South.
Daniel Greene and Daniel Joseph (2015) underline three distinct levels of spatial fixes that relate to the process of datafication: (1) the primitive accumulation of time in the social web, (2) the annihilation of time by space in high-frequency trading, and (3) affect rent in virtual worlds. They demonstrate the ‘social labour-time’ accumulated by social media platforms is then financialized through algorithmic trading (which has resulted in increasing demands for ‘real-time’ data) before finally being realized as ‘use value’ in the so-called Internet of Things. They argue that digital spaces are not separate from their material existence, but are directly connected to the material world and that “informationalization of production is always reliant on “dirty” labour-intensive industries elsewhere on the supply chain”.
The ‘digitality’ of digital labour can be understood, not in the production of affects, but in the displacement and dislocation of value from ‘productive’ endeavours towards ‘speculative’ ones. Algorithmic labour such as data annotation work that helps build various AI tools attempts to utilise temporally antecedent labour to mimic aspects of real-time service. We can identify these investments as speculative because many of these innovations rely on future demand based on a presumption of accelerated growth of consumption. For example, despite the lack of any practical applicability of ‘Generative AI’ like LLMs, big tech platforms have invested heavily into the development of the development of larger and larger language models disregarding environmental and financial costs (Bender, et al. 2021). We can place the role of neo-colonialism (and its expression in coloniality) here as structured and strategic violence maintains the lower levels of productivity of the Global South, which in turn makes labour ‘cheap’. Such cheap digital labour constitutes one of the key aspects in the flow of value from the Global South towards the Global North. The relationship becomes clear when we examine ad revenue metrics like CPM where views from the Global North are considered worth more than views from the Global South.
4. Accumulation by Digital Dispossession
Before we examine accumulation by digital dispossession let us attempt to understand by example an instance of digital labour and how it is evaluated. Let us suppose that a particular influencer we follow posts a three-minute-long video of their trip to New York on Facebook. We like the video leading the recommendation algorithm to recommend other such influencers who have visited New York, other trips made by the influencer we follow, and cheap flight ticket websites to New York. The influencer may receive a share on the ad revenue varying in value depending on our geographical location. Each iteration of the same interaction contributes towards an exponential form of accumulation given that the rate of ‘productivity’ of labour time on the Internet increases with time.
The disruptive power of digital capitalism has been venerated and valorised in media by investors as evidence that ‘legacy’ industries will be displaced by ‘tech’. This digital dispossession in turn is preceded and succeeded by the acceleration of accumulation of value in the form of prosumer data and ‘efficient’ algorithms. What enables these disruptions are not the inherent competitive advantage provided by technological advances, but rather massive investments that subsidize operational costs in anticipation of profits that could only be realised in the form of monopoly rents (or offloading overvalued shares on retail investors at IPO). As a working definition, accumulation by digital dispossession can be identified as a growing tendency within neoliberal capitalism whereby profits, whether existing or even speculated, are realized through the disruption of more stable relations of production and with digital labour, specifically financially uncompensated prosumer labour providing the basis of these profits. In order to substantiate our argument, we can take a look at the transformation caused in two industries.
4.1. Monopoly Rents and the Rise of Foodtech
Foodtech remains one of the most visible manifestations of the gig economy in the public spaces of both the developing and developed world. In India, Zomato emerged as a major player in the foodtech sector alongside Swiggy, in the second half of the 2010s. It originated as FoodieBay.com in 2008 primarily as a restaurant listing and recommendation portal, focusing on generating revenue for the promotion of restaurants. Soon it discovered that its userbase could be leveraged to gain monopolistic control over a segment of the market and thus eventually realise monopoly profits. In fact, much before Zomato or its peers had started online food delivery services fast-food chains like Pizza Hut, Domino’s, and McDonald’s all had delivery services. Mumbai’s “Dabbawala“ system formally known as the Mumbai Tiffin Box Suppliers Association, too, has an efficient network of food delivery largely relying on public transport services (Baindur and Macário 2013). On the other hand, online food aggregators have forced expenses upon gig workers including fuel charges, vehicle maintenance, Internet expenses, unpaid waiting time, and at the same time subverting employment regulations.
Figure 1: Doordash
Marketplace GOV and Revenue (Doordash 2022, 2023)
While online food delivery services have emerged in almost all major urban centres in the world, one would be pressed to find a ”profitable“ model in the conventional sense. Major foodtech companies rely on a distinct set of operational metrics, namely Marketplace Gross Order Value (GOV), Net Revenue Margin, and Contribution Margin, in order to create an illusion of a profitable business model. Marketplace GOV is used to represent the total value of transactions on a platform, with only a fraction of it representing the actual revenue. In the case of Doordash for example, the share of revenue in the marketplace GOV represented approximately 12% between Q2 2021 and Q3 2023. Net Revenue Margin i.e. the ”Take Rate“ refers to the fee charged by a “marketplace“ on a transaction performed by a third-party seller or service provider. Once again in the case of Doordash, the partnership plans offered are tiered based on the level of service i.e. 15% delivery commission for Basic, 25% delivery commission for Plus, and 30% delivery commission for Premier (DeForest 2020). The “dashers“ are compensated through a combination of base pay, “tips“, and incentives, which are mainly provided by consumers through the delivery fee and tips. Doordash (and other similar companies) expect that as their platform grows, they will be able to demand a greater share of the total order value. Yet the fact remains that it has failed to expand its take rate without losing its market share significantly.
Figure 2: Doordash Net
Revenue Margin “Take Rate“ (Doordash
2022, 2023)
The contribution margin is the third operational metric used by Doordash, which is important because it reflects the metrics that the platform itself wants to promote to its investors. Contribution profit is the revenue that is retained by Doordash after expenses on sales and marketing and therefore excludes R&D and GA expenses. The inclusion of the latter categories in the calculation would bring us to the conclusion that Doordash operates at a quarterly loss of 5% even as its revenue growth saturates. Conventional wisdom suggests that such business models may be unsustainable, which might as well be the case, but it continues to play a crucial role in capital asset destruction.
Figure 3: Doordash
Contribution Profit and Net Loss (Doordash 2022,
2023)
The path to profitability for Foodtech needs to be understood in the overall disruption of capital flows in favour of monopolist tendencies. At the time of writing most online food aggregators have lost more than 50% of their value since IPO, including Doordash (-65%), Zomato (-59%), Deliveroo (-68%), which would indicate financial structures analogous to a classic ”pump and dump“ scheme. The first stages of almost all such platforms aim to grow the number of ”monthly active users“, often at the expense of profitability. This growth leverages the existence of the attention economy as a means to simulate the traditional trappings of profit, thus drawing investors in multiple waves. As a result the valuation of the company skyrockets, which is when a large number of investors unload the assets through an IPO onto smaller retail investors. In some cases, we find online food aggregators aggressively acquire other delivery-oriented services such as quick commerce. The idea is that they can leverage a readily available workforce and their significant share in the attention economy into domination of a significant share of the market. This means that their “value“ does not lie in any innovative form of delivery of goods or services but rather in a virtually captive userbase, which can be leveraged into a wide range of businesses.
Accumulation by digital dispossession can be understood as a fundamental mechanism that has enabled the rise of foodtech and online food aggregators. In this case, investments are mainly geared towards the acquisition of a user base of prosumers and thus developing a share in the online attention economy. The prosumer labour, subsidised to an extent by institutional investors and promoters, enables the impression of an “asset lite“ company, which receives enormous valuations compared to its actual profits (or losses). These valuations in turn draw speculative investments from low-level investors, disrupting potential capital flows into other more stable sectors. Finally, the foodtech platform leverages the prosumer labour and available capital to other sections of the market.
4.2. Disinformation and Clickbait as Digital Dispossession
While we have examined the rise of the ”attention economy“ within the context of monopoly capitalism, we have not paid any significant attention to the role of disinformation and clickbait. The decline of a paid readership and the shrinking share of revenue that “legacy“ media outlets have forced many of them to cut down on staff (Bakir and McStay 2018). Literature on fake news and disinformation has sought to establish the link between falling revenue margins and the comparatively low barriers to entry into digital media (Martens, et al. 2018). Livingston and Bennett (2020) note the disruption of public sphere institutions through intensified flows of dark money and illicit campaign finances that has led us to the current juncture. The alignment between the right-wing political establishments, media houses, business interests, and think tanks has championed attacks on public sphere institutions that they viewed as detrimental to the operation of a free market. While we may agree with the general outline of the process, it may be argued that the qualitative and quantitative transformation of disinformation in the digital age has its specific characteristics that may be linked to digital dispossession.
Market-based rational choice systems have been used to explain the proliferation of fake news through an equilibrium between consumer choice and profit motives of news purveyors (Basuchoudhary and Razzolini 2021). Daniel Williams (2022) points out that the demand for fake news is generated by consumers’ attempts to rationalise beliefs held for non-epistemic reasons. However, rationalisation markets in themselves are not ”perfectly competitive“, meaning that rationalisation producers hold some market power. Producers of disinformation (or ”rationalisations“ as Williams identifies them) have an incentive to encourage their consumers to completely exclude alternative sources of information that may challenge biases. On the other hand, the elimination of market competition enables newer market entrants to appear on the newsfeed since the lack of diversity in the sources of content would result in lower engagement. This results in demand for a form of content that performs the task of rationalisation while being relatively low investment and capable of saturating the newsfeed sufficiently. Prior to the election of Donald Trump in 2016, the Macedonian town of Veles came to the fore as a number of viral pro-Trump websites were created by its residents. Those who operated these websites regularly copied content from “alt-right“ media like Infowars, disseminating disinformation through repetition and mimesis (Hughes and Waismel-Manor 2021).
Kevin Munger’s (2019) examination of “clickbait media“ explains to some extent how fake news operates in the contemporary media and social media markets. Drawing upon William Baumol, Munger argues that online media markets closely approximate “contestable markets“ as there are no barriers to entry and exit in the market, minimal technological specialisation, and relatively little requirement for capital investments. Here he argues that given the low capital cost of digital media, media firms do not have an incentive to develop a reputation for news accuracy. But this does not mean that the credibility of the news source is rendered irrelevant. Instead, he theorises the existence of “credibility cascades“ i.e. people consuming a particular content believing it to be credible because others they know have shared it.
Munger’s basic assumptions regarding the ease of access to digital media markets however may be premature, and the ”contestable market“ of digital media may also be headed in the same direction of oligopolic consolidation as commercial airlines, upon which Baumol had premised his theory. A survey of disinformation trends in six Eastern European countries by Judit Szakács (2020) demonstrates how while the dissemination of misinformation and fake news is attributed to small media outlets, the process of ad-driven misinformation has slowly led towards conglomeration, in many cases driven by mainstream news sites that also engage in the proliferation of misinformation in Eastern Europe. While it can definitely be argued that the relative cost of entering the media market has been lowered, this has not guaranteed the kind of access to engagement that could sustain independent media houses. Given the collapse of subscription-based news, the pace of consolidation has increased drastically. Bernie Sanders, the US Senator, pointed this out in 2017:
In 1983, the
largest 50 corporations controlled 90 per cent of the media. Today, as a result
of massive mergers and takeovers, six corporations
control 90 per cent of what we see, hear, and read…These powerful corporations
also have an agenda, and it would be naive not to believe that their views and
needs impact coverage of issues important to them. (Sanders 2017)
Those media houses that are not acquired directly by one of the largest corporations are forced to compete within the highly consolidated attention economy. Faced with a growing demand for content, journalists who are increasingly under stress by deadlines rely on “editorial subsidies“ from PR practitioners (Jackson and Moloney 2015). With PR campaigns becoming increasingly more and more sophisticated, and journalistic resources becoming increasingly scarce, the power relationship between journalists and PR practitioners has shifted in favour of the latter. In his examination of the revival of the conservative movement after the US presidential election of 2008, Richard Meagher (2012) pointed out that dominating and controlling the media narrative has been an aim of the right-wing/conservative establishment since at least the 1970s. The more recent iteration of this ongoing process is visible in the rise of alt-tech and the discourse on the “parallel economy“ being proposed through platforms like Gab, Parler, and Truth Social (Johnson 2022).
In the case of the contemporary rise of the disinformation industrial complex, accumulation by digital dispossession can be seen in investments made by PR firms and right-wing establishments towards the dismantling of public sphere institutions. Prosumer labour drove online media’s dependence on ad revenue, which in turn was favourable to a fractured landscape of media consumption. In turn, this fractured landscape allowed disinformation websites to operate supplementing the increasing amount of disinformation of mainstream or legacy media. Finally, we find efforts to increase monopoly profits in the social media landscape by completely creating new websites and financial structures catering to exclusively right-wing consumers.
5. Conclusion
The problem of theories of data colonialism lies in its obsession with the value of data extracted from users, to the exclusion of the labour performed by prosumers on the Internet. Returning to the proposed working definition of accumulation by digital dispossession, we find a few key features that can help in identifying the phenomenon altogether. First, the presence of uncompensated prosumer labour remains the basis for this accumulation as it is congealed into the algorithmic function of these platforms. This accumulation forms the basis of algocratic power on the Internet that is able to subject prosumers to an invisible (or even sometimes visible) regime of labour organisation. Second, investments made into these platforms are usually made in anticipation of monopoly profits, meaning that these platforms effectively are provided extensive subsidies by the market. Their unit economics is simply irrelevant to their entire model of profitability. Third, the disruption of existing models of profitability and the relations of labour means that cartelisation remains the only viable outcome and conclusion of the combination of these processes.
My intention here is not to provide a conclusive argument regarding what is being described as accumulation by digital dispossession, but rather an open-ended theorisation. The approach can be used to explain a wider range of phenomena in the digital economy including Fintech, E-Commerce, online dating platforms, and other digital marketplaces. Datafication itself is to be understood as an extension of historical processes of accumulation of surplus value in the aftermath of neoliberalisation. As a growing number of platforms are compelled to either transition towards increasing reliance on subscription-based services or rely on increasing ad revenue, the exchange value of user data is forced to track closely to its use value as a commodity for speculating on and influencing consumer demand. Its wider impact on the Internet remains to be understood.