Deep in the aisles of a vast Amazon fulfillment center, one aggrieved young worker mounted a very tiny rebellion. A comic book fan, he would squirrel away intriguing titles as they arrived at the warehouse, stealing glances as he stocked the shelves. After finishing a book, he would hide it in plain sight. He wouldn’t scan the barcode on the book or its corresponding shelf, so it was forever lost to Amazon’s electronic inventory system. Only he knew its location. It wasn’t exactly political sabotage, says Alessandro Delfanti, a University of Toronto communications professor who heard the story while interviewing warehouse workers for a book about Amazon. “It was more of a tiny little revenge, a tiny little way to reappropriate a tiny bit of time.”
You might call such covert micro-mutinies an American pastime. Martin Sprouse’s 1992 book Sabotage in the American Workplace: Anecdotes of Dissatisfaction, Mischief and Revenge features hundreds of similar tales. There’s the pickle packer who privately pitched pickles into the pickle plant’s conveyor belt until it popped. There’s the disgruntled reporter (unimaginable!) who addressed his editor’s demands for brevity by penning the headline “DEAD,” followed by the story, “That’s what Harry Serbronski was after his car hit a telephone pole at eighty-six miles an hour.” Continuing this grand tradition, Amazon workers have set the home screens of their company-issued devices to photos of Jeff Bezos howling maniacally, or scrawled “unionize” on the dust-caked window of delivery vans. Private, perhaps momentarily satisfying mini-revolts in an increasingly automatized, surveilled, lopsided world.
Listen to enough Amazon workers, and you will hear the refrain “We are not robots.” While the company calls its warehouse associates the “heart and soul” of its operations, many workers say they feel like cogs, inhuman appendages of a machine at best. At worst, they become kinks in the system, when their flesh-and-blood functions—fatigue, the wear or tear of a ligament, the call of nature—impede their ability to keep pace with robots. This refrain has grown louder over the years, culminating with the union fight in Bessemer, Alabama.
By Friday morning the union was trailing badly in the vote count, with “no” votes outnumbering “yes” votes more than two to one. Some 500 votes remain disputed, mostly by Amazon, but there are too few to close the deficit. The results are a blow to organizers and the Retail, Wholesale and Department Store Union, which was hoping to represent workers, but the election still represents a milestone—Bessemer is the first US facility to reach this stage in a system that heavily favors employers. The RWDSU announced Friday that it plans to file charges against Amazon for allegedly violating labor law, which could throw the results into question.
Meanwhile, a wave of innovations has been putting the squeeze on workers of all types, tracking them in ever-more sophisticated ways, pushing them to perform at increasingly robotic rhythms. It seems to be working: US productivity grew nearly 70 percent over the last four decades. That’s more than six times the rate of wages, owing partly to the erosion of collective bargaining. Since 1979, the US union membership rate has nosedived from 27 to 11 percent.
The impulse to wring the maximum amount of value from workers at the least amount of cost is nothing new, of course. In the 1880s, an industrial engineer named Frederick Winslow Taylor dreamt up a new form of management consulting, later dubbed “scientific management.” Applying engineering principles to industrial labor, Taylor would roam factory floors, stopwatch and slide rule in hand, looking for ways to shave time off tasks. Numerical tracking was necessary, he argued, to beat back workers’ “natural laziness.” It became gospel among the major steel and shipbuilding companies of the time and influenced Henry Ford’s famous assembly line processes.
Today’s barcode scanners, employee tracking apps, and algorithmic management tools have ushered in an era of “digital Taylorism,” says Delfanti. “Many corporations are renewing that model.” Amazon sits at the cutting edge. The company automatically counts every product scanned, and its “Time Off Task” system logs every second during a 10-hour shift that a warehouse worker is not scanning, including bathroom breaks. Too few scans per hour or too much TOT are grounds for discipline, or even termination.
The ability to datify everything adds a new dimension to Taylor’s project. “Back in the day, you would feed the machine raw materials, or you would use your muscles to make it work,” says Delfanti. “With new digital technology and robots, your knowledge is also being turned into data and fed to the machine.” That information can then be used by the company to refine and improve the system in its quest for ever-more efficiency.
For example, instead of storing items by category—Star Trek zines here, toothpaste tubes there—Amazon stores items randomly, then relies on barcoding to locate them, reducing empty shelf space and saving the time it takes to criss-cross an 855,000-square-foot warehouse fetching or delivering various types of products. “As the worker shoots with the barcode scanner, that information is lost to the worker because the warehouse is so immense and difficult to navigate. It’s captured by the algorithm and stored in Amazon’s machinic memory.” The more data the machine gobbles up from workers, the more strictly it can control their labor process. Think of it like the alien plant from Little Shop of Horrors. The more bodies Rick Moranis feeds it, the more power it gains over him.
Delfanti has researched Amazon’s patents, reading them like tea leaves signaling possible futures. Down the road, he says, workers might transfer not just their knowledge but their senses to machines. One patent for a kinematic, pressure-sensor-studded glove would let workers teach robots how to perform a task, like picking up a product. Another patent for an “enhanced interaction system” could use facial recognition and augmented reality to identify workers and project stats about them, like efficiency, onto a supervisor’s visual field. Of course, many patents never see the light of day. Amazon didn’t indicate whether it intends to develop these specific technologies, but signaled an overall commitment to robotics that make their facilities safer and more efficient.
While the screamiest headlines of the past few years have warned of robots replacing workers whole cloth, the more common scenarios emerging involve a hybrid human-robot model. At Amazon’s re:Mars conference in 2019, then-VP of robotics Brad Portner held forth about the warehouse of the future. Rather than robots replacing humans, Portner described, “a symphony of humans and robots working together.” Humans, after all, still outshine robots at certain tasks. In many instances, they’re cheaper.
Inside Amazon’s newer, roboticized warehouses, which comprise 50 out of the more than 175 fulfillment centers worldwide, workers no longer walk the aisles retrieving products. Instead, motorized shelves zip over to them. This speeds up the rate of work, allowing associates to pack as many as four times more items, according to a September investigation by Reveal. The thing is, when robots set the pace, there’s less room for slack. More packing means more repetitive motion. Bodies rebel. Reveal found that serious injury rates were 50 percent higher at Amazon’s roboticized warehouses, compared to their lower-tech counterparts. This may contribute to the high turnover at the company’s fulfillment centers—around 100 percent per year on average by some estimates.
Amazon says its attrition rate is on par with industry averages, but didn’t provide supporting data. It did cite the billions the company invested last year in new safety measures, the thousands of safety inspections it conducts daily, and the hundreds of improvements made based on employee feedback.
The company says that automation frees up their skilled workers to perform more sophisticated tasks and arranged for WIRED to speak to Enlly Gramajo, a robotics tech worker in New Jersey. Gramajo wears a sensor-studded vest that signals the robots zipping around the warehouse floor to stop moving, so she can go fix problems, like retrieving fallen items. She likes working with advanced technology and prefers it to her previous role as a picker, which involved standing in one spot performing repetitive tasks. She says it’s one of the more coveted roles; about a dozen of these jobs exist at her fulfillment center. (Amazon refused to say how many people it employs as pickers or robotics techs.)
Early-20th-century labor leaders decried Taylorism for making automatons of workers, simplifying previously complex processes, and devaluing work while simultaneously ratcheting up its intensity. By the 1930s, the gospel of scientific management culminated in a breaking point for workers, says Alex Colvin, dean of Cornell’s School of Industrial and Labor Relations. Entrenched Taylorism, combined with massive economic inequality, prompted a wave of union organizing. “Even though in some ways Amazon is a very new type of company, an online technology company, it’s also got a huge operation sorting and delivering packages, so there is a resonance with the old assembly lines. It’s interesting that we’re seeing a parallel reaction of workers organizing unions in response to that.”
So far, Bessemer is the only US Amazon warehouse that’s made it to a union election, although organizers say other facilities are nipping at its heels. Some workers there have said they want a more humane system than the rigid, algorithmic management tools like Time Off Task. They’ve bemoaned the auto-generated disciplinary notices they receive when their TOT exceeds the limit and want a more human-centric grievance process. Women, particularly pregnant women, are disproportionately impacted by a system that counts bathroom breaks against TOT. They’ve criticized the paltry break time—two 30-minute periods during a physically grueling 10-hour shift, including the time it takes to cross the massive warehouse to reach the break room.
Technology needn’t be used to make work more miserable, Colvin notes; it could improve flexibility. Take scheduling. Amazon optimizes unproductive time out of its schedule through mandatory overtime during surges and voluntary (unpaid) time off during lulls. This sometimes leaves parents scrambling to find childcare when a late-breaking mandatory OT notice pops up. As ecommerce demand rose over the past year, the company has also unilaterally rolled out 10-and-a-half-hour overnight “megacycle” shifts. Labor activists have called the shifts “inhumane,” and workers in Chicago organized a walkout in protest Wednesday. (Amazon says mandatory overtime notices go out up to three weeks ahead of time, but no later than lunchtime the day prior. Employees can opt to swap shifts with a coworker or use their allotment of unpaid or paid time off, if available.)
Alternate software exists that lets workers plug in their constraints and preferences upfront, then spits out schedules based on both worker and employer needs. “Within scheduling software, there’s a real variation between ones that are dehumanizing and ones that are empowering,” says Colvin.
At the AFL-CIO’s Technology Institute, which just launched in January, labor leaders are dreaming up ways to include workers in the development of algorithms that govern their work. As a model, they cited work by the hospitality union Unite Here on behalf of a group of hotel cleaners, whose routes at work were controlled by algorithms. Engineers originally optimized the routes with the needs of VIP guests in mind. As a result, the algorithm sent cleaners ping ponging along inefficient paths, heavy carts in tow. Through collective bargaining, they helped rejigger an algorithm; engineers designed new routes to be more efficient and less taxing for the cleaners. “But without a union contract, most workers are crying out in the dark, trying to get the attention of management,” says AFL-CIO treasury secretary Liz Shuler.
As long as regulation trails technology, even unionized workers face an uphill battle. “The technology we have now was not conceived of when we were creating laws about how the workplace would function,” says Ifeoma Ajunwa, founding director of the Artificial Intelligence Decision-Making Research Program at University of North Carolina’s Law School. Some surveillance tech, like AI cameras with facial recognition, relies on shaky science that studies have shown is even more inaccurate for people of color. Others, like wellness apps, could be used to glean information like when a woman might become pregnant. Wearables are yet another frontier. Amazon has a patent for a haptic feedback bracelet that could buzz when a worker reached for the wrong item.
Electronic tracking is not limited to the warehouse floor, either. The pandemic-driven shift to remote work has prompted a surge in the use of employee monitoring software. One company literally named Controlio offers employers searchable keylogging, individualized productivity scores, and a “stealth mode” that lets bosses spy on their employees’ screens. With some local exceptions, particularly around facial recognition, all these devices are perfectly legal. (Even Controlio!) “We need to update laws to take into account the way new technology surveils workers,” says Ajunwa.
Out of this symphony of machinic memory and algorithmic control, a very human process has emerged. Unsure about how to fight back at first, one Bessemer worker discovered the RWDSU through a simple Google search. He and others snuck off to hotels and bars to meet with organizers, stood on sidewalks at twilight holding poster board signs, gathered around card tables telling their stories to Bernie Sanders, masks sliding down noses, their resistance by that point a national movement. They’ve rebelled in ways neither tiny nor private against the trillion-dollar company hurtling them into the future.
- 📩 The latest on tech, science, and more: Get our newsletters!
- When the boss of all dating apps met the pandemic
- Get moving with our favorite fitness apps and services
- Why covering canals with solar panels is a power move
- How to keep nearby strangers from sending you files
- Help! Should I tell my colleagues I’m on the spectrum?
- 👁️ Explore AI like never before with our new database
- 🎮 WIRED Games: Get the latest tips, reviews, and more
- 🏃🏽♀️ Want the best tools to get healthy? Check out our Gear team’s picks for the best fitness trackers, running gear (including shoes and socks), and best headphones