According to the speculations of techno-futurologists, left and right, the machines are here to liberate us. Most of the discourse is dominated by the neoliberal right such as Erik Brynjolfsson and Andrew McAfee and Andrew Haldane, chief economist of the Bank of England. Their arguments, avoiding questions of exploitation, are naturally popular with the establishment. McAfee’s best-selling book The Second Machine Age has been lauded by leaders at the World Economic Forum.
On the left, however, Paul Mason welcomes our new robotic overlords, in an intellectual synthesis that spans Marx’s 1858 ‘Fragment on Machines’ (treated by Mason as a prophecy), Bogdanov’s 1909 novel Red Star and Martin Ford’s 2015 Rise of the Robots, not to mention Andre Gorz. Nick Srnicek and Alex Williams offer a more qualified welcome to the possibility of full automation and a workless future. But even the best of these analyses, and even the most alluring visions of networked insurrection and high-tech communist utopia, have to face up to how these technologies have been used, historically, to deepen exploitation rather than overcome it. It is far more likely, in short, that new technologies will intensify drudgery and further limit human freedom. And it on this basis that we have to evaluate the impacts of platform technologies on the capitalist mode of production.
In Platform Capitalism, Nick Srnicek provides one of the first systemic Marxist interventions into the discourse around data-driven digitalisation, automation and the future of work. ‘Platforms’ are ‘digital infrastructures that enable two or more groups to interact’ within the constraints of the capitalist system. ‘Platform capitalism’ does not simply refer to the rise of alternative work arrangements such as temporary, independent, or other forms of precarious labour contracts, but rather an organisational shift in the system as a whole due to financialisation, increasing inequality, and the tech boom. According to Srnicek, the evolution of internet-era behemoths like Google, Facebook, Amazon, and Uber, as well as radically modernised pre-internet companies like GE, Siemens, and Rolls Royce has fundamentally altered the landscape of capital accumulation and property relations between firms. It is important to remember that the US military and other state-funded bodies made much of the original technological innovations for computing and logistics. The emergence of platform capitalism is essentially the commercialisation and industrial maturation of data-based social relations, theorised in the 1980s as ‘information capitalism’. Does the emergence of platform capitalism constitute a new mode of exploitation? In order to address this question, we must situate the empirical fact of their existence within a historical and theoretical context.
The evolution of these firms is inextricably bound to the history of asset-price Keynesianism. From approximately the mid-1990s, bubbles in asset-prices temporarily drove investment and created jobs and growth where there would have been known. This was inaugurated with the dot.com bubble. During the economic boom of the 1990s, huge financial investments were poured into telecommunications infrastructure. Millions of miles of new cable and major advances in software and network design allowed for the commercialisation of the previously non-commercial Internet. After the dot-com bubble burst in 2001, the combination of financial deregulation and an ever-increasing demand for financial assets led to another crisis in 2007-8, triggered by complex mortgage-backed securities. The crisis response of central banks, including quantitative easing and the lowering of interest rates, weakened returns on the more traditionally secure financial assets. This encouraged investors to look toward other asset containers – mostly property and the tech sector or what would soon be known as the emerging platform economy. Largely as a result of this staggering amount of new investment, the technology and connectivity required to transform every day human activities into digitally recorded data became relatively less expensive and widely available. Srnicek claims that this marked the twenty-first century shift toward the period of ‘platform capitalism’ in which data collection and monetisation is standard business practice.
Platforms are defined by four attributes: they provide an infrastructure for mediating exchanges between different groups; they follow monopoly tendencies driven by network effects; they strategically cross-subsidise different parts of the business in order to diversify user groups; and they maintain a proprietary architecture that mediates interaction possibilities. These attributes are too broad to tell us anything about the mode of exploitation involved; however, Srnicek’s typology of ‘platforms’ is based on their methods of revenue-generation: advertising, cloud-based service, industrial production, product-rental, and lean or gigging hubs. Advertising platforms (Google, Facebook) extract information on user behaviour, analyse that data, and sell it to advertisers. Cloud platforms (Salesforce) own hardware and software that are rented out to digital-dependent (read: nearly all) businesses. Industrial platforms (GE) are modernised hybrids of traditional manufacturing and contemporary logistics that use proprietary hardware and software to provide services and lower production costs. Product platforms (Rolls Royce) also transform traditional goods into rented services by collecting fees for the use of their products. Finally, lean platforms (Uber, Airbnb, Deliveroo) outsource all asset ownership other than software and data analytics, then profit as digitally savvy middlemen disrupting established markets (the impact of which will be discussed in more detail below). Each type of platform often combines one or more revenue models to make a profit; however, the most important asset for platforms is their intellectual property – company software, algorithms, and user data.
The reliance on diverse revenue models raises questions firstly about the structural position of platforms in the overall circuit of capital accumulation, and secondly about whether in the future we will continue to regard the worker as central to production. To address this, it is important to understand the rise of the platform in relation to different forms of exploitation or means of profit-making. The late twentieth century produced a procession of post-capitalist prophets who sought support in Marx’s writings, going back to the Grundrisse, to justify the idea that the workers of the world might eventually ‘step to the side of the production process instead of being its chief actor’. As Tessa Morris-Suzuki argued in the 1980s, the exploitation of surplus labour as the primary source of profit was not, according to Marx, ever intended as an eternal economic law. It was the defining characteristic of industrial capitalism, a particular historical system that evolved out of merchant capitalism, which in turn evolved from feudalism. Marx established that at the most abstract level, aggregate profit is essentially the monetary expression of aggregate surplus value; however, within the circuit of capital, firms can also accumulate profits through unequal exchange and redistributive phenomena between social spheres. Profiting through the former mode is called ‘primary exploitation’ while profiting through the latter is called ‘secondary exploitation’ (this distinction will be further elaborated below). Does the rise of the platform simply indicate a shift from primary to secondary exploitation, or does it represent a new mode of exploitation entirely? Has labour ceased to be the main source of surplus value?
In Platform Capitalism, Srnicek offers an innovative framework through which to address this question in his conception of data as ‘raw material’. Data is defined as ‘information that something happened’. It is distinguished from knowledge or ‘information about why something happened’. The act of recording data is either labour carried out by a human or a function of a human-programed computer algorithm – or, often, both. The production of data thus relies on labour power and a material infrastructure. Data is the ‘raw material that must be extracted’ from the ‘activities of users’ which are the natural source of this raw material. For Marx, raw materials are those parts of nature that have been filtered through previous labour (for example, ore that has been extracted from the earth). Nature is any environment that can exist independently of humanity and serves as the ‘universal subject of human labour’. For example, water is found in nature; yet when it is separated from a river, filtered, and stored in tanks, it serves as a raw material. Srnicek extends this Marxian distinction beyond flora and fauna into the realm of human activity itself. Nature becomes any potential activity humans perform in their daily lives: economic transactions, consumer tastes, user movement, location, and so on. The mining and processing of these activities as data transforms it into the raw material, which can be used in the production of service commodities.
The production of service as a commodity is just like the production of a good as commodities, except for the fact that in the production of service commodities, use-value might vanish with the cessation of the labour-power itself due to the relative simultaneity of production and consumption. We should remember that for Marx, the commodity, as a materialisation of labour in the form of its exchange value, is an imaginary and ‘purely social mode of existence’, which has ‘nothing to do with its corporal reality’. All that matters is that the labour process is subsumed into the capitalist form of primary exploitation. Primary exploitation takes place in the labour process itself and can be productive of surplus value, which is translated into profits through markets and competition. It is the human and social process of ‘exploitation of the workman’ by the capitalist, which relies on a classed monopoly of power over the means of industrial production. The rate of surplus value extracted in the labour process is ‘an exact expression for the degree of exploitation of labour-power by capital, or of the worker by the capitalist’. It is important to note that the rate of surplus value is not an expression of the ‘absolute magnitude’ of the exploitation because not all exploited labour produces surplus value. Platforms can profit from exploitation of surplus labour yet not produce any new value because that labour is socially unproductive. Yet, they can also profit through the production of surplus value.
Harry Braverman, writing in 1974, pointed out that when a worker does not offer labour ‘directly to the user of its effects’, but rather sells it to a capitalist, who re-sells it on the market, this is ‘the capitalist form of production in the field of services’. Many services, such as education and health care can also be productive of surplus value if they take a capitalist form. Marx himself referred to the transport industry as a service that was productive of surplus value. As service producers, platforms like Deliveroo are value-productive since the commodity produced is the change of place itself. Amazon provides a similarly productive service in their logistics centres. Google and Facebook sell advertising space and consumer behaviour data. These platforms are arguably a further development of a longer trend away from producing goods as commodities toward producing services as commodities. Over the past several decades, nearly all developed economies have seen a gradual decline in manufacturing production and a rise in services as a share of employment and GDP. The UK in particular has seen a sharp rise, with ‘services’ now comprising 79% of value-added GDP.
Cloud, and product platforms do not produce services as commodities; rather, they accumulate profits in the form of rents or other means. They are not productive of value and signal a potential shift from primary to secondary exploitation in profit making. Secondary exploitation or ‘profit upon alienation’ takes place primarily through financial and property relations that facilitate the collection of interest payments, rents or profits through unequal exchange (merchant’s capital). This form of exploitation appropriates surplus labour performed elsewhere and in doing so, merely redistributes a portion of the total surplus value of society. The existence of secondary exploitation allows for two things: first, it explains how commercial or financial capitalists can profit from non-capitalist spheres without the creation of new value; and second, it allows Marxian economics to account for the difference between the sum of profits and the sum of surplus values that emerges as values are transformed into prices. This is because profit is not the same as surplus value, though the rate of each tends to equalise over time.
The developmental arc of a successful platform generally begins with the technological disruption of an existing industry, and ends with the platform achieving the status of an industry gatekeeper or quasi-monopoly status. As platforms expand, they capture an increasingly large amount of data. Their quests for gatekeeper status lead them to diversify and encroach on one another. The rapid expansion of platforms has resulted in new monopolies, which now provide the basic digital and logistical infrastructures upon which much of the economy operates. The increasingly privatised ownership and management of public services and business infrastructure is indicative of the aforementioned shift from primary to secondary exploitation.
The enclosure of electronic ecosystems is a particularly interesting instance of secondary exploitation. Facebook has pursued a strategy of ‘funnelling of data extraction into siloed platforms’ in Africa and other lesser-developed areas of the world. Their ‘Free Basics’ program has brought Internet access to over 37 countries and 25 million people; however any service other than Facebook that wants access to these users is required to partner with the company and operate through their network and software platform. This combines monopoly (a sole producer) and monopsony (a sole buyer) power to reproduce the exploitative dynamics of accumulation by dispossession. Through these programs, the extractive apparatuses of imperialism have found their contemporary counterpart in the global enclosure of digital infrastructure and mining of data. Facebook’s ideology of connectivity as a good in itself simply serves the company’s interest and reproduces the exploitation of the economic periphery.
Each platform’s relation to a given mode of exploitation ultimately depends on its concrete form. Advertising platforms like Google and Facebook as well as lean platforms like Uber and Deliveroo use their intellectual property to mine data as raw material, which becomes one of the elements of constant capital in the selling of service commodities. Advertising platforms sell access to billions of users through sophisticated communication and consumer behaviour patterns, while lean platforms sell their particularly efficient means of transportation and user base and profit through a combination of fees. They also tend to rationalise informal economies of petty commodity producers and consumers into a formal economy mediated through proprietary means. Industrial platforms have a more traditional means of profit-making. Cloud and product platforms like Saleforce or Rolls Royce primarily extract rents from the use of proprietary technologies and infrastructure. Each platform uses a combination of primary and secondary means of exploitation to make a profit.
It is difficult to abstract from the concrete relations of the labour process. However, the classification of concrete human labour and its place in the industrial circuit of capital allows us to understand the relation between the production of value one the one hand, and the mere of accumulation profit on the other. It is for this reason that the analysis of platforms must avoid fatalistic metaphysical claims like Negri’s insistence that ‘there is no outside to capitalism’. Accounts of socio-economic phenomena that flatten distinctions between labour and non-labour under capitalism (for example, Beller’s claim that ‘looking is labour’, that merely glancing at an advertisement is productive of surplus-value), or between industrial capitalist and merchant or financial capitalist relations, serve to obscure rather than clarify the underlying processes. Contra autonomist-inspired approaches, which have tended to characterise all activities as potentially free labour, our approach retains labour’s specific meaning in relation to capital, which is neither omniscient nor omnipresent. As Srnicek reminds us, it is precisely because of the fact that, ‘most of our social interactions do not enter into a valorisation process’ that companies are competing to build platforms and capture monetisable data. This is a crucial point. On the one hand, the individual activities of users cannot be classified as free labour, since they are ‘naturally occurring’ and become ‘raw material’ through recording and processing. If it were a source of surplus labour, Srnicek points out, capitalism would have discovered an abundant new frontier of value, resulting in a global boom that shows no sign of appearing. On the other hand, the activities of those who produce the means of extraction and process those raw materials – those who design the user interfaces, write the algorithms, package and sell the analytics – can be classified as labour.
Of the five types of platform, the lean platform has had the most visible and immediate effects on the labour market and workers. Lean platforms have expanded rapidly in sectors that require intensive non-routine manual labour, which is notoriously difficult to automate (this is also one reason why service-sector productivity continues to lag behind that of manufacturing). Lean platforms essentially extend the low-tech model of temp-agencies or informal networks of day-labourers into really subsumed and digitally-mediated service sectors. They offer a private technological ‘fix’ to labour market precarity, taking advantage of job polarisation and the displacement of individuals into the relative surplus population. Companies aim to classify workers as independent contractors and pay piece-rates whenever possible, because they ensure a specific rate of exploitation for service-commodities that hourly wages cannot. Lean platforms reliance on low margins and the pursuit of market expansion over short-term profitability means that workers bear the brunt of any problems. Uber and Deliveroo have had disputes with their workers, who have repeatedly challenged the companies over poor treatment, contracts and wages. Uber’s recent dispute with Transport for London over its contemptuous attitude to regulations and the law is another manifestation of the same model of accumulation.
The broader effects of lean platforms on the labour market are a controversial topic of debate. Some organisations have significantly downplayed the importance of non-traditional work arrangements through digital platform-based enterprises, while others have hailed them as significant. In an online survey of 2,238 UK adults aged 16-75, the Foundation for European Progressive Studies [FEPS] found that 21 per cent of respondents (proportionally equivalent to 9 million people) tried to find work on platforms during the past year and 11 per cent of respondents (4.9 million people) were actually successful in doing so. Based on these results, The Work Foundation claimed that a reasonable estimate of the proportion of the workforce finding jobs through digital platforms would be between 5 and 6 per cent. However, recent research from the McKinsey Global Institute revealed that 20 to 30 per cent of the working-age population in the United States and the EU-15, or up to 162 million individuals, are engaged the ‘independent work’ typical of ‘gigs’ provided by lean platforms. Of the individuals in the McKinsey survey, 30 per cent had chosen their work and derived their primary income from it; 40 per cent had chosen independent work to supplement their income; 14 per cent preferred a standard employment relationship, but were primarily independent workers; and 16 per cent engaged in supplemental independent work out of pure necessity. It is important to note that nature of gig-working may lead to under-reporting and there is a large amount of overlap between different job categories, which makes it difficult to compare platform-based gig working with traditional employment. However, what is clear is that, regardless of the extent of these changes, they constitute an acceleration of existing trends toward casualization and precarity.
At the moment, huge amounts of venture-capitalist investment into technology, automation, and artificial intelligence means that firms like Amazon and Uber can continue expanding without actually making a profit. With the rise of platform capitalism, there is a strong possibility that we will see a corresponding rise in the organic composition of capital i.e. a larger share of constant capital or the inert elements (tools, materials, equipment) compared to variable capital or living labour. In Capital vol. 3, Marx argues that this has direct implications for industrial profitability, which might explain the move by many platforms away from service models of production, towards a model that allows them to profit through collecting rents from the use of their infrastructure or appropriating a share of profits from other sectors. This is not sustainable; however it is likely too early to tell what this means for the future. On the one hand, advertising platforms like Facebook show no signs of slowing down. Facebook reported second quarter net income for 2017 at $3.89bn, a 71 per cent increase compared with the previous year. On the other hand, the lean platforms that have driven the rise of the gig economy are already showing signs of slowdown. JPMorgan Chase Institute has found that participation in labour platforms has levelled off and that workers’ monthly earnings from labour platforms have fallen by 6 per cent since June 2014 as a result of wage cuts and lower participation. Despite these findings, some of the world’s leading think tanks are recommending that ‘a platform strategy and the business know-how to exploit it is more important than ‘owning’ an ecosystem’. By 2018, the International Data Corporation predicts that more than 50 per cent of large enterprises will either create or partner with industry platforms and that the number of industry clouds will reach 500 or more by 2018. Time will reveal the veracity of this claim, but the shift toward a rentier form of accumulation through secondary exploitation shows no signs of stopping.
The conceptual development of the term ’platform’ as a new type of firm that relies on the strategic mining of data is a useful contribution to both Marxian and technological discourses that marks a novel economic phenomenon. However, we should be wary of claiming that this is a new mode of exploitation within capitalism. Most of the industry leaders in the platform economy don’t actually produce anything other than a means to profit from proprietary advantage or the sale of advertising and commodifying social data. Rather than signalling a fundamental shift in production and the condition of possibility of a technological utopia, they actually represent a regressive shift back toward what Marx referred as ‘antediluvian’ forms of accumulation i.e. secondary exploitation.
The evidence indicates that, contra the digital dreams of liberal Californian ideologues or post-capitalist utopians, platform capitalism will not provide the technological impetus to a future free of exploitation and drudgery. It might not even provide the robotic libertarian future romanticised by the Silicon Valley entrepreneurs. Sales of industrial robots to the UK have fallen between the period of 2014-2015 while only 14 per cent of business leaders are investing in AI and robotics. Ultimately, many low-margin service platforms will fail over the next few years; monopoly tendencies and cross-subsidisation will push other firms into luxury markets providing expensive convenience on demand; and those remaining will be forced to amalgamate their model into more traditional business models that rely on product or industrial platforms. The rise of platforms may inspire technological utopian rhetoric, yet it retains the same basic forms of twenty-first century global capitalism. Unless we collectivise and ‘nationalise the platforms’, changing their very form, there is little hope for a utopian future.
Matt Cole is a PhD candidate at Leeds University Business School, researching the labour process and politics of service work. He is the coordinator of the IIPPE Political Economy of Work Group and an affiliate member of the Autonomy Institute. He has written for the Independent, Vice, and Novara Media. He tweets at @NoFutureFutureIf you like this article, please subscribe or donate.