“If I’m doing something useful for the company, I should be paid for that time,” Mark says to me as he drives me over the Brooklyn Bridge. “That’s what work is, right?”
It seems like a simple enough principle. And yet when it comes to the nature of work in the digital platform economy, getting paid for that time is anything but a simple proposition.
Videos by VICE
Mark has a special appreciation for what constitutes value to a corporation. In a city where most Uber and taxi drivers are recent immigrants, he’s an anomaly, a former Wall Street banker who was laid off in the recession and has turned to Uber for part-time work. (Mark and other Uber drivers I spoke to for this story, both in person and online, have requested that their real names not be used out of fear of reprisal from Uber or other employers.)
The usefulness Mark refers to is the data he generates for Uber—not when he has a fare, but when he is waiting to be summoned and isn’t making any money. Uber drivers call that time without fares “dead miles.” Drivers may spend that time roaming around waiting for their next request from the Uber app. Or they may drive from a low-density area where they dropped off their last passenger back to a high-density area where they are more likely to find a new passenger.
While those dead miles are unpaid, the data Mark generates during that time is immensely valuable to Uber. Interviews and research conducted last year by my colleague Alex Rosenblat at Data & Society Research Institute and Luke Stark of New York University, illustrated how Uber collects data from drivers even during their unpaid time.
“Uber drivers continue to generate useful data for Uber even when they are not carrying a fare,” they write, “because they relay data back to the central platform from which inferences can be drawn about traffic patterns, and which feed into supply and demand algorithmic analyses.”
That dataset feeds into the company’s algorithms for understanding traffic patterns and driver safety, for instance, as well as for estimating—and manipulating—supply and demand through surge pricing and other techniques, including what some have described as “phantom cabs”. The data that Mark and other drivers produce is also an invaluable business asset, helping Uber develop new partnerships with both municipalities and other corporations, and for maintaining its competitiveness.
“They not only direct every aspect of a driver’s workday, they also profit off the entire day through data collection.”
Amid a growing recognition that data is an essential component of running a technology business, some drivers and labor advocates are considering drivers’ dead miles as unseen labor, and are wondering if Uber’s data collection constitutes a new kind of wage theft.
“Uber is the closest thing to an employer we’ve ever seen in this industry,” said Bhairavi Desai, founder of the New York Taxi Workers Alliance. “They not only direct every aspect of a driver’s workday, they also profit off the entire day through data collection, not just the ‘sale of a product.’”
Unlike Facebook and Google, which trade users’ data in exchange for their free services, riders and drivers may be thought of as more traditional customers and employees, paying and receiving a monetary fee for a service. Drivers pay a hefty cut for the ability to accept rides, between a 20-30 percent fee to Uber. But, like Uber’s users, its drivers are also creating unseen value outside of those transactions.
The data e-hail platforms collect from drivers. Graphic by Motherboard
“Uber sees its drivers the same way as it sees cars or roads or passengers or anything: as a source of potential value to be extracted,” said Douglas Rushkoff, professor of media theory and digital economics at Queens College. “Asking whether drivers should get compensated for the data they create when they’re ‘off duty’ is certainly a valid question.”
Uber is not the only on-demand service to gather data about its contractors even during “dead miles”: Lyft also routinely collects data from its drivers even when they’re not engaged in a fare. But questions about Uber’s collection of driver data coincide with growing disputes over the nature of piecemeal work on its platform: as observers raise concerns about the company’s size and competitive advantages over its rivals, it has also become the focal point of legal challenges from workers in the so-called “sharing” or “on-demand” economy.
Amid privacy and security lapses, drivers have also complained that Uber has not done enough to protect the data it collects about them. And in a grim turn, some Uber drivers note, the data they generate may be used to eventually supplant them. Uber’s CEO Travis Kalanick has made it clear that the future of the company is to replace drivers with a fleet of self-driving cars.
Dead Miles, But Data-Rich
On online forums for Uber drivers, many message threads are devoted to how to minimize dead miles. According to the nearly two dozen drivers I interviewed in New York City and around the country, dead miles take up somewhere between a third and half of the time that they are working. “Working” here meaning in their car, with the Uber app open. For some drivers, like Arjun, an Indian immigrant who drives for Uber 10 to 12 hours every day of the week, dead miles can take up most of his work day.
“If I’m driving for 12 hours in a day, I can spend maybe 7 hours without a passenger,” he says. The 12 hours does not include his commute into New York City from New Jersey.
For other drivers, dead miles could be as low as 10 percent of their time. But only if they work selectively at the most high-demand hours of the week: rush hours and weekend nights. In general, drivers who use Uber as a source of supplementary income (for instance, students or retirees) were more selective about when they drove and had fewer dead miles. But drivers who rely on Uber to pay their full cost of living tend to work more hours but with more down time between fares, in a cycle of diminishing returns. Some have questioned whether Uber drivers are making minimum wage after accounting for gas, insurance, and depreciation of their car. (On Monday, hundreds of drivers protested recent fare cuts in New York City outside the company’s local office.)
Drivers worry that data mined during dead miles may be resulting in yet more dead miles.
Uber declined to share its data on driver dead miles with Motherboard, but said that it is devoting resources to reducing them for drivers.
“When drivers log into Uber, they want to spend as much time as possible earning money and transporting riders,” an Uber spokesperson told Motherboard. “We’re constantly looking at ways we can use technology to decrease down time between trips. For example, we recently added a feature so that drivers can accept a new ride request even before they’ve ended their current one.”
One easy way for Uber and Lyft to reduce drivers’ down time between trips, some drivers argue, would be to limit the number of cars on the road in over-crowded markets. Uber has explicitly rejected such limits, claiming that restricting Ubers would have an impact on local job creation.
“Gathering data especially during dead miles also explains Uber’s vitriolic opposition to capping their number of vehicles,” said Desai of the taxi workers’ alliance, referring to an unsuccessful proposal put forward by City Hall last year. “It’s not just that they simply don’t care drivers are burning fuel empty, it’s that they are still benefiting from it.”
The fact that Uber collects data from drivers that it can turn into profits wouldn’t be an issue if Uber drivers were considered employees, rather than “independent contractors” or freelancers. In addition to getting benefits and certain legal protections, as employees, Uber drivers would earn an hourly wage. So their time driving dead miles would be compensated. But the company has gone to great lengths to avoid having its drivers legally classified as employees. In California, where it is based, the company is currently the subject of a class-action lawsuit that seeks to have drivers reclassified as employees.
Uber drivers at a protest in New York this week over the company’s decision to lower prices. Image: Evan Rodgers/Motherboard
Much of Uber’s argument for classifying drivers as independent contractors rests on the notion that it is a technology company focused on “ride-sharing” and not a taxi company. Uber’s legal team first outlined this identity in a filing with the California Public Utilities Commission in 2012: “Uber is a technology company that licenses the Uber App to transportation service providers. The transportation service providers pay a fee to Uber to use its software technology; the passenger of the transportation service provider pays the transportation service provider for transportation services received.” As Uber’s Kalanick explains, “Uber is a technology platform that connects riders and drivers.”
That innocuous sentence has enormous ramifications for how Uber is regulated. It means that Uber drivers are considered independent contractors who simply utilize the Uber platform to conduct their business. And it means that Uber doesn’t need to abide by the regulations that would otherwise apply to a taxi company.
This self-identity—a tech company—also suggests the particular value of data to Uber: its product isn’t rides per se but the software that is built upon data from its riders and drivers. On its hiring page this week, Uber lists dozens of open positions in its “Advanced Technology Center” and seeks dozens more engineers and scientists for its data science team.
The Internal Value of Data
Uber extracts value from the wealth of data that it collects from drivers and passengers in a number of ways. Most directly tied into its core transit product is Uber’s routing algorithm, which identifies patterns in traffic to find the most efficient route for a vehicle to take and to determine a driver’s estimated time of arrival. Related to that are algorithms that decide which driver to send a pickup request to, knowing where other drivers are now, and where future passengers are likely to be. Supplied with enough data, the routing algorithm can make basic inferences about traffic, like which roads are best at which times of day.
“If Uber is going to be true to its model and say ‘we are a technology company,’ then their business model needs to reflect that,” says Spencer, a former Uber driver who quit when the company cut rates in his local market last spring. He’s now studying to become a data scientist. “And if they’re going to be collecting information on how drivers are getting around while they don’t have a paying customer, then any data the company receives from an independent contractor should be compensated.”
“Uber lives or dies by data. Their ability to increase profits is all dependent on that.”
Spencer used to drive for Uber on weekend nights and evenings, when he could earn a good amount of money quickly and without much dead time—drunk twenty-somethings are an Uber driver’s bread and butter. But with the rate cut, he decided that it wasn’t worth it for him anymore. “I realized I was basically trading equity in my car for cash now,” he says. His rare combination of familiarity with big data companies and time working as an Uber driver gives him unique insight into Uber’s practices.
“Uber lives or dies by data,” he says. “Their overall mission and their sustainability is completely dependent on how good their data is. The more data they can collect, the more information they can derive from patterns and behaviors. Their ability to increase profits is all dependent on that.”
The more data the algorithms have to work with, the more accurate they get. These algorithms are essential to Uber’s business and help to ensure Uber’s competitive advantage in its own marketplace against rivals like Lyft and Sidecar, the latter of which recently folded partly, its co-founder said, under pressure from Uber.
Surge Pricing And Data Sharing
Drivers’ data also helps Uber determine surge pricing, a feature that has drawn particular scrutiny from riders and drivers alike. Although the company claims that surge pricing is just a reflection of supply in demand, research by some of my colleagues at Data & Society suggests that the surge is not straightforward: Uber creates the mirage of a marketplace that obscures how its algorithms manipulate supply and demand. Some drivers understand this intuitively.
“There are times when I’m going to a surge area and the surge just stops,” Mark tells me, as we wind through lower Manhattan where he used to work. “I feel like they just say that to get you there, expecting it to get busy.”
By knowing where drivers are and where they’re needed, Uber can use surge pricing to encourage drivers to head to those hotspots, in theory reducing wait times for passengers as well as dead miles for drivers. But the result can be an influx of more cars than are needed, resulting in even more dead miles for drivers. Rather than reducing dead miles, drivers worry that the surge system may be adding more.
Experienced Uber drivers tell me they don’t believe that heading to surge zones is worth it. “I’ve been in surge areas just sitting there without getting any passengers,” Mark says.
Other drivers said they felt taken advantage of by surge pricing and confused by it. On message boards, frustration with surge pricing has even lead to a mantra that veteran drivers try to instill in new drivers: “Don’t chase the surge.”
“The surge pricing doesn’t make sense to me,” says Arjun. “Sometimes I see a surge but I don’t get paid for the surge price.” One explanation: drivers may respond to Uber’s prompts to head to high-surge areas but then get requests from passengers nearby who are by that point outside of surge zones or in lower-surge zones.
In their research, Rosenblat and Stark also identified a phenomenon some have called “phantom cabs,” in which the app misrepresents the location of cars on its map to its users. (Uber has denied its app does this).
Rosenblat and Stark say that phenomena like this points to a deeper concern. “Uber’s digitally and algorithmically mediated system of flexible employment builds new forms of surveillance and control into the experience of using the system,” they write, “which result in asymmetries around information and power for workers.”
Data from drivers’ phones can also be a valuable asset in Uber’s effort to maintain influence over its drivers and dominance in the ridesharing market. By knowing a driver’s location at all times, even when that driver is not carrying a passenger or using the app, Uber can also make inferences about that driver’s behavior, allowing it to discern, for example, if a driver is working for a competing service.
“I am concerned that Uber could be tracking my Lyft ride activity and could use that to discriminate against me when it comes to doling out ride requests,” says Brendon, who drives for both Uber and Lyft. “I don’t want Uber—or Lyft, for that matter—knowing what other activities I’m using my car for, whether it be another ride service or personal use.”
Last week Uber announced an ongoing pilot program in which it is monitoring gyrometer and GPS data in drivers’ phones for signs of reckless driving and to verify customer complaints.
“If a rider complains that a driver accelerated too fast and broke too hard, we can review that trip using data,” Joe Sullivan, Uber’s chief security officer, wrote in a blog post. “If the feedback is accurate, then we can get in touch with the driver. And if it’s not, we could use the information to make sure a driver’s rating isn’t affected.”
Data hasn’t just improved Uber’s core product: the company has also used data to establish lucrative partnerships with its partners and, through selective releases, to build relationships with regulators in new markets.
Last year, amid a push for stricter regulations on Uber’s operations in Boston, the company pledged to turn over its quarterly trip data to the city, helping to secure its ability to operate in a metropolitan area with over four and a half million potential customers. Uber also offered up some of its trip data in New York City, where it was facing scrutiny over its traffic impacts, and has made a general offer of data sharing to any municipality. Data caches like these may help planners improve city planning and reduce traffic; Time.comimagined “how Uber could help end traffic jams.”
Otherwise, the company has been protective of its data, even as it has raised concerns about how it uses it and how well it secures it. In 2014, reports emerged that employees were using a “God Mode” feature to track individual users. Last year, after the company changed its privacy rules to collect more user data, including even when customers weren’t using the app, the Electronic Privacy Information Center warned the changes were unfair, deceptive and posed a “direct risk” of consumer harm.
Security lapses have also raised concerns about the way Uber stores its mountains of personal information. A glitch last year exposed the personal data of hundreds of drivers, and in 2014 a lapse leaked the names and license numbers of 50,000 drivers. In January, New York state ordered the company to pay a $20,000 settlement for taking too long to notify drivers about the larger breach. The company has responded by recruiting more talented engineers, to boost a team that includes Sullivan, formerly chief of security at Facebook, and noted car hackers Charlie Miller and Chris Valasek.
Image: Marco Valtas / Flickr
What’s Data Creation Worth To Drivers?
While most drivers I spoke with believed they deserved to be compensated for the data they generate, not all did. When I explained the ways Uber uses the data it collects even while it’s between fares, Marwan, who immigrated from Yemen 25 years ago, just shrugged. “The guy who came up with Uber is very smart, isn’t he?”
Some drivers also didn’t see much of a distinction between Uber’s data mining and the kind that goes on every day on the web.
“If we’re going to scrutinize the fact that they’re potentially learning from the paths we take—consumers and drivers alike—we need to scrutinize the industry writ large”
“Just about everything we do today is tracked by some entity for data purposes, so I’m not sure if Uber should be singled out for compensation,” says Brendon.
As users of free internet services, we produce valuable data all the time: Facebook and Google are two of the most obvious examples. If you’re not paying for the product, the saying goes, then you are the product.
Gilad Lotan, chief data scientist at Betaworks, sees Uber’s use of data along the lines of that of Netflix or Amazon, which provide paid services while learning from data in aggregate.
“Should we consider drivers any differently?” he asks. “If we’re going to scrutinize the fact that they’re potentially learning from the paths we take—consumers and drivers alike—we need to scrutinize the industry writ large: Google, Facebook, Amazon, Netflix, and many more companies building services based off of aggregate data.”
Uber’s collection of aggregate data has an endgame, and drivers aren’t a part of it. In the future, one of the services Uber intends to offer is self-driving taxis. CEO Kalanick has made that goal explicit. “When there’s no other dude in the car, the cost of taking an Uber becomes cheaper than owning a vehicle,” he told a conference in 2014. In May, Uber poached most of Carnegie Mellon’s robotics department to work on developing autonomous vehicles for the company. Kalanick has no qualms about the fact that his company’s fleet of drivers will eventually be replaced by self-driving cars, saying he would just tell them that “this is the way of the world and the world isn’t always great.”
To have a functioning fleet of optimized self-driving taxis—and compete with potential competitors like Google—Uber needs a massive amount of data. It needs to know the best routes to take all the time, every time, in every condition. By producing this data, Uber drivers may be creating the conditions for their own demise. “People who rely on Uber as their primary source of income are eventually going to be obsolete because of autonomous cars,” says Spencer.
Until then, drivers will continue to generate valuable data for Uber, even when they’re not making money. That kind of data collection has become critical to the business of virtually all technology companies, as essential to their growth as the ads they sell and the rides they arrange. The experiences of Uber drivers makes this relationship with data more physical, if no less complicated: Data doesn’t materialize out of nowhere—it’s made by people simply driving around, trying to earn a living.
Jay is a freelance journalist covering the intersection of technology and politics. He is currently journalist-in-residence at Data & Society Research Institute. Follow him at @jcassano.
