Flexibility. Since the emergence of the gig economy in the early 2010s and through the upheavals of the pandemic, it’s something increasing numbers of Americans have experienced at work.
Unsurprisingly, people tend to like having control over where and when they work, fitting it around their lives’ schedules. At the same time, flexibility has become the lynchpin of a well-traveled myth that corporations have spun to their employees, policymakers, and the public.
While this myth doesn’t survive close scrutiny, it has been incredibly effective. With artificial intelligence being explored as a replacement for many white-collar jobs, and the gig work model continuing its global expansion into new sectors like health care, information technology, staffing, hospitality, and child care, the flexibility myth now threatens the hard-won rights and protections afforded to tens of millions of working people.
It goes something like this:
[W]e’ve advocated for creating a framework that gives independent workers the benefits and protections they deserve, while preserving the flexibility they want. One that offers an alternative to traditional employment and fits the way people are increasingly choosing to work. A framework where people can choose work that fits the rhythms of their lives, not the other way around.
These are the words of Tony West, Uber’s chief legal officer, from a public letter written after Uber and Lyft reached a settlement with the Massachusetts Attorney General Office in 2024. The case turned on whether or not drivers were employees under Massachusetts employment law.
I served as the lead expert witness for the Massachusetts Attorney General Office in that suit, and my task was to explain the Uber and Lyft business model to the court—that is, the methods that the companies use to create sustainable profits.
To fulfill this task, I had access to evidence collected through discovery by the Massachusetts Attorney General over several years. For the many academics and governments that have attempted to get an accurate picture of working conditions in the gig economy over the years, this data is something of a holy grail: Most studies of gig worker income, hours, safety, and such are built on laboriously collected surveys. The companies themselves claim these data as proprietary trade secrets and guard closely against release into the public domain.
For the purposes of this article, I have not revealed anything beyond what was stated directly on the public record in my testimony during that trial, corroborated by findings of other recent studies. Many of my findings have more recently been corroborated by other academic studies and investigative reporting.
After reviewing the data, I came to a very clear conclusion: Drivers don’t have flexibility because they want it. They have it because it is the essential core of Uber’s and Lyft’s business model.
This may not seem earthshaking news. But it contradicts a fundamental plank of every one of the regulatory fights that gig companies have faced over the last decade: that their workers need to be classified as independent contractors, so that they can continue to enjoy flexibility. This is the companies’ narrative to the public, the basis of their recurrent threats to leave local markets when independent contractor status is challenged, and the message blasted to the cellphone of every one of their drivers in the affected area: “The government wants to take your flexibility.”
As the former head of the U.S. Department of Labor agency in charge of labor standards, I can assure you: There is nothing keeping any company from offering flexible work schedules to its employees and no necessary connection between independent contractor status and flexibility.
Uber and Lyft would desire to keep drivers’ flexibility even if the drivers became employees. In fact, this has happened elsewhere when governments have refused to accept the companies’ contention that Uber’s drivers don’t actually work for Uber.
The gig work model has been fine-tuned after years of harvesting, analyzing, and operationalizing real-time data on rider and driver behavior. In recent years, developments in artificial intelligence have greatly enhanced its functioning. The result is a management system that addresses fundamental challenges confronting any profit-seeking company in virtually any sector—setting prices for customers, rates of pay for workers, and optimizing the scale of operations. As a result, Uber and Lyft drivers have found themselves at the forefront of experiments using artificial intelligence to manage a company’s workforce. Uber even touts that in offering AI services to other companies.
These innovations have thus far been accompanied by another, more insidious one: how to convince governments that your employees should be treated as independent contractors because of their (not the companies’) desire for flexibility, and designing new, bespoke labor laws to reinforce that position.
To understand the role of the flexibility myth in perpetuating this situation, it’s first necessary to understand the business model.
Looking Under the Hood of an Uber and Lyft Ride
Let’s say you intend to start a restaurant. Before you can open your doors, you’ll need to find a space and equip a kitchen and dining room, hire staff, and source your supplies. And you’d have to calculate such costs against the number of your prospective customers and how much you think they’d pay for the meals you have in mind.
The details are many, and the uncertainties can be maddening: Long-term considerations and fixed decisions about the scale of your operations. Prices for ingredients that change with climatic conditions. The increased productivity versus the wages of another prep cook you only really need on the weekend, and so on.
Of course, you’d aim to keep your costs low and source only the ingredients you need, and to attract and employ just enough workers with the requisite skills to serve them. It might be tempting to constantly adjust your prices to match your costs plus the profit that makes it all sustainable and worthwhile. That, however, would likely be obtrusive and not particularly appreciated by your customers.
What diners are willing to pay can also vary a lot—person to person and hour by hour, season to season and dish to dish. So you’ll need to consider intangible factors like ambiance and mood in addition to spice and method when you make your educated guesses and chart your course to profits.
Historically, most businesses couldn’t continuously vary prices and wages to reflect real-time costs and willingness to pay, in part because diners and workers alike appreciate predictable prices and wages.
Imagine reading a menu, putting it down, and picking it back up to see that the prices have changed based on what your tablemates said they wanted and the way your eyes traveled the menu that first time through.
In their early years, Uber and Lyft solved the problem of how to turn a profit by simply not doing this. Instead, they convinced enough high-risk investors they’d be able to do this in the future by creating a customer base with low prices and recruiting an army of drivers with the lure of a new way to make money through a side hustle. Investor tolerance for losses greatly loosened both companies’ cost constraints and allowed them to maximize for scale.
After creating and cementing their position in the rideshare market, and around the time that both companies made their initial public offerings, Uber and Lyft shifted their approach. Rather than basing customer prices on a fixed markup of payments to drivers, they “decoupled” the rates they charge riders from the rate they pay drivers. They set these prices independently one from the other, in order to control the difference between the two for every offered ride—that is, their profit margin. A recent study of Uber drivers documented similar decoupling in the U.K.
Decoupled pricing allows the companies to process and analyze the terabytes of real-time information they unilaterally collect and control, and convert them via pricing algorithms to set a price for riders and drivers that maximizes their profit margin. On the rider side, the companies have carefully developed dynamic pricing models to tailor prices offered to riders based on the route they pick, the day of the week and the time of day, and also current demand conditions (how many potential customers are looking for transportation) for any given pairing of starting points and destinations. That means setting millions of prices in any geographic area in the course of a day. Economists call pricing like that “third-degree price discrimination.” It sounds illegal (it is not) but means pursuing profits by varying prices based on customers’ willingness to pay.
On the driver side, the companies determine what to pay drivers—their key cost in getting any given customer from point A to point B—based on equally refined dynamic pricing models geared at coaxing out a sufficient number of drivers to transport the current demand for rides. In this case, Uber and Lyft seek to pay a sufficient amount to summon a workforce—but at the minimum price they have to pay at any moment. In other words, they calculate each driver’s momentary willingness to work.
Economists call this kind of practice in labor markets “wage discrimination” (once again, not illegal)—setting wages so as to compensate workers just the amount needed for them to agree to work, but no more than that.
They are able to do this because of the never-ending streams of data coming in from drivers’ and customers’ phones, using AI and machine learning to constantly refine their algorithms. Pricing methods are further fine-tuned by experimentation as the companies vary pricing and then gauge behavioral responses. Dynamic pricing is the primary reason that the companies—Uber in particular—have been able to steadily increase their margins per ride and overall profitability in recent years. A recent study finds that customer use of platforms tends to be “sticky” in real time—that is, once a rider decides to use one company, they don’t do a lot of price comparison with the other. This rider stickiness nets an estimated $300 million additional annual gross bookings (revenues) in New York City alone.
Why the Decoupled Deck Is Stacked Against Drivers
The system is particularly weighted against drivers, who are only paid once they are offered and then accept (within 15 seconds) a ride, receiving no compensation for the time when they are on the platform but not matched with a rider. The length of time that a driver waits in that uncompensated state is largely a function of how many other drivers are also on the platform and willing to be paid less. Since bringing another driver onto the road is essentially costless to the companies, the supply of drivers can be maintained at the minimum amount needed to fulfill current ride demand.
In restaurant terms, imagine a lineup of unpaid waiters and cooks standing out back at all times, allowing the restaurant owner to offer the lowest wage acceptable to someone in the line to prepare and serve every dish. And not only that. Uber’s and Lyft’s business model also addresses the capacity problem—How big a kitchen? How many tables?—with the aid of AI, algorithms, and gluts of real-time information. Rather than guessing about the number of drivers required to fulfill demand, they can manipulate demand based on their rider pricing decisions. In essence, Uber and Lyft act like a restaurateur who can adjust the size of their kitchen and dining room to meet instantaneous demand.
To illustrate how this works in practice, consider a big concert venue after a show lets out. Large numbers of people stream out of the venue, and many are looking for a ride.
Some people who need or want to be picked up immediately will be willing to pay whatever that ride may cost. If prices are too high, others might wait for them to come down. Uber’s algorithms will decide how long to keep prices high (“surge” pricing), taking advantage of the higher revenues—up until the point that some critical mass of riders will choose to walk or take the bus or call a Lyft instead.
On the flip side, the company must bring in a sufficient number of drivers to fill the large number of immediate, high-priced ride requests and still have drivers available as the prices decrease. The screens of drivers’ phones will display a heatmap overlaying the neighborhood of the venue, indicating the availability of surge-priced rides there. Since drivers shoulder all the costs of fuel and their unmatched time, the cost to Uber and Lyft of drawing too many drivers to the area is zero. Working uncompensated hours is a reality for platform drivers, and the main contributor to their low—and in some analyses, falling—average hourly earnings.
It is as if the supply of drivers and the demand for rides were two balloons. When the demand balloon expands after the venue’s house lights come on, Uber and Lyft expand the supply balloon, to their great benefit. That’s the business model and the reason that the companies value flexibility above all. A balloon without flexibility is not a balloon.
No, They Are Not Just Like Airbnb or eBay
Another aspect of the flexibility myth rests on a category confusion. Since their beginnings, Uber and Lyft have objected to being considered transportation companies. Instead, they insist, they are technology companies and platforms that simply make matches between riders and independent drivers.
Unfortunately, much of the academic, business, and popular writing about the companies has ignored the fundamental differences between Uber and Lyft, on the one hand, and a true “two-sided” platform like Airbnb, eBay, or Etsy. By being put in the same technology box as companies whose business models’ profits come from matching two sides of digital marketplaces, Uber and Lyft can define their relationship with drivers as similar to those other companies’ relationships with the “hosts” or sellers on their platforms.
But unlike companies like Airbnb, Lyft and Uber determine the prices paid by riders and the compensation paid to drivers. “Guests” go to Airbnb to choose from a wide variety of options offered by a range of “hosts” for places to stay. Although Airbnb now offers hosts suggestions on rates, those rates are set by the individual hosts, based on their own assessment of the value of their property. Although many guests pay the rate provided by the host, that rate can also be negotiated between the parties. Not so with Uber and Lyft: Both the rider and the driver are matched to one another by prices set by the companies. And that price—what Uber and Lyft like to insist is the “market price”—reflects the optimization decisions of the platforms.
And whereas Airbnb attracts customers by offering variety, Uber and Lyft spend millions of dollars each year on advertising and marketing to convince their customers that they will receive a service—a ride—that is uniformly dependable, safe, and of an expected quality level no matter which driver picks them up. In the apps, riders see uniform icons of cars nearby, and only learn their driver’s identity after they’ve been matched.
The companies have successfully convinced millions of people to get into cars with strangers (Exhibit A: Uber’s marketing of services for driving teens). Doing so only makes sense if riders associate both companies, regardless of the driver sent to them, with basic qualities: dependability, service quality, and safety. The companies are not providing a digital market connection—they are solving a customer problem: getting from A to B comfortably, reliably, and safely. Those are fundamental differences between platforms that function like a digital, two-sided market (Airbnb, eBay, Etsy) and platforms that serve as an algorithmic, AI-tuned management system (Uber and Lyft).
Flexibility Is About Profits, Not Beneficence
This brings us back to why “flexibility” means very different things for Uber and Lyft than it does for its drivers—and the public.
Uber and Lyft argue that, in order to provide their workers with the flexibility they want, they must classify those workers as independent contractors. Absent independent contractor status, their business models would not be possible and the companies would be forced to adopt rigid schedules like a more traditional employer.
This is the flexibility myth, and it has been propagated whenever the companies have faced legislation, referenda, or enforcement under existing labor laws.
But look under the hood at the business model and it’s clear: Setting millions of prices for its drivers and riders is to these companies what the use of point-of-sale data and advanced logistics is to modern retailers like Walmart and Amazon—the very key to profitability.
Why would these companies sacrifice their sophisticated uses of data on revenues, costs, and capacity needs if they had to pay minimum wages, unemployment insurance, and workers’ compensation—the key provisions that really are at stake in the employee/independent contractor distinction.
They would adjust algorithms to take these costs into account if required to do so. In fact, they have done just that in New York City, where they pay 25 percent above the minimum wage for all drivers’ time and expenses, and where drivers are covered by workers’ comp and Uber drivers by unemployment insurance. They have also kept the same business model in other parts of the world where the companies have been required to accept employment status.
The flexibility myth is a bluff.
Independent contractor status releases the companies from obligations that our worker protection and labor standards laws require. This body of laws imposes costs on employers, and it is no wonder that any business would prefer to be released from them. But Uber and Lyft would operate profitably just as they do now if they were required to recognize those obligations.
The Public Discussion We Need About Flexibility
It’s precisely because businesses do not arrive at these practices on their own that public policies impose minimum wages and labor standards, require provision for safe and healthy work environments and compensation for workplace injuries, restrict practices that discriminate, and provide opportunities for workers to exercise their voices. Granting drivers the rights and protections of employees would undoubtedly raise the costs of labor for Uber and Lyft. But the same is true for any company having to comply with employment law and work protections.
It is one thing for a company to press its advantage by disseminating a false narrative. That’s their prerogative. But it’s yet another for government officials and policymakers to adopt that narrative wholesale. In the first Trump administration, they did so by releasing a regulation on employee status that clearly sought to bend how the Fair Labor Standards Act defines employment, to the advantage of the platform companies. After being reversed during the Biden administration, the second Trump administration has doubled down on it.
In an op-ed published in the Washington Examiner, current acting U.S. Secretary of Labor Keith Sonderling (whom Trump has now nominated for the post of secretary) warned the International Labour Organization (ILO) that “The Trump administration will not sit on the sidelines while some foreign governments push to hamper American innovation in the gig economy worldwide.” This year, the ILO took up the issue of worker conditions in the gig economy for the first time and is expected to issue recommendations on standards. And in Sonderling’s piece, there it is, “flexibility,” right in the first paragraph:
The gig economy is one of the defining innovations of the 21st century. It has revolutionized how people earn a living, offering flexibility, independence, and opportunity on a global scale.
Notably, pushing against Sonderling’s and the U.S. delegation’s (as well as the platform companies’) position, this June the ILO adopted Convention 193, “Decent Work in the Platform Economy,” which is the first binding international labor standard for the platform economy.
The flexibility argument was also central to the campaign for California’s Proposition 22, the most expensive referendum campaign in the state’s history, largely due to spending by Uber, Lyft, and DoorDash. There, ostensibly in order to preserve flexibility, voters approved a bespoke minimum wage and benefits package for gig workers and made their independent contractor status permanent: The proposition stipulated it can only be overturned by a vote of seven-eighths of state legislators.
This year, California lawmakers added a bespoke right to organize to Prop 22’s provisions for gig workers. In Massachusetts, a similar referendum was approved by the electorate in November 2024, and drivers recently won the recognition rights under the referendum’s provisions.
On one hand, this gives Uber and Lyft drivers a recognized right to form a union and bargain collectively. On the other, that recognition is based on their nonemployee status (if that were not true, both laws would be preempted by the federal National Labor Relations Act). It requires a separate discussion on whether or not this is a step forward—particularly because the law in California was drafted in part and fully endorsed by Uber.
And the Massachusetts referendum was only possible because of what appeared to many to be a judge’s probable decision in the Massachusetts Attorney General Office’s suit (where I served as an expert) that Uber and Lyft were employees under the state’s ABC test. In a courthouse-step agreement to resolve the suit, the companies agreed to retract a “Proposition 22”-type referendum from the ballot in return for a settlement that left employee status ambiguous.
And, once again premised on the myth of flexibility as a gift provided for the sake of workers, a similar package of “flexible benefits” for gig workers has now been introduced in the U.S. Senate.
There is no doubt that workers value and need flexibility. Erratic work schedules and unpredictability increase earnings volatility, stress, material hardship, and health problems. Conversely, research shows that policies reducing schedule uncertainty, thereby allowing workers to navigate the complexities of modern life while earning a sustainable living, improve economic security and health outcomes.
Uber’s and Lyft’s self-serving rhetoric, however, has created a myth that worker flexibility can only come at the price of rights and protections. Accepting this premise encourages other businesses to follow suit. They have done so, in industry after industry and country after country. This supposed innovation is causing millions more workers every year to find themselves in situations outside the reach of the basic rights and protections their governments have established as minimum standards.
At its root, this has nothing to do with the wonders of digital technology, or artificial intelligence, or America’s national capacity for innovation. It’s based almost entirely on the power of storytelling, and the willingness of certain audiences to suspend their disbelief. It is long past time that we demand the truth instead.
