The theoretical foundations of totality

Gandini (Reference Gandini2021) argues that ‘digital labour’ has become an umbrella term whose internal referents bear little connection to its original meaning, thereby undermining its utility for precise critical or analytical purposes. We do not deny the expansion of the concept’s connotation, nor do we dismiss the genuine concerns underlying this critique. Rather, we advocate for a new conception of the concept of digital labour from the perspective of totality, demonstrating its unique analytical efficacy, and pointing out that this concept need not be understood as an umbrella term; it can and should be understood as a term of totality.

Totality is a major idea in the Marxist intellectual tradition. Marx (Reference Marx, Marx and Engels1986, 38) held that the ‘concrete totality’ is not a pre-given entity but the result of a specific analytical procedure, ‘the assimilation and transformation of perceptions and images into concepts’. He distinguished this ‘rich totality of many determinations and relations’ from ‘a chaotic conception of the whole’, insisting that the concrete totality is ‘a unity of the diverse’ (Marx Reference Marx, Marx and Engels1986, 37–38). Importantly, Marx (Reference Marx, Marx and Engels1986, 43) also pointed out that within any concrete totality there exists a dominant moment – ‘the general light’, ‘tingeing all other colours and modifying them in their specific quality’. For Marx, capital is precisely this ‘general light’, ‘a special ether’ that determines the specific gravity of all that emerges within it. The method of the concrete totality thus demands that the analyst identify the dominant structural principle that organises the whole.

Lukács (Reference Lukács1972) advanced the inseparable connection between totality, subjectivity, and the critique of reification. The capitalist system generates reified appearances, fragmenting social reality into seemingly independent spheres – production, distribution, consumption, labour, leisure – thereby concealing their structural interconnections. Only through the category of totality can the proletariat penetrate the reified surface of capitalist society and grasp its own position as the ‘identical subject–object of history’. Kosík (Reference Kosík1976) identified the primary obstacle to grasping the concrete totality as what he termed the ‘pseudoconcrete’, the world of everyday appearances, fetishised practices, and routinised thinking that presents social relations as natural, independent, and self-evident. He did not deny the objective existence of everyday phenomena but destroyed their ‘fictitious independence’ by demonstrating their mediated character. He insisted that the concrete totality is not merely a cognitive construction but is generated through practice.

These three theoretical resources provide the scholarly support for the perspective of totality advanced in this article, but we draw upon them selectively and critically. From Marx, we derive the methodological core: the concrete totality as the result of a dialectical procedure, organised around a dominant structural principle. From Lukács, we retain the insight that totality is necessary to penetrate reification and grasp structural interconnections; however, we do not accept his Hegelian assertion of the proletariat as the identical subject–object of history. From Kosík, we appropriate the critique of the pseudoconcrete and the insistence that destroying fictitious independence is a practical-critical act; yet we do not follow him to the phenomenological tendency of equating reality with the totality revealed only by practice. Together, these three elements form a solid framework for rethinking the concept of digital labour, not by taking totality as a static container into which diverse phenomena are indiscriminately placed, but as a dialectical method that penetrates the reified appearances of digital capitalism, reveals the structural interconnections among differentiated labour forms, and seeks to awaken the subjective consciousness of digital workers.

A clarification is needed here regarding the use of the category of ‘totality’. This article employs the category at three interrelated but non-identical levels. First, at the level of reality, there exists in digital capitalism an objective totalising process driven by the logic of capital, which integrates differentiated labour forms into the same structure of valorisation through platform architectures, supply chain management, and data extraction. Second, at the methodological level, totality is a dialectical analytical procedure that penetrates reified appearances by tracing the dominant structural principle and reconstructs the structural connections among apparently isolated phenomena. Third, at the conceptual level, the concept of digital labour constructed through the method of totality ceases to be an empty umbrella term and becomes a ‘concrete totality’ that grasps a multiplicity of determinations – that is, a category possessing totality.

The subject of this article is the totality of the concept of digital labour, but as noted above, the concept of totality cannot be divorced from social reality. Mediated by the method of totality, it is generated under the influence of social reality and acts back upon it. Understanding the concept necessarily involves grasping reality and applying the method. The totality of the concept is inseparable from totality at the levels of reality and method; it depends upon the operation of totality at the methodological level – it is a product of theoretical synthesis, not an a priori property inherent in the concept, nor can it be directly equated with an assertion of a completed state of totalisation of reality. Furthermore, as a product of thought, this category of totality possesses a distinct cognitive function: it enables the analyst to discern the structural interdependence hidden beneath differentiated labour processes, forms of value extraction, and conditions of existence, thereby providing the epistemological precondition for traversing the divisions manufactured by capital and identifying the possibility of shared resistance. Here, the concept is not merely thought’s reproduction of reality but also a cognitive instrument for intervening in reality.

The analytical efficacy of totality

The totality of the concept of digital labour offers three unique contributions and provides new ways of thinking about important issues related to digital labour.

First, totality concretely specifies how capital, as the dominant structural principle, organises differentiated forms of digital labour into a functional structure of mutual dependence. This analysis follows the procedure of Marx in the ‘Introduction’ to the Grundrisse: it starts from the inner imperatives of capital valorisation and traces how, through specific mediating links, capital posits these labours successively as functional moments that require one another. The following analysis performs this movement on the commodity chain of a generative AI foundation model.

The analysis proceeds from capital’s abstract imperative: a technology company seeks to valorise a foundation model – a large-scale AI system such as a large language model – whose capacity to generate text, images, or code on demand is the commodity to be sold. To realise this imperative, the model must first be designed, architected, and trained by research scientists and software engineers. Their skilled technical labour determines the volume and type of training data required, sets the standards for downstream annotation, and specifies the computational resources that infrastructure must provide. Training the model presupposes enormous datasets of text, images, and digital content, which posits data extraction labour – the uncompensated activity of billions of users whose posts, comments, images, and clickstreams are harvested as raw material. This is Terranova’s ‘free labor’: at once social practice and productive act, captured without wages, and functionally posited as the source of raw material for the entire chain. Yet raw data must be cleaned, filtered, classified, and labelled, which posits data annotation labour – waged but precarious work concentrated in the Global South, where workers in Venezuela, the Philippines, and Kenya, among others, perform these tasks under the piece-rate logic of platform labour (Casilli Reference Casilli2025). The platforms from which training data are extracted are not natural data reservoirs but commodified environments actively maintained to capture and monetise user activity. This posits content moderation labour – workers, again in the Global South, who filter harmful material and enforce community guidelines (Roberts Reference Roberts2019). Their labour produces the ‘brand-safe’ space that sustains advertising revenue and, in the AI commodity chain, directly safeguards training data quality by preventing contamination from toxic or spam content. Finally, all these activities depend on the physical substrate of computation – data centres, specialised chips, electricity, and water – which posits infrastructural labour: mining cobalt and rare earths, manufacturing semiconductors, and constructing energy grids. Although these forms of labour are often classified as ‘non-digital’, they are equally posited by the same valorisation logic. The result is a concrete totality of mutual positing: the abstract imperative concretises into a functional coupling structure in which each labour form is posited and required by the others. Technical labour organises the conditions that extraction, annotation, and moderation labour must meet; annotation labour depends on extraction labour’s raw material and technical labour’s specifications; moderation labour sustains the platform environments on which extraction labour depends; and all are grounded in infrastructural labour. Capital, as the ‘general light’, does not erase these differences but organises them into an interdependent hierarchy. This is the functional coupling structure that the perspective of totality makes analytically visible – a structure that escapes single-site labour process analyses and expands as new forms of labour are drawn into the same valorisation architecture. This functional coupling structure is not a neutral technical arrangement. In organising this structure, capital systematically exploits existing inequalities of gender, race, and geographic space (Amrute Reference Amrute2016; Graham et al Reference Graham, Hjorth and Lehdonvirta2017; McMillan Cottom Reference McMillan Cottom2020; Rossiter Reference Rossiter2016; van Doorn Reference van Doorn2017) to construct a differentiated labour market, ensuring that the substitutability and irreplaceability of different types of labour are configured in the manner most advantageous to capital (Mezzadra and Neilson Reference Mezzadra and Neilson2013).

It is this cognitive function – refusing capital’s classifications as the starting point of analysis and instead making those classifications themselves the object of inquiry – that constitutes the irreducible analytical contribution of the concept of digital labour. No composite description of ‘platform labour plus user labour plus their interconnections’ can replicate this function because such descriptions necessarily operate within the very taxonomy that the concept of digital labour, as a category of totality, seeks to subject to critical scrutiny. From this perspective, the diverse forms of labour under digital capitalism do not form a static field to be sorted into ever-finer classifications; rather, they are the product of strategic processes that themselves require explanation.

Second, the unique advantage of totality becomes particularly clear at the methodological level. The existing debates about the sources of digital platform profits provide a paradigmatic case. This debate crystallises a fundamental difference between two analytical paths. One can be termed the factor-analytical path, which seeks to attribute platform profits to one or several decisive sources. Fuchs (Reference Fuchs2012b) insists on an explanation grounded in the labour theory of value, arguing that users’ unpaid digital labour directly creates surplus value; Arvidsson and Colleoni (Reference Arvidsson and Colleoni2012) turn to the logic of financial valuation, contending that platform value mainly stems from the construction of affective networks and their realisation in financial markets; Rigi and Prey (Reference Rigi and Prey2015) point out that platform revenue comes from multiple channels, including advertising rents, the sale of information, and speculative virtual capital; Sadowski (Reference Sadowski2020) relocates users as ‘buyers’, framing the question on the side of rentierism. The shared presupposition of these positions is that platform profits can be analytically decomposed into several independent sources, and the task of theory is to determine which source predominates. Yet the above analysis of the functional coupling structure demonstrates precisely that this presupposition itself obscures the crux of the matter. Within this structure, none of these so-called ‘sources’ generates profit in isolation. Profit is not the simple sum of elements but arises from the functional coupling of these moments – from the relation in which one moment posits another as the condition of its own realisation. The factor-analytical path attempts to carve out an independent ‘ultimate source’ from within the coupling structure; this is itself a category error: it presupposes that something which exists only within the totality can be known prior to the totality. The method of totality transforms the mode of questioning. It does not ask ‘which element creates value’, but rather ‘how the functional interconnections among the various elements constitute the necessary conditions for the realisation of value’. It is precisely through this transformation of the mode of questioning that unpaid user activity, advertising sales, rent extraction, and financial valuation are presented as different moments of the same functional coupling structure, rather than as competing explanatory options.

Current research already includes theoretical efforts to grasp digital labour through the method of totality. Fuchs and Sandoval (Reference Fuchs and Sandoval2014) unify the various labours along the global ICT value chain under the definition of ‘digital labour’ and analyse their structural interconnections; Faraoun (Reference Faraoun2024) provides an analysis of historical stages; Wang and Tomassetti (Reference Wang and Tomassetti2024) focus on the analysis of labour–capital relations. These studies correctly identify the interconnectedness of labour forms, but each focuses on a particular aspect of structural connection and tends to offer the conclusion of totality rather than to demonstrate its constitutive process – that is, how the connections are established step by step through specific mediating links. The method of totality advanced in this article aims to push forward precisely in this direction: to transform the conclusion that ‘interconnections exist’ into an analytical procedure of ‘how they are interconnected’.

This analytical move also distinguishes itself from other major frameworks: Vercellone’s (Reference Vercellone2007) thesis of cognitive capitalism focuses on macro-historical periodisation; Casilli’s (Reference Casilli2025) typology of platform labour concentrates on the mechanisms of classification themselves; Mezzadra and Neilson’s (Reference Mezzadra and Neilson2017) analysis of the operations of capital foregrounds the operational logic of capital. The distinctive contribution of the method of totality lies in the fact that it does not replace the respective analytical functions of these frameworks but reconstructs, between the differences they have already precisely described, the functional coupling relations mediated by the logic of capital valorisation.

Third, the above analysis ultimately points towards a re-understanding of the concept of digital labour itself and demonstrates the political potential of the concept of digital labour as a totality. Having undergone the dialectical reconstruction of the method of totality, the concept of digital labour is no longer an umbrella term that merely collects heterogeneous phenomena but a category of totality that grasps a multiplicity of determinations. What distinguishes it from an empty signifier is not the breadth of its coverage or the refinement of its classifications but the fact that it requires the analyst to proceed not from the extension of the concept to enumerate examples but from the inner imperatives of capital valorisation to trace how various labour forms are posited, through specific mediating links, as mutually required functional moments. It is precisely this that fundamentally responds to Gandini’s critique: the validity of ‘digital labour’ depends not merely on the precise delimitation of its referential scope but more crucially on whether it can accurately grasp and illuminate a core ‘problematic’ (Agamben Reference Agamben2009) of contemporary capitalism – namely, how capital, through digital technologies, transforms life activity in its entirety into a field of valorisation.

This category of totality simultaneously contains the cognitive potential to penetrate reification. The functional coupling structure demonstrated above is not only the objective mechanism of capital valorisation but also constitutes the material conditions of existence for digital workers. Digital labour processes transcend spatial, temporal, and community boundaries. Workers dispersed across different continents and different labour processes are thus objectively placed within the same structure of mutual dependence. It is precisely this structural interdependence that provides the objective basis for the solidarity of digital workers. The cognitive function of the category of totality lies precisely here: by revealing the structural connections among separate struggles, it demonstrates why and in what sense the struggles of different groups are related to one another. This is the dialectics by which totality disenchants labour fragmentation, revealing that the seemingly isolated ‘monadic’ suffering is in fact a node in a web of shared fate.

Nevertheless, the category of totality must also take seriously the question of the heterogeneity of digital workers. Digital labour is a field criss-crossed by different stakeholders who engage in vastly different occupations, face different forms of exploitation and labour conditions, and may be unable to understand one another in many respects (Genovesi Reference Genovesi, Genovesi, Kaesling and Robbins2023). On this basis, Englert et al (Reference Englert, Woodcock and Cant2020) have already conducted important research on the organising strategies and struggle practices of different types of digital workers. Yet an overemphasis on heterogeneity risks making common composition impossible and heterogeneity may become a trap that forecloses revolutionary possibility (Roggero Reference Roggero2014). This is precisely the political challenge to which the category of totality responds. The category of totality does not deny this heterogeneity, nor does it attempt to erase concrete differences with an abstract concept of the ‘collective worker’. On the contrary, it demands that we traverse heterogeneity to reach the structural interdependence hidden beneath differentiated labour processes, forms of value extraction, and conditions of existence, thereby creating the cognitive possibility for bridging the divisions manufactured by capital and constructing real solidarity.