Several decades ago, when I first moved to New York City, I answered an ad to be a personal assistant to a writer. I imagined myself as amanuensis, translating inspired pronouncements into poems. Instead, I ordered and returned sweaters, scheduled haircuts, and made three-course-meal seating plans for members of the literati whom I never got to meet. My boss, her money-manager husband, and their children lived on Park Avenue, in a penthouse with Georgian drapes and triple-insulated soundproof windows. She collected bespoke services: personal trainers, personal shoppers, a personal poetry trainer, a personal opera coach. I was one of four full-time staff, along with two live-in Irish nannies and a French maid. During our thirty-minute lunchtime, the four of us would hurry into the kitchen to use the small gold-handled faucet that produced instant boiling water to make tea and soup. We slurped and laughed and complained about our boss.

During one of these meals, the chief nanny began a call on the phone in the corner, then quickly slammed down the receiver. Pointing to its golden handle, she mouthed that she thought our boss was listening in. As we huddled over our soup, I said that our boss was always asking me for reports of what we talked about, and the nanny whispered that she was pretty sure she had seen her lurking outside the kitchen door. This was funny for a moment, and then not—a thin skein of anxiety started winding its way across the room.

A few weeks later, the maid was fired. It wasn’t clear that her dismissal was related to anything that had been said. But once paranoia gets its claws in you, it doesn’t let go easily. Our wages and raises were all unpredictable. Two of the staff relied on green cards. These circumstances, which had been the subject of so many conversations, suddenly became the source of insecurity. We gradually, then all at once, stopped having lunch together.

I have lately been thinking of that small discouraging experience as we live through an explosion of corporate investments in workplace surveillance. The year 1995, when I had that job, seems almost embarrassingly quaint, an era of surveillance innocence. There was no Facebook or Google following people everywhere they went, no spooky personalized ads. Back then, Americans spent an average of thirty minutes a month online, and 24/7 intimate surveillance was reserved for targets of FBI investigations.

At the dozen-plus places I had worked by the age of twenty-four, I punched in and out, sped up my dishwashing when the supervisor came through, weighed the beans I picked, bargained to get off early in exchange for cleaning extra bathrooms, and wrote reports for the third-grade teacher I assisted in the classroom. Even the tips I received while waitressing were my business, not the restaurants’. My bosses knew me superficially—my clothes, my general productivity—not what I thought or felt outside the workplace, unless I chose to share it.

As it happened, the 1980s and 1990s were a major turning point in surveillance, the period when companies went on their first buying sprees for electronic performance-monitoring. In 1987 approximately six million workers were watched in some kind of mediated way, generally a video camera or audio recorder; by 1994, roughly one in seven American workers, about 20 million, was being electronically tracked at work. The numbers steadily increased from there. When videotape technology was supplanted by digital devices that could scan multiple locations at once, the cameras first installed to protect businesses from theft shifted their insatiable gaze from the merchandise to the workers.

The second big turning point in electronic performance-monitoring is happening right now. It’s driven by wearable tech, artificial intelligence, and Covid. Corporations’ use of surveillance software increased by 50 percent in 2020, the first year of the pandemic, according to some estimates, and has continued to grow.

This new tracking technology is ubiquitous and intrusive. Companies track for security, for efficiency, and because they can. They inspect and preserve and analyze movements, conversations, social connections, and affect. If the first surveillance expansion was a territorial grab, asserting authority over the whole person at work, the second is like fracking the land. It is changing the structural composition of how humans relate to one another and to themselves.

Some long-haul truckers have to drive a fifty-foot flatbed truck six hundred miles a day with a video camera staring them down the entire time, watching their eyes, their knuckles, their twitches, their whistles, their neck movements. Imagine living in front of that nosy boss-face camera for months on end as it scans your cab, which serves as your home most of the time. On one of many angry Reddit forums about driver-facing cameras, a trucker wrote that he’d put up with one only “if the company owner gives me a 24/7 unrestricted stream in to his house.” “Those few hundred miles a day is the only time I completely have to myself and I feel as if it is being tainted,” added another. “I just want to pick my nose and scratch my balls in peace man.” A bus driver described the human desire to


pull a weird face or talk to yourself or sing along with a song….

I could feel how much less cortisol was flowing through my body in my second job where the buses were older and did not have cameras inside. It makes you unhealthy and run down.

Employers read employees’ e-mails, track their Internet use, and listen to their conversations. Nurses and warehouse workers are forced to wear ID badges, wristbands, or clothing with chips that track their movements, measuring steps and comparing them to coworkers’ and the steps taken yesterday.

The wristbands that now commonly encircle your skin, caressing your median nerve, might in the future be used to send signals back to you or your employer, measuring how many minutes you spend in the bathroom. Amazon, which minutely tracks every moment of a warehouse worker’s activity, every pause and conversation, has a patent for a wristband that would, the Times reported, “emit ultrasonic sound pulses and radio transmissions to track where an employee’s hands were in relation to inventory bins” and then vibrate to steer the worker toward the correct bin. A “SmartCap” used in trucking monitors brainwaves for weariness.

Off-the-shelf human resources software can monitor workers’ tone of voice. One major firm, Cogito, touts its product as “the AI-informed coach [that] augments humans through live, in-call voice analysis and feedback.” While workers are making fifteen dollars an hour fielding angry consumer complaints in a cubicle, they must pay heed to a pop-up screen that starts flashing if they talk too fast, if there is overlap between their voice and the customer’s voice, or if a pause is too long. “Empathy at scale,” the company boasts.

In one sense, intimately tracking behavior is old news: the business model of tech companies like Facebook and Google, after all, relies on tracking users on- and off-site. The commodification of data is in its third decade. But surveillance and automatic management at work are different. Workers can’t opt out without losing their jobs: you can’t turn off the camera in the truck if doing so goes against company policy; you can’t rip the recording device off your ID card. And worker surveillance comes with a powerful implicit threat: if the company notices too much fatigue, you might get overlooked for a promotion. If it overhears something it doesn’t like, you could get fired.

The political implications of ubiquitous employment surveillance are monumental. While bosses always listened in on worker conversations, they could only listen rarely—anything more was logistically impossible. Not now. Employees have to assume that everything they say can be recorded. What does it mean when all the words, and the tone of those words, might be replayed? Whispering has lost its power.

In many cases, worker surveillance is installed for ostensible safety reasons, like the thermal cameras installed to protect customers and coworkers from a worker who has a fever. But it is not, it turns out, good for our well-being. Electronic surveillance puts the body of the tracked person in a state of perpetual hypervigilance, which is particularly bad for health—and worse when accompanied by powerlessness. Employees who know they are being monitored can become anxious, worn down, extremely tense, and angry. Monitoring causes a release of stress chemicals and keeps them flowing, which can aggravate heart problems. It can lead to mood disturbances, hyperventilation, and depression. Business professors from Cornell and McMaster Universities recently conducted a survey of electronic monitoring in call centers and showed that the stress it caused was as great as the stress caused by abusive customers. Workers felt that monitoring was used for discipline, not improvement, and that the expectations were unreasonable and the use of monitoring unfair. They preferred a human boss to an ever-present robot spy with the power to affect their paychecks.

Is it any surprise that truckers’ mental health is suffering? Or that call center employees are breaking down? Truckers and call center workers report a kind of destabilizing fog, a constant layer of uncertainty and paranoia: which hand gesture, which bathroom break, which conversation was it that caused me to lose that bonus? “I know we’re on a job, but, I mean, I’m afraid to scratch my nose,” an Amazon driver told Insider for a story about the company’s driver-facing cameras. She didn’t share her name for fear of reprisal.

In 2011 Travis Kalanick, the founder of Uber, invited A-list Chicagoans to a party at the Elysian Hotel. On a large screen, he displayed what he initially called “God View” and later renamed “Heaven,” a map on which the company was able to track drivers, unbeknownst to them. The partygoers watched in amazement as hundreds of cars zipped around the city in real time, giddy in their perch on top of the world.


This anecdote, from Mike Isaac’s Super Pumped: The Battle for Uber, depicts Kalanick as elated, relishing the domination. More often we see him, and the company, exhibiting habitual paranoia, spying to protect the fortress. The book opens with Kalanick responding to regulatory resistance by hiring “ex-CIA, NSA, and FBI employees” to build a “high-functioning corporate espionage force” that “spied on government officials, looked deep into their personal lives, and at times followed them to their houses.” Once the corporate spies identified regulators who were trying to build a case that Uber was breaking local laws, Uber built code to make sure the regulators never actually matched with Uber drivers, and therefore could not investigate violations of local for-hire laws. Instead, Uber would serve up a mock model of the application with fake cars. The regulator would appear to match with a driver, but the driver would never appear. Uber called the program “Greyball.”

Isaac is a tech reporter at The New York Times who writes frequently about Silicon Valley. He broke the story on Greyball in the Times in 2017, two years before his book was published. His engrossing portrait of Uber follows the company from its early days to Kalanick’s final ejection, returning repeatedly to the ways in which Uber used surveillance to build power. The company tracked customers after they left their ride-share, mined credit card data to learn about competitors, and spied on drivers who drove for rival companies. It built a Strategic Services Group that used “virtual private networks, cheap laptops, and wireless hotspots paid for in cash.” Uber also impersonated drivers in private group chats to learn about rivals, took pictures of officials, had people followed, and recorded the private conversations of competitors.

Isaac demonstrates how Kalanick spent tens of millions of dollars on spying and related activities. In July, Mark MacGann, Uber’s former chief lobbyist in Europe, the Middle East, and Africa, leaked to The Guardian more than 124,000 documents that showed the extent of Uber’s disregard of law between 2013 and 2017 and how executives courted heads of state to build their empire.

In 2017, after a series of sexual harassment and workplace discrimination scandals, Kalanick was replaced as CEO by Dara Khosrowshahi, previously the head of Expedia. Khosrowshahi has changed the company in some ways since he took over, but the surveillance drivers face does not appear to have diminished.

For the period covered by Isaac’s book, Uber reportedly took 20 to 25 percent of the cost of the ride, plus add-ons. This year, Khosrowshahi launched a new system for paying drivers and pricing rides in some cities. The fare is based on “several factors”—including, according to the online investigative news site The Markup, “base fares, estimated trip length and duration, real-time demand at the destination, and surge pricing”; the pay for drivers is opaque and can vary. A driver shared screenshots of his payments with The Markup. One showed him getting fourteen dollars for a ride while Uber got thirteen dollars, another had him getting six dollars while Uber took nine dollars. No one knows why driver pay fluctuates. We do know that Uber tracks metrics down to the speed at which drivers brake, where the drivers go, their ratings, the rides they accept and cancel, how long they take to get somewhere—it seems likely that pay is connected to all this data.

Other companies, like the food delivery platforms DoorDash and Instacart, have been doing something similar, using opaque systems to dole out personalized payment. Instacart used to pay delivery workers a base amount, but now the pay decisions are a black box. Workers worry that the company is using all it knows in order to pay them as little as possible. But they can’t prove anything.

All of this is demoralizing and dystopian, but what does it have to do with democracy? Elizabeth Anderson’s lively and persuasive 2017 book, Private Government, offers a partial answer. Anderson, a political philosopher at the University of Michigan, shakes the reader by the shoulders to get us out of the strange rigidity that pervades public discussion of government. Employment is a form of government, she argues, one that is far more relevant and immediate for most people than the Washington, D.C., kind.

A powerful company like Amazon, for instance, sets its own terms of employment—and in so doing impacts those of UPS drivers and the broader logistics industry. Private employers with industry-wide influence have coercive power—what Anderson calls governing power. Private government, personified by private guilds or by state-sanctioned economic monopolies in soap, salt, and leather, was the central target of intellectuals and activists like John Locke and the Levellers. Anderson sees in Locke, Adam Smith, and others a belief that the arbitrary power to debase and discipline is a threat to a free society, wherever it appears, and that public, accountable government should protect against private tyranny.

Many modern “thinkers and politicians,” she argues, are “like those patients who cannot perceive one-half of their bodies”: they “cannot perceive half of the economy: they cannot perceive the half that takes place beyond the market, after the employment contract is accepted.” As a result, companies are generally treated as wholly private.

Many private-sector workers, Anderson writes, live under

dictatorships in their work lives. Usually, those dictatorships have the legal authority to regulate workers’ off-hour lives as well—their political activities, speech, choice of sexual partner, use of recreational drugs, alcohol, smoking, and exercise.

For her, service workers who clock out, or technicians and realtors and cooks who seem endowed with substantial freedoms, are burdened by a legal system that allows corporations to fire a worker based on off-the-clock activity.

The speech rights of workers are practically nonexistent except as they explicitly relate to labor organizing, which, Anderson argues, is effectively a dead letter these days because of the difficulty of enforcement and the fear of challenging the boss’s tactics.

How did things get so bad? Anderson believes the root issues that enabled the current dystopian workplace go back generations. When the Industrial Revolution shifted the “primary site of paid work from the household to the factory,” it imported the long tradition of wholly arbitrary power within the household, in which children did not have freedom vis-à-vis their parents, and wives had limited freedom vis-à-vis their spouses. The Industrial Revolution could have provided an escape from the private tyrannies of home life, but instead it replicated them.

During the heyday of the Ford Motor Company, its Sociological Department began inspecting workers’ homes. Anderson writes:

Workers were eligible for Ford’s famous $5 daily wage only if they kept their homes clean, ate diets deemed healthy, abstained from drinking, used the bathtub appropriately, did not take in boarders, avoided spending too much on foreign relatives, and were assimilated to American cultural norms.

Anderson points out that while Apple does not visit people’s homes today, it does require retail workers to open their bags for inspections before coming into work. We take this for granted, she notes, but should we? Nearly half of Americans have undergone a suspicionless drug test. And many workers have no protection from getting fired for what they say on social media. To those who claim the workplace isn’t government because you can quit, Anderson retorts, “This is like saying Mussolini wasn’t a dictator, because Italians could emigrate.”

Anderson isn’t focused on surveillance, but her work suggests two things. First, that to address the constant spying, we should focus on power, not just the technology. Labor rights and antitrust enforcement must be first-level responses to the current—and worsening—structures of power. Second, we should treat employer surveillance as we do any governmental surveillance—in other words, with deep suspicion. It is a truism that governmental surveillance chills speech and debate and erodes the public sphere; once we can perceive the workplace as a site of government, we can perhaps build a political movement for greater freedom in the places where working Americans spend most of their waking hours.

To make sense of the reality we are in, we need to be able to talk to one another without fear of our conversations being used against us. The private conversations among workers—and friendships, debates, questions—are part of the cohesion and connection that enables not just labor organizing but public life. When everything we say is being listened to—especially by a smaller and more powerful cadre of employers—it can become easier not to speak. This is not unlike the political totalitarianism that Hannah Arendt warned against, where the state aims to disintegrate both the private and the public by submerging the private into the public and then controlling the public. The logical conclusion of workplace surveillance is that the private sphere ceases to exist at home because it ceases to exist at work, where visibility into the worker’s life is unrestrained.

Three years ago, while I was reporting my book about monopolies and how they act as private governments, I spoke to chicken farmers who were paid different prices every month by big poultry distributors. One farmer stands out in my memory. A loving, angry, and dispirited man, he described what it was like to examine a month of payment from his poultry distributor and not know whether the pay reflected fair competition with other farmers, retaliation for speaking out, or evidence that he was part of an experiment. He recounted fellow farmers’ confessing they got so angry they wanted to murder the distributors.

This system of pay is called the tournament system. The farmers compete to be the most productive; theoretically, they are paid based on how productive they are relative to other farmers. However, there’s no accountability or fact-checking mechanism: the distributor keeps all the data, and when it passes out the paychecks, the farmers, wholly reliant on the distributor to stay solvent, have to accept that it is being honest.

A similar structure—a tournament—is the blueprint for how Amazon exerts power over its counterparties, be they governments, sellers, or workers. In Brad Stone’s second book about Amazon’s corporate growth, Amazon Unbound, he covers the company’s enormous expansion over the past decade and the growth of its political power.*

Jeff Bezos, in Stone’s telling, was deeply involved in all aspects of HR, both in the corporate offices and in the warehouses. He embraced a reimbursement and promotion system called “stack ranking,” in which midlevel managers ranked their employees and fired the lowest-ranked ones. Managers had quotas of how many people they needed to fire and were expected to rank them to get there. After a front-page story in the Times about how the company culture pitted workers against one another, the practice was ended.

But the philosophy—force people to fight for scraps, kick out the lowest performers—keeps reappearing in different parts of the company. When Amazon was relying on contractors to deliver packages, it developed an app called “Rabbit” that tracked the delivery. The Rabbit team watched drivers, Stone writes, “skip meals, rush through stop signs, and tape their phones to their pant legs so they could easily glance down at the screens, all to meet the challenging delivery deadlines.” Those who didn’t meet them were fired. When Amazon decided it wanted to build a new headquarters, it announced a tournament to determine the location—getting competitive data on 238 different cities for free in the process.

According to Stone, a tech reporter at Bloomberg News, Bezos was furious when Amazon’s head of operations tried to get the company to incorporate the “Lean” approach from Toyota, in which workers developed trust and relationships with their managers with the goal of long-term employment. When the HR deputy from the same department presented a paper called “Respect for People,” Stone reports, “Bezos hated it. He not only railed against it in the meeting but called [the deputy] the following morning to continue the browbeating.” Instead of a stable workforce, he wanted warehouse workers to stay for a maximum of three years, unless they got a new job internally. He severely limited raises after three years.

For the warehouse workers, the company sets extraordinary demands: it bans talking, tracks everything, fires workers who fail to meet their quotas, and expects that conditions are bad enough that workers will quit. Pre-pandemic, the Times reported, “the turnover among its work force was roughly 150 percent a year.”

“You spend 10 hours on foot, there’s no windows in the place, and you’re not allowed to talk to people—there’s no interactions allowed,” one worker told Vox for a story on the growing number of 911 calls from Amazon warehouses. “I got a sense in no time at all that they work people to death, or until they get too tired to keep working.”

“It’s one of the big reasons people want to unionize,” Chris Smalls, the leader of the Amazon Labor Union, which organized a Staten Island warehouse this year, told The Washington Post last December. “Who wants to be surveilled all day? It’s not prison. It’s work.”

It may be tempting to see the Amazon surveillance as purely a warehouse problem, and surveillance-driven variable pay as purely a gig problem, but employers face no legal limit to incorporating new types of variable pay into formal employment—and abuses faced by independent contractors are merging with those faced by formal employees. This is one of the core arguments of Your Boss Is an Algorithm by Antonio Aloisi and Valerio De Stefano, both European law professors. Gig work, where the worker is weakest, is a site of experimentation in new managerial techniques. Those experiments become the proving ground for strategies that are then brought into other forms of employment.

The future, Aloisi and De Stefano argue, is in combining the tracking and rewarding tools from gig work with employment contracts that allow for changing pay. The existing toolkit is vast:

Activtrack inspects the programs used and tells bosses if an employee is unfocussed, spending time on social media. OccupEye records when and for how long someone is away from their workstation. TimeDoctor and Teramind keep track of every task conducted online. Similarly, Interguard compiles a minute-by-minute timeline that monitors all data such as web history and bandwidth utilisation and sends a notification to the managers if workers pick up anything suspicious. HubStaff and Sneek routinely take snapshots of employees through their webcams every five minutes or so to generate a timecard and circulate them to boost morale. Pragli synchronises professional calendars and music playlists to create a sense of community; it also features a facial recognition that could display a worker’s real-world emotion on their virtual avatar’s face.

Right now, there may be limited proof that these tools are used to vary pay in traditional workplaces. But the authors argue that these technical tools are not hard to combine with legal innovations in work contracts. Contracts that allow for adjusted wages can easily bring many of the conditions of gig work to traditional employment. Corporations may soon jettison the fixed-wage model that has been a feature of blue collar employment for decades.

It is no coincidence that routine work surveillance followed closely on the heels of the Reagan antitrust revolution and the collapse of private sector unionization. Nothing except unionization or new laws would stop an employer from taking all the data it is gathering from sensors and recordings and using them to more precisely adjust wages, until each worker gets the lowest wage at which they are willing to work, and all workers live in fear of retaliation. This is no more sci-fi than Facebook and Google serving users individualized content and ads designed to keep us on their services as long as possible, allowing them to sell as many ads as possible.

The bespoke clothing that my Park Avenue boss wore was a mark of privilege, a step above mass manufacture—suits made to fit her individual body, shoes tailored to the grooves and arches of her feet. The modern promise of tech personalization, building on a romanticized notion of individuality and authenticity, is that we can all live in similarly tailor-made worlds, with newsfeeds adjusted to our preferences and professional and leisure interests. You may be one of few listeners who loves both Kenny Rogers and the Cure, but Spotify knows you, and can bring you songs that speak to your singular soul.

But extending this tailor-made ethos is exquisitely unromantic: these eyes may have the intimacy and memory of a lover, but they lack all affection. Modern surveillance technology means that tailor-made wages are coming for all workplaces. The mass-produced, nonunionized depressed wages of the late twentieth century were already alarming, but the new, specially commissioned AI wages of the twenty-first century enable a new level of authoritarianism. To stop it, we’ll have to outlaw particular forms of spying, and use antimonopoly and labor laws to restructure power.

Tracking technology may be marketed as tools to protect people, but will end up being used to identify with precision how little each worker is willing to make. It will be used to depress wages and also kill the camaraderie that precedes unionization by making it harder to connect with other workers, poisoning the community that enables democratic debate. It will be used to disrupt solidarity by paying workers differently. And it will lead to anxiety and fear permeating more workplaces, as the fog of not knowing why you got a bonus or demotion shapes the day.

This matters because work is not an afterthought for democratic society; the relationships built at work are an essential building block. With wholly atomized workers, discouraged from connecting with one another but forced to offer a full, private portrait of themselves to their bosses, I cannot imagine a democracy.