The Age of Surveillance Capitalism

by Shoshana Zuboff

  Schmidt was, in fact, merely paraphrasing computer scientist Mark Weiser’s seminal 1991 article, “The Computer for the 21st Century,” which has framed Silicon Valley’s technology objectives for nearly three decades. Weiser introduced what he called “ubiquitous computing” with two legendary sentences: “The most profound technologies are those that disappear. They weave themselves into the fabric of everyday life until they are indistinguishable from it.” He described a new way of thinking “that allows the computers themselves to vanish into the background.… Machines that fit the human environment instead of forcing humans to enter theirs will make using a computer as refreshing as taking a walk in the woods.”2

  Weiser understood that the virtual world could never be more than a shadow land no matter how much data it absorbs: “Virtual reality is only a map, not a territory. It excludes desks, offices, other people… weather, trees, walks, chance encounters and, in general, the infinite richness of the universe.” He wrote that virtual reality “simulates” the world rather than “invisibly enhancing the world that already exists.” In contrast, ubiquitous computing would infuse that real world with a universally networked apparatus of silent, “calm,” and voracious computing. Weiser refers to this apparatus as the new “computing environment” and delights in the possibilities of its limitless knowledge, such as knowing “the suit you looked at for a long time last week because it knows both of your locations, and it can retroactively find the designer’s name even though that information did not interest you at the time.”3

  Schmidt was not describing the end of the internet but rather its successful unshackling from dedicated devices such as the personal computer and the smartphone. For surveillance capitalists, this transition is not a choice. Surveillance profits awakened intense competition over the revenues that flow from new markets for future behavior. Even the most sophisticated process of converting behavioral surplus into products that accurately forecast the future is only as good as the raw material available for processing. Surveillance capitalists therefore must ask this: what forms of surplus enable the fabrication of prediction products that most reliably foretell the future? This question marks a critical turning point in the trial-and-error elaboration of surveillance capitalism. It crystallizes a second economic imperative—the prediction imperative—and reveals the intense pressure that it exerts on surveillance capitalist revenues.

  The first wave of prediction products enabled targeted online advertising. These products depended upon surplus derived at scale from the internet. I have summarized the competitive forces that drive the need for surplus at scale as the “extraction imperative.” Competition for surveillance revenues eventually reached a point at which the volume of surplus became a necessary but insufficient condition for success. The next threshold was defined by the quality of prediction products. In the race for higher degrees of certainty, it became clear that the best predictions would have to approximate observation. The prediction imperative is the expression of these competitive forces. (See Figure 3 here.)

  Google/Alphabet, Facebook, Microsoft, and many more companies now drawn to surveillance revenues have staked their claims on the internet’s “disappearance” because they must. Compelled to improve predictions, surveillance capitalists such as Google understood that they had to widen and diversify their extraction architectures to accommodate new sources of surplus and new supply operations. Economies of scale would still be vital, of course, but in this new phase, supply operations were enlarged and intensified to accommodate economies of scope and economies of action. What does this entail?

  The shift toward economies of scope defines a new set of aims: behavioral surplus must be vast, but it must also be varied. These variations are developed along two dimensions. The first is the extension of extraction operations from the virtual world into the “real” world, where we actually live our actual lives. Surveillance capitalists understood that their future wealth would depend upon new supply routes that extend to real life on the roads, among the trees, throughout the cities. Extension wants your bloodstream and your bed, your breakfast conversation, your commute, your run, your refrigerator, your parking space, your living room.

  Economies of scope also proceed along a second dimension: depth. The drive for economies of scope in the depth dimension is even more audacious. The idea here is that highly predictive, and therefore highly lucrative, behavioral surplus would be plumbed from intimate patterns of the self. These supply operations are aimed at your personality, moods, and emotions, your lies and vulnerabilities. Every level of intimacy would have to be automatically captured and flattened into a tidal flow of data points for the factory conveyor belts that proceed toward manufactured certainty.

  Just as scale became necessary but insufficient for higher-quality predictions, it was also clear that economies of scope would be necessary but insufficient for the highest quality of prediction products able to sustain competitive advantage in the new markets for future behavior. Behavioral surplus must be vast and varied, but the surest way to predict behavior is to intervene at its source and shape it. The processes invented to achieve this goal are what I call economies of action. In order to achieve these economies, machine processes are configured to intervene in the state of play in the real world among real people and things. These interventions are designed to enhance certainty by doing things: they nudge, tune, herd, manipulate, and modify behavior in specific directions by executing actions as subtle as inserting a specific phrase into your Facebook news feed, timing the appearance of a BUY button on your phone, or shutting down your car engine when an insurance payment is late.

  This new level of competitive intensity characterized by scope and action ratchets up the invasive character of supply operations and initiates a new era of surveillance commerce that I call the reality business. Economies of scale were implemented by machine-based extraction architectures in the online world. Now the reality business requires machine-based architectures in the real world. These finally fulfill Weiser’s vision of ubiquitous automated computational processes that “weave themselves into the fabric of everyday life until they are indistinguishable from it,” but with a twist. Now they operate in the interests of surveillance capitalists.

  There are many buzzwords that gloss over these operations and their economic origins: “ambient computing,” “ubiquitous computing,” and the “internet of things” are but a few examples. For now I will refer to this whole complex more generally as the “apparatus.” Although the labels differ, they share a consistent vision: the everywhere, always-on instrumentation, datafication, connection, communication, and computation of all things, animate and inanimate, and all processes—natural, human, physiological, chemical, machine, administrative, vehicular, financial. Real-world activity is continuously rendered from phones, cars, streets, homes, shops, bodies, trees, buildings, airports, and cities back to the digital realm, where it finds new life as data ready for transformation into predictions, all of it filling the ever-expanding pages of the shadow text.4

  As the prediction imperative gathers force, it gradually becomes clear that extraction was the first phase of a far-more-ambitious project. Economies of action mean that real-world machine architectures must be able to know as well as to do. Extraction is not enough; now it must be twinned with execution. The extraction architecture is combined with a new execution architecture, through which hidden economic objectives are imposed upon the vast and varied field of behavior.5

  Gradually, as surveillance capitalism’s imperatives and the material infrastructures that perform extraction-and-execution operations begin to function as a coherent whole, they produce a twenty-first-century “means of behavioral modification.” The aim of this undertaking is not to impose behavioral norms, such as conformity or obedience, but rather to produce behavior that reliably, definitively, and certainly leads to desired commercial results. The research director of Gartner, the well-respected business advisory and research firm, m
akes the point unambiguously when he observes that mastery of the “internet of things” will serve as “a key enabler in the transformation of business models from ‘guaranteed levels of performance’ to ‘guaranteed outcomes.’”6

  This is an extraordinary statement because there can be no such guarantees in the absence of the power to make it so. This wider complex that we refer to as the “means of behavioral modification” is the expression of this gathering power. The prospect of guaranteed outcomes alerts us to the force of the prediction imperative, which demands that surveillance capitalists make the future for the sake of predicting it. Under this regime, ubiquitous computing is not just a knowing machine; it is an actuating machine designed to produce more certainty about us and for them.

  This gradually accruing, smart, and muscular apparatus is gradually being assembled around us. No one knows what the real magnitude is or will be. It is a domain plagued by hyperbole, where projections frequently outrun actual results. Despite this, the planning, investment, and invention necessary to draw this vision of ubiquity into reality are well underway. The visions and aims of its architects, the work that has already been accomplished, and the programs that are currently in development constitute a turning point in the evolution of surveillance capitalism.

  Finally, I want to underscore that although it may be possible to imagine something like the “internet of things” without surveillance capitalism, it is impossible to imagine surveillance capitalism without something like the “internet of things.” Every command arising from the prediction imperative requires this pervasive real-world material “knowing and doing” presence. The new apparatus is the material expression of the prediction imperative, and it represents a new kind of power animated by the economic compulsion toward certainty. Two vectors converge in this fact: the early ideals of ubiquitous computing and the economic imperatives of surveillance capitalism. This convergence signals the metamorphosis of the digital infrastructure from a thing that we have to a thing that has us.

  Futuristic as this may sound, the vision of individuals and groups as so many objects to be continuously tracked, wholly known, and shunted this way or that for some purpose of which they are unaware has a history. It was coaxed to life nearly sixty years ago under the warm equatorial sun of the Galapagos Islands, when a giant tortoise stirred from her torpor to swallow a succulent chunk of cactus into which a dedicated scientist had wedged a small machine.

  It was a time when scientists reckoned with the obstinacy of free-roaming animals and concluded that surveillance was the necessary price of knowledge. Locking these creatures in a zoo would only eliminate the very behavior that scientists wanted to study, but how were they to be surveilled? The solutions once concocted by scholars of elk herds, sea turtles, and geese have been refurbished by surveillance capitalists and presented as an inevitable feature of twenty-first-century life on Earth. All that has changed is that now we are the animals.

  II. The Tender Conquest of Unrestrained Animals

  It was a 1964 international expedition to the Galapagos Islands that presented a unique opportunity to explore telemetry, a frontier technology based on the long-distance transmission of computer data. A new breed of scientists who combined biology, physics, engineering, and electronics championed this new tech, and chief among these was R. Stuart MacKay, a physicist cum electrical engineer, biologist, and surgeon who was known among his scientific peers as the experts’ expert.7

  MacKay viewed telemetry as a means of enhancing and protecting the well-being of animal populations. A photo of MacKay from the Galapagos expedition shows him poised tenderly beside a giant tortoise that had swallowed his tiny machine; in another he gently holds a rare marine iguana with a sensor attached to its torso, all of it to measure the animals’ internal body temperatures. He emphasized the key element that distinguished telemetry from other forms of monitoring: the possibility of capturing behavior in its natural habitat with sensors of such compactness that they could disappear into the body without triggering the animal’s awareness:

  The use of a radio signal from a transmitter, in or on a subject, to carry information to a remote receiver for recording allows flexibility of movement and permits disturbance-free exploration of otherwise inaccessible parts of the body without the subject even being aware of the measuring process… the methods leave the subject in a relatively normal psychological and physiological state, and do not interfere with the continuation of normal activities.8

  MacKay’s published work focused primarily on the technical aspects of his studies, although occasionally there is a glimpse of larger purpose. Telemetry created the possibility of huge data sets and the opportunity for correlational studies at the scale of entire animal populations. He noted that the same techniques could be applied to the static world: forest canopies, the curing of concrete, chemical reaction vessels, and food processing. MacKay envisioned whole populations of connected data-emitting individuals. His first-generation “wearable technologies” made it possible to study “unrestrained animals” among every species, including people. Biomedical telemetry, he stressed, was uniquely suited to gather information that would be otherwise impossible to collect “in the wild.” The key principle was that his telematics operated outside an animal’s awareness. This was especially useful in solving problems such as the difficulty of measuring “uncooperative animals” and the need to gather data even when herds roamed through “inaccessible regions.” In other words, MacKay’s inventions enabled scientists to render animals as information even when they believed themselves to be free, wandering and resting, unaware of the incursion into their once-mysterious landscapes.

  MacKay stressed that the transmission and monitoring of sensor data were only part of the story. The route was not enough; it had to be the routing. He argued for a “reverse process” of telestimulation that would not only monitor behavior but also reveal how it could be modified and optimized, providing what he regarded as “a remote dialogue between the subject and the experimenter.”9

  MacKay’s broad vision has come to fruition in the digital age. Satellite acuity combined with the explosive growth of computational power fitted onto tiny slivers of silicon, advanced sensors, internet-enabled networks, and “big data” predictive analytics have produced extraordinary systems that reveal the meanings and movements of whole animal populations and their individuals: anywhere, anytime. The same wearables traveling on and in the bodies of animals have also become broad sensors of the planet’s climate, geography, and ecology, enabling “a quorum sensing of our planet, using a variety of species to tap into the diversity of senses that have evolved across animal groups,” producing a “sixth sense of the global animal collective.”10 As you will already have guessed, there is little reason to suppose that these capabilities will remain trained on nonhuman species.

  Indeed, the threshold has already been crossed.11 In 2014 a team of University of Washington researchers led by Jenq-Neng Hwang announced a “super GPS” assembled from municipal surveillance cameras “to enable the dynamic visualization of the realistic situation of humans walking on the road and sidewalks, so eventually people can see the animated version of the real-time dynamics of city streets on a platform like Google Earth.”12 If this were a novel, then Professor MacKay’s brilliant work, along with that of the many dedicated scientists who followed in his footsteps, would serve as foreshadowing.

  In a metamorphosis that MacKay did not foresee, the science of animal tracking that grew from his pathbreaking vision became the template for surveillance capitalism’s next phase of evolution as telematics now applied to human behavior succumbed to the thrall of a new and lucrative logic of accumulation. The requirements of prediction that would later merge into an economic imperative were already evident in MacKay’s work. The need for economies of scope, both in extension and depth, is reflected in his foundational framework that aimed to render information about populations and the details of individuals, reaching into the distant
corners of previously inaccessible regions. Today those regions include the front seat of your car, your kitchen, and your kidneys. His “reverse process” of telestimulation is resurrected in the economies of action that automatically stimulate behavior, not to save the human herd from catastrophe but rather to heighten the predictability of its behavior.

  MacKay yearned for discovery, but today’s “experimenters” yearn for certainty as they translate our lives into calculations. MacKay’s animals were unrestrained and innately uncooperative because they felt themselves to be free, sheltering and roaming in unknown terrain. Now, the un-self-conscious, easy freedom enjoyed by the human animal—the sense of being unrestrained that thrives in the mystery of distant places and intimate spaces—is simply friction on the path toward surveillance revenues.

  III. Human Herds

  MacKay’s legacy is reimagined for our time in the work of Professor Joseph Paradiso of the MIT Media Lab, where some of surveillance capitalism’s most valuable capabilities and applications, from data mining to wearable technologies, were invented.

  Paradiso’s brilliant group of data scientists, engineers, musicians, and artists reconceives the world through the lens of Google Search by applying the same disciplines that mastered the web—datafication, indexing, browsing, and searching—to master reality itself. Without “ubiquitous sensate environments,” Paradiso writes, “the cognitive engines of this everywhere-enabled world are deaf, dumb, and blind, and can’t respond relevantly to the real-world events that they aim to augment.”13 In other words, ubiquitous computing is meaningless without the ubiquitous sensing that conveys the experience for computation.