These days, it’s difficult to read about workplace policies without encountering “algorithmic management.” We hear that companies are increasingly controlling their employees through various digital “tricks.” These companies record our keystrokes, track our location, and even monitor us through our webcams. This same story is heard in academic journals, government reports, and the general press. In fact, this story is even reflected in federal regulations, specifically the U.S. Department of Labor’s current rules regarding independent contractors. As with most popular accounts, this rule assumes that algorithmic management is prevalent. and treats that practice as a form of “control.”
There’s just one problem. Algorithmic management is not real. It’s a term used to describe a business practice so old and so ordinary that it would be boring if it didn’t have a scary label attached to it. These practices include monitoring performance, providing incentives, and tracking work. Critics of these practices have successfully promoted them as if they were new and bad for American workers. And in recent years, that misleading narrative has shaped Labor Department policy.
Fortunately, this story may finally have come to an end. The Department of Labor recently proposed removing this idea from its regulations and returning to a more traditional definition of workplace “control.” This definition at least implicitly recognizes that incentives and monitoring are not themselves forms of control. Properly understood, these are ways to manage suppliers when a company has no control. That is, they are not evidence of control, but the opposite. And that’s true even when it’s packaged under a scary label like “algorithm-managed.”
Algorithm management: null set
Thermal algorithm management has always been a bit vague. The term was coined in a 2015 academic paper that focused on the then-novel ride-sharing platform. In this paper, we studied how these platforms use digital tools to coordinate a vast number of independent drivers. This paper focuses on three of these tools: passenger-driver matching, customer rating systems, and contingency pricing. In this paper, we find that these tools can not only help platforms balance driver supply and rider demand, but also keep drivers largely satisfied. Algorithmic management doesn’t seem to be such a bad thing.
But since then, the term has taken on a life of its own. It’s now being applied to tools ranging from scheduling software to chatbots to AI, the latter pushing vigilance to new heights. Writers such as Norm Scheiber of the New York Times, Sam Levine of the Federal Trade Commission, and Veena Duvall of the University of California, Irvine, have painted algorithmic management in toxic terms, arguing that companies use algorithms to trick people into working longer hours for less money. Duvall even compared it to modern-day Jim Crow.
These arguments are not just rhetoric. They are beginning to shape public policy. In 2024, the Department of Labor adopted regulations specifying the difference between employees and independent contractors. The distinction is important because under the country’s main wage and hour law, the Fair Labor Standards Act (FLSA), only employees are entitled to minimum wage and overtime pay. The FLSA distinguishes between these classes of workers based on multiple factors, including who “controls” the work. The 2024 regulations define this type of control to include some type of algorithmic management. In particular, he pointed to “technical” surveillance such as GPS tracking. According to the regulations, this type of monitoring counts as control even if it is not combined with other procedures. Technical surveillance was itself a means of controlling the work of others.
This approach was a departure from the department’s traditional rules. Traditionally, control has meant some kind of positive action, requiring companies to direct, deter, or punish employees when they do something. However, the 2024 Rules expand this concept to mere observation. If a company simply collects information about its operations, it could be considered to be managing its operations through “technology.”
Analog logic for a digital world
This kind of thinking puts problems on the back burner. Incentives, monitoring, and similar techniques are not signs of control, but of their absence. In economic terms, these are just ways for companies to solve the “principal-agent problem.” The basic problem is that when companies hire contractors, they have different incentives. Businesses want the best service at the lowest price, while contractors want the best price with the least amount of effort. Therefore, enterprises must take protective measures against idle labor to protect their profits. If you are using your own employees, you may be able to direct risk management simply by telling them what to do. However, this is not possible when using a contractor, as the details of the work itself are often not fully known. (That’s why you hired the contractor in the first place.) Instead, you might require the contractor to report when certain milestones are reached (monitoring). Alternatively, you may offer additional payments (incentives) to the contractor for high-quality deliverables.
This type of indirect influence is never, and should not be, considered a “control” for purposes of classifying workers. If so, it would be difficult to classify any work as independent. Since all principals monitor agents to some degree, monitoring itself tells us nothing about how to classify relationships. For example, when a company ships a package and pays for delivery confirmation, they are, in a sense, monitoring that operation. However, no one believes that the law should treat all logistics suppliers as employees of all customers.
Nothing has changed with modern technology. Even when companies and contractors use technology, they still face principal-agent problems. And the best tools for dealing with these problems remain monitoring, incentives, and similar indirect methods. These methods may work faster in the digital world, but the dynamics are the same, even when wrapped in scary labels like “algorithmic management.”
Of course, this increased efficiency of algorithmic tools is exactly what makes people nervous about them. People worry that if an employer can monitor an employee’s keystrokes or access their webcam, the employer will use that information to micromanage the employee’s work. But what matters is not whether an employer collects information, but how the employer uses the information. When an employer uses keystroke data to evaluate, compensate, or punish an employee, it is the use, not the data, that counts as control. The legal question is whether the principal has control over the work, not whether the principal knows how the work is being done. This is true whether the data is collected through technology, delivery notes, or observation with your own eyes.
Fortunately, the Department of Labor takes that view. In late February 2026, the company announced new rules to distinguish between contractors and employees. The new regulations say nothing about algorithmic management or other “technical” controls. Instead, it goes back to first principles. This means that if a person manages their own work, they are probably a contractor. If not, she’s probably an employee. This remains true even as work moves into the digital world.
