For a long time, a certain set of assumptions dominated our digital imagination. These assumptions should be familiar enough. Information wants to be free. Anything that connects people is good. The government is bad. The internet is another world, where the old rules don’t apply. The internet is a place of individual freedom, which is above all the freedom to express oneself.
Such ideas were never 100 percent hegemonic, of course. They were always contested, with varying degrees of success. Governments, for one, found several ways to assert their sovereignty over online spaces. Scholars sounded the alarm on the rise of the white supremacist web — the notorious neo-Nazi site Stormfront launched in 1996 — and presciently observed that the internet’s connectivity could also make the world worse.
Even so, these assumptions and the intellectual traditions they emanated from — techno-utopianism, cyberlibertarianism, the Californian Ideology — largely kept their grip on the common sense. The long 1990s is said to have begun with the fall of the Berlin Wall in 1989 and ended with the attacks of September 11, 2001. But, when it came to our popular discourse about the internet, the long 1990s lasted a lot longer.
Then came Snowden. In 2013, the former NSA contractor revealed that the internet was a vast spy machine for the American security state. A tremor of tech pessimism crept into public consciousness. Then came Trump. The media’s failure to anticipate the possibility of his victory in 2016 led it to amplify the significance of Russian influence operations via social media — operations that clearly existed, but which, at a moment of supreme disorientation, metastasized into the deus ex machina that could explain an inexplicable result. Yet this coping mechanism had a silver lining: it provided the initial spark for what has come to be known as the “techlash.”
Journalists and politicians began to pay closer, less credulous attention to the internet and the companies that control it. Disinformation remained a key concern, but far from the only one: a long series of tech scandals have fed the fire, too many to keep count. The right has also joined the fray: the (laughable) notion that the big platforms silence conservative voices has taken root in the reactionary mind, turning a range of right-wing figures into harsh critics of Silicon Valley.
The resulting shift is stark. A sharper tone prevails in the New York Times and on Fox News, in statehouses and on Capitol Hill. Criticisms once confined to scholarly circles, or to more oppositional outlets like The Baffler and Valleywag, have become conventional, even banal. One could be uncharitable about the heavy Kool-Aid drinkers who abruptly sobered up — there is no shortage of annoying figures among the late converts to tech critique — but the techlash has been a very good thing. We are at last having a more honest conversation about the internet. The long 1990s are over. The old gods are finally dead.
Who are the new gods? This is what makes our moment so interesting: the conventional wisdom is cracking up but its replacement hasn’t quite consolidated. As James Bridle says, something is wrong on the internet — and something is wrong with the way we have thought about the internet — but there is not yet a widely accepted set of answers to the all-important questions of why these things are wrong, or how to make them right.
Different camps are now competing to provide those answers. They are competing to tell a new story about the internet, one that can explain the origins of our present crisis and offer a roadmap for moving past it. Some talk about monopoly and antitrust. Others emphasize privacy and consent. Shoshana Zuboff proposes the term “surveillance capitalism” to describe the new kinds of for-profit monitoring and manipulation that the internet and associated technologies have made possible.
These analyses have important differences. But they tend to share a liberal understanding of capitalism as a basically beneficent system, if one that occasionally needs state intervention to mitigate its excesses. They also tend to equate capitalism with markets. Sometimes these markets become too consolidated and need to be made more competitive (the antitrust view); sometimes market actors violate the terms of fair exchange and need to be restrained (Zuboff’s view). But two articles of faith always remain. The first is that capitalism is more or less compatible with people’s desire for dignity and self-determination (or can be made so with proper regulation). The second is that capitalism is more or less the same thing as markets.
What if neither belief is true? This is the starting point for building a better story about the internet.
The Archipelago and the Network
If capitalism isn’t (only) markets, then what is it?
There have always been markets. Capitalism, by contrast, is relatively new. Its laws of motion first emerged in Europe in the fifteenth and sixteenth centuries, and reached escape velocity with industrialization in the eighteenth and nineteenth.
If capitalism didn’t invent markets, however, it did make markets much more important. The historian Robert Brenner observes that capitalism is defined above all by market dependence. Pre-capitalist peasants can trade and barter, but they don’t depend on the market for life’s necessities: they grow their own food. In capitalist societies, on the other hand, the market mediates your access to the means of subsistence. You must buy what you need to survive, and to have the money to do so, you must sell your labor power for a wage.
Market dependence doesn’t exist for its own sake. It serves an important function: to facilitate accumulation. Accumulation is the aim of any capitalist arrangement: to take a sum of value and make more value out of it. While markets are certainly central to capitalism, they aren’t what makes it tick. Accumulation is. To put it in a more Marxist idiom, capital is value in motion. As it moves, it expands. Capitalism, then, is a way to organize human societies for the purpose of making capital move.
There are a few different methods for making capital move. The principal one is for capitalists to purchase people’s labor power, use it to create new value in the form of commodities, and then realize that value as profit by selling those commodities. A portion of the proceeds are reinvested into expanding production, so even more commodities can be made at lower cost, thus enabling our capitalist to compete effectively with the other capitalists selling the same commodities.
This may seem entirely obvious, but it’s actually a very distinctive way of doing things. In other modes of social organization, the point of production is to directly fulfill people’s needs: think of subsistence farmers, growing food for their families to eat. Or the point is to make the rulers rich: think of the slaves of ancient Rome, doing the dirty work so that imperial elites could lead lives of luxury.
What makes capitalism so unusual is that production (and accumulation) isn’t for anything exactly, aside from making it possible to produce (and accumulate) more. This obsession gives capitalism its extraordinary dynamism, and its revolutionary force. It utterly transforms how humans live and, above all, how they produce. Capitalism forces people to produce together, in increasingly complex combinations of labor. Production is no longer solitary, but social.
This dynamic is most vividly illustrated by the factory. The modern factory was largely born in nineteenth-century Manchester, where Friedrich Engels’s father co-owned a cotton mill. This gave the young Engels the opportunity to observe the birth of the factory up close. He saw hundreds, even thousands of workers, crammed into vast buildings, arrayed around machines, and performing different roles within a complex division of labor in order to work as one. What they made, they made together.
In pre-capitalist Europe, one person or a few people could plausibly claim credit for producing something. This wasn’t the case in the capitalist factory, however. “The yarn, the cloth, the metal articles that now come out of the factory were the joint product of many workers, through whose hands they had successively to pass before they were ready,” Engels wrote. “No one person could say of them: ‘I made that; this is my product.’”
Yet there was a contradiction lurking here. If no one worker could claim sole credit for a product, the owner of the factory could still claim sole ownership of everything the workers made together. Wealth was being created socially, on a new model — but still owned privately, on the old model.
The contradiction became even sharper when zooming out to consider the wider economy. As many workers as it took to run a Manchester mill, it took even more workers to make that work possible, from the machinists who manufactured and maintained the power looms and the other machines to the slaves in the American South who picked the cotton that kept those machines fed. The collective labor inside the mill was sustained by many concentric circles of collective labor outside of it.
The pre-capitalist economy looked like a cluster of islands — an archipelago. It involved a collection of small producers relatively isolated from one another and producing mostly for personal use. (Marx memorably compared the French peasantry to a sack of potatoes.) By contrast, the capitalist economy looked like a network. The network of capital concentrated masses of people into larger nodes of production and linked them through countless threads of interdependence. Yet the wealth that this network generated didn’t flow to the many workers who collectively created that wealth. It flowed to the few who owned the network: the capitalists.
Before capitalism, when production happened on a more personal basis, such an arrangement might’ve made sense. If the economy was a cluster of islands, it followed that each island would own what it made. But capitalism, by revolutionizing production, introduced a contradiction: wealth was now made as a network, but still owned as an archipelago. Capitalists like Engels’ father became rich. The workers of Manchester earned starvation wages, and lived in cholera-infested slums.
The New Manchester
What does this have to do with the internet?
The internet, and the constellation of digital technologies that we call “tech” more broadly, intensifies the fundamental contradiction in capitalism between wealth being collectively produced and privately owned. It takes the Manchester model and elevates it to the nth degree. It makes the creation of wealth more collective than ever before, piling up vast new fortunes in the process — fortunes that, as they did in Engels’ day, accrue to a small handful of owners.
A worker in a Manchester mill couldn’t point to a finished piece of yarn and say, “I made that,” but a few thousand workers (and slaves) probably could. Tech’s wealth, on the other hand, is woven out of the contributions of billions of people, living and dead.
This helps explain why the tech industry is so ludicrously profitable. Take Facebook. In 2018, Facebook reported a net income of $22 billion with an operating margin of 45 percent. The company only has about 40,000 full-time employees, along with an undisclosed number of contractors. In other words, relative to its costs, Facebook makes an absurd amount of money. And Facebook’s power isn’t just about money: as the dominant media ecosystem in many countries around the world, it also embodies what Frank Pasquale calls “functional sovereignty.” It operates like a government — which is particularly evident in the case of Libra, its new global digital currency. And this government is quite explicitly autocratic given a shareholder structure that preserves Mark Zuckerberg’s personal control of the company.
It’s hard to imagine a more extreme form of the contradiction on display at Manchester than a social network of more than two billion people ruled by a single billionaire. The network of capital has become denser, and more literal, than Engels could’ve possibly imagined, while its control has become concentrated in even fewer hands.
To observe that Facebook has relatively few workers is not to suggest that the work they perform is not important. Without content moderators, data center technicians, site reliability engineers, and others, Facebook’s product would become unusable and its business would collapse. But their collective labor, like that of the workers within Engels’ father’s factory, depends on many concentric circles of collective labor outside of it. And, for Facebook and the other firms that fall under the umbrella of tech, the share of value supplied by these external layers is especially vast.
One source is the workers who invented the software, hardware, protocols, and programming languages that laid the basis for today’s tech industry. These were developed over the course of several decades, starting with the creation of the first modern electronic computers in the 1940s, and relied heavily, often exclusively, on US military funding. Another source is the workers who, in the present day, continue to make and maintain the stuff on which tech profits depend. While this work takes many forms, most of it is menial or dangerous. It includes manufacturing smartphones, mining rare earth elements, and labeling training data for machine learning models.
As varied as these jobs are, though, they still look like traditional labor. People work and get paid, whether they’re inventing the internet protocols or laying fiber-optic cable. Tech, however, also manages to draw value from activities that don’t look like traditional labor. To return to Facebook, those more than two billion users create value for the company by supplying the site with its posts, comments, and likes. This content, paired with the rest of their activity on the platform, also furnishes Facebook with the personal data it uses to sell targeted advertising, which makes up the vast majority of its revenue.
It’s a contested theoretical question whether all this posting and clicking should count as “labor” — and if so, what kind. In her canonical article on the subject, the theorist Tiziana Terranova uses the term “free labor” to describe the various unwaged activities that propped up profit-making in the early days of the commercial internet, from volunteer moderators on America Online to open-source software developers. But the scope of such activities has grown dramatically since Terranova published her piece in 2000, and they look less and less like labor. Increasingly, tech is able to harvest value from us simply for existing.
A good example comes in the form of a cafe in San Francisco called Brainwash. This cafe, since closed, had a camera inside of it that filmed customers. A group of researchers obtained the footage, and turned it into a dataset to train machine learning models for detecting heads and faces. Published in 2016, this dataset was then used by the Chinese firm Megvii, a global leader in facial recognition, to refine its own software. Megvii also happens to be implicated in the totalitarian surveillance state that the Chinese Communist Party is constructing in the western province of Xinjiang. In other words, by walking into a cafe one day in San Francisco, you may have helped a tech company make money by selling the Chinese government a product it uses to repress millions of its citizens some six thousand miles away. (Megvii is currently valued at $4 billion, and hopes to raise as much as $1 billion in an IPO expected for late 2019.)
These kinds of strange and tangled value chains will only become more common in the coming years. As small networked computers burrow ever more deeply into our homes, stores, streets, and workplaces, more data will be made. Meanwhile, advances in machine learning and the growth of cloud-based processing power will continue to make data more valuable, as the fuel that feeds automated systems for everything from recognizing faces to predicting consumer preferences.
The upshot is a world where the creation of wealth becomes more collective than ever before. In the nineteenth century, Engels reflected on how capitalism transformed production “from a series of individual into a series of social acts.” The total enclosure of our world by computing means that those social acts can now happen at the scale of entire societies. The industrial factory has become what Terranova and others, building on a term from Italian autonomism, call the “social factory.” The new Manchester is everywhere.
The Difference Engine
Capitalism connects. In its perpetual push to accumulate, it draws people into new sites and circuits of collective wealth-making. But if capitalism is a connector, it is also a differentiator. If capitalism is a network for making wealth, it is also an engine for making difference.
To watch this differentiating dynamic at work, let’s return to Manchester for a moment. The people who collectively created the city’s wealth were not a single homogenous mass. Quite the opposite: they were divided into men and women, English and Irish, white and Black. And these divisions were constantly being reinforced, since they served a valuable purpose: they helped make exploitation seem justified, even natural.
Thus it was natural for the Irish to be paid less and live in appalling slums. It was natural for women to be paid less while also performing the unpaid work of raising children — children who went into the mills as young as five. It was natural to enslave human beings of African origin and put them to work harvesting the cotton that those mills turned into textiles. It was natural to dispossess and exterminate the Indigenous people who had formerly inhabited the land that became those cotton fields.
Capitalism doesn’t invent human difference, of course. Humans look different; they speak different languages; they come from different communities and cultures. But capitalism makes these differences make more of a difference to people’s lives. Differences become more differential. They become differences of capacity and value — differences in how much a human being is worth, or if they’re even considered human at all.
The political scientist Cedric J. Robinson argued that this difference-making has been a core feature of capitalism from the beginning — he called it racial capitalism for this reason. Feudal Europe was highly racialized, Robinson said. As Europeans conquered and colonized one another, they came up with ideas about racial difference in order to justify why, for instance, Slavs should be slaves. (In fact, Slavs were so frequently enslaved in the Middle Ages that they supplied the source of the word “slave,” in English and several other European languages.)
If racial thinking saturated the societies where capitalism first emerged, capitalism subsequently picked up these concepts and extended them. It generated deeper and more varied ideas about racial difference in order to justify the new relationships of domination that the imperative of accumulation demanded — particularly as Europeans began carving up Asia, Africa, and the Americas. “The tendency of European civilization through capitalism,” Robinson wrote, “was thus not to homogenize but to differentiate — to exaggerate regional, subcultural, and dialectical differences into ‘racial’ ones.”
Robinson’s insight helps clarify another crucial aspect of how tech operates. If tech intensifies capitalism’s contradiction between wealth being collectively produced and privately owned, it also intensifies capitalism’s tendency to slice people into different groups and assign them different capacities and values. Indeed, the two operations are closely related. “Capital can only be capital when it is accumulating,” says the theorist Jodi Melamed, “and it can only accumulate by producing and moving through relations of severe inequality among human groups.” The network for making wealth, in other words, relies on the engine for making difference.
That engine is now made of software. Differentiation happens at an algorithmic level. The abundant data that flows from mass digitization, combined with the ability of machine learning algorithms to find patterns in that data, has given capitalism vastly more powerful tools for segmenting and sorting humanity.
Way back in 1993, the media scholar Oscar H. Gandy, Jr. offered an extremely prescient view of how this works. He called it “the panoptic sort,” in a book of the same name. “The panoptic sort is a difference machine that sorts individuals into categories and classes on the basis of routine measurements,” he wrote. “It is a discriminatory technology that allocates options and opportunities on the basis of those measures and the administrative models that they inform.”
Gandy was looking at how corporations and governments collected and processed personal information at a time when computing was widespread, but fairly primitive by today’s standards — the commercial internet was still years away. Even so, Gandy discerned a logic that by now feels very familiar. Data was being drawn from many sources — thus the “panoptic” part — in order to sort people “according to their presumed economic or political value.” And this operation wasn’t peripheral or incidental to capitalism, but absolutely integral to it: the panoptic sort, Gandy argued, was “the all-seeing eye of the difference machine that guides the global capitalist system.”
Today, this all-seeing eye sees much, much more. And the stakes of the sorting are even higher. Algorithmic differentiation helps determine who gets a loan, who gets a job, who goes to jail. Moreover, Gandy observed how the panoptic sort amplified existing disparities, racial and otherwise. This is far truer today, thanks to the mainstreaming of machine learning systems.
In recent years, scholars and journalists have called attention to the problem of “algorithmic bias.” Such bias is endemic to machine learning because it “learns” by training on data drawn from our social world — data that inevitably reflects centuries of capitalist difference-making. Thus “predictive policing” algorithms trained on data that shows that the police arrest a lot of Black people suggest arresting more Black people. Or an Amazon algorithm trained on the resumes of its mostly male workforce advises against hiring women.
The role of these systems is not just to reproduce inequalities, but to naturalize them. Capitalist difference-making has always required a substantial amount of ideological labor to sustain it. For hundreds of years, philosophers and priests and scientists and statesmen have had to keep saying, over and over, that some people really are less human than others — that some people deserve to have their land taken, or their freedom, or their bodies ruled over or used up, or their lives or labor devalued. These ideas do not sprout and spread spontaneously. They must be very deliberately transmitted over time and space, across generations and continents. They must be taught in schools and churches, embodied in laws and practices, enforced in the home and on the street.
It takes a lot of work. Machine learning systems help automate that work. They leverage the supposed authority and neutrality of computers to make the differences generated by capitalism look like differences generated by nature. Because a computer is saying that Black people commit more crime or that women can’t be software engineers, it must be true. To paraphrase one right-wing commentator, algorithms are just math, and math can’t be racist. Thus machine learning comes to automate not only the production of inequality but its rationalization.
The New Barcelona
Anything that moves has an ideal medium for its motion. A fish moves best in water; a car moves best on pavement. Capital is value in motion, so it must always be moving. And it moves best through a particular kind of social fabric, one that is both webbed and fissured, linked and sliced, connected and differentiated.
This helps make sense of what we call tech. Tech is an agent and accelerant of these dynamics, of “densely connected social separateness,” to borrow a term from Melamed. This explains its tendency to generate immense imbalances of wealth and power, and to heighten the hierarchical sorting of human beings according to race, gender, and other categories.
For our analysis to be useful, though, it needs to have not only a descriptive but a prescriptive element. It needs to offer some answers to the question of what is to be done.
This is where things get murkier, as one might expect. But there is clarity on at least one point. If tech intensifies the contradiction between wealth being made by the many and owned by the few, then the obvious solution is to resolve the contradiction: to turn socially made wealth into socially owned wealth. Or, as Marx and Engels put it in The Communist Manifesto, to convert the “collective product” of capital into “common property, into the property of all members of society.”
The logic is appealingly simple: if the network makes the wealth, then let the network own the wealth. But how, precisely? What does it mean to transform the wealth that society makes in common into the common property of society? This is the most bitterly debated question in the whole history of the radical Left. For most of the actually existing socialisms of the twentieth century, the answer was full nationalization on the Soviet model. This answer hasn’t aged well.
Another approach, and one that is currently enjoying renewed popularity, draws from the tradition of worker self-management. This tradition comes in many flavors, but perhaps its most heroic moment occurred in revolutionary Catalonia during the Spanish Civil War, when people seized factories, farms, even flower shops and, for a brief period, ran everything themselves. A young Marxist from Kentucky named Lois Orr would later remember the thrill of strolling through anarchist Barcelona and seeing its “cafés, restaurants, hotels, and theaters lit up red or red and black [with] banners saying Confiscated, Collectivized.”
Barcelona, then, is one alternative to Manchester. But what would self-management mean for tech? A number of different experiments offer preliminary materials towards an answer. There are small, cooperatively owned platforms for everything from ride-hailing to social media. There are municipally owned broadband networks governed by local communities. There is an initiative to create a socially owned smart city in, of all places, Barcelona. There are also more ambitious but less immediately feasible schemes for democratizing the big platforms, whether by converting them into cooperatives of some kind or socializing their data.
These projects and proposals have the virtue of being concrete. As working hypotheses, they are immensely valuable. But they remain necessarily incomplete and provisional, particularly when considered as possible directions for moving beyond capitalism. Cooperatives under capitalism often behave like normal firms, since they are subject to the same market imperatives as everyone else. There is no straight line, then, from experiments in self-management to the broader goal of breaking with the logic of infinite accumulation and rebuilding society on a radically different basis.
Neither is there a direct relationship between democratizing ownership and combating the various oppressions implicated in capitalist difference-making. A cooperatively owned platform wouldn’t put an end to algorithmic racism, for instance. This brings us to another important point: sometimes the most emancipatory option isn’t to transform how infrastructures are owned and organized, but to dismantle them entirely.
Thinking in Motion
Consider the Stop LAPD Spying Coalition, an alliance that has been organizing against police surveillance in Los Angeles for years. They have successfully pushed the LAPD to abandon two predictive policing programs — programs that led to increased police violence against working-class communities of color. The organizers did not want these programs reformed, but stopped. They were not demanding that the ownership of the algorithmic policing apparatus be “democratized,” whatever that might mean, but abolished.
Here is an organization that is taking on tech’s tendency to intensify capitalist difference-making, and using the framework of abolition to do so. One can see a similar approach in the emerging movement against facial recognition, as some city governments ban public agencies from using the software. Such campaigns are guided by the belief that certain technologies are too dangerous to exist. They suggest that one solution to what Gandy called the “panoptic sort” is to smash the tools that enable such sorting to take place.
We might call this the Luddite option, and it’s an essential component of any democratic future. The historian David F. Noble once wrote about the importance of perceiving technology “in the present tense.” He praised the Luddites for this reason: the Luddites destroyed textile machinery in nineteenth-century England because they recognized the threat that it posed to their livelihood. They didn’t buy into the gospel of technological progress that instructed them to patiently await a better future; rather, they saw what certain technologies were doing to them in the present tense, and took action to stop them. They weren’t against technology in the abstract. They were against the relationships of domination that particular technologies enacted. By dismantling those technologies, they also dismantled those relationships — and forced the creation of new ones, from below.
Machine-breaking is often a good idea; for more ideas, we can turn to other movements. Tech workers are taking collective action against contracts with the Pentagon and ICE, and demanding an end to gendered discrimination and harassment. Gig workers for platforms like Uber are organizing for better wages, benefits, and working conditions. Within these movements we can find more useful materials to think with, materials that might disclose the contours of a society organized along different lines.
The intellectual is not the only one who thinks. Masses of people in motion also think. And it is the thinking of these two together, in the creativity that results from their continuous interaction, that furnishes the form and content of anything worth calling socialism. This process is messy and circuitous, with many blind alleys and false starts. It involves more time spent moving contradictions around, and creating new ones, than resolving them. But it is the only path to a future where capital’s motion finally grinds to a halt, and a different set of considerations — human need, a habitable planet — comes to coordinate our common life. This is how the Left will answer the question of what is to be done, about tech and about everything else: by thinking en masse and thinking in motion, while traversing difficult terrain.