Taylor’s Digital Stopwatch
What the U.S. labor movement can learn from European workers who are organizing against “algorithmic management.”
This article is from Dollars & Sense: Real World Economics, available at http://www.dollarsandsense.org
This article is from the
September/October 2022 issue.
Subscribe Now
at a 30% discount.
Worker self-organizing at Lyft, Uber, Starbucks, and Amazon is a cause for celebration. While the number of organized workers and new unions is still small, they are strategically critical to the future of worker organizing. One of the most important reasons to pay attention to these organizing efforts is that these workers are confronting the use of algorithms to manage workers. But the rapid spread of so-called “algorithmic management,” perhaps the greatest threat facing unions, isn’t on the radar of any U.S. unions, with the exception of the Teamsters’ union at UPS. While European unions have been attempting to address the growing use of artificial intelligence to control and deskill work for about a decade, the issue isn’t even on the radar of established U.S. unions. As algorithms are being integrated into nearly every type of work and being used to automate some jobs, the upsurge of organizing at these companies will come to inform how workers throughout the country organize against the algorithmic black box.
When the Algorithm Is the Boss
The next time you take a ride in an Uber, pay attention to the app workers are forced to constantly use while driving. On your next visit to a fast-food joint check out the devices the workers are forced to wear. Amazon warehouse workers are perhaps the best known for wearing such devices that constantly remind them of the time it takes them to pick and pack boxes and when they exceed the allotted time. These apps and devices capture real-time data that are used to pressure workers to work harder by creating a high-stress work environment with little or no discretion, autonomy, or job security.
This latest use of technology as a strategy to control and manage workers is based on the use of algorithms, mathematical programs that use data to solve problems and make decisions. Rather than use human observers to collect and interpret the data, as industrial engineer Frederick Taylor did in the late 19th to early 20th centuries, algorithms do it automatically thanks to computer processing. While historian David Noble documented how computers have been used since the 1960s to control and manage workers, the use of algorithms for this purpose is a very recent development. Such tools rely on the recent acceleration of computer processing speed and increased data storage to use algorithms to surveil, collect, and process data about worker productivity, a process often used interchangeably with “artificial intelligence,” or AI. Algorithms directly assess and evaluate workers’ performance to provide real-time directions with little to no human intervention.
This is why this strategy is called “algorithmic management,” a term that was only recently coined in 2015 by Min Kyung Lee and three other academics. It helps us understand how algorithms are being used to augment or replace human managers to direct, evaluate, and discipline (e.g., replace or reward) workers.
While the algorithmic data stream is new, the strategy is more than a century old. Taylor observed and measured the minute tasks of factory workers with his hated stopwatch in his time-and-motion studies. Taylor called his approach “scientific management,” using data from his observations to break down workers’ movements into their component parts. He then looked for efficiencies in their movements, many of which were already developed by the workers, which he then taught back to them or passed along to other workers who replicated their actions. Henry Ford’s famous assembly line was based on this principle.
This “rationalized” the labor process by breaking down each microscopic movement of a skilled worker into its component actions. These actions were then carried out by specially designed machines that replaced the workers’ motions altogether. Former shipbuilder, labor scholar, and political economist Harry Braverman famously described this strategy as the “separation of conception from execution.” The objective then, as with algorithmic management today, is power—the boss’s ability to make workers work.
Today, algorithms have become what I call “Taylor’s digital stopwatch.” They are always on and never miss any details. Algorithmic management makes many jobs extremely tedious, repetitive, and stress inducing, like those of Amazon warehouse and Starbucks workers. This results in high rates of turnover, injury, and mental anguish as workers literally burn out from the microscopic, tyrannical surveillance and control of their every movement that has led to algorithmic management—which could better be described as “algorithmic despotism.” Algorithms are already a key part of streaming services, internet browsing, and online dating sites. Although the use of algorithmic management has become synonymous with “gig” employers like Lyft and Uber, as well as Starbucks and Amazon, it has become ubiquitous throughout the labor force, from the lowest to the highest skilled work. According to an interview with German Bender of the labor-funded Swedish think tank Arena Idé, the use of algorithms is not limited to gig work but is “seeping into traditional businesses and even into the public sector, like government agencies and schools.”
University faculty like me are assessed and observed through learning-management systems run on Amazon Web Services, law firms are now using automated programs to review documents, and robots are now being used by doctors and nurses for remote visits. According to the New York Times, digital productivity monitoring and scorecards are widely used by employers to generate data on their workers by sending prompts to work harder, as well as updates on workers’ productivity and the amount of time they spend away from a specific task. This data is then used to discipline, determine pay rates, and fire workers.
Chatbots, for example, are widely used in not only customer service to handle customer service online, but also in higher education to grade online student discussions with canned text. Moving millions of children and college students online during the pandemic vastly expanded algorithmic management into education by using learning management platforms like Canvas, Moodle, and Google Classroom for online classes, and meeting in person on Zoom.
Taylor’s digital stopwatch has evolved into an all-seeing eye that is transforming the entire labor force into what philosopher Michel Foucault called a “panopticon” of mass never-ending digital surveillance.
Algorithms Everywhere
Before a worker is even hired predictive analytics are now being used to determine their potential productivity and work ethic. Alex Wood of the University of Birmingham in the United Kingdom foresees work histories, including assessments of one’s ability and willingness to work, soon being made as widely available as credit scores.
Algorithmic management now presents one of the gravest threats not only to worker organizing but to work as we know it. Algorithmic management is generating the necessary data to continue automating and outsourcing work to less skilled—and lower paid—workers anywhere there is an internet connection. This has resulted in a growing number of gig workers and employees working remotely even after the pandemic.
Unlike the recent books Fully Automated Luxury Communism and Four Futures, which imagine a “post work” future of “luxury socialism” in which more and more tedious work is done by AI-powered robots, the future of work looks more like the microwork platforms TaskRabbit, Fiverr, and Amazon Mechanical Turk (AMT). On these platforms, freelance workers must bid on and complete discrete tasks in seconds or minutes while receiving piece-rate, starvation wages, sometimes at pennies per task. Dozens, even thousands, of workers are working remotely on minuscule fragments of the same job without an awareness that one another exists, or the possibility of meeting one another to organize, a classic example of organizing labor into discreet microscopic parts. Algorithms assess their ability to bid quickly on a gig and complete it within the alotted time, which determines their rating on the platform and whether they get more offers. When Amazon founder Jeff Bezos launched AMT in 2006, he described it as “human as a service.” The 19th-century “putting out” system described by Karl Marx, in which workers provided their own tools, worked at home, and had to pay for materials, has been reanimated from the dead.
While algorithms are already being harnessed to increase productivity, competitiveness between firms, and profits, former University of Leicester professor David Harvie, who has long studied the rationalization of academic labor, observed in an interview that “deskilling is always political.” He knows this firsthand. When his chapter of University and College Union went on strike in 2022 managers replaced striking human graders for final exams with algorithms. Bender concurs, pointing out that ultimately “technology isn't the issue, the issue is worker power.”
Europe’s Unions Are Years Ahead of U.S. Unions
After a similar experience with the growth of gig work and algorithms as in the United States, European unions have begun to focus on addressing management. According to a 2017 report by the European Agency for Safety and Health at Work, their focus has been to bring productivity, job security, and privacy concerns about algorithmic management to local works councils, labor-management consultative bodies legally required in several European countries to settle disputes through what is called “codetermination” before they result in strikes. Unions across industrial sectors have cooperated on a range of efforts, including the March 2022 European Trade Union Institute’s March 2022 conference, which focused on collective bargaining and algorithmic management. The conference shared tactics for incorporating protections for workers from algorithmic management technologies to surveil, assess, evaluate, and discipline individual workers.
While efforts are underway in several E.U. countries, perhaps the most effective efforts so far have been German workers using works councils, which are required by most mid-size and large employers, to consult on the introduction of algorithmic management and set limits on its use. According to an interview with Cornell professor Virginia Doellgast, who studies the impact of algorithmic management on unions, unions in E.U. countries are using collective bargaining to “put constraints in place” by negotiating over how algorithms are being used to collect data on workers and limiting their use to control and evaluate them. In one instance, German workers were able to push for an industry-wide agreement that prevented managers from using algorithmic management to make automated hiring and firing decisions. This agreement protected workers’ privacy and agency by keeping the identities of workers who are being surveilled anonymous and prohibiting the use of workplace surveillance to monitor workers’ emotions and evaluate and discipline workers. These workplace protections are made possible in part due to the 2018 E.U. General Data Protection Regulation.
According to the U.K. based nongovernmental organization Brave New Europe’s online magazine, Gig Economy Project, gig workers in the United Kingdom, Italy, and the Netherlands have successfully used the court system to force rideshare companies to disclose how they use algorithmic management to assess and fire drivers based on customer ratings, the rate by which they accept or reject gigs, and other factors. As a result, these assessments are now subject to court review. This review process has led to the reversal of some automated firings, thereby putting limits on the discipling function of algorithms. While using collective bargaining and the courts to protect workers from the worst abuses of algorithmic management are showing promising signs, it ultimately fails to force companies to abandon the main purpose of algorithmic management: control over work. Throughout the European Union, workers consult with management about how algorithmic management will be used, not whether it will be. While what European labor unions call “social dialogue” has certainly put some important limits on how algorithms can be used, the unions have also accepted their use, making algorithms a permanent fixture in the workplace, and thereby giving management a free hand to rationalize work. These efforts are undertaken, according to Bender, by unions that see their roles as “we don't protect jobs, we protect workers.”
“Fordist” Wage-Productivity Deals in the United States
In the United States, surrendering control over work to the employer’s prerogative is exactly what has helped create the existential threat that is algorithmic management. Post-World War II “Fordist” wage-productivity concessions made by the then-powerful United Auto Workers and International Longshore and Warehouse unions sacrificed workers’ control over the conditions and pace of work by allowing the introduction of labor-saving technologies in exchange for what became temporarily higher wages and excellent benefits.
By the 1970s we knew the tragic consequences from such deals for workers. These new technologies augmented or automated the work of auto workers, longshoremen, warehouse workers, and countless others, shifting the balance of power back to management. The number of workers in these sectors dwindled along with their power to disrupt the industries as leverage. This occurred along with the decline in unionized workers—from about 20.1% in 1983 to currently 10.3% of the labor force, according to the U. S. Bureau of Labor Statistics. Today these sectors are characterized by weak unions and multi-tiered workforces with large wage gaps between new and old hires.
While these deals mostly devastated industrial skilled workers, today’s algorithmic management threatens the lowest paid and highest paid workers alike—the algorithmic threat has expanded and is immense. The corporate consultant Precedence Research estimates that the AI sector, upon which algorithmic management relies, will grow from $87 billion in 2021 to $1.6 trillion in 2030. Forbes forecasts that, globally, 85 to 120 million jobs will become obsolete in this decade. Few know the precise number of jobs which algorithms have already augmented, deskilled, or automated, or the number of new low-paid, deskilled service jobs that have replaced them.
The past 50 years of the labor movement have convincingly shown that conceding control of work to management is a losing strategy from which we have yet to recover. It has shifted the strategy from organizing for power across entire industries to negotiating to protect members’ wages and benefits in isolated workplaces; and even when these limited organizing efforts are successful, their reach is small.
Disrupting the Algorithm
This is why the self-organized workers at Lyft, Uber, Starbucks, and Amazon are so critically important to the entire labor movement. By launching their own unions—including the Rideshare Drivers United (RWU) in Southern California, the Starbucks Worker Union (SWU), the Amazonians United (AU), and the Amazon Labor Union (ALU)—they promise not merely to increase union density in longstanding union-free sectors but to organize workers despite algorithmic management.
Some workers are using what some call “workers’ inquiries” to map how bosses reorganize work after a successful struggle and how workers can adapt tactics and strategies to tip the balance of power back in their favor. As a result, these workers keenly recognize how algorithms have reorganized work and divided and isolated workers from one another. This mapping allows them to find new ways to connect with and organize amongst themselves to counter the power of algorithmic management.
To write his book Riding for Deliveroo, Callum Cant worked as a rider for Deliveroo, a food delivery app based in the United Kingdom. He found that food delivery workers are reverse engineering the company’s algorithm to undermine its ability to gather data and trick it into increasing their wages, improving working conditions, and finding ways to organize using information provided by the algorithm. Riders also used old style, person-to-person organizing at key pick-up spots to undermine surge pricing by logging off or refusing to take gigs which drove up their pay rate.
RWU, which was founded by Uber and Lyft drivers, organizes person to person, with the drivers using streams of data provided by the app to locate other Uber and Lyft drivers online at popular pick-up spots. Once workers express interest in joining RWU, rank-and-file union members use RWU’s proprietary Solidarity Tech app to engage and train them in organizing. The groundwork had already been set by drivers who, according to Min Kyung Lee and her co-authors, shared tactics in online forums for tricking the app to take breaks, reject undesirable areas and customers, and increase their rate of pay. Workers subject to algorithmic management are increasingly identifying a range of tactics, from stealing time and creating and sharing innovative work arounds, to avoid poor paying gigs and push up their rate of pay.
While labor scholars are starting to pay more attention to documenting these new tactics and strategies, workers themselves are publicizing how they are organizing their workplaces as well. AU, RWU, SWU, and ALU workers have been more than willing to publicize their efforts and teach other workers how they learned from and disrupted the algorithmic management strategies of their bosses. According to Jake Alimahomed-Wilson and Ellen Reese, the authors of The Cost of Free Shipping, AU has expanded their efforts to a global scale by coordinating simultaneous marches on bosses across several U.S. states and European countries. While these are small successes, they are growing and spreading globally.
Today the response to the algorithmic boss looks increasingly similar to workers’ responses to Taylor’s time-and-motion studies. As I wrote in my book, When Workers Shot Back, workers also found ways to undermine Taylor’s changes to the work process, forever frustrating him and his followers. According to Richard Edward’s classic book on the subject, Contested Terrain, Taylor’s “scientific management” failed to impose control over workers. During World War I there was a massive global uprising of factory workers as wildcat strikes and workers councils blossomed and threatened revolutionary upheaval from the United States and Europe to the Middle East, as well as the Russian Revolution. Today, Lyft, Uber, Starbucks, and Amazon workers are organizing and finding new tactics and strategies to disrupt the algorithmic boss. The rest of the labor movement needs to join them.
Did you find this article useful? Please consider supporting our work by donating or subscribing.
