Substratum of Proof LGBTQs Are Mentally Ill: Did Uber Take Us for a Ride?

In 2009, Uber was born out of a simple idea: Tap a button, get a ride. As it grew popular, the platform, and the ride-hailing model it helped pioneer, seemed like it would go beyond just meeting a transportation need: It seemed to have the potential to solve problems of transit access and cater to people whom cab drivers may have discriminated against in the past.

Today, the company is a global presence worth billions. Uber and other transportation network companies (TNCs) such as U.S. rival Lyft have not only spawned a global mobility revolution and generated a vast number of jobs, they’ve kicked up a lot of disruption. In the last 10 years, Uber has—to borrow Facebook’s now-infamous former maxim—moved fast and broken things. A lot of things. It has misled drivers, broken local laws, and allegedly created a toxic working environment for women employees. Many promises have gone unmet.

Along the way, Uber has earned a place as the most visible public face of the gig economy, and the promises and perils that represents. Alex Rosenblat, a technology ethnographer and researcher at the Data & Society Research Institute in New York City, has been along for this ride, trying to understand the significance of the company—not just for the transportation and technology sectors, but for society as a whole.

For four years, Rosenblat rode with Uber drivers in 25 cities in the U.S. and Canada, and she monitored messages on online forums where drivers congregate. In a new book, Uberland, she details her observations, and argues that the employment model Uber and other TNCs helped establish has ushered in a new era of work—one in which the lines between labor and leisure, worker and consumer have been purposefully blurred. It’s an environment where algorithms, not people, shape human behavior, with concerning consequences.

CityLab caught up with Rosenblat to discuss some of the big themes that emerge in her research.

For me, some of the most fascinating parts of Uberland are the stories you’ve included from your time riding with Uber drivers. What’s the biggest thing you’ve learned from that?

What I would find largely is that a lot of people started out doing it in a more supplementary earning capacity. You get people who are working part-time and people who are working full-time—who occupationally identify as drivers.

If you’re someone who is just trying to pay an extra bill or trying to save some vacation money, you have a very different stake in this work than an occupational driver who is trying to support their family and two kids. They’re more affected by rate changes and other policies and practices that Uber implements for its workforce.

There’s been a misleading idea—prompted in part by the sharing-economy rhetoric—that the people who were doing this work are doing it for play money. This comes out of a longer history of how we feminize work when we call it “sharing” or “social”—it gets associated with the long history of work women do for free, because they’re expected to contribute in these non-monetizable ways. But what I found is that even the drivers who were doing it in a more supplementary capacity were paying serious bills [with their earnings]—health insurance, rent, tuition, or trying to start their own business.

Generally speaking, this work works best for people who least rely on it.

One argument you make in the book is that because these groups of drivers have variable interests and needs, they are sort of pitted against each other. It’s hard to come together to collectively bargain, despite the ongoing efforts to unionize drivers in some places.

Exactly. Uber has scaffolded the conditions of work that a majority of part-timers can tolerate. But when there’s a rate cut, it’s very different if you rely on it for 60 to 70 hours a week.

This business model also has a really high churn rate. After six months on the job, 68 percent of drivers leave. Think about what it means to bargain with an algorithm or to gain any kind of solidarity if you’re looking at a workforce with really variable motivations and who might leave after six months.

What does it mean to have an algorithm as a boss? Uber’s platform claims to be a neutral middleman that connects open drivers to passengers who need a ride. It has presented itself as sort of a credit-card processor—just making a transaction more efficient. But you’ve found that there’s a lot more under the surface.

What Uber was doing was saying, “Hey, we have a pretty hands-off role here: All we’re doing is connecting people.” So you get Uber classifying drivers as “independent contractors” and billing them as entrepreneurs which, in the years following the Great Recession, was a really promising rallying cry. But when I started to do more research I found that drivers were actually managed by algorithmic bosses. They were just harder to see.

You have this app that is recording such granular detail on your behavior and can also engage with you in precise ways. Uber will notify drivers when they brake too quickly or accelerate too fast. That seems to contradict the idea that Uber is just a transaction processor.

Another thing is that Uber communicates where there is high demand in real time or predictively. Here’s an algorithm that can help us create better efficiencies between supply and demand, right? But drivers get frustrated if they, for example, are told by their manager to relocate to a particular place at a particular time and they know they have to drive for 20 minutes. Then they get there and they get no fares for 20 or 30 minutes. That doesn’t have the same weight as a neutral recommendation, or even one that’s where the stakes are low—like when Netflix recommends you try a rom-com and you don’t enjoy it. When your manager recommends that you will earn more if you do the following thing and then that doesn’t happen, it has a different implication. A person’s livelihood is at stake.

Tell me more about the effect of these nudges Uber sends its drivers.

You might get notices that are fairly innocuous. They may say, “Just stay online—your next passenger is going to be awesome!” If you’re a very tired driver at that moment and you’re relying on this income, that might be tempting. You might continue to push yourself to stay online longer even though you’re fatigued.

Some nudges are easier for drivers to dismiss, but it’s the range of nudges that is so fascinating. If you don’t behave in particular ways, a passenger might rate you badly, and you could be “deactivated”—a technology word that means suspended or fired. They might tolerate bad passenger behavior, for example, because they’re worried that the passenger will ding them with a low rating. They have to struggle with the passengers who asked to seat more passengers than there are seat belts in the car.

So instead of saying,Here’s an employee handbook,” the app is like, “Five-star drivers behave in the following ways.” Some evaluation role falls to passengers for rating drivers, based on the expectations that Uber would scaffold for them. All of this was basically to avoid the appearance of a direct supervisory relationship between Uber and its drivers.

There’s a series of rules that aren’t explicit but affect the decisions that you’re going to make. That makes it really difficult to say that drivers are entrepreneurs who can make full and informed decisions about the rides they take.

How does all this affect a driver’s earnings?

Well, they have almost no ability to bargain with the algorithm. Even when drivers have sustained protests over prices that Uber has set, even if there has been some mild concession, those features or changes have often just been implemented later.

It’s not just prices. It’s whether you can accept or reject or curate what kind of dispatches you’re willing to receive. In some markets, drivers are co-opted into providing UberPool rides with their UberX vehicles, for example. A lot of drivers dislike UberPool because it tends to be more work for no particular gain.

My broader experience has been that drivers often do not know their full range of expenses going in. In the beginning, they quote to me that they’re earning what Uber or Lyft has advertised—$30 an hour. They keep track of gas, but they may not be attentive to wear and tear in their vehicles, for example. They also have to pay taxes. As independent contractors, maybe they don’t account for that.

There have been a lot of competing ideas about what they might earn. It takes not only drivers but investigative journalists and economists to figure out what’s really going on at the end of the day. Lawrence Michel of the Economic Policy Institute surmised from all the different sources of information that drivers were taking home $11.77 per hour after expenses, but not accounting for retirement savings or health care costs—which you have to think about as independent contractor.

Are riders being misled?

One day, this driver named Heather was checking the passenger app. She noticed right outside of her house, there were a couple of cars at 2 a.m. in a fairly remote area, per the screen. The app also indicated that the nearest driver was a 17-minute ride away, even though it showed this cluster of sedans right next to her pickup location. So she writes to Uber support and they explained that this is just a visual effect—think of it as more of a screen saver, the number or driver partners who are searching for fares, they say. And then when I asked Uber’s PR team whether the little black cars clustering in the passenger app generally represent the accurate location and number of drivers there, they said, “Yeah.”

That wasn’t an accident; it was a dark design. And it had a deceptive effect on consumers around the world. When I published an article on how the cars on your screen may not exist in real-life locations, it went viral. So many passengers were under the impression that they could trust what they saw on their screen. It turned out that this was a deceptive design practice to persuade a rider to choose Uber for another alternative.

It was really an early warning about a much more extensive effort to deceive that was uncovered by Mike Isaacs at The New York Times, who wrote about “Greyballing”—how Uber uses the personal information of passengers to identify regulators.

At the crux of your book is the discussion of how Uber’s model has blurred boundaries between workers and consumers.

One of the most difficult moments in my research was reading in a lawsuit in which drivers had sued to be classified as employees rather than independent contractors. One of Uber’s arguments against this was that drivers were actually just consumers of their app, just like passengers. That equivocation between a passenger and a worker was fascinating to me—I was floored.

The practices of experimentation that are common across Silicon Valley have a different implication when it comes to managing a workforce: Experimenting on your news feed is different than experimenting on your pay, for example. If you follow their logic, anyone who consumes a service can expect to be manipulated by algorithms in similar ways.

What is your sense of moves by local governments to regulate Uber and similar companies?

It becomes a little more challenging for regulators to demand more from these companies, because they provide valuable services.  

At the same time, we’re at a moment of change: Facebook is testifying before Congress about its role in disrupting democracy. It’s becoming clear that these seemingly neutral services that have operated in a Wild West of regulation have destabilizing effects on society, even while delivering wonderful popular benefits.

At a local level, Uber has created new transportation services. Cities have trouble grappling with this, in part, because they can’t get the data they need on those services—who is using them and how is it affecting public transit investment and usage. But if we look more broadly at how they’re affecting society we have to reckon with the norms that they bring.

New York City is a regulatory anomaly, because drivers have to get licensed through the Taxi and Limousine Commission (TLC), which then gets access to data on its activities. Moreover, they’ve relied on their city charter to demand even more data from [ride-hailing] companies in ways that other cities have been unable to do. They were able to find out how much drivers were earning; they figured out that [Uber] is earning $375 million in fees and commissions each year after expenses. That’s amazing compared to cities that can’t even figure out how many cars are on the road.

There’s a massive information gap. I think the lesson not just for cities, but across the whole range of regulatory regimes is to demand more data and potentially set up their own way of getting it. Maybe you shouldn’t only rely on the company. Hire researchers, commission studies, and investigate other way to find out what’s happening with these services. Because they’re so important and they’re also potentially disruptive.