The telegraph pole, the Christian cross, and the rifle arrived all at once for the Bororo people of Mato Grosso. The rifle of the soldier and the settler served to seize the Bororo's land in the name of industry and progress, the cross "pacified" and "civilized" them, and the telegraph integrated them into the rest of the newly wired Brazilian republic in the mid-nineteenth century.

Yet we know that the effects of colonialism continue to be felt, as indigenous people even today resist dispossession, cultural invasion, and genocide. Consider next another starting point, the Idle No More movement, a campaign by indigenous peoples in Canada to protect their ancestral resources. Like many activist movements, Idle No More has become a smart user of social media to promote its cause and enlist supporters. The telegraph pole used to link the Bororo into networks of colonial power has given way to a tool on which even the victims of colonization would now seem to depend. … Reflecting on the use of social media during the campaign's protests, Leanne Betasamosake Simpson, a scholar, writer, and artist of the Nishnaabeg people, wrote that "every tweet, Facebook post, blog post, Instagram photo, YouTube video, and email we sent during Idle No More made the largest corporations in the world . . . more money to reinforce the system of settler colonialism. . . . I wonder in hindsight if maybe we didn't build a movement, but rather we built a social media presence that privileged individuals over community, virtual validation over empathy, leadership without accountability and responsibility."

Following a series of fainting spells, doctors advised her to increase her hydration because, like many busy people, she sometimes forgets to take a water break. Fortunately, as the cliché goes, there's an app for that. The makers of Water Minder offer a program to bypass the part of the brain that regulates thirst, reminding you to drink regularly to meet predefined quotas while tracking your progress. Like many apps, the program claims to turn what would otherwise be an insignificant private act into a social celebration, allowing you to earn achievements that you can share with your friends.

The simple, daily act that every individual body does of monitoring whether it has drunk enough water has suddenly become something that happens in a competitive social space. The human body has been reworked into something that requires a distant infrastructure, from which, incidentally, profit can be made. In Rooney's own words, "I have contracted out one of the essential functions of my body to a piece of software." But this is just one small example of something much bigger: the systematic attempt to turn all human lives and relations into inputs for the generation of profit. Human experience, potentially every layer and aspect of it, is becoming the target of profitable extraction.

By tracing continuities from colonialism's historical appropriation of vast territories, such as contemporary Brazil, all the way to data's role in contemporary life, we suggest that although the modes, intensities, scales, and contexts of today's dispossession are distinctive, the underlying function remains the same as under historical colonialism: to acquire large-scale resources from which economic value can be extracted.

If historical colonialism annexed territories, their resources, and the bodies that worked on them, data colonialism's power grab is both simpler and deeper: the capture and control of human life itself through appropriating the data that can be extracted from it for profit. If that is right, then just as historical colonialism created the fuel for industrial capitalism's eventual rise, so too is data colonialism paving the way for a capitalism based on the exploitation of data. Human life is quite literally being annexed to capital.

Data colonialism requires the creation of a new social and economic order that is potentially as enduring as the order that enabled capitalist market societies from the nineteenth century onward. The third point concerns how the power relations generated by this emerging order work: data colonialism appropriates not only physical resources but also our very resources for knowing the world. This means that economic power (the power to make value) and cognitive power (the power over knowledge) converge as never before. Therefore, what is happening with data can be fully understood only against the background not just of capitalism but of the longer interrelations between capitalism and colonialism. The exploitation of human life for profit through data is the climax of five centuries' worth of attempts to know, exploit, and rule the world from particular centers of power. We are entering the age not so much of a new capitalism as of a new interlocking of capitalism's and colonialism's twinned histories, and the interlocking force is data.

For our very specific purposes, the concept of data cannot be separated from two essential elements: the external infrastructure in which it is stored and the profit generation for which it is destined. In short, by data we mean information flows that pass from human life in all its forms to infrastructures for collection and processing. This is the starting point for generating profit from data. In this sense, data abstracts life by converting it into information that can be stored and processed by computers and appropriates life by converting it into value for a third party.

Data colonialism is, in essence, an emerging order for the appropriation of human life so that data can be continuously extracted from it for profit. This extraction is operationalized via data relations, ways of interacting with each other and with the world facilitated by digital tools. Through data relations, human life is not only annexed to capitalism but also becomes subject to continuous monitoring and surveillance. The result is to undermine the autonomy of human life in a fundamental way that threatens the very basis of freedom, which is exactly the value that advocates of capitalism extol. … These fundamental transformations of human life have dramatic consequences for the social world too. They enable what we call social caching, a new form of knowledge about the social world based on the capture of personal data and its storage for later profitable use. As social relations are thus transformed, we see the emergence of the Cloud Empire, a totalizing vision and organization of business in which the dispossession of data colonialism has been naturalized and extended across all social domains.

Today’s attempt to extract economic value from human lives through data has a systematic integration and depth that we argue is, in some respects, without historical precedent. But we see its features most clearly through their continuity with past relations between colonialism and capitalism.

This sector has been growing for a long time, in part through marketers’ accumulation of consumer data, such as credit card data, which began in the 1980s. In the past fifteen years, however, the social quantification sector has achieved a new depth and complexity.

This sector currently includes the manufacturers of the digital devices through which people connect. By this we mean not just well-known media brands such as Apple, Microsoft, and Samsung but also the less-well-known manufacturers of “smart” (that is, internet-connected) fridges, heating systems, and cars through which we never imagined we would communicate. Still less did we imagine that, in the quickly expanding Internet of Things, such devices would communicate with other devices about us. The sector also includes the builders of the computer-based environments, platforms, and tools that enable us to connect with and use the online world, including household names such as Alibaba, Baidu, Facebook, Google, and WeChat. There is also the growing field of data brokers and data processing organizations such as Acxiom, Equifax, Palantir, and TalkingData (in China) that collect, aggregate, analyze, repackage, and sell data of all sorts while also supporting other organizations in their uses of data. And, finally, the social quantification sector includes the vast domain of organizations that increasingly depend for their basic functions on processing data from social life, whether to customize their services (like Netflix and Spotify) or to link sellers and buyers (like Airbnb, Uber, and Didi).

A few words of clarification about the term social quantification sector are needed. When we say social quantification sector, we use the word social to refer to that constantly changing space of relations and interconnections on which the quality of human life depends but whose meaning is endlessly contested in political and civic struggle. When we use the word social, we do not necessarily mean a well-ordered or well-integrated way of living together. For at issue in this whole transformation is precisely the quality of life that human beings will have together in the new capitalist social order. The capture of personal data through social caching and its storage for later profitable use—normal today but two decades ago barely imaginable—has major implications for our quality of life as human beings.

This was the deeper meaning of Edward Snowden's revelations of US and UK security services' data-gathering in 2013. Suddenly, citizens became aware that today's pervasive state surveillance would be impossible without the continuous social-caching operations of commercial corporations. Since then, fears of a new "corporate governance of everyday life" have been growing. But other fears have started to overtake them, including the "fake news" scandals that have gripped politics since late 2016 in the United States, United Kingdom, and elsewhere. There are also fears that, because of polarizing forces online, "social media is ripping society apart"; fears from other commentators of a social dystopia driven by platforms' search for advertising income; and, finally, the fear that the targeting of news via social media platforms represents "the most lethal political weapon ever invented." Calls for the regulation of social media platforms and other information technology giants are becoming familiar. But none of these highly charged debates answers the underlying question on which this book will focus: should human beings in the twenty-first century accept a world in which their lives are unceasingly appropriated through data for capitalism?

This is where the long history of colonialism's entanglement with capitalism helps us move beyond the sound and fury of contemporary scandals and grasp the longer pattern of resource appropriation that gives shape to today's developments. Long before Karl Marx identified it as a force in the world, capital was already expanding in the sixteenth century, in the period most commonly identified with historical colonialism and the emergence of the Spanish, Portuguese, and British empires. Capital was acquiring new territories from which to extract resources and new bodies from which to extract labor. Those close relations between colonialism and capitalism (indeed, between colonialism and modernity in general) are important to our story; they help us grasp what is distinctive about the current expansionary phase of capitalism.

Capitalism has been expanding from the start, in the sense of exploiting new resources, finding new ways of making profit, and finding new markets. But until recently, this expansion has been based on the exploitation of human production through labor relations, as classically theorized by Marx, resulting in the ever-greater transformation of physical nature as an input to capitalism. But the appropriation of human life in the form of data (the basic move that we call data colonialism) generates a new possibility: without ending its exploitation of labor and its transformation of physical nature, capitalism extends its capacity to exploit life by assimilating new or reconfigured human activities (whether regarded as labor or not) as its direct inputs. The result, we argue in chapter 1, is the expansion of the practical scope of capitalist exploitation, but in ways that can be linked back to Marx's own sense of capitalism's expansionary potential. In this emerging form of capitalism, human beings become not just actors in the production process but raw material that can be transformed into value for that production process. Human life, in the form of profitably abstracted data, becomes more like the seed or manure that Marx noted became factors of capitalist production, having once just been part of human beings' cycle of interaction with the land. This transformation of human life into raw material resonates strongly with the history of exploitation that preceded industrial capitalism—that is, colonialism. … The very concept of raw material has deep colonial roots. Raw, in this context, means available for exploitation without resistance rather than a substance that needs no processing. The natural environment itself first had to be reconstructed so that it became available for value extraction. Then, through the practice of slavery (which preceded colonialism but which reached a massive scale under colonialism), human bodies were transformed into a raw material for capitalism in the form of slaves. Historians have warned us against treating slavery—usually imagined as a premodern practice—in isolation from the development of industrial capitalism. The plantation and the factory coexisted for a long time. The treatment of human beings as mere property stimulated the rationalities of profit maximization, accounting precision, and data optimization that we now tend to associate with modern rationality.

The very concept of raw material has deep colonial roots. Raw, in this context, means available for exploitation without resistance rather than a substance that needs no processing. The natural environment itself first had to be reconstructed so that it became available for value extraction. Then, through the practice of slavery (which preceded colonialism but which reached a massive scale under colonialism), human bodies were transformed into a raw material for capitalism in the form of slaves. Historians have warned us against treating slavery—usually imagined as a premodern practice—in isolation from the development of industrial capitalism. The plantation and the factory coexisted for a long time. The treatment of human beings as mere property stimulated the rationalities of profit maximization, accounting precision, and data optimization that we now tend to associate with modern rationality.

What if comparable processes for abstracting human life are today enabling new, extended forms of economic extraction? … The world of internet-based connection is a world in which new types of corporate power, with privileged access to data streams extracted from the flow of life, can activate a metaphorical button—an apparatus of extraction, whether platform, app, or AI system—that reconfigures human life so that it contributes continuously to the collection of data and thereby, potentially, to the generation of profit.

Human data is not actually raw, but business often imagines it is. Human life first needs to be configured so as to “naturally” generate a resource such as data. Momentary data about one individual’s actions or properties is worthless unless it can be combined with data about other individuals, actions, moments, and properties to generate relations between data points, and that is why an infrastructure has been built to ensure this.

But that should not blind us to what remains a disturbing line of continuity: as happened historically, but under new conditions, human life today is becoming the object of appropriation for extraction and, in the process, enabling capitalism to move to a still higher scale and integration of operation. Our intent is not to make comparisons to the detailed contents or form of historical colonialism but to focus on colonialism’s enduring function, which we see as enacting illegitimate appropriation and exploitation and as redefining human relations so that the resulting dispossession comes to seem natural.

More explicitly defined, data colonialism is our term for the extension of a global process of extraction that started under colonialism and continued through industrial capitalism, culminating in today’s new form: instead of natural resources and labor, what is now being appropriated is human life through its conversion into data. The result degrades life, first by exposing it continuously to monitoring and surveillance (through which data is extracted) and second by thus making human life a direct input to capitalist production. Data colonialism is, in other words, an emerging order for appropriating and extracting social resources for profit through data, practiced via data relations. Unlike historical colonialism, whose vast profits helped create the preconditions for what we now know as industrial capitalism, data colonialism emerges against the backdrop of the entire intertwined history of colonialism and capitalism. This means that the basic colonial move of appropriating data from human life (data colonialism) works hand in hand with social arrangements and technological infrastructures, some that emerged during earlier capitalism and some new, that enable that data to be transformed into a commodity, indeed, a direct input to contemporary capitalist production.

Although the site of data colonialism that we most often notice is the social quantification sector, and particularly the personal data extraction practiced by social media platforms, the basic extractive principles underlying data colonialism have been gestating in the business methods of modern capitalism for three decades. The original context was industrial capitalism’s progressive globalization in the late twentieth century through trade liberalization and extended supply chains as well as its financialization through an explosive growth of debt (both corporate and personal) and the acceleration of global capital flows. In this general context, the information infrastructures emerged that enabled people and processes to be connected to each other under conditions that facilitate data extraction.

Take an example far from social media: the discipline of logistics. The goal of logistics is to use continuously connected data flows to organize all aspects of production across space and time within global commodity chains. It was enabled, decades before social media, by software inventions such as the relational databases that help businesses find patterns in huge data flows across diverse sources and locations.

If logistics aims to rationalize production, the recent growth of the Internet of Things (IoT) aims to incorporate consumption—what we do with products after we buy them—into an extended chain of profit extraction through the processing of data. The bigger vision of the IoT, and of data colonialism as a whole, has been expressed by a company normally seen as the enemy of the entrepreneurs who built social media platforms—IBM. By turning the human environment into a network of listening devices that capture data about all activities, IBM suggests that they can “liquify” areas previously inaccessible to capital.

Far from being a feature of the West only, the growth of the social quantification sector, logistics, and the IoT in the context of rapidly expanding uses of artificial intelligence is a core development and policy goal in China too. The social quantification sector emerges as an arena of commercial and geopolitical competition between the West (particularly the United States) and China.

We can capture the core of our double argument most succinctly by characterizing data colonialism as an unprecedented mutual implication of human life and digital technology for capitalism. None of this would have been possible without the emergence in the past three decades of a radically new technological infrastructure for connecting humans, things, and systems, known generally as the internet.

This book is an attempt to disentangle a historical moment of great complexity. As the first subjects of an emerging data colonialism, we can make sense of this complexity only within the much longer history of colonialism and capitalism. We also need to do a considerable amount of interpretative and analytical work across today's societies and economies. In writing this book, we have been fortunate enough to build on much excellent work within the past decade that has begun to uncover the shape and dynamics of human life's appropriation through data. But in searching for a larger theoretical framework to make sense of the whole sweep of these developments—the framework of data colonialism—we have had to forge some new connections and take an eclectic path in terms of theory. Rather than positioning our argument exclusively within Marxism, postcolonialism, Foucauldian analyses of biopolitics, or critical-information science, our aim is to put these perspectives into conversation, taking the best from each to forge a new starting point adequate to today's complex transformations and their hidden violence.

"THIS IS WHAT MODERN COLONIALISM LOOKS LIKE." So tweeted Christopher Wylie, the whistle-blower who kicked off the Facebook/Cambridge Analytica scandal in March 2018.1 Wylie was referring to Cambridge Analytica's plans to expand its operations in India for using social media targeting to influence the political process there. But the scale and scope of data colonialism is much wider than the malfeasance of a few overweening data marketers and their in-house psychologists. It extends much wider even than Facebook's normal practices of data extraction and data licensing that the scandal opened to view.

Yet the scandal was important. It was as though a side deal by Facebook with independent data prospectors had accidentally left open a hole in the ground that allowed the general public, for the first time, to see clearly into an underground anteroom. There, in that anteroom, visible for all to see, was the entrance to social media's real data mine, although few understood exactly what lay behind that subterranean door in Facebook's exclusive domain, let alone the planetary scope of capitalism's data mining. The long-anticipated "techlash" had begun but, as yet, without a map of the wider pattern of exploitation, whose traces had suddenly become visible.

To start, we must recall what made historical colonialism distinctive. Colonialism was a form of economic and social organization dominated by major colonial powers such as Britain, France, Spain, and later the United States. It is now usually regarded as historically closed, ended by the decolonizing movements of the later twentieth century, although in politics and other areas, neocolonial forms of power live on (a more detailed discussion of historical colonialism can be found in the interlude following chapter 2). Our interest in this book is in the continuities from that older colonialism to a new form of colonialism—data colonialism.

There were four key components to historical colonialism: the appropriation of resources; the evolution of highly unequal social and economic relations that secured resource appropriation (including slavery and other forms of forced labor as well as unequal trading relations); a massively unequal global distribution of the benefits of resource appropriation; and the spread of ideologies to make sense of all this (for example, the reframing of colonial appropriation as the release of "natural" resources, the government of "inferior" peoples, and the bringing of "civilization" to the world).

The discovery of new forms of raw material is what makes the current moment distinctively colonial. If historical colonialism expanded by appropriating for exploitation geographical territory and the resources that territorial conquest could bring, data colonialism expands by appropriating for exploitation ever more layers of human life itself. … When we refer to data practices as colonizing human life, we refer to the appropriation of data, potentially for profit, in any and all of these areas. But we also intend the term human life to refer to the as-yet-still-open horizon of exploitation over which data colonialism claims future rights: as ever more of our activities and even inner thoughts occur in contexts in which they automatically are made ready for appropriation as data, there is, in principle, no limit to how much of human life can be appropriated and exploited.

The result of this convenient failure is not just to renew colonialism but also to expand the scope of capitalism too, that is, the capitalism developed on the basis of historical colonialism. Through data colonialism, contemporary capitalism promises to consume its last remaining “outside,” dispossessing human subjects of their capacity as independent sites of thought and action. Resisting data colonialism becomes the only way to secure a human future not fused indissolubly with capitalism, indeed, the only way to sustain the value that capitalism claims to promote: human freedom.

This progressive opening up of human life to externally driven data extraction is what we mean by the capitalization of human life without limit. In this phrase, we recognize Marx’s long-standing insight that capitalism has always sought to manage human life for the maximization of profit; at the same time, we emphasize that data colonialism absorbs new aspects of human life streams directly into the productive process. It is not that social limits to life’s capitalization can no longer be imagined—indeed, the whole point of this book is to argue for the necessity of such limits—but that, as things currently stand, much corporate discourse fails to recognize any limits except those that it sets itself.

Capitalism can exploit any number of data sources. Any computer, any device with an embedded computer, or any entity readable by a sensor with computing power can generate data for this purpose. Data sources may be processes, things, or people as well as the interactions between any of these sources. The extraction of value from data is equally indifferent to its origin. Capitalism as the systematic organization of value extraction has only one goal in relation to data—to maximize the production of value through data extraction—and so in principle cares little about the sources and types of data exploited.

As we show in detail in chapter 5, all notions of autonomy, until now, have assumed that individuals have access to a minimal space of the self that is its space of becoming. But the goal of continuous data appropriation intrudes on this space and changes humanity’s relations to external infrastructures decisively, erasing, potentially forever, the boundary between the flow of human experience and the environment of economic power that surrounds it.

Davenport distinguishes three phases in the development of data analytics: whereas early analytics was essentially “descriptive,” collecting companies’ internal data for discrete analysis, the period since 2005 has seen the emergence of the ability to extract value from large unstructured and increasingly diverse external and internal data sets through “predictive” analytics that can find patterns in what appears to have no pattern. Today’s “analytics 3.0” uses large-scale processing power to extract value from vast combinations of data sets, resulting in a “prescriptive analytics” that “embed[s] analytics into every process and employee behavior." Once the world is seen by capitalism as a domain that can and so must be comprehensively tracked and exploited to ensure more profit, then all life processes that underlie the production process (thinking, acting, consuming, and working in all its dimensions and preconditions) must be fully controlled too. This principle, made possible by technological connection, is the engine that drives the capitalization of human life in its twenty-first-century form, and its remit goes much wider than social media platforms.

Surveillance is not obviously a benefit to citizens (except as a temporary means to counter serious threats), so it must be repackaged or disguised. Here is the source of data colonialism's most interesting contradictions. It is eerie but not uncommon to find the language of personal freedom melded with the logic of surveillance, as in the motto of facial-recognition software manufacturer, Facefirst: "Creating a safer and more personalised planet through facial recognition technology." Google, meanwhile, is marketing Nest Hello, a video doorbell that includes facial-recognition technology, and Amazon has its own facial-recognition service called simply Rekognition. In China, facial-recognition software is becoming the cool new way for customers to pay for fast food and a (less cool!) way for city authorities to monitor public spaces.

Yet Apple tracks its users for many purposes and so does not contradict the trend of data colonialism, except that its business model does not generally depend on the sale of this data. Indeed, Apple receives substantial sums from Google for allowing it privileged access to iPhone users. A controversy developed when Apple's iOS and MacOS systems were shown to collect information on user location and search activity. Apple's subsequent Privacy Policy states that collected data “will not be associated with [a user's] IP address,” yet iPhone features still support the surveillance needs of marketers.

Like historical colonialism, data colonialism would be inherently unstable if it could not translate its methods into more enduring forms of social relations. As Nick Dyer-Witheford pointed out in an early analysis, capitalism has always approached the internet as a domain in which control over the communicative capacity of individuals would allow capital to appropriate not just labor but also, as Marx himself put it, “its network of social relations.” Data colonialism extends this network well beyond communicative capacities, ensuring the continuity of data appropriation across an expanding array of social relations.

At the core of data colonialism is the creation of a new type of social relation that we call “data relations,” as defined in detail later. Data relations make the appropriation of human beings’ data seem normal, just the way things are. Data relations are of many sorts, but all share one basic feature: they ensure informational resources (data) for capitalism in areas of human life that, previously, were not considered direct inputs to production. Far outside the sphere of normal productive activity, ordinary social interaction is increasingly lived in environments of continuous data collection, behavior prediction, and choice shaping. But this is possible only because social actors now enter more or less voluntarily into data relations that secure regular data flows for capital. The drive to expand data relations explains a lot about contemporary capitalism. So, for example, it is the need to make data relations routine everywhere that drives Facebook’s offer of simplified internet connection (Facebook Free Basics) in more than twenty countries with weak or uneven internet infrastructure, principally in Africa. The emergence of data relations increasingly complements labor relations’ contribution to capitalism’s reproduction.

Meanwhile, existing social relations (including labor relations) for hundreds of millions of people are increasingly datafied—that is, managed through data, including data gathered from the surfaces or insides of workers’ bodies. Work, for many, increasingly occurs within the sort of corporate environment of sensors that Davenport imagined; in its absoluteness, this recalls the continuous surveillance, if not the violence, that slave labor endured on a large scale under historical colonialism. Labor relations are becoming more directly and continuously extractive, whatever the formalities that cloak them. This has implications too for inequality, as exposure to (or freedom from) continuous surveillance becomes a key factor that distinguishes lower-status from higher-status jobs.

New forms of labor are also emerging under data colonialism in the “sharing economy.” Here data relations are fused with labor relations, although the existence of labor relations is controversially denied by platforms such as Uber. These new hybrid data/labor relations encompass a huge variety of more or less formalized work, including data processing. Labor is captured in a seemingly “scale-less” business model that detaches workers from institutional supports but richly rewards platform management. Rhetoric cannot disguise the potential for exploiting low-level work skills at a distance and therefore at a scale and speed without historical precedent (for example, the repetitive coding and data inputting necessary for the training of artificial intelligence in so-called Machine Learning).

Data relations give corporations a privileged window into the world of social relations and a privileged handle on the levers of social differentiation. States have become increasingly dependent on access to what the corporate sector knows about the lives of those states’ citizens, reversing the long-subsisting direction of knowledge transfer (from states to corporations).

Because data reinterprets the traces of the everyday world, large-scale data processing recalibrates social knowledge, with the consequences depending on the type of market society under analysis. In liberal market societies, the new data infrastructure involves for the first time market institutions in producing basic social knowledge. But in state-led market societies, such as China, the state acquires a remarkable new tool to direct the production of social knowledge in its own interests.

Another way in which capitalism’s new interrelations with data deserve to be called colonial is the massively unequal global distribution of economic power on which they are based.

There are many additional layers of colonial-style ownership. One is domination over production of the “tethered” devices through which human beings connect to the infrastructures of data collection (by “tethered,” we mean a device whose use ties users to reliance on a particular operating system or set of proprietary products: think of Apple but also of the use restrictions built into most portable devices). Another is power over the computing capacity that enables large-scale processing and storage of data, usually known as the cloud, a mystificatory term. (Amazon by the end of 2017 already had 51.8 percent of global market share, rivaled only by Microsoft with 13.3 percent, the Chinese Alibaba with 4.6 percent, and Google with 3.3 percent.)

These layers of power concentration offer to citizens and consumers a world that is highly connected, but the quality of connection is distinctly uneven. Beyond questions of individual access, countries vary in the robustness of their internet connectivity, while intercontinental cable laying is becoming a growth area for the social quantification sector itself. Although the exact distribution of the benefits of these power concentrations will no doubt evolve, at this point the global domination of a web of US companies and a small group of Chinese companies is secure.

There is also the ideology of datafication, which insists that every aspect of life must be transmuted into data as the form in which all life becomes useful for capital. Practically, this means not just attachment to a computer connection but the removal of any obstacles to corporate extraction and control of data once that connection is established. The point is not that data itself is bad but that the compulsion to turn every life stream into data flows removes what was once an obstacle to extracting value from those life streams.

The marketing ideology of personalization makes such tracking and surveillance seem attractive. Who after all would not want a service more geared to his or her particular wants and needs? The argument, very simply, is that personally targeted messages require prior information that can come only from . . . targeted surveillance! Yet evangelists for data-driven personalization still feel the need to defend this ideology aggressively, as when Acxiom, in a 2015 opinion survey, characterized acceptance of continuous data collection as “pragmatist” (fair enough) and opposition to it as “fundamentalist” (itself rather an extremist statement).

These specific ideologies are examples of “dataism.” They are tools for shaping practice and thus highly specific in their applications. They are often combined with an overarching imaginative claim that we might call myth: the myth that all this is inevitable and that today’s infrastructures of connection and data extraction fulfill human beings’ collective potential in some transcendent way. The term community, as something seemingly inherent to human life, can easily fuel this myth. Think not just of Mark Zuckerberg’s frequent appeals to Facebook’s “global community” but also of the Tencent CEO’s invocation of a “digital ecological community.” This ideology of inevitability also has its counterpart in ideas on the anticapitalist left that treat networks as the basis for a leap toward overthrowing capitalism.

Data colonialism has a distinctive geography. Like historical colonialism, data colonialism is global in its ambitions, but unlike it, it penetrates the life conditions of individuals across all societies, wherever they are, that are being reshaped around digital infrastructures of connection. Data colonialism’s expansion is therefore both external (geographical) and internal (social). Given that the drive to appropriate data is fueling China as much as, if not more than, it is the West, it makes no sense to read data colonialism as exclusively a Western project.

Indeed, capitalism’s standard ways of understanding itself as a rational order are effective in masking the colonial scale of appropriation that is underway. Much of today’s data appropriations are not seen in daily life as appropriation at all but as part of everyday business practice. As such, data relations introduce into business management a new bureaucracy, as first theorized by Max Weber. This bureaucracy is based on the apparent “rationality” of organizations’ continuous data extraction.

Older critiques of historical capitalism can provide valuable insights here. One important idea is that of the “fetish.” For Karl Marx, the fetish (classically the “commodity fetish”) made the economic relations of capitalism seem necessary. Marx delineated a trinity of fetishized objects: interest as the fruit of capital, rent as the fruit of land, and wage as the fruit of labor. Now we can add a fourth object: data as the fruit of life processes. This is exactly how business now speaks (think of the mantra of data as the new oil).

Another deep continuity with classic Marxist accounts comes from the data-driven possibilities of management control over production. The core of how industrial capitalism transformed work was not technological but social. The organization of work came under the exclusive control of management and, as such, became controllable at every stage of performance. Traditional industrial management remained limited, however, to mechanical data collection based in observation or form filling. By contrast, data-driven logistics, built on infrastructures of connection and fueled by artificial intelligence, convert all aspects of production—far beyond the factory walls and in every corner and moment of a transnational supply chain—into a managed assembly line. The new infrastructure of data processing that sustains data relations extends so-called scientific management in ways that recall classic diagnoses of industrial capitalism: “The image of the process, removed from production to a separate location and a separate group, controls the process itself.” Through data, capitalism begins to govern the whole social domain with management logic, another aspect of its annexation of human life.

Every aspect of data colonialism can be understood in terms of its contribution to a fundamental principle of capitalism: the drive to maximize control of the inputs to its production process. This requires a reasonably ordered state of the social world that makes markets possible and functional. In all societies characterized by increasingly overwhelming information flows, the premium on society's legibility, even its basic countability, grows. In capitalist societies that are not democracies, such as China, capitalism's desire for the control of production inputs works hand in hand with government's unrestrained desire to control all inputs to the social order.

This newly intensified role for both capitalism and governments in managing social life became possible only through the unprecedented technological opportunities for connection that we know collectively as the internet. Their origins lie in investment from the US Cold War military-industrial complex and subsequent massive corporate investment that transformed the internet into a global space for corporate networking and transnational market coordination. Through the internet and its embedding into everyday life and business emerged something else: the possibility of a new type of economic and social order.

Move forward another ten years, and, through portable devices with connectivity (smartphones and then tablets), the internet stopped being something statically available from particular points in space and became a dimension that overlays social space continuously. As Rob Kitchin and Martin Dodge put it, “Software . . . alters the conditions through which society, space and time, and thus spatiality, are produced.” As a result, the internet, understood as an infrastructure of connection, has reconstituted social space in a fundamental way.

Meanwhile, Google’s alliance in 2008 with online advertising company DoubleClick brought together Google’s mastery of linking search activity to searchers’ interests and DoubleClick’s tracking of individuals based on their unique data profiles. Under Google’s control, the combination of the two approaches greatly increased the information that could be matched with unique individuals.

Combine vastly extended interconnection with that basic feature of computers (they capture data), and the result is the beginning of a radical new possibility: that each computer’s archive of its own history becomes available to be tracked and influenced by any other computer anywhere in the internet’s vast universe of networks. Human beings themselves, as regular users of those computers and their archives, became, in principle, “trackable and tractable.” The actualization of this theoretical possibility resulted from multiple commercial innovations, spurred by various forms of necessity and opportunism. It was marketers, plagued by the perennial difficulty of reaching customers with their messages, who took advantage first with the humble cookie, capable only of tracking people on single devices, and more recently with techniques that link online data with offline behavior or use a social media ID to combine data about various online selves.

The invention of social media platforms in which every user move is automatically tracked was a major further step. The data benefit unleashed to social media providers such as Facebook was huge: through its Open Graph program, Facebook gained access to what users did not only on Facebook but also on every platform that users accessed via their Facebook ID—for example, Spotify. The result is what the industry, without irony, calls “people-based marketing.”

The growth of computer-to-computer monitoring for marketing purposes was only part of a more general growth, starting in the 1980s, in the monitoring and shaping of life at a distance. Much of this data gathering requires no subterfuge, because it is a basic mode of operation in corporate intranets, networks of environmental sensors, and scientific data collection—that is, “new types of sensors are constantly becoming available.”

Yet a more connected social world will not necessarily be more secure. Indeed, there is no limit to the instability that an interconnected social order can generate. As Bruce Schneier argues, it is a fallacy to believe that more connection generates more security. On the contrary, since the possibilities of bad actors are infinite, more connection must generate more, not fewer, problems of security. The average security robustness of connected space tends to fall as more points get connected. Devices in the Internet of Things (such as those that control fridges or thermostats) simply lack the computing power to have their security repatched remotely, unlike, say, desktop or laptop computers. People cracking the code of baby monitors to speak to and scare children supposedly under digital protection are just graphic illustrations of a much deeper security problem, which in fact serves the interests of data extraction.

The purpose of collecting data is always to differentiate: data's purpose is to generate information that usefully discriminates between entities. Again, discrimination in itself is not bad: we want a heart recorder that correctly distinguishes between a heartbeat and other bodily signals. The issue is how data discriminations are applied within the framework of existing structures of social discrimination. Today's vast infrastructures of connection, like all forms of power, build on existing inequality. But they also create new forms of inequality through various means: inequalities of data capital, exploitation of the human labor that sustains Machine Learning, and the ability of data-driven categorizations to reinforce those existing inequalities. Just as the capitalization of work through labor relations has over time produced a highly unequal distribution of the profits generated by labor, so over time we can expect the capitalization of life through data relations to introduce new forms of inequality into human life itself. We can increasingly expect human beings to be socially managed in terms of their data value—that is, their value as data inputs to capitalist production.

Platforms are structured online spaces, made possible through elaborate software, that offer services of various sorts: a space to sell things, meet people, share information, find specialist resources, and so on. But they are also fundamentally spaces for data extraction. As Tarleton Gillespie puts it, platforms are "built on an infrastructure . . . for processing data for customer service, advertising, and profit" that makes them "oriented toward eliciting more data, and more kinds of data, from [their] users." Platforms generally operate without explicit payment. The actual "payment" that the platform operator receives is usually based directly on the data extracted, though often also in part on data-driven advertising. The data "payment" is more interesting than the advertising payment. It can be either the value that the operator generates by packaging and selling to third parties the data it gathers about transactions and users (the commodity value) or the value that the operator raises by giving others access to its data (what we might call the rent value).

Platforms create an interface without historical precedent: an interchange in which social life in its open-ended variety interfaces seamlessly with the forces of economic extraction. This seamlessness is not natural any more than data is naturally raw. It must be constructed through the painstaking removal of barriers to data flow within and between platforms; seamlessness is an achievement, in part, of software that enables platforms to produce "the social" for capital. This explains the deeply ambiguous status of the largest and most successful platforms, such as Facebook, which is expected to speak for and even regulate the social world because of the power that running a platform with two billion users gives it. Major platform power in China (Alibaba and Tencent) is even more ambitious, developing a triple interface across social, economic, and financial worlds to provide a new practical infrastructure for the world's largest society. These models of economic power and social order depend on the unspoken principle of seamlessness: every barrier to data flow is a barrier to the unlimited production of social life in forms that can generate data for capital.

The largest platforms distribute their presence across the internet via plug-ins that allow platform users, wherever they are, to link back to the resources of the platform, creating even more possibilities for aggregating data continuously across space and time. These power concentrations depend on the emerging social centrality of these platforms in a genuine sense, for we really do go onto platforms because we know others will be there (the so-called network effect). That is why platforms are fast becoming meaningful social infrastructures too. In these various ways, platforms are a principal organizational form through which capitalism’s connected social and economic order is being realized. They are a key tool for ensuring that the colonial-scale appropriation of data for human life becomes the norm. As networks, some platforms are larger than most nation-states. The largest platforms conceive of themselves accordingly. As Zuckerberg put it, “In a lot of ways Facebook is more like a government than a traditional company.” The longer-term implications for legal, political, and social power are as yet unknown.

By “data relations” we do not mean relations between data but the new types of human relations that data as a potential commodity enables. In time, data relations are likely to become as naturalized as labor relations.

But in many other cases a relation has to be constituted in such a way that it enables data to flow in the first place. Once constituted, the seeming naturalness of the relation frames the resulting data as something that can be validly extracted from the flow of life. Without the act of extraction there would be no identifiable item of data and thus no separate right to extract that data. It is the stability of data relations that enables data extraction to seem both valid and beyond challenge. This is another consequence of the basic point already noted, that data is not natural but a resource whose extractive possibilities must themselves be socially constructed, just as physical nature had to be reconfigured so that it could be exploited by capital.

Indeed, Zuboff plausibly argues that much data extraction today is not a contract or relation at all, in that data just seems to get extracted at a distance. But the wider set of arrangements on which data colonialism relies would be ineffective unless they operated as something like a social form that seems both natural and normal, quite unlike an arbitrary act of theft or extraction. It is this social form we call data relations; data relations enable the basic processes of social construction on which the stabilization of data colonialism depends for the long term.

We can expect the varieties of data relations to grow massively as data colonialism itself expands. To get at the core of what we mean by a data relation, imagine yourself entering a spiral. The entry point is that at which, for whatever reason, you enter the ambit of computer monitoring. Computer tracking is now so basic to the social world that it might seem hard to imagine one not already being within that ambit, but, even so, at any moment there is always likely to be at least one specific transaction in which you are engaged that motivates the capture of particular life traces from you. It could be using a search engine, investigating a particular website, searching for a product at a good price, using a platform, or dealing with an institution (say, a school, corporation, or government).

Yet, in agreeing to or accepting data collection, people may have a pressing need to secure a service of some sort (just as a labor contract is held together by the worker's need to sell her labor power for money so she can buy the necessities of life). Through data relations, for example, we access connection (platforms where everyone else we know is a user) and services that are basic to daily life, such as insurance.

Sometimes there is no consent to data relations at all, just brute necessity. One example is when people wear a sensor or tracker, on order of their employer, as they move around their work space. Another example is when, as in China's Xianjiang province in February 2017, the government banned petrol sales to anyone without GPS tracking installed in his or her car. The logical limit of consent is when—as is increasingly common in China today—paying for transactions or just walking around public space involves facial recognition by algorithmically enhanced tracking systems linked to national databases. The development of systems that manage these barely consensual relations is a growth area: the Beijing company Face ++ was valued at $1.5 billion before its latest fund-raising from the Chinese and Russian sovereign wealth funds.

Whether immediately commodifiable or not, the data extracted has potential value in a wider space of equivalence in which the individual's distinctive properties as a data source can be ignored (this abstractness is exactly how Marx understood commodification to operate). The commodification of exchangeable data is the end goal of data relations and is, as Matthew Crain notes, "the root of [their] power imbalances."

Data subjects often attempt to modulate their behavior in order to influence the algorithm that they believe is categorizing them. But however well meaning, this ignores the basic point: that just by going on with the activity that brought us into data extraction's ambit, we confirm data relations' continued force.

We are concerned here with the annexation to capital of life processes whether or not they are labor, evidenced when many of the life streams from which data is extracted for value are not seen by those involved as part of any productive activity.

On this reading of Marx, at the core of industrial capitalism’s longterm transformation of the social world was how work (the everyday productive activities that have gone on since the beginning of time) acquired an abstract dimension that enabled it to be commodified, that is, turned into a commodity that could be exchanged. It is this possibility of abstracting value from life processes, even when they are not directly productive activities, that leaves open the prospect of new types of commodified social relations and, through them, a new social order for capitalism.

But the outcome is hardly something to celebrate. For the unwelcome truth is that, just as in Marx's eyes capitalism had disrupted human beings' relations with physical nature, in the era of data colonialism, capitalism risks disturbing humanity's relations to its nature—that is, our lives as reflexive, relatively autonomous human beings. Data colonialism interposes infrastructures of data extraction directly into the texture of human life and so risks deforming human experience in a fundamental way, invading the space of the self on which the values of autonomy and freedom in all their forms depend.

Why is it that so far we have talked simply of capitalism and not digital capitalism, informational capitalism, communicative capitalism, platform capitalism, or surveillance capitalism, to name some rival terms? The reason is straightforward. No convincing argument has yet been made that capitalism today is anything other than what it has always been: the systematic organization of life so as to maximize value, resulting in the concentration of power and wealth in very few hands. Contemporary societies are marked by the ever-increasing importance of the circulation and processing of information. As we emphasized earlier, this accelerated circulation and extraction of data and information has had profound impacts on the management of business, on the organization of work, and on the integration of social life into the economy. But that is no reason to say that the fundamental drivers of capitalism have suddenly changed; they have not.

Surveillance is certainly part of this, again as we have emphasized, but not sufficiently to brand today's capitalism as surveillance capitalism. For, within the longer history of colonialism and capitalism, surveillance has often been the accompaniment to the direct appropriation of laboring bodies for value (think of the slave plantation). What is new today is not so much surveillance but rather the networks of social relations in which vastly extended modes of appropriating human life through data work to order economic and social life as a whole.

But Autonomists were rather vague about the mechanisms of this general intensification of capitalism's influence over social life, except when they claimed the structure and norms of capitalist work somehow expanded out into social life. That idea gives us no grip at all on data colonialism, which, as we have shown, appropriates life as raw material whether or not it is actually labor, or even labor-like.

And it ignores the power of data relations to restructure life for capital in societies such as China and Russia, where data colonialism proceeds under the auspices of an authoritarian state.

IN 1999, YEARS BEFORE FACEBOOK WAS FOUNDED and just as Google was getting out of a garage, Josh Harris not only predicted but enacted the future of the internet in an art project he called Quiet: We Live in Public. Harris was a brilliant yet somewhat disturbed entrepreneur who made a fortune by creating platforms that integrated user content and new media in visionary ways. For Quiet, he set up an underground bunker where volunteers lived for a month in hive-like conditions, cameras recording their every moment. Room, board, and amenities were provided free of charge. The experiment, which included Stasi-style interrogations and plenty of sex, drugs, and guns, ended in near catastrophe, but not before Harris had come up with the credo for the social quantification sector, captured in the preceding quotation.

As discussed in the previous chapter, data colonialism is what happens when human life becomes an input for capitalism. That human life should acquire this particular status under a capitalist system is not strange or new. After all, capitalism seeks to commodify everything, so everything potentially becomes its input. But how “everything” is defined at a specific historical moment is what requires attention. What is unique about our historical moment, we argue, is that human life as organized through data relations becomes the direct input of capitalism, a condition that we describe as specifically and newly colonial.

One factor that makes the Cloud Empire colonial is the logistical and martial ethos of the enterprise of expansion. Colonialism requires an infra structure that facilitates the movement of resources, and what makes the Cloud Empire colonial is the scale and scope of this worldwide network of extraction and distribution, which is managed through increasingly sophisticated logistics. And we should not forget the debt that the concept and sometimes the practice of logistics owes to the military spirit. The smooth movement of resources along global supply chains is a movement in which corporations and armies have often supported each other during colonial times. To be sure, Amazon does not require its own army, like the East India Company did. But as Deborah Cowen argues, logistics "[map] the form of contemporary imperialism," and as Ned Rossiter points out, "Increasingly, logistics infrastructure is managed through computational systems of code and software," that is, through data. Thus, we can see data relations as part of a highly organized system of economic extraction and market penetration, a system founded on logistics.

At the same time, as Immanuel Wallerstein points out, capitalism (and by extension, colonialism) cannot flourish without the active participation of the state, which controls the relations of production by legalizing (or refusing to regulate) certain forms of labor, such as slavery in the past or temporary or gig-based work nowadays.

Another factor that makes the Cloud Empire colonial is the enormous role that transnational corporate networks play in keeping the system running. In such an environment, we see the emergence of what Philip Stern called the "company-state": a corporation with immense powers to regulate not just trade but also law, land, and liberty—in short, to regulate life. At the same time, as Immanuel Wallerstein points out, capitalism (and by extension, colonialism) cannot flourish without the active participation of the state, which controls the relations of production by legalizing (or refusing to regulate) certain forms of labor, such as slavery in the past or temporary or gig-based work nowadays. Thus, it is powerful corporations operating in collaboration with powerful states that are defining the new colonial geographies and constructing a different social and economic order. The world is conquered and divided roughly along old lines but with new motives. A clear axis is emerging that reflects the divisions of political and economic power and influence in today's world, with China and the United States at opposite ends of that axis but equally engaged in rampant data colonialism.

Problem 1: As inequality increases and the majority of people have less money to spend, capitalism needs to find new ways to exploit them that don't involve buying things. Wealth is becoming less evenly distributed, which means most of it is accumulating at the top while the majority of people experience a loss of wealth. If people buy fewer and fewer things, this can become a problem for capitalism. Fortunately (for capitalism), social quantification introduces a new model in which people can generate profit for corporations simply by participating in social activities, without having to buy commodities.

Problem 2: If, as some argue, the rate of profit is falling because of technological advances, social quantification represents a valuable new source of profit, at least for the immediate future. As new technologies are introduced into the production process—or as workers are replaced with machines and artificial intelligence—things can be produced more cheaply, which brings down their prices, reducing the profit that corporations can derive from selling their goods. Although opinions vary on how much effect this is having on the global economy, what is clear is that capitalism seeks opportunities to maximize profit, and the social quantification sector represents an opening, at least in the short term, to exploit new revenue streams.

Today’s data empires are not necessarily interested in land, but they are interested—at least allegorically—in air. If there is one metaphor that captures the structure of this nascent empire, it is that of “the cloud.” Essentially, cloud computing enables on-demand access to a shared pool of computing resources. Because information is stored on third-party servers, not on the owner’s computer, the data is said to live “in the cloud.” More than just a convenient and evocative metaphor, the concept of the cloud operates at different levels to help shape our social realities.

The concept of the cloud functions in a similar way. The cloud can be a metaphor to suggest that data is gathered together in a remote location. It can be a template to organize how data is extracted, stored, and analyzed so that it can be managed together “in” the cloud. And the cloud, especially the Cloud Empire, can act as an episteme, a model for knowledge that organizes our social realities—even those that are not datafied—into structures that can facilitate their datafication.

One reason the Cloud Empire is so successful at appropriating all of human life is because its biggest corporations can dominate our data relations both as sellers and buyers. Companies such as Google act as monopolies, “selling” us (sometimes providing to us for “free”) every device or service we need to live our datafied lives. But Google also acts as what economists call a monopsony, that is, the single buyer of all the data we produce (the identification of monopsony as the dominant market structure of the internet was made by one of us back in 2013 and has since become commonplace). A monopsony means that although digital tools have empowered us to become producers instead of mere consumers of media, there are limited options for choosing which platforms to participate in. Those of a certain demographic wanting to join a social networking site will probably join Facebook, because that is where their friends are more likely to be. Those wanting to upload videos of cats are highly incentivized to use YouTube, because that is where most users are likely to watch; and so on. We—the producers of data—are legion, but the buyers of what we produce are few. A platform, in particular, is the perfect medium for monopsony, a means to produce social content for capital.

Some might celebrate the power of the internet to subvert capitalism and decentralize cultural production, but this power means little if a handful of corporations control the entire infrastructure of data appropriation and use. The internet, which originally represented a rejection of mass media, is being remassified. On a regular day, Netflix and YouTube account for half of all internet traffic in the United States, with massive implications for their ability to “buy” and distribute content for money or otherwise. The one-to-many model of mass media dissemination has simply been replaced by a many-to-one model, not the democratic many-to-many model we once hoped for.

The monopoly-monopsony hybrids we are describing represent massive concentrations of power. These companies currently exercise a hold over the infrastructure that makes social quantification possible; they control everything from undersea cables to satellites to the “last mile” architecture that delivers internet service to individuals. They also control the environments or platforms in which data is being generated, the design and production of the devices through which data is collected, and the computing capacity necessary to analyze the data, including through increasingly sophisticated machine learning methods that do not require explicit human intervention. Finally, they also control the content.

Appropriation and exploitation are the reasons we believe it is still relevant to talk about colonialism in the age of data. They are not outdated features of an old system but relations that newly define our present age. Marx, in an attempt to identify the characteristics of industrial capitalism, differentiated what he was trying to describe from previous modes of production, including those forms of primitive accumulation that can be said to include historical colonialism. But as David Harvey points out, the problem with this approach is that it relegates “accumulation based upon predation, fraud, and violence to an ‘original stage’ that is considered no longer relevant or . . . as being somehow ‘outside of’ the capitalist system.” Our thesis is that primitive accumulation does not precede capitalism but goes hand in hand with it. We follow authors such as Shoshana Zuboff, Julie Cohen, and Saskia Sassen25 in recognizing an emergent phase of primitive accumulation so unique and historically significant that it deserves treatment as a new stage of history that we call data colonialism.

But if colonialism can be understood, among other things, as a process that allows one party to occupy the living space of another and appropriate his resources, overpowering him through a combination of ideological rationalizations and technological means (which include the use of surveillance and dominance), then we propose that we have entered a new phase of colonialism. To say this is not in any way to forget the racist, sexist, exploitative, and Eurocentric legacy of historical colonialism (which, after all, continues to shape the way we relate to each other today, even online).

If data in the cloud is stored in data centers and not in personal computers, this brings up the very important question of who owns it. Whereas earlier models of the internet allowed for distributed ownership of resources that could be used collectively, the cloud centralizes ownership and establishes very specific parameters for how data is shared. Instead of data being stored at each individual computer and each individual node deciding what data to share or not share, the cloud represents a model in which data is stored in "banks" that are owned by private corporations.

Once data generated by individuals is stored in the cloud, it is the corporations that decide what data other users are allowed to "borrow" from the bank, without those users being able to permanently store it in a way that would let them freely reproduce or transform it. That is why streaming—or downloading content in a format intended for short-term storage and access—is the norm instead of storing content locally and permanently. In this model, users are no longer the proprietors of any data they produce or acquire, and they don't get to decide what is collected or shared or for what purposes. From producers and owners, they pass into being renters of data, laborers who generate data to be sent back to the servers or, as earlier discussed, who are the source of life streams from which such data can be automatically extracted.

Our use of the label "social quantification sector" is intended precisely to expand the discussion beyond infrastructure and to single out corporations and other actors operating at the intersection of larger regions of the economy, including the information and communication technologies (ICT) sector and the arts, entertainment, and recreation sector.

Although our main focus is companies that seek to profit from the capitalization of our social lives, it is important to keep in mind the relation that the social quantification sector has to the larger ICT sector, and indeed to the wider business world, in its use of ICTs. Plainly stated, without the tools and business models of the ICT sector, the social quantification sector would not exist. … The point is that the social quantification sector is more of a subsector of the entire ICT business sector, useful in identifying specific activities and approaches that revolve around the capturing of everyday social acts and their translation into quantifiable data, which is then analyzed and used for the generation of profit. The point also is that in the era of data colonialism, the ICT sector is working ever more closely with all forms of business within the wider Cloud Empire.

Such a degree of control is seen as authoritarian by some and as the epitome of good design by others. Regardless, this stance allows Apple to distinguish itself from its competitors by, for instance, building ad blockers right into its Safari web browser or claiming that they do not sell any of the data they collect from users. Even if this is the case, Apple must take some responsibility for the labor conditions under which its devices are manufactured, for its collusion with the media industry in ways that are not favorable to consumers, or for its innovative ways to avoid paying taxes: the company pays 9.8 percent in taxes, whereas global corporations typically pay around 24 percent.

What is unique about these big five monopoly-monopsony hybrids is that they have carved out the social quantification space among themselves under an informal Pax Britannica–type of noncompetitive agreement. Yes, Microsoft spends about $5 billion a year keeping up with Google Search, and Apple spends about $1 billion45 competing only with Google Maps (just as Google tried unsuccessfully to launch Google+, an alternative to Facebook). But what this really means is that there are no other corporations out there with the resources to compete at the same scale in this sector.

The hardware area includes manufacturers of digital devices that extract and use social data, from laptops and tablets to phones and watches to gaming consoles, robots, cars, drones, and so on. A manufacturer of a “dumb” television set would not be part of the social quantification sector, but any manufacturer of a “smart” device (a product that collects and analyzes information about its user) would be included. This sector also includes the manufacturers of the infrastructure required to run the social quantification sector (routers, servers, transmitters, and so on) and the service providers that deploy hardware to deliver internet and phone access to individuals.

Digital platforms give gatekeeping power to their owners, much as the navigation routes of historical colonialism empowered the towns near where goods had to land. But because of the connected nature of the online environment, today’s gatekeeping power has consequences on a vast global scale, not just a local or national one.

Data brokerage firms are also companies that collect data but more with an intention of “packaging” and selling any personal information related to financial, demographic, criminal, political, recreational, and social activities as well as data about general interests, purchase behavior, and the health of individuals. … There is even the practice of onboarding, which involves matching data from one’s offline activity (gathered, for example, if you give your email address in the middle of a transaction that you thought was offline) with the much larger amount of data available about one’s online activities.

In theory, all this data collection must work around privacy and consumer-rights constraints, but often it does not. Although some countries offer more protection than others, Katherine Strandburg's observation that "US law for the most part has not regulated the datafication or repurposing of information that is acquired as a byproduct of providing service" is applicable beyond North America and particularly in developing countries. In India, for instance, one can easily buy lists of one hundred thousand individuals (car owners, retired women, or any number of categories) for less than $250; the lists are generated from information collected by mobile companies or even by enterprising individuals working at banks or hospitals looking to make some extra cash. In China, regulation about what data companies can collect and what they can do with it is intentionally relaxed, which the government hopes will make Chinese companies more competitive in the global market.

Given that 2017 saw 1,120 data breaches with more than 171 million records exposed in the United States alone, no one can deny that criminal players are part of the social quantification sector. The role of crime is particularly important when, as noted in chapter 1, the overall insecurity of infrastructures of connection is increasing.

Finally, when considering the size and composition of the social quantification sector, we should also consider its economic impact. In terms of actual revenue, it is important not to overestimate the size of the social quantification sector in the wider economy. Oil and gas, financial services, automotive, and pharmaceuticals are in fact larger sectors than ICT. For instance, in 2016, oil conglomerate Saudi Aramco made $311 billion dollars in revenue, compared to Amazon’s $107 billion; Toyota made $237 billion compared to Google’s $89.5 billion; Apple posted a revenue of $234 billion, less than Walmart’s $482 billion; and Facebook’s $27.64 billion and Uber’s $6.5 billion of revenue are considerably less than Volkswagen’s $237 billion. But it is equally important not to underestimate the prominence of the ICT sector in terms of stock market capitalization and not to underrate the considerable growth in this sector: in the first quarter of 2008, the top four corporations in terms of market capitalization (the value of the shares of a corporation) were Exxon/Mobil ($452 billion), Petro China ($423 billion), General Electric ($369 billion), and Gazprom ($299 billion). A decade later, in the first quarter of 2018, they were Apple ($851 billion), Alphabet/Google ($717 billion), Microsoft ($702 billion), and Amazon ($700 billion). In that same year, Apple broke the $1 trillion record in capitalization for a brief period.

By itself, the social quantification sector might be less economically significant than others and certainly not as significant as mainstream opinion would lead us to believe. But it plays a key role in the overall structuring of the Cloud Empire and the shaping of the social world around it. Its smaller economic role is more than made up for by its enormous influence on data processing and business models within the wider social order built around data colonialism.

India's approach in the IT sector, meanwhile, has been to leverage cheap labor and engineering expertise to attract global investment. But in contrast with yesterday's image of the Indian call center providing customer support for Western companies, India's current IT sector represents a burgeoning ecosystem of start-ups, technology developers, and service providers of global reach. This infrastructure is being developed in close collaboration with the state through public–private partnerships. For instance, India's data centers, which provide server hosting and colocation services to companies all over the world, are supported by the government through land allotments and tax rebates; furthermore, the IT sector and the government work closely on national data-management projects such as census or citizen-identification programs.

The most comprehensive example of internal colonization is, of course, China, a Cloud Empire in and of itself that is as advanced as the West's and could in the near future surpass it. In fact, it is already extending to other parts of the world not just through hardware but also through cloud computing, financial services, and artificial intelligence. Companies such as Baidu, Alibaba, and Tencent are sizeable conglomerates with more diversification than their Western counterparts.

On closer inspection, however, such initiatives are problematic not so much because they are uniquely antidemocratic (the West has its own crises of civil rights) but because they perfectly illustrate where datafication might eventually be headed everywhere. In the name of safety and security, other governments will find it easier to claim they must not be left at a disadvantage and will escalate the initiatives already in place (in fact, these governments might find it appealing to import ready-made systems of surveillance along with telecom infrastructure, as some African countries are doing from China). Ultimately, the close collaboration between a government that grants economic and legal advantages to tech companies75 and companies that help the government conduct its surveillance is a global feature of data colonialism, not one restricted to China, Russia, or the United States.

To reiterate, data colonialism is what happens when life becomes the input of capitalism and becomes organized through data relations. As part of this process, data colonialism comes to depend on an emerging transformation of social relations both within the formal economy and outside of it. Urgent calls for us to submit to this new order are portrayed as inevitable, when in fact they are not.

Immanuel Wallerstein observed that “profit is often greater when not all links in the chain are in fact commodified.” By this he meant that capitalism has always relied on indirect modes of exploitation that can operate in addition to the direct modes of exploitation related to wage labor. Put differently, unpaid or underpaid work is of utmost importance in capitalism, whether that work is the work of the slave, the housewife, or the selfie poster. Historically, this has served both to externalize costs so that it is members of society, not employers, who have to bear them and to extend capitalism beyond the formal economy. Today, social quantification represents the most extensive attempt to construct a whole economy based on the free ride that capitalism can extract from our lives, so that modes of unpaid and underpaid work that were unimaginable before are legitimized, normalized, and, in the long run, naturalized.

There is a lot of unpaid work performed by humans on social quantification platforms, even when this work provides value to the network and allows the platform owners to enrich themselves. Consider, for instance, all the ways in which users can contribute free labor in the Local Guides program that is part of Google Maps. This program asks individuals to “Review a place,” “Rate a place,” “Add photos to a place,” “Answer questions about a place,” “Edit a place’s information or Edit a road on the map,” “Add a missing place,” “Check facts about places nearby,” and “Respond to Q&As.” In return, the platform offers recognition; working for them can help “your contributions get noticed” and valued in this “gamified” system. The work can also be remunerated not in cash but in perks that can be used to further integrate workers into the platform (such as “3 months of Google Play Music and 75% off a movie rental on Google Play” or something similar). Many social quantification platforms have similar ways of rewarding users, but in general, the more “social” a platform is, the less expectation there is that users will be recompensed for their work, since the joy of social interaction facilitated by the platform is presented as enough payment. Julian Kücklich refers to this phenomenon in which participation itself generates value as playbor (part play, part labor), suggesting that from the perspective of capitalism, the means of production here are the players themselves.

Workers are seemingly empowered because, as contractors, they set their own hours and choose their own tasks. But employers are empowered too, since they don't have to pay payroll taxes, can avoid regulations concerning minimum wage and overtime work, and can get millions of microtasks performed that regular employees might resist doing.

Turkers are freelance independent contractors who perform small tasks on Amazon's Mechanical Turk platform, or MTurk, a crowd sourcing marketplace launched in 2005 where Requesters (businesses or individuals) can post tasks that computers can't currently perform or that are cheaper for humans to do. Armies of Turkers can then earn small amounts of money (usually less than one cent per task) to perform the jobs, which might include doing things such as categorizing, sorting, testing, rating, generating content, or training artificial intelligence algorithms that will eventually make the human Turker redundant.

The gig economy, for all its limitations, has significance as an exemplar of where capitalism is heading. Platforms and apps do the work of connecting demand (people looking to hire someone to do a job) to supply (temp workers willing to perform that job for a price set by the platform). Because the economic transactions happen within and through the platform, companies are able to benefit by taking a cut from each transaction, through advertising, or by extracting data from the transactions that they can then sell to third parties, all the while washing their hands of any messy contractual obligations between themselves and the users (some platforms simply charge giggers to bid on jobs, regardless of whether they get the job). This is monopsony in action.

Summarizing in a way that is applicable to the whole sharing economy, Melissa Gregg puts it well when she says that "liberation [from employee contracts] frees up Uber from the traditional responsibilities of being an employer. It is ingenious to make employment less predictable, and thus less costly, in the name of independence and choice." As such, regardless of its size or long-term success, Uber remains an important exemplar of a projected direction for global capitalism: it represents the potential to create new forms of exploitative work, to substantially increase the levels of exploitation within established forms of work, and to circumvent human labor altogether (the company is currently investing heavily in driverless cars and trucks). Data-gathering and data tracking are essential to all these functions, the fact of which makes exploitation seem like the "natural" outcome of algorithmic processes. Gone are the narratives of a benign form of capitalism in which institutions still acknowledged a social responsibility. Instead, we have institutions that merely engineer "processes" or "interfaces," and society has to live with the results.

Despite the challenges that gig workers face, those who celebrate the "disruptive" effects that the sharing economy is having on traditional sectors (hotel and taxi companies, for instance) could perhaps be excused for seeing in all this a benefit for society—cheaper services. What is missing from such an analysis, however, is any discussion of long-term effects. The buyers of gig services today may enjoy lower costs, but if more and more of them are forced to become the providers of gig services tomorrow, then the "disruption" that the sharing economy is creating is clearly negative over the longer term.

Not only are workers’ conditions becoming more precarious, but their moment-to-moment activities are also becoming increasingly monitored. A large transformation underway is datafication’s role in changing how workers are managed, which entails an increasing degree of monitoring. As Larry Catá Backer argues, “Surveillance has become not only a technique of governance, but its substitute,” with assumptions and objectives beyond passive data collection. Extending beyond the workplace to all aspects of a worker’s life, this new kind of surveillance is very much a domain of the social quantification sector. Its evolution must be placed in the longer history of surveillance within capitalism, which, as Christian Fuchs points out, was already noted by Marx. Alongside the consumer surveillance, discussed more fully in chapter 3, and the mutual surveillance between competitors that is inherent to market economies, capitalist labor relations have—according to Fuchs—always been characterized by at least three types of surveillance: applicant surveillance, workplace surveillance, and workforce surveillance.

By 2014, an estimated 60 to 70 percent of US prospective employees faced online personality tests before they could even be seen or spoken to by a human being. Personality testing dates back at least to the 1930s, but greater data processing power has massively increased the capacity to integrate such testing and the results it produces into the categorization of actual and potential workers. One estimate suggests there might be as many as seventy-five-million assessment tests administered to potential employees each year in the United States. A form of anticipatory management, personality tests can flag any criterion an employer chooses—for example, candidates who are prone to join a labor union, who display potential markers for mental illness, or who are likely to ask for salary increases or, conversely, remain on the job without asking for a raise.

Once an employee is recruited, both workplace and workforce surveillance take over. The line between the two is in fact disappearing, as the workplace becomes any and every site the worker might happen to be at, while being pervasively measured and monitored through continuous streams of data collection and processing.

Smart scheduling. Work shift-management systems are increasingly being used by businesses such as retail stores and restaurants to schedule work periods, allowing corporations to replace a full-time workforce with a “just in time” army of temps whose schedules are irregular and unpredictable. The benefit for the corporations is that they can increase profits by avoiding overstaffing and full-time contracts. As with the gig economy, the benefits to employees are much less clear. As of 2017, only 17 percent of US retail employees had a set or constant work schedule.

Remote desktop surveillance. This method includes systems such as those used by online staffing platform Upwork to keep an eye on freelancers doing work from home (about a third of the US workforce). The system, installed voluntarily by the freelancers, takes pictures of their desktops at random times and tracks their minute-by-minute activity, including keystroke and mouse movement (are they checking personal email? posting on Facebook?). This is a way for clients to make sure workers are not “wasting time,” and freelancers face pressure to accept this level of surveillance because doing so guarantees that they will be paid immediately after the job is done, while refusing might mean having to wait a long time for a client to pay an invoice. In other words, submission to surveillance is directly tied to economic benefits.

Telematics. This management approach involves using data generated by sensors to continuously monitor a worker’s performance. A growing sector in which this approach is being applied is transportation, where quantification provides new tools to monitor and control drivers. UPS package-delivery trucks, for instance, are now equipped with around two hundred sensors that monitor everything from location and speed to whether the seatbelt is engaged and, of course, how much time the driver is taking to deliver each package. Likewise, 30–50 percent of the US trucking industry is now using electronic onboard recorders that radically change truckers’ relations to their work, fostering competitive individualism and influencing their behavior.

Bodily surveillance. This includes the use of wearable devices and self-tracking technologies (WSTT), which employ Bluetooth, triangulation algorithms, and infrared sensors to measure and track mental and physical manifestations of arousal and performance, allowing for ubiquitous monitoring beyond the workplace and the workday. McDonald’s, for example, has introduced “sociometric badges” that monitor everyday behavior and emotion. Leisure self-tracking devices such as Fitbit are increasingly becoming requirements of employment contracts within the context of “wellness programs.” This sudden focus on well-being stems less from humanitarian concerns than from corporations’ efforts to reduce the influence of stress on productivity: according to the UK’s Health and Safety Executive, the number of days lost due to stress, depression, anxiety, and headaches is rising dangerously. The assumption that the constant monitoring and reporting of health statistics will reduce (rather than increase) stress seems counterintuitive to anyone except the designers and enforcers of such programs. Yet by 2015, a total of 580,000 US companies had already implemented wellness programs involving such devices.

As Ranabir Sammadar argues, “For capital this is the desirable history of labour—labour at work but not visible, ready at hand but not always necessary, labour living but, whenever required, soon dead.” The last part of the quote might sound like an exaggeration, except when we begin to consider the role of automation. The prediction that technology would bring about the end of unfulfilling labor does not seem to be unfolding according to the hopes of economists. Automation and AI technologies, which many corporations in the social quantification sector actively develop and promote, seem to be designed to eliminate laborers, not liberate them.

With the arrival of the “second machine age,” the destruction of many jobs is a given, even if the scope of the transformation is still up for debate. According to some estimates, digital automation puts up to 47 percent of all US employment at “high risk” of vanishing. In the retail sector alone, this translates into 7.5 million jobs out of a total of 16 million. Other estimates that focus on tasks instead of occupations present a much smaller figure of 9 percent of jobs at risk of disappearing. Looking at historical data instead of predictions reveals that adding one robot to a workforce of one thousand human workers has increased unemployment and reduced wages, even if the shifts are very small thus far.

As more people become underor unemployed, social controls for managing discontent need to be tightened. There are two main approaches we are seeing within the Cloud Empire for dealing with this. First, there are, in various countries, increasing alliances between the state and the private sector in supporting each other’s goals. We see a similarity in approaches across the political spectrum—from open to closed societies. These approaches include deregulation (if previous regulation even existed) of the media and technology industries in a manner that gives more market power to favored corporations; collaboration between the government and the private sector to develop and implement technologies for surveillance; expansion of the state’s power to impose special measures of surveillance during increasingly permanent periods of emergency; and comprehensive secrecy about what governments and corporations do with data collected from citizens, all in the name of security and antiterrorism. These practices are replicated in practically every country with a modern ICT infrastructure, regardless of whether they are democratic or authoritarian in character.

Second, there is a targeting of vulnerable social groups as part of the strategies of data colonialism. One thing that makes the Cloud Empire colonial is the approach of corporations and states to the oppressed. Under the Cloud Empire, everyone contributes as a raw input to capitalism, but as with old forms of colonialism, the poor (always a racialized and gendered category) continue to pay a heavier price. Here we allude to the work of authors such as Oscar Gandy, Seeta Peña Gangadharan, Virginia Eubanks, and Safiya Umoja Noble, who argue that when it comes to the poor and the vulnerable, digital “inclusion” comes at the cost of extensive data profiling, including practices that are predatory, discriminatory, exploitative, or simply degrading.

The merciless power of the Cloud Empire comes from an overarching rationality in which data shadows—and eventually stands in for—the very thing it is supposed to measure: life. Apps, platforms, and smart technologies capture and translate our life into data as we play, work, and socialize. AI algorithms then pore over the data to extract information (personal attributes from “likes,” emotions from typing patterns, predictions from past behaviors, and so on) that can be used to sell us our lives back, albeit in commodified form. Science, technology, and human ingenuity are put at the service of this exploitation, making it likely that, unless resisted, the Cloud Empire will in a few decades seem like the natural order of things. This is indeed a bleak picture, but it is not entirely new. The larger pattern harks back to a previous era of rampant extraction—that of historical colonialism, as we will explore in the next chapter.

Historical colonialism required first and foremost an external colonizer, a group of people who took land and resources that were not theirs. "Appropriation" implies that the land was already occupied by someone, although these colonized, "primitive" inhabitants were not treated as the real owners of the resources. Speaking in general terms, then, colonialism can be said to consist of a global system in which colonizers use power to dispossess the colonized of valuable resources, sending the wealth back to the metropolis or the motherland and all the while justifying their actions with a discourse of hierarchical differentiation between peoples and cultures.

A dichotomy of us versus them was established in which the European self was assumed to be a complete individual and the non-European other an incomplete one. The colonizer thus assumed the "obligation"—the "burden," even—of exploiting the colonized in the name of civilizing them. In the eyes of the colonizer, the violent cost of the enterprise to "modernize" the "primitives" was inevitable yet necessary, even if this meant working individuals to death. Indeed, the most inhuman treatment was not seen as unethical because it was not believed to be inflicted on full human beings; violence was merely a feature of the economic mode of production.

If the first three centuries of colonialism were marked by territorial conquest, wealth extraction, and an increasingly interconnected global mercantile network, it would take until the industrial revolution of the eighteenth century for a mode of production not directly based on tribute to fully emerge. Colonialism slowly created the conditions for the transition from the tributary mode of production (plantations or mines with slaves) to something different: modern capitalism (factories with workers, developed national markets, and growing international trade cycles). The capitalist mode of production can be described as one in which people are paid to work instead of merely forced to work. Nevertheless, exploitation is still present in capitalism: workers generate an output value greater than what it costs to hire them, value that is converted by the capitalist into profit. Additionally, workers have no other choice but to sell their labor; since they do not own the means to produce anything themselves, they are obligated to buy what they need to subsist from the capitalist. A geographically extended labor pool—available and willing workers all over the world—was formed. This new labor market was created in large part by advances in manufacturing technologies linked with growth in the transportation, finance, and communication sectors that characterized the industrial revolution, and it allowed factory owners to easily replace skilled workers with interchangeable unskilled workers, which meant workforces could be increased or decreased (hired or fired) in response to fluctuations in the market. The resulting social relations that sustained both labor and commodity exchange in industrial capitalism were radically different from the feudal and family-based forms of production that preceded them. Despite the continuities between colonialism and capitalism, capitalism did indeed transform the social and economic order that had characterized Europe during early colonial times.

An important point in this discussion, however, is that wage labor did not spread evenly across the world. In other words, the capitalist revolution did not look the same or have the same results outside of Europe. The “coloniality of power,” as described by Aníbal Quijano, means that opportunities for labor were always linked to race and geographically differentiated: while wage labor became the norm for white Europeans, their non-Western counterparts continued to be exploited under earlier models (slavery, serfdom, and independent-commodity production) that were nonetheless compatible with capitalism. Colonial labor had always had a racial component added to the other categories of worker classification; one’s place within the system and the kind of labor one engaged in was determined by one’s geographical and geopolitical location as much as one’s race, sex, and labor power. This did not stop being the case under capitalism, at least not for the colonies. Although we may speak of a general transition during the colonial period from a tributary to a capitalist mode of production, the truth is that this transition was not experienced by everyone on equal terms. What defines the history and development of capitalism is, in fact, its ability to simultaneously operate with multiple modes of production: a capitalist metropolis in which workers earn wages can coexist with tributary colonies in which workers are enslaved through mercantile systems of exchange, setting in motion large-scale flows of commodities and people that progressively extended across the world the inequalities inherent in the capitalist system.

The economic development of the West cannot be understood without slavery, although this reality is neglected in the standard histories of capitalism. As late as 1800, other nations such as China or Japan could have emerged to occupy the place that Europe and the United States did in global capitalism, but it was the advantage provided by slavery that allowed the colonizer nations to emerge as superpowers.

Strictly speaking, historical colonialism is over. Most colonies gained their independence by the 1960s. But independence, acquired at a bloody cost, did not overcome dependency. A combination of internal conflict, debt, corruption, and poverty—all direct or indirect legacies of colonialism—meant that liberated colonies seldom achieved authentic independence in the form of generalized prosperity. Consider the case of the 1791 to 1804 revolution in Haiti, an uprising by black slaves that completely transformed the social and racial politics of the nation. Although the revolution managed to expel the French, it came at a cost of 162,000 lives and the imposition of an external debt of 150 million francs, or around $20 billion by today's standards, a debt that was eventually paid back at huge cost, the effects of which have burdened the country ever since.

The legacy of colonialism explains not only debt and poverty but also other important issues of our times such as racism, migration, lack of educational opportunities, and terrorism. The remnants of colonialism will linger as long as former colonizers continue to try to impose their economic and political models of development as the only worthy goals for the rest of the world. Thus, although historical colonialism might be over, we can still speak of coloniality, that is, of legacies of colonialism that have outlived colonial rule per se.

Neo colonialism is a term generally associated with a twentieth-century discourse that assigned to the former colonial powers the blame for the stunted development of newly independent nations. It is also a concept that allowed intellectuals to begin talking about the internal colonialisms that divide former colonial societies along ethnic, and not just racial, divides—for instance, with lighter-skinned people having a higher status than darker-skinned people. Today, neocolonialism is used more frequently to suggest a new form of colonialism whereby a powerful country might exercise control not through territorial occupation but at a distance through economic or political domination, cultural monopolization, or the threat of military force. Contemporary US imperialism might be said to be a form of neocolonialism—in other words, a form of military and economic dominance without direct political control.

Since colonized elites played a crucial part in exploiting their own people (even if they did so under the protection of the colonizer), it is too simplistic to assign all the blame to the colonizers. Thus, postcolonialism attempts to offer a more nuanced analysis of colonialism and its repercussions, particularly with respect to the ways in which the relationship between former colonizer and colonized continues to shape the realities of both groups in the contemporary world.

Other writers such as Edward Said have linked issues of representation to the exercise of power. Said argues that "the Orient" was formulated in the minds of Europeans to possess characteristics that were based not in a search for truth but in a desire to dominate, and he was mostly concerned with how these stereotypes, which could serve not only to oppress but also to romanticize or patronize, were projected and reproduced through scholarship and culture.

Authors such as Aimé Césaire and Albert Memmi, writing in the 1950s, argued that there was no real human encounter in colonialism, only thingification, an objectification of the colonized that justifies violence and racism. The colonized were forced to occupy a place in which their culture and the culture of the colonizer were constantly in conflict and in which they not only suffered subjugation but came to accept it as a natural social order (although theorists such as Memmi have always seen in this opportunities for creation, not just destruction).

Postcolonial studies also attempt to examine how history and science were weapons in the quest to use knowledge to subordinate the colonized politically and economically and to get them to accept their subordination. In order to accomplish this, history and science needed first to be portrayed as the domain of Europe, part of the heritage of the Enlightenment that the West deemed necessary to bring to the rest of the world. History (the particular histories and trajectories of colonial appropriation) thus became a universalizing Eurocentric endeavor, describing every thing not just from the perspective of European eyes but in terms that made sense only in relation to European values. This kind of historicism situates progress and “universal” principles as emanating from the West and then flowing to the Rest: modernity emerges as a one-directional arrow pointing to Europe, something all “nonmodern” societies must strive toward. The history of human civilization is seen as an evolutionary trajectory that starts with man in a natural state and culminates with European “civilization.” Whatever inferior samples are encountered along the way (in the form of non-European humans) can have their differences explained through natural (that is, racial) means and not through a history of power. Whereas all knowledge systems can be said to be ethnocentric, this modern European ethnocentrism is the only one that claims for itself universality across time and space. Postcolonial studies help us see how the language of modernity and universal knowledge, inscribed in historical discourses, becomes a tool of imperialism to civilize and educate colonial subjects.

The military, mercantile, and bureaucratic exploits of colonial Europe would not have been possible without scientific and technological tools. These included new means of communication and transportation (the telegraph and the railroad) as well as new ways of collecting and analyzing information for the purpose of the administration of populations and commodities (statistics, accounting, economic theory, and so on). The success of these inventions in facilitating empire-building was seen as further validation of the superiority of the West. But beyond the creation of tools for managing territories and populations, discussed from a different perspective in chapter 4, colonial science provided its own new rationalizations for conquest. A whole new obsession with classification, particularly during the eighteenth and nineteenth centuries, helped to establish the boundaries not only between human and beast but between perceived higher and lower forms of humans, all in the name of the expansion of science and, through it, the betterment of “man.” Immanuel Kant, a central figure in modern Western philosophy, provided the first scientific definition of the concept of race in his 1775 essay “Of the Different Human Races.” Subsequently, a branch of anthropology became invested in the elaboration of “racial types,” whereby certain physical attributes were linked to certain mental and behavioral characteristics—for instance, skull size and other arbitrary measures were used in an attempt to establish the superiority of white men over people of color and also over white women.

This “scientific” racism and sexism would become essential in colonial governance. Well-meaning white people believed (and continue to believe) that Western science and technology were the only means of improving the condition of the destitute “savages” in the colonized territories. Other less well-meaning individuals saw the absence of “advanced” knowledge in the colonized as a good reason for their annihilation. Frederic W. Farrar (a cleric born in India who became friends with Charles Darwin) argued, for instance, that the extermination of the Tasmanian people was justified because these were people who had not “added one iota to the knowledge, the arts, the sciences, the manufactures, the morals of the world.”

These are goals that the project known as decoloniality also shares, but its approach to realizing them is somewhat different. To begin with, decolonial thinkers critique certain aspects of postcolonialism. They reject Western traditions more emphatically and instead draw their inspiration from the work of nonwhite and mostly non-Western thinkers such as Frantz Fanon. They accuse postcolonialism of being too aligned with the antihumanist thought of postmodernism: in their broad distrust of narratives, rationality, universalism, and constructs such as freedom and human rights—or, rather, in their association of all of these things exclusively with Western values—postcolonial theorists (the critique goes) become complicit in the continued dominance of Eurocentrism. Instead, decolonial thinkers are inspired by and engaged with grassroots movements such as the Sem Terra (Landless) movement in Brazil or the Zapatistas in Chiapas, gender-equality movements such as Chicana and Muslima feminism, projects such as the Caribbean Philosophical Association’s efforts to shift the geography of reason, and the creation of institutions such as the World Social Forum.

First, as already mentioned, it seeks intellectual sources beyond the Western canon and is inspired by political movements from the Global South. Second, it is based on a critical dialogue between diverse political and ethical projects that moves toward a pluriversal rather than a universal worldview (based on an abstract Western universalism). And finally, it recognizes that the decolonization of knowledge can be achieved only through critical thinking “from and with subalternized racial/ethnic/sexual spaces and bodies.”

IN SID MEIER'S STRATEGY VIDEO GAME Civilization IV: Colonization, you—the human—can choose to play Spain, England, France, or the Netherlands in a battle to colonize the New World. The colonized, though, are represented by "nonplayer characters" controlled by the game's algorithms. Your role is to oversee an expanding territory from an omniscient "God view" perspective, allocating resources and squelching insurrections. Meanwhile, the only reason nonwhite people and their lands exist is to be conquered and used, making at least that aspect of the simulation seem authentically colonial. Rated E for "Everyone," the game whitewashes the truly horrendous aspects of colonialism: the slavery, the rape, the genocide. And perhaps because of this level of sanitization, the game-play quickly becomes boring and repetitive, falling into the established 4X pattern that characterizes strategy video games: explore, expand, exploit, and exterminate.

It is not surprising that strategy video games, which are all about the expedient management of resources in the context of colonization, should follow this 4X model. But what of data colonialism? In this chapter, we use the framework of exploring, expanding, exploiting, and exterminating to conduct a transhistorical comparison of dispossession that will reveal continuities between historical colonialism and today's data colonialism. To truly recognize the injustices of today, we need to look at them through the eyes of yesterday. By understanding the detailed parallels, we can uncover fully the coloniality3 of data relations, a concept that points to the broader continuities in how the legacy of historical colonialism resonates in the present—in this case, through social relations as abstracted and commodified by data.

On the American continent alone, of the 145 million people living in the hemisphere before the conquest, 90–95 percent had been exterminated by 1691.4 Most of the survivors were condemned to poverty. Around the time Columbus arrived in America, a typical European had an average per capita income three times that of someone living in what is now sometimes called the Global South. By 1850, that ratio had increased to five to one. By the early to mid-twentieth century, when around 85 percent of the world was under some form of colonial rule, the ratio was fourteen to one. The inequality that plagues us today is not entirely the result of colonialism but at the same time cannot be explained without it.

In this new world, corporations act as colonizers that deploy digital infrastructures of connection to monetize social interactions, and the colonized are relegated to the role of subjects who are driven to use these infrastructures in order to enact their social lives. Just as, to meet the requirements of industrial capitalism, all economic relations were transformed into market relations via the medium of money, today social relations and human life in general are increasingly transformed into potential market relations through the medium of data. Because it is very often the stuff of a subject's own life that is being exchanged, data relations comprise what in the perspective of colonial history can be seen as an embryonic “mode of production,” a new form of tributary relationship between colonizer and colonized.

Instead of looking at them as discrete and separate moments in history, it is more productive to look at colonialism as an “uncivilized” mode of production allowed to operate at the margins of a more “civilized” capitalism. Let us then conceive of colonialism and capitalism as two sides of the same coin: one side enacting dispossession in a brutal and rampant manner and the other normalizing this process by relegating it to the outside (outside the present, outside the civilized, outside the measurable, and so on). All the while, however, the process of dispossession itself is presented as a necessary and beneficial step in human progress. Various critics of neoliberalism and datafication9 have already recognized in our times a renewed race to amass wealth through new kinds of dispossession. But to understand why this form of accumulation is a continuation and not a break with the historical arc of the past five centuries, we need to look more closely at the history of colonialism and its role within capitalism.

Colonialism, in its old and new forms, operates through the dispossession of resources. This dispossession happens through the appropriation of things that belong to someone else and through the extraction of value from the appropriated resources. By doing this, the natural and the social worlds are implicated in new modes of production, transformed into factors in ever-changing capitalist cycles of production. Put differently, appropriation and extraction convert life, particularly human life, into profit centers for capitalism.

For personal data to be freely available for appropriation, it must first be rendered as a natural resource, a resource that is just there, ready to be extracted. This is where the logic of colonialism comes into play; the idea is reinforced that the depletion of resources by those who find them is natural. There is a parallel here with how capitalism has appropriated natural resources, which is by framing their state of readiness for exploitation through legal, linguistic, and other means. Jason Moore10 describes how this process has worked, arguing that capitalism has so far depended on the availability of cheap nature: natural resources that are abundant, easy to appropriate from their rightful owners, and whose depletion is seen as unproblematic. The resources are not cheap per se, but legal and philosophical frameworks are established to rationalize them as such. Exterminating natives and chopping down trees is accomplished "cheaply" only when a certain ideological, legal, cultural, and even religious system of beliefs is put in place.

Like oil, data needs to be processed. But unlike oil, data is not a substance found in nature at all. It is the by-product of social interactions that are mediated by digital technologies, a by-product that is captured and processed by a third party that is not intimately involved with the people in the interaction and indeed can operate at many removes from them through the market for selling data. The first use of the metaphor of data as the “new oil” is attributed to Clive Humby, the conceptual grandfather of the loyalty shopping-club card, who in 2006 said, “Data is the new oil. It’s valuable, but if unrefined it cannot really be used. It has to be changed into gas, plastic, chemicals, et cetera to create a valuable entity that drives profitable activity; so must data be broken down, analyzed for it to have value.” Data must thus be presented as an ownerless resource that can be exploited only by certain parties (what Julie Cohen calls data refineries).

Data are an intangible asset; like other information-related goods, they can be reproduced and transferred at almost zero marginal costs. So in contrast to the concept of ownership of physical goods, where the owner typically has exclusive rights and control over the good—including for instance the freedom to destroy the good—this is not the case for intangibles such as data. . . . Data ownership can be a poor starting point for data governance, and can even be misleading. . . . The important question isn’t who owns the data. Ultimately, we all do. A better question is who owns the means of analysis? Because that’s how . . . you get the right information in the right place. The digital divide isn’t about who owns data—it’s about who can put that data to work.

These rationalities are more practical and specific than are the broad ideologies of data colonialism mentioned in chapter 1. They help make all the small choices involved with data relations seem rational and common sense.

Extractivism began at the point of contact between the colonizer and the colonized. This initial encounter contained within it the terms of exploitation to be deployed as the colonizers explored and "discovered" new resources. Consider the Spanish Requerimiento of 1513, a document read by conquistadors to newly found subjects, who were abruptly informed that their lands belonged not to them but in reality to the spiritual leader of the explorers, someone called the pope, successor to Saint Peter and leader of the Catholic Church: Of all these nations God our Lord gave charge to one man, called St.

Of all these nations God our Lord gave charge to one man, called St. Peter, that he should be Lord and Superior of all the men in the world, that all should obey him, and that he should be the head of the whole Human Race, wherever men should live, and under whatever law, sect, or belief they should be; and He gave him the world for his kingdom and jurisdiction. … But, if you do not do this, and maliciously make delay in it, I certify to you that, with the help of God, we shall powerfully enter into your country, and shall make war against you in all ways and manners that we can, and shall subject you to the yoke and obedience of the Church and of their Highnesses; we shall take you and your wives and your children, and shall make slaves of them, and as such shall sell and dispose of them as their Highnesses may command; and we shall take away your goods, and shall do you all the mischief and damage that we can.

The irony, of course, is that voluntary submission was not even a real option, because this was a document read in Spanish to non-Spanish-speaking peoples. Even if they could have accessed a limited translation, they probably would not have understood the legal and theological concepts by which they were suddenly and forcefully dispossessed of their property, or even understood "property" in the same way as did the invaders. The quick annexation of resources was justified by means of these abstract rationalizations that needed to make some sort of sense only to the colonizer.

The Requerimientos of our times are known as end-user license agreements (EULA) or statements of rights and responsibilities (SRR). These documents spell out how newly discovered resources and their presumed owners are to be treated by colonizers. … At least according to European law, “opting out” does not meet the requirements for legally valid consent.

Users give Facebook permission to use, for commercial or noncommercial purposes, whatever content they upload to the platform, including photos and videos. What is less well known is that the license that users give Facebook, as specified in the SRR, is transferable (which means companies affiliated with Facebook also own the content), worldwide (which means it applies to all Facebook users, regardless of the laws in the places where they live), and royalty-free (which means users are not entitled to any remuneration).

Such blanket and overreaching clauses—which are standard in the social quantification sector—are not only legally dubious but as full of hubris as the Requerimiento was. Companies reserve the right to modify agreements at any time, and users are expected to preemptively agree to any changes, no matter when and how often they are implemented. Because the language is so incomprehensible, most users simply scroll to the bottom of the document and click the “I agree” button without so much as a quick glance.

Following the “discovery” of the Americas, a global network for the flow of goods and information began to take shape, a network whose infrastructure continues to be visible today and whose development was part and parcel of the growth of Western science and technology. Francis Bacon, for instance, made an explicit connection between long-distance travel (that is, colonial expansion and the knowledge transfers it unleashed) and the growth of various scientific disciplines. Since the colonies could not be managed directly by the metropolis, they had to be objectified through scientific means and represented through emerging forms of records that could travel through time and space and thus conquer complexity. Principal among these new methods of management were two important technologies: maps (representations of the space to be colonized and managed) and surveys (mechanisms to extract data that can be analyzed in order to manage colonized spaces). … But according to Cole Harris, modern maps “conceptualized unfamiliar space in Eurocentric terms, situating it within a culture of vision, measurement, and management.”

Like the network visualizations of today, colonial maps illustrated a form of knowledge that can be described as nodocentric. The maps showed the location of mines, ports, towns, and other resources that were crucial to the operation of the empire; they isolated and represented the points or nodes that had value and rendered everything else invisible. A network diagram performs very similar functions. Only nodes and their connections are prioritized, giving us a reality in which nodes can only “see” other nodes, and only nodes deserve to be accounted for. In essence, the colonial maps of the past and the network diagrams of the present function as spatial representations of a power structure that in each case submits to a capitalist and a colonialist logic. They connect peripheral resources to processing centers or network hubs while depicting intermediary spaces as empty or dead. Locations that hold no value to the colonial enterprise, just like individuals who are not users of the platform, practically disappear from the map or diagram.

Speaking of processing centers, it is important to note the role in exploration that the metropolis played as the centralized data bank in which information about colonized territories would be collected, analyzed, and used to assemble maps and other documents. Institutions such as the Casa de la Contratación de las Indias and the Consejo Real y Supremo de las Indias in Spain or the Royal Geographical Society in London would function as information repositories and map production centers at which standardized survey data would be organized into chorographic records and reports on the natural and social history of colonized territories. Other institutions such as the Dutch East India Company maintained extensive botanical gardens to collect knowledge about plant species from the colonies. Centers such as the Society of Jesus in Rome served as clearinghouses of new medical knowledge based on remedies made from New World ingredients, knowledge that would then be disseminated through the connections of these centers to universities. Today, data centers serve similar functions by storing data and "mining" it—a pertinent colonial metaphor—to produce new knowledge for the benefit of corporations.

For example, Britain forced companies that operated transatlantic submarine cables to use London as their connection hub (after World War I, the epicenter for global communications shifted from England to the United States). Ownership of the new communication infrastructure continued to follow colonial models as well by concentrating it in the hands of a few multinational consortia of state and private-sector corporations. When independence was finally achieved, former colonies simply inherited the infrastructure of the colonizers and in many cases were pressured to allow foreign companies to continue to control and profit from that infrastructure. To attempt to intervene with that model could have potentially disastrous results, as evidenced by the extreme example of Chile, where the US-based International Telephone & Telegraph corporation played a crucial role in the CIA-backed overthrow of the government in 1973 after democratically elected president Salvador Allende threatened, among other measures, to nationalize the telecom industry.

We would do well to remember, however, that even when infrastructure came under the control of independent governments in post colonial times, it did not necessarily represent gains for the average citizen. As Miriyam Aouragh and Paula Chakravartty40 remind us, calls by former colonies for the democratization of global media systems often went hand in hand with efforts by those same governments to silence dissent at home. In most cases, the censorship and antidemocratic regulation undertaken by authoritarian regimes was backed by the United States and other Western powers; commitment by the West to “freedom of information” consistently took second place to the support of oppressive regimes that pushed the West’s agenda of modernization.

One need only consider Facebook’s Free Basics project. According to its own PR, Facebook is “helping people around the world access impactful local services, including health resources, education and business tools, refugee assistance sites, and more.” But the NGO Global Voices calls the program an instance of “digital colonialism” that, under the guise of bringing “free” internet to underdeveloped countries, locks its fifty million users into the company’s extraction model. The Global Voices report43 looked at the success of the program in Colombia, Ghana, Kenya, Mexico, Pakistan, and the Philippines and found that Free Basics is not always customized to feature local content or use local languages. Additionally, the program breaks net neutrality principles by preventing users from browsing the internet openly, and, more importantly, it collects data about all the activities of users (restricted as they are), not just those conducted on Facebook. Because most users of Free Basics are people already connected to the internet who merely want to lower their data-plan bills, the report concludes that the program is motivated more by the desire to capture new audiences and less by a desire to help people bridge the digital divide.

Indians have plenty of reason to be suspicious of foreign corporations. Founded in 1600, the East India Company by the year 1833 controlled five hundred thousand square miles of territory in South Asia, encompassing 93.7 million subjects paying 22,718,794 pounds a year to the British empire in taxation. At its height, the Company controlled about half of the world’s trade, specializing in basic commodities such as tea, textiles, and spices. The power of the East India Company was secured by its own military and paramilitary armies, which served not only to protect its interests and enforce compliance with its decrees but also to collect revenue and perform police duties.

The effect of the railroad on the expansion of empire stemmed from its role not just as transportation technology but also as a means of economic development, state building, and information distribution.

The railroad as transportation and communication technology enhanced colonial governance by improving and extending tax collection and information flows, making centralization more efficient and allowing for the rapid deployment of military resources.

Karl Marx himself believed the railroad and the newspaper would help to politically unify India. This list of effects and hopes seems to closely mirror those surrounding the internet.

Along with other forms of identification that were introduced in the colonies, the point of fingerprinting was not primarily to identify unknown subjects (who hadn't been fingerprinted yet) but to extend police power over groups who were already under suspicion and already indexed by the system.

Similarly, obedience to authority in data colonialism is promoted through a quasi-paranoid self-monitoring feeling that “we live in public” but that we need not worry too much about it because everyone else is also being watched. We need not worry, that is, as long as we have “done nothing wrong.” If the panopticon effect relied on our inability to tell exactly when we were being watched, so that we behaved all the time as though we were, the “inverse panopticon effect” of pervasive data surveillance relies on us knowing that we are being watched all the time but lapsing into behaving as though we are not, thus naturalizing acceptance of a world in which surveillance and continuous tracking operate unnoted in the background.

In this manner, the technologies of data colonialism become what Ashis Nandy called (in the context of historical colonialism) our most intimate enemies—internalized overseers that we carry with us even when we are not connected to the network. As Nandy observes, colonialism is a “state of mind” that requires a “shared culture” between the colonizer and the colonized. Our comfortable acceptance, at least until recently, of the extractive rationalities of data colonialism indicates the extent to which our subjectivities had already been utterly occupied by data colonialism’s norms and ideologies.

Natural resources, or cheap nature, could be exploited by colonialists and capitalists only through employing cheap labor, which framed the productive capacity of a certain class of human beings (first slaves and then workers) as exploitable also. Like cheap nature, cheap social data can be fully capitalized only through the exploitation of cheap labor. But as we argued in chapter 1, data relations are not labor relations in the strict sense of the word. Jason Moore explains that “capitalism must commodify life/work but depends upon the ‘free ride’ of uncommodified life/work to do so.” Similarly, with data colonialism, the exploitation that occurs through social quantification has less to do with the kind of exploitation that happens in a factory and more with the “free ride” that capitalism benefits from when extracting wealth from our broader social lives.

Words such as exploitation and labor have very specific meanings in Marxist theory, so it is worth discussing how they might be employed in the context of datafication. Strictly speaking, as Nick Srnicek observes, internet users do not receive wages, and their work is not used to lower production costs, increase productivity, and so on, so “it is hard to make the case that what they do is labor, properly speaking.” Thus, from a Marxist perspective, internet users cannot be said to form part of a working class, and we can’t say they are exploited in the traditional sense. But theorists argue that media consumers are exploited not because they literally work for media companies but because as an audience they become a commodity sold to advertisers without themselves benefiting at all (for researcher and activist Dallas Smythe, for instance, watching advertisements was a form of labor and exploitation).

More bluntly, if the price of new tools for "overthrowing capitalism" is to annex all of social life to capitalism, then the deal is a bad one.

There seems to be, however, an imminent shift in the struggle for global dominance at a time of accelerating ambitions for artificial intelligence. According to the McKinsey Global Institute’s Connectedness Index, which measures the size of inflows and outflows of goods, services, finance, people, and data relative to a country’s GDP, China has climbed from number twenty-five to number seven in recent years.

As with previous stages of colonialism and capitalism, exploitation creates and exacerbates inequalities between metropolis and colonies, between “developed” and “underdeveloped” worlds. The flow of data between the Global North and the Global South exemplifies the inequalities that still plague neocolonial relations, especially since the circulation of digital data is becoming just as important as the circulation of material goods in international trade. Cross-border data exchanges represented an overall traffic flow of 211.3 terabytes per second (TBps) in 2014 (up from 4.7 TBps in 2005) and were valued at $7.8 trillion in that year alone. But a closer look reveals that exchanges are highly uneven: while flows between North America and Europe jumped from 1,000 TBps to more than 20,000 TBps between 2005 and 2014, the flow between North America and Latin America, for example, increased only from 500 TBps to 5,000 TBps during that time, and the flow between Asia and Africa (to illustrate flows between the Global South) went from less than 50 TBps to no more than 500 TBps. An overwhelming majority of all cross-border data traffic passes through the United States and is subject to whatever surveillance and collection methods that government deems necessary. Thus, the Global North still assumes the role of gatekeeper, as it did in the days of the telegraph and the telephone, and data flows continue to replicate the movement of resources from colony to metropolis.

While enjoying global reach, most data empires are concentrated in the United States and China; other countries may contribute a few notable competitors (Spain’s travel website ODIGEO and Brazil’s e-commerce platform B2W, for example), but their international reach is much more limited. According to 2017 figures, three out of the top four revenue-earning IT companies in the world were US based, and one was Chinese; four out of the top five “traditional” media companies were also US based. The extractive reach of these companies may be global (more so for US companies at the moment, with China close behind), but it is somewhat misleading to think of them as “transnational” corporations: most of their decision power, employment, capital expenditures, and profits are kept close to home.

In a study on development and growth, Steven Weber75 describes the economic imbalance that can result when one side (the colonized) provides the “raw” data, and the other does all the capturing, the processing, the analysis, and the creation of value-added products that the colonized cannot develop on their own and which they must buy at a disadvantage. As has been the case throughout the history of colonialism, this results in the colonizers becoming richer and more powerful, and the colonized underdeveloped and dependent. “The users in [disadvantaged] countries get to consume the products but are shut out of the value-add production side of the data economy,” Weber writes.

It is estimated that 185,000 kilograms of gold and 16,000,000 kilograms of silver flowed from Latin America to Spain during the sixteenth and early seventeenth centuries, coming principally from mines in Bolivia and, later, Mexico. A Catholic priest described these mines as veritable mouths of hell, in which slave or indentured laborers—men, women, and children treated “like stray animals”—died from exhaustion, abuse, or poisoning within four years of arriving. In an ironic and cruel twist, however, Spanish law (the Compilations of the Laws of the Indies) prohibited the exploitation of the “Indians” working in the mines and granted them the right to own and profit from them. In other words, while in reality indigenous people were being worked to death, the letter of the law treated them as rightful owners of the mines, explicitly denying the fact of their extermination.

The mission statements of social quantification companies do not mention anything about extermination of natives, but like Spanish law, they adopt a very generous and helpful attitude toward those whom they seek to subjugate. They purport to “bring the world closer together” (Facebook) and to “enable people and businesses throughout the world to realize their full potential” (Microsoft). They also aim to “give everyone the power to create and share ideas and information instantly, without barriers” (Twitter). Google has policies that promote values of transparency, user choice, and responsibility—including a “responsible manufacturing” policy and even a “conflict minerals” policy,” both of which regulate the sourcing of manufacturing materials from war-free locations. Amazon wants to be a place where “customers can find and discover anything they might want to buy online,” and Airbnb imagines a “magical world” in which “everyone of us can belong anywhere” and we “can be anything we want.” Intuit, which sells financial software to individuals, declares that they want to change their users’ lives “so profoundly, they couldn’t imagine going back to the old way.”

Granted, none of these companies is setting out to commit genocide. But while physical violence is the feature we most often associate with colonialism, economic violence can also have disastrous and lasting effects on a society. Undeniably, extermination was a key component in the process of extraction and appropriation during historical colonialism. But as the system “evolved” to one characterized by the organized economy of industrial capitalism, the need for applying physical violence was reduced. Systemic violence is a costly strategy, to be applied only when there is a demonstrable “return on investment,” (when the costs of the large-scale brutalization of native populations still translates into profits for the empire). But a point can be reached when these costs become too high, economically but also ethically, since empires can present themselves as righteous while engaging in genocide only for so long. At such moments, direct systemic violence is gradually replaced with indirect and interpersonal forms of localized cruelty (as between masters and slaves) and eventually eliminated and replaced with entirely different social relations, such as labor or data relations.

It is well-known that an automaton once existed, which was so constructed that it could counter any move of a chess-player with a counter-move, and thereby assure itself of victory in the match. A puppet in Turkish attire, water-pipe in mouth, sat before the chessboard, which rested on a broad table. Through a system of mirrors, the illusion was created that this table was transparent from all sides. In truth, a hunchbacked dwarf who was a master chess-player sat inside, controlling the hands of the puppet with strings.

Imagine a product designed to enforce the long-distance economic dominance of one group over another, a product that creates passivity and dependence on the part of those who consume it, that can be used to destabilize the economic and political relations within a society, and whose usage entails an exploitative chain of transnational dimensions. The fact that we could just as easily be talking about opium or the latest app from the social quantification sector suggests how much things have not changed in colonial relations over the last two centuries.

Around 1800, Britain was consuming a lot of Chinese tea and other Chinese products and was therefore becoming increasingly indebted to that country (while goods flowed from China to Britain, considerable amounts of silver flowed from Britain to China). In order to fix this trade imbalance, Britain decided to try something different. Instead of the territorial colonization they had practiced in India and the Americas, the East India Company was mobilized to attempt a financial conquest of China through the trade of a highly addictive drug that “targeted the biology of the Chinese population.” Since opium could not be grown legally in China, the East India Company moved all of its production to India, becoming a monopsony for the poppy plant: “If a peasant decided to be in the business of producing opium, he had no option but to deal with the Company. . . . The entire output of the drug would have to be handed over to the Company through a contractor at a price determined unilaterally for the year.” Two wars later, the Chinese had been forced to open their territory to trade and had ceded Hong Kong to the British, and the East India Company had saturated the market with the drug, devastating Chinese society. By 1825 the opium trade was generating enough funds to pay for all the Chinese tea Britain consumed, and on top of material commodities the British were taking large numbers of indentured laborers from China (“coolies”) to work in their colonies.

Consider the teen drama Skam, a social media series that started on Norwegian public TV and is now a very successful Facebook franchise airing in the United States, France, Germany, and Italy (and watched all over the world by millions of teenagers). Skam did not invent the use of personal surveillance as a mode of social interaction, but it definitely promotes and rewards it by encouraging viewers to stalk or “play detective” with fictional characters on Facebook and Instagram as a way to drive the narrative forward (Instagram, currently a very popular platform with young people, is owned by Facebook). The fact that Facebook can create realistic profiles for fictional characters, complete with histories and timelines that can be modified by the corporation to augment credibility, illustrates Facebook’s ability to manipulate reality for profit.

Skam’s ability to create addiction in vulnerable audiences, reorganize reality for profit, reduce social life to social media life, derive value from the resources of users, and mobilize global markets makes it a good example of data colonialism in action.

If we fail to do that, the term digital natives (generations who have always grown up with the internet) starts to take on a more sinister meaning: those whom data relations have dispossessed.

In their responses to historical colonialism, postcolonial studies have helped us frame counterhistories, or alternative versions of history that reclaim elements erased by Eurocentrism. Similarly, a decolonial approach to data can help us frame counterpresents, alternative understandings of our current social realities that allow us to see ourselves as more than mere inputs for capitalism.

In order to articulate these positions, what are needed are methods for denaturalizing the logic of data and critiquing data colonialism. We must then use the results of those methods to reimagine different forms of social and economic relations and connections outside of Western notions of modernity. The project of creating these alternate versions of modernity is what Enrique Dussel calls a “worldwide ethical liberation project” that challenges the universalism of a Western form of reason. In essence, creating counterpresents implies engaging in media practices that question how notions of “progress,” “civilization,” or “innovation” are equated with data relations as conceptualized by corporate interests. Creating counterpresents would also mean reclaiming the agency of the colonized data subject. The colonial project granted subjecthood to dominated individuals, but it did not grant them equality.

An important reason for undertaking all of this urgently is that data colonialism is a period of transition. It is not here to stay; its patterns simply announce an emerging order that is yet to come. As discussed in chapter 1, historical colonialism served as a prelude to modern industrial capitalism. The eighteenth-century abolition of colonial slavery solidified the British empire's transition in the nineteenth century to a new form of imperialism based on liberal values such as free trade. This in turn allowed the empire to reimagine itself at the center of a global network of commerce in manufactured goods—the beginning of what we would recognize today as modern capitalism. In short, historical colonialism was not just about systematic and global extraction and appropriation but about reimagining those things as progress and enforcing a new world order in which colonizers could continue to rule with decreased (yet still prevalent) levels of violence but also with increased legitimacy. They would rule, in fact, by disavowing violence and declaring themselves the defenders of human values. In light of this history, we must reassess recent statements made by commentators following the Cambridge Analytica scandal expressing dissatisfaction with the corrupt data extraction system but assuring the public that it can be reformed and salvaged so that the march of progress may continue.

The challenge that data colonialism poses for social life and for critical social research is entirely new and disturbing. Because this new colonialism's mode of operation involves producing the materials from which information, and ultimately knowledge, are made, its operations cannot be insulated from how we understand the worlds we share, from our social imagination and even our politics. Data relations generate value in the very same movement in which they redescribe and remap the world, fusing knowledge production and value extraction.

As a way into this complex topic, we consider social knowledge forms in the North America and Europe of the nineteenth and early twentieth centuries. There is a special reason for this comparison. One of the most powerful challenges to economics’ detachment from the social world was made by the historian Karl Polanyi. In the 1940s, Polanyi rethought how market societies came into existence. The advent of industrial machinery changed everything, requiring long-distance markets to supply the parts that kept machinery moving and sufficient customers to buy what machines produced. Since markets are oriented to exchange for gain regardless of scale, the expansion of national markets across every sector of life severely damaged local social relations; far from being natural, human life organized exclusively around markets proved almost unlivable in seventeenthand eighteenth-century England, the first economy to industrialize, as older ways of organizing resources (for example, the family workplace and the local market) were pulled apart.

Polanyi concluded that “a market economy can only exist in a market society,” which must be created through “highly artificial stimulants administered to the body social.” This happened in the nineteenth and twentieth centuries through what Polanyi calls a double movement; first, a profound institutional reorganization made markets (in particular in labor, land, and finance) into social realities, turning “all transactions . . . into money transactions”; second, over time, a countermovement of compensatory mechanisms emerged to control market relations and make them socially livable. Key developments were the regulation of labor, land markets, and the financial sector.

Polanyi’s analysis has provocative implications for today. As he makes clear, the market transformation that data colonialism involves will not succeed by itself: it will require its own social transformation that, initially, is likely to be violent, or at least dislocating. At the core of any such social transformation will be “the commodity concept.” Capitalism, to reap the benefits of data colonialism, must make the social domain into a marketized space whose every point has value and is available for commodity exchange. Data relations, and the new forms of “knowledge” to which they give rise, are the key means to this.

The value of data colonialism's extractive processes depends on the comprehensiveness of the data generated. As a result, and certainly not by accident, a new form of knowledge emerges that appears to cover the whole of the social world, but only because no part of that world is exempt from processes of data extraction. This new social knowledge begins to override older forms of social knowledge. In making a historical comparison here to North America's and Europe's nineteenth century, we have no wish to romanticize the fierce social and political conflicts of that era, still less to romanticize the term social. What is new today is not the quantification of the social or its measurement but the depth at which datafied social knowledge conducts its operations and how the resulting social knowledge is (or is not) accountable to the social actors it describes.

The social quantification sector is fast building its own social knowledge. That knowledge is commercially controlled, generally opaque in its production, not open to public debate, and thoroughly oriented first to capitalist drives before it is oriented to the interests of government. It shapes everyday actions quite differently than did earlier forms of social knowledge. The result, as we already hinted in chapter 1, is a different type of social order in which the agency and control of corporations is enhanced and that of individual citizens is reduced and sometimes even erased.

The notion of "society" is not natural; it had to be constructed at a certain time and place. The nineteenth century's (indeed, also the early twentieth century's) construction of society emerged out of what Max Weber called "the peculiarity of modern culture, its calculability."

Dry details perhaps? Every process of abstracting knowledge from every day life needs practical tools for formulating it in public as well as devices to disseminate it. This “concrete history of abstraction” is of enormous significance for understanding nineteenth-century Western politics and social reform. Numbers, and the statistical science that developed through them, enabled a model of social control, with the statistical norm operating as an instrument of normalization, within a wider vision of “government from the social point of view.” This was the period when the modern psychological disciplines and the practices of social care and urban policy developed. Statistical knowledge brought with it, over time, more complex mechanisms of description, what historian Dan Bouk calls “data aggregates” such as “charts, tables, maps, or even algorithms.” For example, Charles Booth’s London poverty maps (prints of which can still be seen on the corridors of the London university where one of us works) provided a tool for managing neighborhoods as part of the wider regulation of social life.

These social uses of numbers were a crucial site for contesting power. An emerging social state aimed at the betterment of all created “experts of the social” with new types of professional authority. A key dispute within statistical science was whether “the norm” (for example, the norm of average family size or income) was a real-world reference point around which human variation actually turned or an abstract mathematical point that random variation implied. As Hacking writes, “Normality” emerged “to close the gap between ‘is’ and ‘ought.’” Science apart, statistics’ emergence as a forum for debating “the social” did not escape contamination by more sinister dreams of managing life—indeed, the evolutionary prospects of the human race. The most scandalous example was eugenics, the pseudoscience of managing births to secure better genetic outcomes for the overall population. Here the apparently objective science of numbers merged with the racial hierarchies between peoples that had already shaped colonialism for centuries.

First, the key actors in developing social knowledge and social statistics in this earlier period operated, with few exceptions, at the state’s initiative (an exception would be the data-gathering of the commercial insurance industry). It was not until the 1970s and 1980s in the United States that large private companies (marketers) acquired a data-collection capacity to rival that of the state. As a result, the data collectors of that earlier period necessarily faced some requirements for public accountability in how data was collected, categories adopted, and results applied.

Second, the forms of discrimination that resulted from data collection were also largely public—and visible as such—resulting, for example, in disputes discussed by historian Dan Bouk about insurance premiums with US commercial insurers. With relatively few variables in play and calculation tasks far less complex than those of today, the reason for someone’s higher insurance premiums (often their ethnicity, gender, or “statistical pasts”) was clear enough, and the correlation crude enough, to be publicly challenged. In the nineteenth-century United States, African Americans succeeded in lobbying for adjustments in insurance-industry practice.

Third, the search for statistical generality in no sense required an abandonment of the human individual. One of the most intense debates of nineteenth-century intellectual history was about whether statistical laws, operating at the level of the whole society, were compatible with individual free will. Some, like the French social theorist Adolphe Quetelet, insisted the answer was yes; others, like the British historian Henry Buckle, insisted the answer was no, with popular culture (Charles Dickens in his novel Hard Times) weighing in to mock statistics’ mechanistic view of humanity. Statistics’ significance for human self-understanding was debated in public. Perhaps this is unremarkable in itself, but it is sharply different from today’s development of Machine Learning tools for computing the social world, largely unobserved and entirely under the control of commercial corporations.

The invention of poverty, for example, as a social, not moral, providential, or economic, category—as something based in the unequal social distribution of resources and opportunities—shows how a social dimension to our common knowledge can make a political difference.

As Theodore Porter put it, “Quantification is a technology of distance.” Some distance is inherent to the abstraction involved in producing all quantified knowledge of the social. Such distance is more than a distance from one’s subjective opinion—practices of quantification also involve a social separation, the development of “forms of expertise and power relations” that allocate the power to quantify to some actors and not others. It is easy to forget “the social conditions in which science is possible,” as French sociologist Pierre Bourdieu once noted, but that is exactly what we must not do in the era of the God view.

Reversing the earlier pattern in which the corporate sector relied on the state’s statistics, social quantification today is principally a capacity of the commercial sector, with states the dependent users. Only in this era did it become imaginable that a government might, as the United Kingdom did in 2013, float the idea of abolishing its census and relying instead on privately acquired data.

In the predatafication era, the state “enjoyed a unique combination of legal and moral authority to carry out statistical inquiries,” but its exercise was restricted by the frequency with which it could approach its citizens for comprehensive data (for example, a ten-year census) and by the number of permanent categories it could impose (Social Security number, passport number, and so on). The commercial data collection of our times is continuous and on myriad dimensions.

For subjects who interact with it, the social quantification sector’s actions are extraordinarily complex, highly distributed, massively comprehensive, continuously updated, and, for these reasons, almost entirely opaque. Social quantification is contestable by individuals or groups only with the greatest difficulty. This is the challenge of “the black box society,” in which, as legal scholar Frank Pasquale puts it, “the spaces of our common life that are supposed to be the most open and transparent are being colonized by the logic of secrecy.”

The continuity with the nineteenthand twentieth-century European state’s attempt to make society more “legible” is misleading. Unlike the Charles Booth poverty maps of London, no one today can print and put on their wall even a simplified map of how Facebook categorizes the members of a small town or a how a credit rating agency differentiates the customers on its books. Social quantification today produces myriad microcategorizations, but these are privately controlled and largely inaccessible to public view. The social quantification sector exists not to make society generally more legible (that is, readable as and by society as a whole) but to give corporations special and private access to particular forms of knowledge from which economic value can be generated. It is these privileges that giant monopoly/monopsonies such as Facebook and Google can exploit.

There is a strange asymmetry here. Anita Gurumurthy, head of Indian civil society organization IT for Change, writes that “as the citizen becomes more and more legible to the state, the state becomes more and more opaque to the citizen,” although we might add corporations to the state in that observation. When the goal of data extraction is the capitalization of human life, we should not be surprised that the resulting knowledges have at best an ambiguous public status.

Opacity is also a practical consequence of operations. Data is collected, processed, and purchased so that particular actors can do particular things. The computing complexes available to corporate and state actors today are capable not just of automatically tracing the relations between large numbers of variables and categories but also of finding patterns in massively large data sets in which no order appears to exist. Underlying the much-hyped Big Data is a new model of knowledge called Machine Learning.

How does Machine Learning work, in broad terms? The calculative power involved in today’s social knowledge is massively greater than that employed in the nineteenth and twentieth centuries; it also operates on different principles. Whereas the earlier social knowledge was built from specific calculations performed on so-called structured data (for example, entries in statistical tables and databases), today’s social knowledge can be built from unstructured data, drawn directly from the traces left in the flow of everyday life. The structure of the data has to be found through repeating and gradually refining initial “rules” (guesses) for how to categorize those traces. New rules of categorization are generated in the course of applying those initial rules. As a result, the significance of the analysis is nonlinear, “emergent . . . , [and] consisting of previously unknown patterns and relationships.” The end product (the knowledge generated) is not explicable in terms of rules at all; it is just a complex correlation that emerges from a calculative process operating at many different levels. As such, the operations of data analytics are necessarily opaque, and no one—not even the engineers who run the process—can account exactly for how that knowledge was generated.

Some, such as Wired editor Chris Anderson, turn this opacity into a positive. Anderson argued controversially in 2008 that Big Data (the principle that says N must be all) undercuts the basic idea that historically underlay statistics as a science: that of formulating and carefully testing hypotheses to establish precisely stated correlations and, where possible, causal relations between variables. … Out with every theory of human behaviour, from linguistics to sociology. Forget taxonomy, ontology and psychology. The point is that they do it, and we can track and measure it with unprecedented fidelity. With enough data, the numbers speak for themselves.

Big Data practice is in fact considerably more complex than Chris Anderson's cartoon picture. N may well not be all, or anything like all, for many producers of social knowledge, not least because this data is competitively harvested. No one corporation has access to the all, and few, if any, are willing to give up their share of that all to build a collective data pool, although corporate data sets are sometimes shared on a limited basis. In the infinitely large domain of data extraction, each company has a very particular vantage point tied to the functionality and market position of its data-gathering platform.

If Big Data reasoning relies on the predictive power that comes from repetitive processing of unstructured data, this data can be generated, directly or indirectly, from whole populations. It no longer relies on the carefully constructed samples of the general population on which classical statistical knowledge relied. This new model of knowledge requires collecting data continuously, so that N might as well be all.

Potential inputs that are not machine readable become irrelevant, since the goal is always to increase N, the aggregate of what can be counted as data. Meanwhile, data subjects are generating material for machine reading all the time. We are constantly encouraged to act in ways that stimulate further counting, gain us more followers, and achieve better analytics. … This “Truth” is not the truth about society at which nineteenth-century statisticians were aiming—it is a branded, tethered, contract-specific, only conditionally sharable, personally targeted truth that is both the reward and price of continuous connection.

What is actually personalized in this new social knowledge is the fit to commercial targeting. The fit is intimate. As Bruce Schneier puts it, data colonialism today offers “a very intimate form of surveillance.” Tracking provides an “auto-fitting mould,” in Deleuze’s prescient phrase from three decades ago. … That world would provide marketing opportunities without limit: “Sensor revealing that you’re thirsty? Here’s a coupon for smart water.” Or as a marketing professional more bluntly put it, “Treat your customers like dogs.”

There is a deep paradox here. Although quantification is a technology of distance, today’s data relations increasingly strive to erase the distance between the social actor’s body and the precisely targeted sensor system that serves it (or more accurately, that the body serves). … But a real and important distance will remain: the distance comprised by the power of the social quantification sector against which individual actors currently have little or no agency. There is a divergence here from social categorizations as we have known them in the past. As Ian Hacking explained, the distinctive feature of social categorizations, at least until the era of Big Data, was that they were interactive—they operated as categories with which people could interact and negotiate as they were used on them. Humans can interact with today’s algorithmic categorization processes, if at all, only derivatively, through an indirect sense of what data might have already done to them.

Many of the accepted infrastructures for valuing people today are managed by platforms and are driven by those platforms' incitements to action and the types of data they do or do not value. Insofar as we care about our reputation (and few people don't), we are continuously oriented to what counts as reputation for platforms. … Such soft incitements are formalized in marketers' strategy of gamification. Gamification makes data extraction playful but far from formal—as with games such as Pokémon Go whose playing requires the revealing of location and other data. Gamification, as the president of loyalty marketer Points puts it, creates "a relationship of exchange where consumers are . . . incentivized to share more of themselves: their time, their attention, their information, and ultimately their loyalty."

Even though the new social knowledge is produced through operations that bypass human beings, it is actual human beings who are tethered to the discriminations that such knowledge generates. Based on algorithmic reasoning, it is a real person who gets sent to jail, refused health insurance, or offered a favorable price in the supermarket or an opportunity for social housing. Indeed, the gap between the uncertainty of cause and the certainty of personal impact is what makes algorithmic power so unnerving.

Almost every time we do something online, and sometimes when we are offline too, a small data packet for later use is grabbed via infrastructures of connection. Let's call those packets "caches." In computing, caches are, as Christian Sandvig notes, stores of content that are held so as to be available for later use; an important practical application of the term cache is the stored version of a video on a local server, available when a local customer wants to download it for streaming. The gathering together of data caches and their processing into more useful material generates a live-streamed vision of social reality for the social quantification sector's institutional players. Such social caching is data colonialism's extractive dynamic in action.

The goal is to cache the social at every available opportunity and then store the cached fragments for use in later targeted messages, offers, or ads. The official justification given by the marketing industry for this deepened personal tracking is not economic advantage but personalization. Because it is now standard to send marketing messages to people's personal devices, it is argued that those messages need to be personalized—that is, adapted to what available data suggests that person wants right then and there. But the logic is circular, and there is no obvious way out of the circle for the consumer. Increasingly, products (from expensive cars to regular consumer goods) go on transmitting information of marketing value about their usage and user after purchase. The goal for marketers becomes the "persistently connected product" that maximizes revenue per user throughout its life. Marketers describe this personalization as "one-to-one communication" and "one-to-one dialogue." Yet there is nothing dialogic about the individual's emerging relation to capitalism or the continuous caching of social life on which it relies.

Even if this banal trust in data-hungry devices seems harmless, there is the risk of function creep: the more we become used to tracking as the natural way to find out what to do in our personal lives, the more likely we will be to accept tracking as the infrastructure for knowledge in areas of our lives in which we have long expected less control, such as work.

Continuously caching the social enables corporations and all forms of authority to gain the sense that the social world, or at least selected slices of it, can be continuously streamed just for them—not for their entertainment but to enable that world's modulation or management in their interests.

Continuously caching the social enables corporations and all forms of authority to gain the sense that the social world, or at least selected slices of it, can be continuously streamed just for them—not for their entertainment but to enable that world's modulation or management in their interests. When the CEO of Shotspotter, makers of algorithmically processed surveillance mechanisms for the law enforcement sector, was asked to disclose his company's data to the US public, he responded that it would be like "taking someone else's Netflix subscription." So social caching is more than an academic metaphor: it is the practice that de facto—and, increasingly, de jure—authorizes the privileged viewpoint on human life that corporate power claims for itself in the era of data colonialism. Just as historical colonialism installed numerous extractive and other privileges for mining, trading, and other companies, so too data colonialism brings legibility privileges for the social quantification sector over the lives of data subjects.

We are not saying that all uses of social caching are bad or misguided. Some uses of social data, perhaps by civic or public institutions, may be responsible, modest, and oriented to organizational self-reflection. At the same time, the normalization of data extraction as the general means to monitor the world reinforces the "desire for numbers" (provided they are machine readable). New forms of professionalism and expertise around data capture are reshaping the world of work under data colonialism. Before the internet, however, as Bruce Schneier notes, data sources about social life were limited to company customer records, responses to direct marketing, credit bureau data, and government public records (and, we might add, insurance company data on their insured). Now, a varied universe of data collectors and services within a complex web of data processing functions extracts data from everyday life at a depth far exceeding that found in earlier forms of social knowledge.

The Shared Value group of insurers, which includes the South African company Discovery and its product VitalityDrive, claims to change "individual behaviour through incentives for fitness and healthy eating." Beneath the inspiring words is a clear-eyed focus on what Discovery calls "the behavioral nature of risk." Once insurers' risk becomes defined as principally about the insured's ongoing behavior (not occasional accidents), the relationship of insurer to insured becomes one that requires tracking. But do insurers' commercial interests really make them good life trainers? There are signs—even in the United States, which leads this development—that many now find this particular type of trade-off unacceptable. Unacceptable or not, data relations are already becoming for many a compulsory dimension of both labor and life.

Human inputs are only part of the territory that data colonialism seeks to annex to human capital. Machine-to-machine connections significantly deepen the new web of social knowledge production. Consider the fast-growing Internet of Things (IoT). The goal is clear: to install into every tool the capacity to continuously and autonomously collect and transmit data within privately controlled systems. Cisco predicts that by 2022, machine-to-machine connections will be 45 percent of all internet connections. Since machines can communicate with each other continuously, as human beings cannot, we can expect within a few years that the everyday world will have become a communication space in which human beings are minority participants and interpersonal communication a mere subset of what gets profitably tracked. The language of capture implies an unending series of territorial expansion: from the Internet of Things to "the internet of medical things," from "the internet of services" to "the internet of everything."

Executives in the social quantification sector, whether or not they planned to, are becoming social theorists, or at least trying to sound like them. “Thank you for believing in this community,” wrote Mark Zuckerberg in his full-page apology in British Sunday newspapers on March 25, 2018, in the wake of the Cambridge Analytica scandal. But behind the bland term community lay a lot of background theorizing. Certainly executives’ habit of speaking like social theorists did not start in the past decade. Just before the 2012 Facebook international public-share offering, Zuckerberg claimed that Facebook is “a fabric that can make any experience online social.” In 2010, he had already dubbed the 2007 invention of the social graph (Facebook’s way of mapping the network connections between users) “our way of explaining the phenomenon we thought was happening in the world.”

Can the human and social sciences correct for the limits of the social quantification sector’s emerging model of the world? Or might much of the social sciences already be complicit in this hollowing out of the social? This claim might shock those who see in “datafication . . . an essential enrichment in human comprehension” or “a great infrastructural project that rivals” the Enlightenment’s Encyclopédie. But let’s get beyond the hype and look at what is actually going on in the social sciences today.

To be fair, Pentland cares about data privacy and has developed an “Open Personal Data Store” that separates the sending of information for authentication purposes from the underlying private data that enables authentication, hopefully minimizing risks of reidentification. But what matters more are the silences built into Pentland's redefinition of social science. He says nothing about the close connections between ubiquitous data collection and the new direction of capitalism, and he says little about the familiar ways of understanding the social that this new “social physics” casts into permanent shadow.

Whereas critics quickly saw Nudge as a way of locking in market-oriented behavior, Thaler and Sunstein claim that the nudge is just a tilting of the “choice environment” that helps individuals reach the outcomes they would want anyway “as judged by themselves.” But how many times does an environment have to nudge you before it starts to govern what you want? Sunstein's claim that Nudge was always about enhancing individual “autonomy” carries little weight when the nudge is continuous, systematic, and opaque. The nudge is also a useful technique for data colonialism, generating at scale not the gentle subtlety Thaler and Sunstein intended but the force of what legal scholar Karen Yeung calls the “hypernudge”: dynamic and interactive behavior modulation enabled by continuous connectivity and continuous data flows.

It is paradoxical when Bruno Latour, one of the founders of the wonderfully illuminating perspective on human beings’ relations with technology called Actor Network Theory, writes that “purposeful action and intentionality may not be properties of objects, but they are also not properties of humans either. They are properties of institutions, apparatuses.” Whether this is plausible as an account of human purpose or intention, it is hardly helpful to those who want to defend the role of human agency in our understanding of the social world.

There is, however, one bright note: the rise of critical information science that seeks to systematically establish the distortions woven into social caching’s presentations of the world. There is now an emerging movement for algorithmic justice, which has raised awareness of many specific issues. The critical movement within data science concerned with “Fairness, Accountability and Transparency” has regular conferences, and a number of universities have focused programs for investigating how algorithms cover the social domain. More generally, an important intersection between critical information science, legal theory, and social theory is opening up the question of how the social qualification sector presents the social world for action by powerful institutions. US civil society has generated some effective campaigning. For example, ProPublica’s demonstration that it was straightforward to build implicit racial exclusions into housing advertisements led to Facebook adjusting its practice on monitoring ads.

But the playing field for this crucial work is hardly level: as we write, a legal case is going through the US courts in which researchers from the University of Michigan and Northeastern University are challenging the constitutionality of the US government’s interpretation of the Computer Fraud and Abuse Act that prima facie would block researchers from running fairness tests on how platforms present the world. The defense of earlier reference points to social knowledge and critical social research risks becoming engulfed in the disturbing trend in many countries toward political authoritarianism.

It is not enough to condemn today’s “advances” in social knowledge for not caring about human subjects. Critique must begin from the fact that what now passes for social knowledge is produced through social relations that have entirely different power implications from those that underlay earlier forms of social knowledge.

Today's social knowledge is produced principally through privately controlled data extraction whose goal is to discriminate between social actors for economic advantage. Algorithms help “fram[e] problems” and provide the classifications on which broader social differentiations are constructed. But “many classifications are now embedded in markets,” a notable example being credit rating systems. Data-driven social classifications have profound consequences for the fairness or otherwise of the social world. Indeed, various entrenched features of the new social knowledge incline it toward enhancing not justice but inequality.

Software is being used by police to predict reoffending and crime distribution in a number of countries (especially the United States and China, with experiments reported in Israel, Canada, the Nether lands, and the United Kingdom). In the United Kingdom, algorithms are being used by cash-strapped local governments to assess which children need protection from child abuse. Powerful new networks are being built to "connect devices, apps, and people for more than half of the [US's] major city police agencies," with Axon AI being a leading network provider and Palantir (the company of Peter Theil, a Facebook board member) being a leading provider of analytics software.

The larger implications for the management of law, crime, and policing are as yet unclear. This radically privatized production of social knowledge compares poorly with the historical processes for generating social knowledge discussed earlier in the chapter. There are growing concerns in the United States that, whatever else it is, algorithmic processing cannot be a substitute for human judges’ discretion exercised on behalf of the state. In February 2018, the European Commission for the Efficiency of Justice launched a review of the ethics of using algorithms in justice systems. Meanwhile, a very different vision is emerging in China, that of an “intelligent court” that will use “cloud computing, big data, artificial intelligence . . . to promote modernization of [the] trial system and judgement capability.” Contested or not, a new version of social governance is emerging from the daily operations of data colonialism.

It is not just that many instances of data discrimination occur far from the contexts in which the original data collection was authorized and its meaning shaped. Data’s very exchangeability depends on abstraction and “the stripping away of narrative” that “facilitates the . . . insertion of numbers in new locations and their adaptability in new contexts.” Once in circulation, classifications can generate not just first-order but “second-order corrections,” indeed, corrections in any higher domain without limit. Those “corrections” shape the life chances of human subjects. When a person gets categorized as a high-risk loan recipient in one context, it is likely to affect how they are categorized in other contexts.

There is first the problem, already noted, that decision makers may not understand their data sources. The increasing use of data streams with insufficient contextualization in government decision-making through so-called dashboards has provoked concern in the United Kingdom. The problem is deepened by the distinctive way that data generates knowledge through Machine Learning. Data processing generates proxies—regular correlations that stand in for other variables of social knowledge—out of which larger models of the world to be governed are built. The whole point of proxies is to black-box a more complex social domain while facilitating intervention. When an insurer assumes that driving after midnight equates to choosing to take more risk even if a person’s working conditions require it or when algorithms treat geographical areas as proxies for levels of crime risk, complexities on the ground become unreadable.

When patterns in current data are assumed to be reliable proxies for future events that may (or may not) happen, preventive action will be authorized whose necessity, by definition, never gets demonstrated. The counterfactual—what if no preventive action had been taken?—simply falls out of the picture.

There is a bigger danger too. If lack of accountability to context is becoming a principle of the new social knowledge, then this serves data colonialism (the mutual implication of human life and digital technology for capitalism) rather well.

If lack of accountability to context is becoming a principle of the new social knowledge, then this serves data colonialism (the mutual implication of human life and digital technology for capitalism) rather well. Abstraction means removing from view the textures of social life that might otherwise provide the reference point for social actors to hold social knowledge to account. Above all what risks being hollowed out by the new social knowledge is the space from which human actors could once imagine themselves with a license to speak truth to power. When the social world, in all its possible layers of action, thought, and expression, becomes merely an input to capital, the once resounding Foucauldian principle that for every power there is resistance starts to ring hollow.

Voice—the unmodulated, nonpredictive accounting of experience, once valued as part of social life—is by design excluded from Big Data analytics. People’s reflections on the world are unlikely to be part of the data that makes up their data doubles. The capacity for voice does not easily scale, and that is its point.

Let's return for a moment to Karl Polanyi's vivid account of the social unfolding of industrial capitalism. Data colonialism, it is clear, still awaits its countermovement: there is as yet no major change in social thinking or government policy to challenge its cruelties, as those of industrial capitalism were challenged in the nineteenth century. Recent calls for more openness about data practice offer only the very start of this change.

If no resistance is forthcoming, the capitalization of life is destined to reconstruct Hobbes's "state of nature"—a battle of all against all—via an algorithmic simulation that preserves, indeed stimulates, social appearances while regrounding them in capitalism's general drive to profitable extraction. The process will be heavily disguised. Popular narratives of "smartness" (for instance, the smart city) are an example: in the smart landscape, there are only connected individuals, natural and willing targets of commercial messaging and sources of commercially useful data, functioning within the larger cycle of capitalized life. Everyone else is by definition invisible.

Recall that the whole trajectory of data colonialism started out from the banal fact that to function, computers must capture data about their changes of state. It is a curious irony that B. J. Fogg, a leading developer of unconscious behavioral influence through online choice architectures, named his new science of intervention "captology." Knowledge by capture, influence by appropriation. When the cutting edge of social knowledge is a tool kit for achieving power at scale (or what Walmart more comfortingly calls "personalization at scale"), then the stakes of the social-knowledge game just got higher and its potential contribution to social justice more tenuous.

But ethics must start out from an understanding of the self. What if the boundaries of the self too are under sustained attack by data colonialism? If so, we may require philosophical resources to identify with precision the violence that data colonialism does to the self and, by extension, the social world as well. That is the topic of the next chapter.

Surely the most wretched unfreedom of all would be to lose the ability even to conceive of what it would be like to have the freedom we lack, and so dismiss even the aspiration to freedom, as something wicked and dangerous.

WHAT DOES IT MEAN to be human—to remain distinctively human—in the twenty-first century? A decade and a half ago, it still made sense for one of the United States's leading privacy advocates, Philip Agre, to assert that “your face is not a barcode.” But in May 2018, the black rap artist Donald Glover, a.k.a. Childish Gambino, released a video that went viral called “This is America.” As the song fades, he sings, “You just a barcode.” The chasm between these two sentences says much about the recent trajectory of poverty and race relations in the United States, already touched on in the last chapter. But it also opens an even larger story about the violence that data colonialism does everywhere to the core human values of freedom and autonomy.

A trucker from the same study went further and insisted that “any piece of electronics that is not directly hooked up to my body cannot tell me [when I am tired].” But other employees have been “hooked up” to devices that monitor at least their external movements for more than a decade (for example, store and distribution-center workers of the UK supermarket chain Tesco). If the second trucker's imagining comes true, it will not be his body telling him he is tired but a calibrated data system linked up to the bodies of a whole workforce. That system will monitor working bodies through their contracted activities and much of their off-work activity, using the latter as evidence for judging the former.

There is a long history of surveillance at work, disproportionately affecting low-paid jobs and disadvantaged populations, but the intimacy of monitoring in today's datafied workplace is without historical precedent. It fits well, however, with the colonial approach to social knowledge that chapter 4 uncovered in which individual reflection and voice matter little, and only aggregated data counts.

Hegel was unique among European Enlightenment philosophers in anticipating moments of breakdown when social formations based on certain values come to be organized around practical goals and institutions that directly undermine and oppose those values. The result is the deepest form of alienation.

The reason both traditional surveillance and datafied tracking conflict with notions of freedom derives from something common to both: their invasion of the basic space of the self on behalf of an external power.

We must immediately note that our standard image of surveillance—a person, generally representing the state, watching or listening in to the full stream of another person’s life—is unhelpful for various reasons. Today it is generally corporations, not the state, doing the tracking. The medium of corporate tracking is unlike a human sense (seeing or listening). It is, instead, the accumulation of data from multiple sources, what scholars have called “dataveillance.” At any one point in time, the surveiller is unlikely to have a full “picture” of the surveilled; only over time and through aggregation does a sufficiently detailed picture accumulate to motivate action. Finally, as noted in chapter 4, the target of surveillance is not the whole person but rather a montage of data doubles that probabilistically identify a real individual.

The argument is not that the data extracted from me is already my “property” or “possession” and as such must not be appropriated by another. For until the act of extraction there was no boundedness to the information relating to my life stream—as we have noted many times, data itself is not raw material. What previously was just part of the flow of my activity becomes separately identifiable and so potentially transferable only through the act of extraction. This transformation has nothing to do with whether the individual is already identified. Rather, since life (like land in Karl Polanyi’s historical account of industrial capitalism) is not the sort of thing that is naturally commodified, data requires for its creation institutional processes of extraction and demarcation, from which potential commodities can be generated.

The deepest problem is the violence that the very fact of data collection through surveillance does to the minimal integrity of the self. We derive this latter notion from the concept of autonomy.

The self’s minimal integrity is the boundedness that constitutes a self as a self. Often that boundedness is experienced in defending a minimal space of physical control around the body, but it can also be invaded without physical incursion—for example, by acts of power that intimidate, shame, harass, and monitor the self. By “the space of the self” we mean the materially grounded domain of possibility that the self has as its horizon of action and imagination. The space of the self can be understood as the open space in which any given individual experiences, reflects, and prepares to settle on her course of action. Hegel describes this space as both external and internal, like a circle that never-endingly turns in on itself. This, he says, is what makes “the free will . . . truly infinite, for it is not just a possibility, a predisposition; its external existence is its inwardness, its own self.” It is this continual interplay between internal and external, self and other, that enables the deeply social understanding of the self that, as a philosopher, Hegel advocated. From this understanding of the self, we salvage some ideas in this chapter as tools for challenging data colonialism.

Insight into this social concept of freedom (or autonomy) is not unique to Hegel, however, if we look to philosophical traditions outside Europe and North America. Argentinian-Mexican philosopher Enrique Dussel's philosophy of liberation has as its goal the need to articulate a concept of human freedom from outside the West's power centers. Freedom, for Dussel, is the substance of a person with "all its uniqueness, its proper indetermination, its essence of bearing a history, a culture; it is a being that freely and responsibly determines itself" and so is always situated "beyond" what Dussel calls "the horizon of totality." In insisting on a space of the self beyond any horizon of totality, Dussel goes one step further than Hegel did, explicitly defending the space of the self against any attempt of power to own and absorb it.

It has been fashionable to forget this and to feign indifference. Some question the "special" moral status of human beings and even compare them to ants and bees, whose collective life has certainly fascinated humans for millennia. Insect metaphors have long been adopted into the technical language of computing; we have swarm robotics, registry hives, and the like. Perhaps there is some force to such analogies when infrastructures of connection enable the aggregation of human-data traces on a vast scale. But there is a danger in such metaphors. No one has yet shown that human beings are collective animals in a zoological sense; pretending they are may divert attention from the human costs of capitalism's new social order.

Philosopher Luciano Floridi goes further, renaming the very stuff of the self in terms of information. To him, we are not exactly humans anymore but hybrid creatures that Floridi calls "inforgs." … Inforgs are the perfect creatures for the hypernudge to rule when they have been refashioned to always be open to data flows and so continuously available for modulation. But by what? Presumably by some form of intelligence? We could bet everything on artificial intelligence, but few have taken this step. Perhaps modulation by human intelligence exercised at the collective level? US West Coast evangelist Kevin Kelly comes close to this when he celebrates the "technium" (his term for the supposed collective "autonomy" that develops from "the feedback loops in the technological system"). But why believe humans are intelligent at a collective level while ignoring or recklessly overriding humans' ability to be independently intelligent at the individual level?

As we noted in chapter 4, the poor are likely to be particularly disadvantaged. Digital platforms, apps, and sensors provide infrastructures for seamlessly installing data-derived performance monitoring into all aspects of work, but they are being installed most energetically in lower-status and low-wage work. Employees such as truck drivers or warehouse workers already have continuous surveillance installed as a basic feature of their lives. There may soon be no workers left in those sectors who remember what it was for their bodies and their every action to be unmonitored. In higher-status jobs, the intensity and continuity of surveillance may be limited to aspects of the recruitment process in which a candidate’s social media history is now pored over. In today’s “fissured workplace,” those in powerful positions will seek privileged control over who is surveilled (and how much) and who is not surveilled. Social power may come to mean, in part, being a net controller of the benefits and costs of surveillance, just as under industrial capitalism management power meant controlling the flow of knowledge and information that made up work.

People are also differently placed to repair the costs inflicted through surveillance systems. Imagine the costs when every object, most processes, and much of one’s work and other life opportunities are managed through automated monitoring. You may need to buy a new tool when its built-in tracking system breaks down; you may need to find alternative sources for education, health care, or insurance when denied them on algorithmic grounds; or you may need to correct inaccurate data held about you in a system—for example, a credit rating. People’s ability to bear those costs will depend, as ever, on a preexisting (likely unequal) distribution of resources. And the more opaque the processes of data sorting are, the more resources will be needed to contest them effectively.

We will certainly not be claiming that all forms of damage to the space of human autonomy are the same. There is a world of difference between slavery's direct and violent ownership of bodies and minds and the more indirect forms of surveillance and dataveillance that characterize contemporary data relations. Still, they have something crucial in common. We seek to identify what that is, building on excellent work done by many authors in sociology, communications, and legal theory, including Mark Andrejevic, Julie Cohen, Oscar Gandy, Karen Levy, and David Lyon.

Our argument, however, tries to go deeper. The minimal self-integrity under threat is not the sort of thing that human beings can trade without losing touch with what constitutes them as selves rather than as inputs to a collective intelligence or information-processing system. Two senses of self are in play here. Capitalism affirms, for sure, the uniquely identifiable reference point on which all notions of the self hang (it is, after all, this bodily subject whose market potential gets traded in proxy form). But at the same time, capitalism invites us to lose control over a core element in the content of that identified self whose continuity (and change) we value with time. By installing automated surveillance into the space of the self, we risk losing the very thing—the open-ended space in which we continuously monitor and transform ourselves over time—that constitutes us as selves at all.

Hegel saw freedom as "possible . . . only if one is also already in a certain (ultimately institutional, norm-governed) relation to others." In other words, freedom is actualized through our social relations, which means that the loss of freedom is actualized through distortions in social relations. This argument, for the first time, opened up philosophical thinking to sociological evidence and, at its most radical, acknowledged that at particular historical moments, institutional forces may come to undermine the inherited values and norms that govern social relations. When this happens, people start to feel themselves "part of a practice that has either gone dead . . . or requires of [them] further commitments incompatible with others necessary within some form of life." Put bluntly, when our form of life starts to conflict with our underlying values, "the lives in a form of life become uninhabitable."

But equally important is corporate power. There is no reason to assume that a corporation, unless specifically checked from doing so, won't use the personal data it collects or buys in its own interests and against an individual's interests. Many forms of personal data are the source of validly competing interests: when individuals insure against risks to themselves, the interests of insurers and insured are often, quite legitimately, opposed.

Automatic data collection poses a danger to autonomy at a second, deeper level, not because its actual uses are always harmful (indeed, some might be beneficial) but because the possibility of harmful use distorts the space in which individuals live and act. This is what some writers call the "chilling" effect, which is the impact that the power to collect and hold data on a large scale has on individuals' ability to think of themselves as free actors. Political theorist Quentin Skinner's comment on the Snowden revelations is decisive: "Not merely by the fact that someone is reading my emails but also by the fact that someone has the power to do so should they choose leaves us at the mercy of arbitrary power. . . . What is offensive to liberty is the very existence of such arbitrary power."

This takes us to a third and more fundamental disruption to autonomy. What if capitalist data relations trouble the very basis of autonomy, the minimal boundedness that underlies the very possibility of being a self? What if datafication, not just potentially but in principle and by design, interferes with the self’s space of movement that is its “own”?

Two characters, a keen young recruit (Mae) and her more experienced mentor (Annie), each want to live a life that harmonizes with the surveillance practices on which the Circle’s business model depends. They want to but cannot—at least not without a sanity-threatening struggle. At risk is not just the ability of these characters to direct their lives but their basic sense of themselves as selves: the minimal integrity without which they fail to be selves at all.

The employee says, “My problem with paper is that all communication dies with it. It holds no possibility of continuity [. . .]. It ends with you. Like you’re the only one who matters [. . .] but if you’d been using a tool that would help confirming the identity of whatever birds you saw, then anyone can benefit. [. . .] Knowledge is lost every day through this sort of shortsightedness.” The problem was that Mae chose to reflect her experience “only” back to herself rather than relay it to the Circle’s data pool. Her choice of recording technology (paper versus smartphone or tablet) and the refusal to connect to wider data systems that this choice implied, is interpreted back to her in moral terms when Mae is told that “knowledge is lost every day through this sort of shortsightedness.” The senior Circle employee who criticizes Mae believes that there is no space in which an individual is free to reflect on herself that is not already a space over which her corporation, acting in the name of social knowledge, has right of access.

This is the space, as Rössler puts it, in which “processes of self-description, self-definition, self-discovery, or indeed self-invention” take place, all of them “dependent upon a person bringing herself face-to-face with herself in conditions in which she can really be sure that she is protected from the eyes of anyone else.” The liberal philosopher Charles Fried had anticipated the problem when in 1970, with what would now be a primitive surveillance technology in mind, he imagined a parolee kept under constant and total monitoring in case he commits another crime; is this, he asks, consistent with any notion of freedom? Fried argues no, because such an arrangement overrides the basic element of interpersonal trust fundamental to human life: trust that respects the independence of the other person, as an other person, to make his own errors. If trust, he argues, were just a matter of risk management, we would probably delegate it to “cheap and efficient surveillance,” but “man can’t know he is trusted unless he has a right to act without surveillance” and so prove he is worthy of trust. … This is exactly the sort of trust-less social order toward which data colonialism is taking us.

Capitalist data relations redraw the boundaries around the subject’s space of action and reflection, dismantling the effective limits on which depends the very possibility of a self that can trust itself and others. When we strip down autonomy to this minimal space whose integrity is necessary for the self to be a self at all, we touch on the roots of the many and varied forms that privacy takes across the world and that human beings have defended throughout history. This is the philosophical and practical core of contemporary debates about privacy. Privacy, according to surveillance expert David Lyon, is “a pivotal concept that helps to throw light on what is wrong with mass surveillance.” Privacy is a matter of the self’s minimal integrity whose necessity persists beneath all cultural constructions.

Working under the influence of the European ideology of possessive individualism, Hegel calls this space of freedom a "property," a property that is inalienable and nontradable. But we can challenge data colonialism in a way that does not depend on possessive individualism by reformulating Hegel and defending the minimal space without whose boundedness the self would not be a distinct self at all. This is the space from which the self listens out and experiences the world. We fear the moment when that space is ruptured; George Orwell's novel 1984 provides the most famous example of this in literature. Hegel himself discusses slavery as an external capture of this space and insists on "the slave's absolute right to free himself." Continuous surveillance of the self from the outside is not as absolute and violent as slavery—it does not amount to the direct transfer of all control of the self's space of freedom—but it unquestionably violates that space.

Evangelists of technological connection such as Kevin Kelly see things differently. In his recent book The Inevitable, he names tracking as one of the twelve dimensions of technology's transformation of society. By attributing this transformation to “what technology wants” (the name of his previous book), Kelly seeks to quarantine today's profound transformations from ethical critique. His goal is to help us “work with [technologies'] nature, rather than struggle against it.” The entire edifice of data colonialism and capitalist data relations is obscured by this rhetorical move.

Since it is the abstracted, commodified form of data relations that binds us into capitalism, it hardly matters that self-tracking relations are often voluntary.

What are the larger transformations at work here? Consumer activity has rarely been exempt from some form of observation, even in the traditional open-air market. But the decisive shift came in the 2010s with the convergent growth of social media platforms, connected mobile devices, and ever more sophisticated “smart environments” that made possible something close to “autonomous” surveillance.

As Nicholas John points out, what is most striking about sharing as a metaphor is the extendibility of what is to be shared: from thoughts and pictures to information sources and life histories to everything and anything. Sharing our lives has come to imply a broader value of “openness and mutuality” but at the risk of mystifying the communities we supposedly form online.

Microsoft CEO Satya Nadella comments on the digital personal assistant, Cortana: “It will know your context, your family, your work. It will also know your world. It will be unbounded.” The resulting forms of self-reflection and community exchange are, as Taina Bucher puts it, “force-relations,” relations with external entities and processes that intimately expose us to the economic forces of capitalism.

“Know thyself” is one of the oldest ethical maxims, but it has been used to market self-tracking devices ever since the days of the humble bathroom scale. As life becomes capitalized, “know thyself” morphs from philosophical principle to profit center. No doubt specific knowledge can be gained from some forms of self-tracking, and there is much to be gained sociologically in better understanding how individuals use such technologies. But what in the long run constitutes valuable self-knowledge here—valuable for whom? The goods generated by capitalist data relations are profoundly ambiguous. They offer us, apparently, the chance to be more autonomous—in the sense of regulating ourselves more intensely—but at the cost of submitting to an external data system whose functioning requires the integration of its devices permanently into the space where, until now, we imagined we were truly, and only, ourselves.

Even when we choose to install apps such as Foursquare and thus share data about where we are, the background data aggregation that goes on through Foursquare’s API is largely outside end-user control.

Even when we choose to install apps such as Foursquare and thus share data about where we are, the background data aggregation that goes on through Foursquare’s API is largely outside end-user control. Foursquare, at least until the introduction of the European Union’s General Data Protection Regulation, had tended to treat information, for example, about the web pages accessed through the app and users’ locations “as non-personal information” except when required otherwise by law; Google’s location API operates, however, without any need for user participation or knowledge. Yet such data acquisition is offered as an enhancement of “experience,” echoing the personalization doublespeak that retail marketers use to justify their intrusion into consumers’ lives.

We might counter by asking defenders of datafication to deny that some measure of control over data relations is important for autonomy, and indeed the Quantified Self movement has been ready to cede individual property rights in data.

For Hegel, it is the reflective relation to oneself that is essential to the development of a free will. Therefore, a free life needs to be a self-sufficient life in which “nothing from outside, nothing not-me, determines my actions.” Hegel does not deny—indeed, he emphasizes—that the individual lives a life that is not free from external constraints; it is a life lived through mediations (lived “in the other,” as Hegel puts it). But at the core of freedom for Hegel (the core of autonomy, in our terms) is the notion of an inner life that remains under one’s own reflexive control. This inner life is what enjoying one’s “right of subjectivity” means. The very thing Quantified Selfers and the social quantification sector see as a benefit (that our moment-to-moment existence is understood better by external data processing systems than by us) collapses the space of subjectivity in which the self enjoys its freedom.

But contemporary data relations involve more than just relations; they involve systems that insinuate themselves within “the self’s needs, desires and other choices.” This is what Hegel in his discussion of free will called the problem of arbitrariness. The data infrastructure on which the self-tracker relies to augment herself contains elements and dynamics that are troubling because they are arbitrary, driven by goals that are not hers, and operate in ways that remain largely opaque and so can never be reflexively integrated into her goals.

An unease with such an unacceptable deal may explain why many self-trackers give up after a while. Some ex-self-trackers report losing their “ability to self-regulate.” Other testimony suggests unease that “getting good data” had become the overriding driver of submission: “From a mental health point of view, one of the things I realized about myself was that I was only really happy when numbers were trending in the right direction.” In Hegel’s term, this person felt he was not really “with himself” in the data, only emotionally tied to the stimuli that the system generated. A self-tracker can find herself genuinely puzzled about whether, as one put it, self-tracking got her “closer in contact or further away from [herself] and the world.”

The risk is clear, however: that since we have no choice but to go on acting in a world that undermines the self's autonomy, one consequence is that we may progressively unlearn the norms associated with it. As the second quotation at the beginning of the chapter head suggests, data colonialism's subjects may come to unlearn freedom in time.

There are issues here for sure. What does it mean to obtain the consent of a whole population (in Iceland, the data deal was done in secret, without consulting Iceland's citizens explicitly, which proved highly controversial)? Barbara Evans, a leading scholar of health and data law, argues that we need elaborate processes for collective management of data gathered in this way. The risks increase when we realize that frequent claims that medical data is securely anonymized may be unreliable. Under any political system, the security of health data must depend on how far its collection can be insulated from market dynamics (since values such as patient confidentiality are not, after all, market values). Health data is, in any case, quite specific and highly contextual. A proposal for everyone to submit to continuous collection of all their health data would be very different in its implications for the self's minimal integrity than would be specific individuals sharing specific health data for a specific beneficial purpose. Legal scholars fear that it is exactly such large-scale aggregations of health data that risk escaping the legal protections that have long been integral to medical ethics and law.

So-called adaptive learning is today being offered by commercial providers as a “personalized” advance on the education of the predatafication era. Older, pre–Big Data education is depicted now as a “factory-model” under which all children had to receive the same content in the same manner (IBM Education), impeding “a deliberate and continuous approach to the improvement of learning and teaching” (Pearson).

In the education sector, by contrast, the principle of confidentiality is less heavily ingrained, and the logic of automated data collection operates with fewer initial constraints. Few roadblocks exist to stop intensified surveillance of students through datafication, and education has always involved monitoring in some form.

The links between continuous surveillance and the education of children are so deeply naturalized here that the panoptic notion of teachers using continuous real-time monitoring is presented as neither chilling nor threatening. On the contrary, it is a selling point, giving teachers “a full bird’s eye view of the entire classroom” (Impero Education), which enables them to “immediately intervene in a highly personalized way” (Blackboard).

The idea [of digital monitoring] is to allow students the online freedom they need to grow, learn and survive in a digital world, with the safety net of keyword monitoring to protect against the risks.

Abandoning the concept of privacy here would be a fundamental mistake, since it still retains some reference to the minimal integrity of the self. Privacy, as leading US privacy campaigner Marc Rotenberg argues, remains "an extraordinarily powerful and comprehensive human-rights claim." But what would effective protection of privacy look like within a broader challenge to data colonialism?

There is a growing global debate about whether the existing frameworks for regulating data processes need strengthening and about how this might be done. Some see little hope for any practical intervention that could halt the expansion of market forces, but such pessimism can be understood as the symptom of neoliberal culture's success in enforcing the dominance of market reference points, particularly in the United States and United Kingdom. The increasingly popular business argument that data collection is so pervasive that it should be exempted from legal regulation, and regulation should be focused entirely on use (an argument Helen Nissenbaum calls "Big Data Exceptionalism"), reflects that US market context, though it also echoes much global business rhetoric—for example, from the World Economic Forum.

Legal scholarship has for more than a century debated the need to protect the physical, communicative, and virtual space of the self. Samuel Warren and Louis Brandeis in the late nineteenth century proposed a "right to be let alone" in response to the incursions of the yellow press and early paparazzi. This "negative" form of liberty was extended in the twentieth century to connect with "positive" forms of liberty such as informational privacy and decisional privacy—respectively, the right to control over one's personal information and over the process of one's decision-making. Capitalism's broader drive to extract value from personal data is clearly in tension with such inherited notions of privacy and the philosophical values that underpin them.

These rules insisted that customers "must be empowered to decide" how providers may "use and share their data." But they were repealed by Congress in 2017, and the possibility of reviving them is uncertain.

The United States lacks a comprehensive legal regime for regulating data ownership or, indeed, privacy more generally, with the US Constitution throwing up conflicting signals. The Fourth Amendment may impose limits on the ability of corporations and states to invade the space of the individual, if not on single occasions, then, according to the so-called Mosaic theory, through a cumulative series of actions. According to many commentators, however, its usefulness remains uncertain and highly contingent. Meanwhile, the famous First Amendment, which guarantees freedom of speech (compare article 10 of the European Convention on Human Rights), has recently been interpreted by the US courts to protect data-gathering and data use themselves as forms of corporate free speech.

Major legal innovation is needed if we are even to begin to address the consequences of data colonialism, and here the legal traditions of the United States and Europe are differently placed.

The GDPR requires data collectors to give an ordinary-language account to every user of what data they hold and for what purpose. This approach assumes that lack of such information damages the underlying right of each person to control her information boundaries, hence the need for an immediate alert when any activity potentially breaches those boundaries. The GDPR's practical requirements interrupt the supposed "naturalness" of data collection from individuals by requiring individuals to be informed of what is going on with their data.

The GDPR changes the rules under which data colonialism operates but leaves unchallenged the commercial purposes for which data is collected. There is also a question about the limits of the “informed consent” principle on which the GDPR relies; this principle is dependent on how situations of consent are configured in everyday life. It is hard to see how this, or any similar general framework that we might imagine, would challenge an environment in which consent to data collection and processing is overridden by the bargaining power of employers or insurers. Meanwhile, it is super-powerful incumbents such as Google that are arguably best placed to bear the costs of adjusting their business to GDPR rules.

We should not forget, however, that the law was a site of extensive conflict over how industrial capitalism’s early excesses could be corrected (this conflict occurred as part of the double movement that Polanyi analyzed, discussed in chapter 4). And there is no doubt that geopolitical battles over competing regimes of privacy and regulation of data flow—for example, between the European Union, the United States, and China—could be a major factor in how data colonialism develops on a global scale over the next decade.

Perhaps it was naive to expect the legal institutions of market societies to be the site at which general resistance to capitalism’s expansion could take practical form. In China, where data platforms have emerged in close alliance with the state, there are signs of the state intervening to insist on some privacy standards, but at the same time, it is well-known that platforms consistently give the Chinese state access to their data.

Today's environment of data collection and processing far exceeds in complexity, continuity, and depth the single violations that were once understood to threaten a subject's privacy. If data colonialism works through an "ecology" of data collection, as the social quantification sector consistently tells us it does, then any response must be environmental too. Driven by private corporations much more than by states, today's "autonomic smart environments," as Dutch legal theorist Mireille Hildebrandt calls them, are saturated with automated agents that have the power to shape the contexts in which human agents can make choices. So pervasive is this shaping and so unpredictable to human beings are its outcomes that the social environment stops being formed through mutual expectations between people and becomes driven exclusively by the goals of datadriven manipulation. This unsettles the individual freedom of choice that is core to all liberal notions of privacy.

A better alternative, proposed by various legal scholars, is to move beyond traditional liberal framing of privacy toward a relational understanding of privacy. This relational approach sees privacy as a feature of the mutuality of social relations, a condition that must be collectively sustained if selves and others are to flourish at all. This approach flows directly from the socially grounded approach to the minimal integrity of the self that we found in Hegel and Dussel. This approach also emphasizes not just control over one's process of making choices but also second-order control: the ability to manage to some degree one's choices about choices, that is, one's values. This is precisely the level of freedom that is damaged, as Hegel predicted, when a social order starts to be organized around practical principles that no longer fit with the values its members thought were theirs.

All values, such as privacy, are socially negotiated. But there is something distinctively complex about privacy and, specifically, the importance of privacy to autonomy (understood as the capacity to "find one's own good in one's own way"). We cannot find that path if we lose the "breathing-room" in which to develop as our selves. In what entity or process exactly are these new, more social notions of privacy or autonomy themselves grounded? Some suggest that "the self has no autonomous, precultural core" to be defended, but if values such as privacy and autonomy are entirely relational, containing in themselves nothing essential, then what entity exactly is it "whose" autonomy is to be protected? Today's new practices and norms of data collection and processing are themselves surely social and cultural contexts too, but that does not mean that something fundamental for human beings is not at stake in the damage data relations do to privacy.

There remains, however, a deep paradox that both Julie Cohen and Mireille Hildebrandt note: that practices of data colonialism are potentially eroding the very notion of the rational, reflexive, law-respecting human subject on which the legitimacy of law itself depends. Meanwhile, courts in various countries have become early adopters of algorithmic processing, and much of the power to regulate infringements of autonomy is now delegated to commercial platforms, with dangerous practical and normative implications. In the age of datafication, legal institutions need to defend some version of the principle of autonomy for their own survival.

Autonomy in this basic sense is the core value underlying any human social order that has the potential to be good. Yet, when all the complex detail of data practices has settled, it is this value that is being threatened today, not by technology itself but by the mutual implication of life and technology for capitalism that we call data colonialism.

Moving decisively away from individualistic notions of autonomy, freedom, and privacy, we can argue that the datafied social governance and incursions into the self’s minimal integrity discussed in part II of the book together constitute an environment that is toxic for human life. The way forward lies less in the detail of legal regulation and more in defending an ecology of human life that connects with the visions of those who have resisted colonialism and capitalism for centuries.

Untouched also would be the basic fact stressed throughout this book that capitalism's data-driven order is being built globally, not just in the West. It is an order that has behind it the full force of the world's two most powerful states (the United States and China), who are competing for leadership of the social quantification sector and its AI hinterland.

It is this larger narrative of transforming life through data that we are told is impossible to halt, because it is driven by a "technological momentum" that is "inevitable." The notion that datafication is inevitable is, as we noted in chapter 1, a myth of data colonialism.

Colonialism is about appropriation; whereas historical colonialism appropriated land, resources, and bodies, today's new colonialism appropriates human life through extracting value from data. We uncovered the basic features of data colonialism: the processing of ever more personal data, the universalization of logistics as a mode of management, the datafication of most aspects of labor, and the creation of data relations that draw us into arrangements whereby data is gathered and value extracted from it, regardless of whether we are at work or somewhere else. Platforms play a key role in making our participation in data relations seem natural, but pressures at work and elsewhere also play a part. We argued that underlying these was something even more fundamental: the drive to capitalize human life itself in all its aspects and build through this a new social and economic order that installs capitalist management as the privileged mode for governing every aspect of life. Put another way, and updating Marx for the Big Data age, human life becomes a direct factor in capitalist production. This annexation of human capital is what links data colonialism to the further expansion of capitalism. This is the fundamental cost of connection, and it is a cost being paid all over the world, in societies in which connection is increasingly imposed as the basis for participating in everyday life. The resulting order has important similarities whether we are discussing the United States, China, Europe, or Latin America. The drive toward this order is sustained by key ideologies, including the ideologies of connection, datafication, personalization, and dataism. The result is not a new type of capitalism but certainly a new means by which and a new scale on which capitalism is operating; this result transcends earlier diagnoses of neoliberalism or the social factory and may, in the long run, provide the basis for an entirely new mode of production.

The result is a rearrangement of life so that it is configured directly under the "hand" of capitalist management and its infrastructures of connection; the goal is, literally, to annex life to capitalism. It is the absoluteness of this ambition to order a world made tractable to capitalism's operations that gives data colonialism its simplicity: the mutual implication of human life and digital technology for capitalism. A capitalism without limitation, obstruction, or remainder. A capitalism that encompasses every scale of human life (across space) at every layer (in depth) of individual and collective experience, all unfolding without limitation or interruption (in time).

Because we are only at the start of this new colonialism, we cannot be sure what this mode of production will eventually look like. We can, however, borrow from postcolonial and decolonial thinking tools to articulate not just counterhistories but counterpresents, or alternative understandings of contemporary social orders.

Instead of the appropriation of cheap natural resources and cheap labor, the new colonialism is premised on the “natural” availability of “cheap” social data. As corporations engage in new forms of exploration, expansion, exploitation, and extermination, a new capitalist mode of production emerges (just as industrial capitalism emerged during the colonial era to replace earlier tributary modes of production).

Meanwhile, much of the social sciences has aligned itself with this devaluation of social knowledge and so provides a poor site from which to resist data colonialism.

A new “social knowledge” emerges that lies entirely under corporate control. We call this “social caching.” This “God view” of the social world, though parceled out among many corporations, coalesces in a corporate social imaginary that justifies new ways of governing everyday life for the benefit and convenience of corporate power (think of insurers’ expanded power to demand self-tracking as a condition of service).

New forms of injustice inevitably result: even accidental discriminations remain opaque to subjects and system operators alike and so are likely to go uncorrected. But datafication also provides the perfect cover for older forms of inequality and discrimination to be reproduced and amplified.

Whereas specific human subjects face danger from the new social hierarchies that are being built through managing extracted data, all human subjects face a threat to the minimal integrity of the self from data colonialism's reliance on ever-expanding mechanisms of surveillance and tracking. Lower-status work brings higher exposure to surveillance, with zero discretion, while higher-status work enjoys more protection from surveillance and greater discretion.

To sum up, the problem and challenge of contemporary data practices is neither data nor simply the particular platforms that have emerged to exploit data. The problem is the interlocking combination of six forces, unpacked in the previous chapters: an infrastructure for data extraction (technological, still expanding); an order (social, still emerging) that binds humans into that infrastructure; a system (economic) built on that infrastructure and order; a model of governance (social) that benefits from that infrastructure, order, and system and works to bind humans ever further into them; a rationality (practical) that makes sense of each of the other levels; and finally, a new model of knowledge that redefines the world as one in which these forces together encompass all there is to be known of human life. Data, in short, is the new means to remake the world in capital's image.

But confronting data colonialism as a whole is seriously inconvenient. That is because data colonialism's model for organizing things underlies countless business models and everyday resources. That is why, as Mark Andreessen inadvertently implied (in the chapter's opening quote), the point of opposing colonialism was never immediate economic success. Worse, over the past two decades, billions of people have started to organize much of their personal lives around the infrastructures of digital platforms and other services that depend, or seem to depend, on seamless flows of data.

The problem of inconvenience drives the temptation toward partial solutions—for example, the idea that we just need better networks, perhaps networks with an element of public purpose or, indeed, that we should follow big tech’s own proposals for how to use their products in ways that enhance “digital wellness.” “Better” data-driven networks will not save us, however. For by attempting to reform a particular network within a wider system of platforms, we are not challenging the foundations of the system but merely finding alternative ways to replicate it. There is a danger here of repeating the error of those early nineteenth-century political economists, mocked by Polanyi for their “mystical readiness to accept the social consequences of economic improvement, whatever they might be.”

But how do we make any progress when our target is precisely the wider economic, social, and political order built through our use of those same systems?

We do not deny that media literacy may have some benefits in the immediate term, but we should not pretend that media literacy is a means to resist data colonialism; at most, it enables us to live more at ease with it, thus normalizing it for the longer term. Worse, like all notions of literacy, media literacy relies on the virtuous "disposition" of the subject, which misses how the new order works to dismantle the autonomy of the subject. In any case, media literacy soon comes up against some practical limits. When US sociologist Janet Vertesi applied her media literacy by leaving no data traces during her pregnancy, she found basic economic activity increasingly difficult, until, finally, her husband's attempt to buy cash vouchers to pay for a baby stroller on Amazon was arbitrarily blocked by a store algorithm.

The European legislation on privacy (the GDPR) represents a serious practical challenge to the details of data colonialism's existing business models. As a result, parts of those models may get reconfigured and rerouted, and the principle of obtaining consent for personal data use is at least now being embedded as a norm of daily practice, even if opportunities to refuse consent to powerful platforms of convenience often remain highly constrained. But even a broken-up and redistributed order of data power is still an order. The problem, as always, is bigger: how to live differently, knowing what we know and having seen the offer of a datafied order, with its benefits and costs.

A Muslim cleric in Indonesia declared selfies haram, saying, "If we take a selfie and upload it on social media, desperately hoping for views, likes, comments or whatever—we've fallen into the OSTENTATIOUS trap." But the fatwa backfired: thousands of reaction selfies promptly posted under the hashtag #selfie4siauw (the name of the cleric). One day, however, another fatwa against a social media company may come along, and people will take it as a battle cry; maybe whole communities will start to feel that particular platforms are no longer for them.

This and similar endeavors are being explored by artists, scholars, and activists from within emerging fields such as postcolonial digital humanities and indigenous digital studies. Such work can provide a window onto a different space, what First Nations writer and academic Leanne Betasamosake Simpson calls a "productive place of refusal." For Simpson, a place of refusal is productive if it generates a "grounded normativity," a place from where we feel entitled and empowered in a sustainable way to say no to what we are told is "necessary" and yes to what for now seems impossible.

Our point is certainly not to negate the value of such experiments in resistance to data colonialism. But if we already know that as partial actions they cannot be sufficient, we must set our sights higher; we must develop a wider vision of resistance, a vision for connecting with one another on different terms that might provide possibilities for solidarity with resistant data subjects, wherever they are. Who knows, after all, where data colonialism's great wall will break in the end?

What is at stake when any of us try to refuse data colonialism? What would it mean to articulate a vision of how individuals, communities, and nations can know themselves other than through the tools of data colonialism? A vision that rejects a rationality that claims to bind all subjects into a universal grid of monitoring and categorization and instead views information and data as a resource whose value can be sustained only if locally negotiated, managed, and controlled. We might call this a socialism of data because of the primacy it gives to social choices over data, but this is not a vision for a global socialist data-state; we must certainly reject the vision of an integrative social order exemplified by the data-driven market socialism of China today.

Capitalism has always sought to reject limits to its expansion, such as national boundaries. But now, as not only human geography and physical nature but also human experience are being annexed to capital, we reach the first period in history when soon there will be no domains of life left that remain unannexed by capital.

And yet this seemingly inevitable order has a fundamental flaw from the perspective of human subjects. That flaw is its incompatibility with a different kind of necessity, the most basic component of freedom, which we call the minimal integrity of the self. The emerging colonialist and capitalist order is therefore, under whatever disguise it operates, incompatible with every political structure built on freedom, including democracy in all its forms, liberal or otherwise.

Vision and imagination become the most practical starting points when we are faced with a social transformation so large, so ordered, and so multilayered. We need a vision that does not turn its back on the future but asserts the possibility of a different future, a vision that does not subordinate human autonomy to system autonomy. We will suggest here some practical steps toward this reimagining of our relations to data and data colonialism, a potential new social imaginary about data and our relations to data.

How does this vision start? It begins by rejecting in principle the premises of the new social and economic order and, for example, insisting at every opportunity that corporations—indeed all actors—who use data can do so legitimately only if that broad usage is based on respect for the human subjects to whom that data refers and for the goals and awareness of those data subjects. For it is not just individuals' right to control their data that is at stake if we don't reject data colonialism's premises, but also the freedom of social subjects to exist without being tracked, indeed, the right for everyone to live in a world that still holds the possibility of freedom for human subjects in general.

The order of data relations relies on the unlimited possibilities of data processing generated by the infinite connectivity of the contemporary world. Data relations rely on removing all limits to data appropriation and thus building an expanding, knowledge-based social and economic order. From simple starting points (computer's data capture, the connectability of computers, and the information processing and monitoring that together they enable), institutions and systems are acquiring the capacity to govern life in a completely new way.

What sorts of data should be collected? What sorts of data should be combined with what other sorts of data? When do we need to impose a limit on algorithmic decision-making? Our answers to these questions will shape the social worlds in which we live over the coming decades. We can go even further: because order is the basis of all forms of life, the social and economic order being built through data colonialism raises questions about the future of human life itself.

Data colonialism proposes a connected world that appears to know itself through its absolute connectability and through unfettered data flows between all its points (whether living or not). Indeed, dataism claims to find a higher force than human life, the force of information processing or algorithmic power, which appears to know human life better than life can know itself. But this ideology clashes with a much older vision of how human life should be, an ecological view of human life, unchallenged until very recently, which assumes that human life, like all life—whatever its limits, constraints, and deficits—is a zone of open-ended connection and growth.

The order of data colonialism damages the minimal integrity of the self as a space of self-relation. Yet human beings' possibility of being "at home" with themselves depends on having access to that space of the self. The damage done by data colonialism is therefore ecological in its nature and not just in its scale. Recall the ancient Greek roots of the word ecology: the task of gathering together a home (legein + oikos). The social order of data colonialism is, in a precise sense, an antiecology that disrupts the basic space of human autonomy and conviviality in a process that might be compared to what biologists call "dysbiosis." This is the opposite of the successful forms of interdependency between life forms that we call "symbiosis." Data colonialism disrupts the ecology of human life not for some higher purpose but in the name of pure instrumentality: to support capitalism's drive to profit.

Insisting on boundaries to the continuous flow of data (the principle of seamfulness) means insisting that human reason should always be able to impose limits on that flow, because data's goal should ultimately be to serve human, not artificial, life. Nothing we have argued in this book rules out the search for social uses of data that genuinely remain under the control and monitoring of citizens themselves. We have instead offered ample evidence that the emerging social order built from seamless data flows in the interests of capital is becoming toxic to human beings through two means: (1) the highly targeted but almost entirely opaque forms of inequality that it authorizes in the name of greater social knowledge and resource efficiency and (2) the uncontrolled surveillance of human life on which it is premised.

The principle of seamfulness therefore becomes a potential tool for resisting social worlds that are less accountable, harsher, and characterized by ever higher levels of inequality, mistrust, and despair. It also alerts us to the danger that without seamfulness, the social costs that data colonialism generates will simply be passed down to the most vulnerable, quite possibly on the basis that they somehow "deserve" it, exactly the form of "moral behaviorism" that has for two decades characterized neoliberal societies.

Here, as Boaventura de Sousa Santos has noted, questions of "global social injustice" become "intimately linked" with questions of "global cognitive injustice." Capitalism's grip on the world, through the mechanism of connection, creates a new modality of injustice, indissolubly both social and cognitive, that is derived from data processes. Data colonialism's true violence is to make human beings the objects of external control through data, together with everything that, until now, has passed for their lives' interior.

Whereas the word order has acquired certain negative connotations within the ambit of Western rationality, as something implying imposed structures, it can also mean "balance," as in harmony, and beneficial forms of interconnectedness and living well together. Order is the abstract term for many elements of human life we are truly thankful for: routine, expectation, the security of knowing that people generally act as they say they will. But one must always ask: Order for whom? Order on what terms?

It is necessary to extricate oneself from the linkages between rationality/modernity and coloniality, first of all, and definitely from all power which is not constituted by free decisions made by free people. It is the instrumentalization of the reasons for power, of colonial power in the first place, which produced distorted paradigms of knowledge and spoiled the liberating promises of modernity.

Outside the West, virtually in all known cultures . . . , all systematic production of knowledge is associated with a perspective of totality. But in those cultures, the perspective of totality in knowledge includes the acknowledgement of the heterogeneity of all reality; of the irreducible, contradictory character of the latter; of the legitimacy, i.e. desirability of the diverse character of the components of all reality—and therefore, of the social. The [better, alternative] idea of social totality, then, not only does not deny, but depends on the historical diversity and heterogeneity of society, of every society. In other words, it not only does not deny, but it requires the idea of an "other"—diverse, different.

There is no empirical test that could verify this vision of social knowledge: it has authority only by virtue of being imposed on a world reconfigured to its own image. As we learned in chapter 4, measures “recreate social worlds.”

It is to articulate a vision that, against the background of Big Data reasoning, will appear counterintuitive to many, a vision that rejects the idea that the continuous collection of data from human beings is a rational way of organizing human life. From this perspective, the results of data processing are less a natural form of individual and societal self-knowledge than they are a commercially motivated “fix” that serves deep and partial interests that as such should be rejected and resisted.

Leanne Betasamosake Simpson reflects on rationality but from the perspective of human meaning. She affirms a meaning that “is derived not through content or data or even theory . . . but through a compassionate web of interdependent relationships that are different and valuable because of difference.” Why difference exactly? Because only by respecting difference do we stand any chance of not “interfer[ing] with other beings’ life pathways” and their possibilities for autonomy.

It is time to call out the contradiction between capitalism’s current business models (based on default data harvesting) and the value of freedom that capitalism’s representatives claim to espouse.

Resistance to data colonialism does not mean gestures made for effect (such as quitting Facebook) that leave the order of data colonialism intact. But it does mean saying no to practices that normalize the order of data colonialism in everyday life. We need to find ways of collectively unsettling our identification with data colonialism's requirements. Power is always social in the end, so discovering new powers of resistance must emerge from the social world too.

More positively, it also means intensifying the pleasures of sociality (solidarity, even) that not even datafication has been able to suppress. One reason we all flocked to social media platforms originally was because they reminded us of the joys of collectivity at a time when mass media seemed to offer only individualistic isolation. That pleasure risks being extinguished now as we become ever clearer that social media outlets are managed by vast commercial interests. But we don’t need to forget everything that has been learned during our time on digital networks. We can develop forms of solidarity that extend what we have learned online but turn them toward values that are no longer aligned to data colonialism.

Recall for a moment some of the actions whose partial nature we criticized earlier: media literacy campaigns, withdrawal from this or that platform, and legislation requiring data companies to actually obtain something like consent for using personal data. What if we approached them not merely as individualistic acts or as technical fixes but as part of a collaborative attempt to build a different social world than the one data colonialism promises us? Acquiring media literacy might then include citizens learning to help one another become less reliant on systems whose negative effects we all know. Withdrawing from a particular platform might involve learning with others how to distinguish between forms of forced coordination that benefit platforms and forms of collaboration that genuinely benefit people and then, on the basis of that understanding, building new communicative forms that prioritize the latter over the former. Legislating for a better data environment might involve going beyond attempts to “tame” data companies and affirming positive principles on the basis of which future social uses of data could be organized for collective benefit and not corporate privilege.

The Indian anthropologist Arjun Appadurai recently made an important call to rethink research beyond the walls of academia. He proposes that we think of research in terms of the "right to the tools through which any citizen can systematically increase the stock of knowledge that they consider most vital to their survival as human beings and to their claims as citizens."

This, incidentally, is what distinguishes the universalizing involved in our argument from the universalizing that colonialisms have helped shape, indeed, the universalizing that underlies the whole vision of Big Data. Decolonizing data means speaking now from the position of data's colonized and imagining a common future for humanity beyond the contemporary project to reduce human life to the inputs and outputs of data processing.

Gayatri Spivak once famously asked, "Can the subaltern speak?" Similarly, today we might ask: Can data colonialism's subjects speak up against the structures that dispossess them? An affirmative answer is possible if we take seriously the temporal dislocation that postcolonial theory grasped so well when it defined itself as an "active transformation of the present out of the clutches of the past." Except that today the dislocation is subtly different. Even though we have explored the roots of data colonialism in the past, the debates started here must address the kind of future that is already being created in the clutches of the present.

Decolonizing data's role in explaining the social world means rejecting the assumption that only data can explain data—that is, that only quantitative methods based on Big Data methodologies can accurately describe the world from now on. While it should not reject the idea of gathering data as such (for example, for clearly defined social purposes), the evolving project to think together about the consequences of data colonialism must insist on close attention to those "aspects of social life that remain unseen, unheard, uncounted or unacknowledged within prevailing understandings of capitalism," as Johnna Montgomerie argues. Put differently, our understanding of the social world needs to be rescued from a narrow understanding that equates it with large data sets and abstract models with "only fantastical links to actual human experience or observation."

The Cloud Empire's myth that the only way to harness new social resources is to surrender to platforms' control of how and where we make knowledge, organize resources, or even just express solidarity and connection must be unlearned, and even more rapidly than it came into being. Recent infrastructures of connection have facilitated new forms of collectivity, but this does not mean that we must accept the forms of exploitation that are indissolubly tied to the data relations associated with those infrastructures.

Axon's triumphal rise is just a small part of the increasing datafication of the US law-and-order sector, which is, in turn, just part of the wider transformation of social governance and individual self-governance in the United States and many other nations. Governed this way, the social world stops being a space of dialogue and debate, let alone of democratic change, and becomes what the data industry calls a propensity and probability spectrum, governed by knowledges that are largely in corporate hands.

The representatives of data colonialism try to convince us that there is nothing hollow in our assent to their order, that we are choosing familiar values of connection and human solidarity when we choose this datafied social order. They tell us that we need to go on connecting with and through them. But that is a ruse. The ruse depends on inducing human beings to forget ways of living that were left behind when we came to spend so much time servicing the social quantification sector.

The greatest threat that data colonialism poses is that, in time, it works too well for us to want to live any other kind of existence, so that our complicity in losing hold on the possibility of freedom becomes complete.

The path away from data colonialism will start when we reclaim the capacity to connect that human beings have always possessed, and we decide that today's costs of connection are neither necessary nor worth paying. The struggle is a global one: it matters for humanity as a whole, not just for market democracies, and it confronts capitalism as a whole, not just social media.

We draw here on William Sewell's definition of the social as "the various 'mediations' that place people into 'social' relations with one another . . . [as] interdependent members of each other's worlds" (Logics of History, 329).

We appreciate that Marx himself had an extremely broad understanding of "labor" to encompass not just productive activity but also the wider activities and pleasures of human life (Berki, "Nature and Origins," 36) or, as Marx put it, "Life activity, productive life itself" (Manuscripts, 113). But this unusually expansive notion of labor risks obscuring the transformation underway whereby new forms of exploitation emerge through the collection and transformation of data from countless human activities that are not understood by human beings as involving any productive activity at all (for example, just chatting online, looking at things other people show us, or monitoring oneself).

China's 2015 "Internet Plus" policy describes itself as "the profound integration of the innovation achievements of the Internet in all areas of the economy and society." It puts special emphasis on the "smartification of the manufacturing sector" and "high efficiency logistics" (see China Copyright and Media, "'Internet Plus' Plan"). China's "Artificial Intelligence Development Plan" proposes "industrial intelligentization," including "smart logistics" (China Copyright and Media, "Artificial Intelligence").

For rare anticipations of this colonial reading, see Cohen, "Biopolitical Public Domain"; Thatcher, O'Sullivan, and Mahmoudi, "Data Colonialism"; Shepherd, "Measured and Mined"; and Shepherd, "Neocolonial Intimacies." Although the analysis done by Thatcher, O'Sullivan, and Mahmoudi is consistent with ours, their usage of the term data colonialism is explicitly metaphorical; rather than colonial, Cohen uses the term "postcolonial" and Shepherd the term "neocolonial." There are also growing debates about neocolonial influence in the technology sector and in the use of data by development and humanitarian agencies; see https://globalstudies.trinity.duke.edu/volume-31-decolonizing-thedigitaldigital-decolonization. Particularly interesting is the work of Joanne Radin ("Digital Natives") on the unreflexive use of health data from First Nation peoples in the United States in the "training" of artificial intelligence. For an analysis of AI practices that opens up a neocolonial perspective and was published as we completed this book, see Crawford and Joler, "Anatomy," especially pp. 17–18. But none of these writers or approaches argue, as we do, that data practices literally represent a new stage of colonialism.

See Grewal, Network Power, on how interlocking forms of consent within digital networks can amount to force. Compare Bratton, The Stack, 43; and Chun, Programmed Visions.

We are not claiming datafied labor relations (or data relations) are slavery; such hyperbole would be offensive. We are making a comparison in terms of the degree and continuity of surveillance that is normal in various regimes of labor.

Polanyi (Great Transformation, 78–79) had a similar formulation when he analyzed industrial capitalism in terms of "fictional commodities": the transformation of work into "labor," land into "real estate," and exchange into "money." Building on Polanyi, Zuboff ("Big Other," 84) argues that "reality" is now being turned into tracked "behaviour."

Capitalism here aims at the state in which, as Marx (Capital, vol. I, 952) put it, "All the means of labour . . . now also serve as ingredients in the valorization process. Where they are not converted into actual money, they are converted into accounting money." We are arguing, however, that the "means of labour" include both labor itself, with its means of reproduction, and the expanding conditions, or factors, of production.

The closest Habermas comes to unpacking the metaphor of "colonization" is on page 355, where he says system imperatives "make their way into the lifeworld from the outside—like colonial masters coming into a tribal society—and force a process of assimilation upon it." But this idea is not developed as a theory of colonialism, and its starting point—the separation of technological systems from lifeworld—is hard to reconcile with life in most, if not all, contemporary societies. More recently, and in a similar vein, Adam Greenfield has written about the Internet of Things as "the colonization of everyday life by information processing" (Radical Technologies, 32).

We generalize here the concept of "tool reversibility" from Couldry and Hepp, Mediated Construction, 132. "Tool reversibility" was already anticipated in nineteenth-century workers' analysis of the factory as making the craftsman into "an animated tool of the management" (cited in Braverman, Labor, 94). In a broad way, today's critique of tool reversal echoes Illich's critique of modern tools (Tools, 34–37) and Marx's account of humankind's alienation from their productive lives under capitalism (Manuscripts, 112–13). But there is something distinctly shocking about contemporary smart tools' direct use of their users.

One respected analyst of people's use of fitness trackers seems to recognize the problem with orthodox Marxism but draws the wrong conclusion, claiming that, while not intuitively labor, such activities must nonetheless be understood as really labor. He says, "Regardless of how we might like to conceptualize exercise activities that are tracked, they are labor, because the corporations are treating them as such. Surplus value will be extracted from these activities by corporations, as if they were work" (Till, "Exercise as Labour," 452).

Our argument starts out more pessimistically, but we also believe more realistically, from the current capitalist drive to treat the human activities that generate data as raw material for production. Being paid for one's data would, in any case, do nothing to address the damage done by the continuous surveillance on which data collection is built (see chapter 5).

Contrast the wages-for-housework argument of 1970s and 1980s feminists (Federici, Wages, 4; Delphy, Close to Home), which had the goal not of commodifying that work but of "denaturalizing" capitalism's wider social relations so as to challenge them for the longer term. So it is today with platforms: the key issue is not securing payment but denaturalizing the abstraction of social relations through data whereby capitalism is able to expand ever deeper into the social world.

As Postone ("Rethinking Marx," 7) put it more precisely, the commodity for Marx is a "structured form of social practice that is a structuring principle of the actions, worldviews, and dispositions of people."

Autonomists draw on Marx’s early notebooks and single out his comments there about the role that human beings’ general productive and creative capacity (the “general intellect”) can play in the development of capitalism (for a more popular version of this argument, see Mason, Postcapitalism, 164–67). Autonomists propose that, as humans become increasingly part of the social factory, particularly online, they become able to turn their creativity against capitalism. But Marx’s words even in the Grundrisse were more ambiguous: “The development of fixed capital indicates to what degree general social knowledge has become a direct force of production, and to what degree, hence, the conditions of the process of social life itself have come under the control of the general intellect and been transformed in accordance with it” (Grundrisse, 706, emphasis added). If by “the general intellect” we mean the organizational and creative power within capitalism (think of Google’s leadership), then Marx’s comment allows for a radically different conclusion from that which the Autonomists draw: “the general intellect” might be the means for capitalism to appropriate social life.

Using figures from 2017, calculations were performed by dividing the market capitalization (the market value of a company’s outstanding shares) by the number of users. Market caps and approximate user numbers on December 29, 2017, were as follows: Alphabet—$731.90B, 2B users; Facebook—$512.76B, 2.2B users; and Alibaba—$727.04B, 488M users.

The market value for Tencent was $519 billion in November 2017, and the number of users in quarter 2 of that year was 963 million. There was a 25 percent fall in the value of their stock in the first half of 2018, however. Regardless of these adjustments, the fact remains that each of these companies benefits greatly from each of their users.

Marx believed technological advances are always making production more efficient, which in turn lowers costs and makes it more difficult for capitalists to maximize profit. Periodical crises can restore some degree of profitability by allowing successful corporations to absorb the resources of failing corporations. The problem, according to some theorists, is that recent government bailouts are preventing that process from fully unfolding.

It is estimated that data centers will use 3 to 13 percent of all global electricity by the year 2030, compared to the 1 percent they used in 2010. Water is another resource gobbled up by data centers. The surveillance facility constructed by the National Security Agency in the United States, the third largest on the planet, consumes 1.7 million gallons of water daily to operate its servers.

58. The value that the information and communication technologies (ICT) sector contributes to national gross domestic product (GDP) reveals relatively small figures, which are at odds with the enthusiasm exhibited by market capitalization. This sector (of which the social quantification subsector would only be a subsegment) has added merely 6.8% of value to the US economy in recent years. In a report that pools data for 2006 and 2014 from developed countries in the European Union, North America, and Asia, only Taiwan and South Korea had percentages in excess of 6% (13.3/15.9% and 9.2/8.9%, respectively). Between 1994 and 2014, ICT was only 2.6% of GDP in twenty countries surveyed by the OECD, with an average growth across those countries of 16.7%. In both the United States and the United Kingdom, two supposed leaders of the digital transformation, ICT investment as a percentage of GDP actually fell in this period, to 3.15% and 2.17%, respectively (OECD, "Share of ICT"). As far as the industry's impact on employment rates, it is calculated that worldwide the ICT sector and the consumer electronics sector together employ nearly two hundred million workers, with another two hundred million if we add mining and labor-intensive work (Maxwell and Miller, Greening the Media, 106.). This might sound like a lot, but when we isolate the social quantification sector, the overall picture changes. Figures for the year 2000 show that the industries involved in the ICT sector accounted for only 3.9% of employment in the European Union and 4.9% in the United States (Doogan, New Capitalism?, 59), with those numbers in fact decreasing over time. According to the OECD report, by 2014 the figure for the European Union had gone down to 2.5%, and the report identifies only Taiwan as showing an employment rate in this sector above 10%. A look at how many full-time employees were actually employed by social quantification corporations reveals a very small number: Airbnb—1,917 employees in the United States, according to 2016 reports; Apple—approximately 123,000 as of September 2017; Facebook—23,165 as of September 2017; Google (Alphabet)—72,053 as of December 2016 (keeping in mind that half of Google's workers are contractors who don't receive full benefits: cf. Bergen and Eidelson, "Inside Google"); Instagram (a Facebook company)—450 as of January 2017; Snapchat—1,859 as of December 2016; Twitter—approximately 3,860 as of June 2016; and Uber—approximately 12,000 as of June 2017 (all figures comes from corporate websites). It should be noted that both Google and Facebook are adding about 10,000 employees each to act as moderators in the aftermath of the fake news scandal (Naughton, "Dirty Work"). Of course, these figures do not take into account all the workers involved in datafication both inside and outside the social quantification sector, including part-time and freelance workers. But to put things in perspective, Walmart and McDonald's each employ about two million people worldwide. Only Amazon comes close to that magnitude, with 541,900 employees as of October 2017.

Acemoglu and Restrepo ("Machine and Man") calculate that adding one robot to every one thousand workers "reduces the employment to population ratio by about 0.18–0.34 percentage points and wages by 0.25–0.5 percent."

Pomeranz argues that "slavery helped make Euro-American trade unlike any between Old World cores and peripheries" (267) by making "the flow of needed resources to Europe self-catalyzing in ways that consensual trade between Old World regions was not: it anticipated, even before industrialization, the self-perpetuating division of labor between primary products exporters and manufacturing regions in the modern world" (24).

Although Said's work has been critiqued for denying agency to the Other (the one who is the object of representation, as opposed to the Self who does the representing) and for somewhat reifying Eurocentrism by situating colonialism as a transhistorical and universalizing force, within postcolonial studies Said opened up avenues for thinking about the historical relationship between knowledge and power and about the specific ways in which colonialism sought to produce "truth" and knowledge, and not just ideology, in order to subjugate.

In response, postcolonial examinations of science and technology have, according to Harding ("Convergences and Dissonances"), encompassed four general areas: (1) they have put forth counterhistories of Western progress, providing a critical reevaluation of their development and impact; (2) they have sought to reclaim and recontextualize non-Western forms of knowledge, which were eradicated or appropriated during colonialism; (3) they have examined the ways in which colonial science and technology continue to shape societies even after their independence; and (4) most importantly, they have promoted the development of alternative science and technology projects using decolonized methodologies, even in the difficult conditions of a neocolonial context. Even mathematics, the ultimate universal discipline, has been deconstructed to show that different cultures have generated mathematical ideas and that it is truly a pancultural phenomenon (Bishop, "Western Mathematics," 72).

The purpose of maps has always been to make it possible to locate the most efficient paths of circulation. Colonial expansion was facilitated by the pioneering of mercantile routes that mapped newer and faster channels of commodity exchange, creating wealth for European colonizers and merchants by enabling price differentials based on differential access to the assets that the maps represented (see Wolf, People without History, 88).

Harris notes that automata such as the Mechanical Turk were referred to as mammets, a word deriving from the proper name Mohamet or Mohammed. He adds that “medieval Christian theologians used the word 'mechanicum' as a synonym for Muslim sorcery: they regarded Islam as a mechanical religion incapable of true life and of a meaningful future, and thus consigned it to a dead, unusable past. Benjamin's Turkish chess-playing machine eerily replays this typological gambit.”

“If the recorded individual has come into full view, the recording individual has faded into the background, arguably to the point of extinction” (Fourcade and Healy, “Like a Market,” 11).

Rainie and Duggan (“Privacy and Information,” 4) found that 55 percent of the US population surveyed would reject a thermostat that collected “basic” data about activities in their house in return for regulating the house’s temperature efficiently.

Some interpretative sociologists have challenged the idea that individual economic calculations are irrational and are reinterpreting them not in terms of numbers but of the complex moral relations in which economic transactions are in fact embedded (Zelizer, Economic Lives, discussed in Wherry, "Relational Accounting").

17. We do not therefore deny that property-based arguments (for example, legal fictions of ownership) can be developed that might support privacy and autonomy; for an interesting US argument along these lines, see Fairfield, Owned. There are also proposals for developing, even within the big-tech industry, some form of personal data portability; see the Data Transfer Project sponsored by Google, Micro soft, and Twitter, at https://datatransferproject.dev and http://dataportability.org. See also Tim Berners-Lee’s proposal for a new web infrastructure that would support data portability: https://solid.inrupt.com. But for caution against this type of argument, see Gurumurthy and Vasudevan, “Societal Need.”

23. For recent examples, see Parikka, Insect Media, in media theory; and DeLanda, New Philosophy, in social theory. The fascination goes far back in literature to Virgil’s bees simile in Book Four of the Georgics and to Mandeville’s Fable of the Bees. Note, however, that Thomas Hobbes based his political philosophy in Leviathan on insisting that human beings are not like bees and ants because they “are continually in competition for Honour and Dignity” (Leviathan, 225).

45. Eggers’s novel is only the latest in a long line of imaginative critiques of corporate life, starting perhaps with Kafk a, but it gains a particular force by having been written in San Francisco, close to the physical location of the power appropriations and work cultures that Eggers dissects. China’s data platforms have yet to attract a novelistic treatment.

For this reason, we are not convinced by accounts of self-tracking that celebrates its “situated objectivity” (Pantzar and Ruckenstein, “Living the Metrics”), let alone its potential as resistance to corporate data practices (Nafus and Sherman, “This One Does Not Go Up to Eleven”). Yes, these practices are richly contextualized, but as data relations they are part of the means to reproduce the new data-driven capitalism.

Arguing that privacy remains a necessary term is not meant here to imply that it is sufficient to address many of the forms of exploitation through data that occur, which is why we spend time on the notion of autonomy and its basic preconditions.

Other countries are introducing parallel legislation, for example, Brazil with its Data Protection Law (August 2018). As a major power, the European Union may be in a position to seek to impose GDPR as a standard on its trading partners.

Hildebrandt (Smart Technologies, 15) expresses this as a shift from the normal "double contingency" of human social reality to a world driven by a "single contingency," what systems and their drivers want. On the challenge through datafication to mutual expectations assumed by classical phenomenology, see Couldry and Kallinikos, "Ontology."

Srnicek (Platform Capitalism, 158), who argues that "rather than just regulating corporate platforms, efforts could be made to create public platforms—platforms owned and controlled by the people."

We hold here onto the term tactical while acknowledging that the underlying distinction from French theorist Michel de Certeau (Everyday Life) between strategy and tactics may have outlived its usefulness. De Certeau argued that within every large-scale structure of strategic power (government, markets, rituals, the order of space itself) there are places for tactics: empty zones, such as the corner of a market square, to which the eye or hand of power doesn’t penetrate. In tactical places, the possibility subsists of “free” activity with real benefits for the actors, benefits that are not envisaged or controlled by strategic power. De Certeau’s notion of tactics, however, assumed some limits to how far power can penetrate into space—a reasonable assumption about every form of generalized power until now. But what if the new order being built is an order beyond tactics, an order that appropriates almost all resources from which tactics might be fashioned except the short-term local gaming of the system that, by its own admission, “plays the game” and so cannot challenge it?

We can find an interesting parallel in comments by the prominent, if controversial, figure of Pope Francis in his recent encyclical, where he warned about the “technocratic paradigm,” criticizing a way of thinking about technology “as if the [technological] subject were to find itself in the presence of something [the world] formless, completely open to manipulation.” The result, he suggests, is to authorize a new and powerful subject “who, using logical and rational procedures, progressively approaches and gains control over an external object” (Encyclical, 66).

"Hopelessness," Paulo Freire wrote, "is a form of silence, of denying the world and feeling from it. [But] the dehumanization resulting from an unjust order is not a cause for despair but for hope, leading to the incessant pursuit of the humanity which is denied by injustice" (Pedagogy, 64).

Simpson (Always Done, 17) writes of the capacity called Biiskabiyingan in the Nishnaabeg language: "The process of returning to ourselves, a reengagement with the things we have left behind, a re-emergence, an unfolding from the inside out." For the idea of "returning to ourselves," compare what H. S. Bhola (Literacy) calls the ontological reintegration of the individual to the world.