Sarah Holder is a staff writer at CityLab covering local policy, housing, labor, and technology.
Uber drivers in Europe and the U.S. are fighting for access to their personal data. Whoever wins the lawsuit could get to reframe the terms of the gig economy.
Over two years of driving for Uber, James Farrar logged thousands of miles on the app. Many weeks, he’d work more than 80 hours behind the wheel of his Ford Mondeo, crisscrossing the streets of London deep into the night. Along with passengers, Farrar was collecting data.
During his time in the car, Uber’s app recorded where he went, how long he stayed, how much money he made, and how many stars he was given by his passengers. It noted how many rides he accepted and how many he cancelled, mapped where trips started and ended, and how long it took him to wind through traffic to get there as he followed the algorithmic cues nudging him around the city.
Being an Uber driver, Farrar found, did not agree with him: “[L]ife behind the wheel,” as he wrote in a recent op-ed for the U.K. Independent, “can become a blur of endless traffic, crushing loneliness, enduring fatigue and relationships strained by absence.” And he grew frustrated by the way the app always seemed to be pushing him to accept more rides, while his earnings kept declining. In 2016, he and another London Uber driver, Yaseen Aslam, brought a worker’s rights claim against the company, arguing that drivers aren’t true independent contractors and should instead be classified under the U.K.’s third employment category of “worker”—entitling them to minimum wage and paid vacation days. Farrar’s team won the classification case.
But during the trials, Uber was able to use Farrar’s personal data as legal ammunition against him, he said; the company argued that the reason he made under minimum wage some days was because he’d declined several rides, not because he was being fleeced by the app. “I decided then that I needed to see all my data,” Farrar told CityLab in an email. “[S]o I could properly assert my rights and eliminate the asymmetry in information power between me and Uber.”
Uber appealed, arguing, as it long has, that it merely connects independent entrepreneurs with riders, and that a change in classification would impede drivers’ freedom. (A judge said that the wording of Uber’s contract, which makes the same claim, contained a “high degree of fiction.”) Still, one judge out of three backed the company, and Uber was given permission to bump the trial up to the U.K.’s Supreme Court.
With the support of Ravi Naik, the lawyer who is also representing plaintiffs in data privacy cases against Facebook and Cambridge Analytica, Farrar and three other drivers have bundled their data requests, eking out more information with each challenge. This March, they filed a lawsuit against Uber for withholding some data, which they say is in breach of the European Union’s General Data Protection Regulation (GDPR). This law gives E.U. citizens the right to request any and all personal data that a platform retains about them.
Though he says he’s not “anti-Uber,” Farrar’s labor rights activism has accelerated: In 2017, he helped found and became the chairman of the United Private Hire Drivers branch of the IWGB union, which represents more than a thousand workers for private hire companies. And as of this summer, he has pushed more than 60 other drivers to file similar data claims. Now he’s pooling their information online as part of an organization he founded and directs called the Worker Info Exchange. Since Uber arbitrates cases from all its worldwide markets besides North America in Amsterdam, an E.U. country, these GDPR-based claims could be replicated in dozens of countries.
With that aggregate, he wants to determine definitively how much (or little) drivers make for their time—and how an over-supply of temporary drivers has saturated the market with idling cars. “The negative effects of these apps are congestion and poverty, and we need the data to show that,” Farrar said.
This tension isn’t an entirely new one. Mounting evidence suggests that all kinds of apps, from the Weather Channel to the selfie-filtering Perfect365, have habitually scraped location data from phones and used it to better predict consumer buying habits. And Uber’s tight grip on information is part of a long pattern for the ride-hailing titan, which has often tangled with local regulators over its vast trove of trip data. But Farrar believes that whoever gains access to Uber’s complete data caches will find something bigger than just tools for better traffic management: It might be the key to creating a more equitable gig economy. And Farrar is prepared to wage a years-long legal battle to ensure that drivers are the ones to get it.
Who’s the algorithmic boss?
For ride-hail companies, the hunger for information is clear: Uber uses real-time location data to route drivers more efficiently through the city, and could eventually feed the troves of geographical knowledge to a fleet of self-driving cars. When collected, aggregated and made available to a broad community of drivers, however, it could help them make smarter decisions about how and where to drive. And in the hands of city governments, it could inform transportation and labor policy-making. In New York City, it was detailed trip data that helped them craft the U.S.’s first ride-hail minimum wage, for example, and a cap on ride-hail licenses.
There’s a variety of trip and payment info already available to drivers through their apps, like when they logged on and off, where they picked up and dropped off drivers, and how much they made per mile versus what Uber made in commission. And not all drivers want (or need) any more than that. “I know when to work [and] how to work, because I’ve been doing it for four years,” said Bill Steigerwald, a former journalist and Uber driver based in Pittsburgh. “I’m not sure if there’s anything that Uber could tell me that would help at all. I feel like I know this market better than Uber does.”
But from Uber’s disclosures so far, Farrar has already found much more: Along with the trip history, he has cumulative GPS data. He also has dispatch data, and details about how “productively” he spent his time on the platform, either waiting for work, traveling to pick up a passenger, or carrying out a fare-paying trip. These last points will prove important, he says, because in its 2016 ruling, the court decided drivers should be considered protected workers from the moment they log in to the platform until the moment they log off.
“Once we win in the Supreme Court—and I believe we will—then the case will be kicked back down to Tribunal level where we and Uber have to calculate what is owed,” Farrar wrote in an email. “It is at this point having our own view of the data will be key.”
There’s still more information out there, though: Uber hasn’t published the GPS data it collects from drivers while they’re not completing a trip (so-called “dead time”), or the driver profiles that Farrar thinks Uber culls from customer service interactions and trip history. (“The only profiles we have for drivers are those visible in the driver app and the rider app,” an Uber spokeswoman said in an email. “This information is always available to drivers in the partner portal, regardless of GDPR.”) The company has held riders’ ratings and reviews tightly, too, citing privacy concerns. Drivers are given a recalculated average rating at the end of each day, but they don’t have a granular sense of what went wrong, when.
Having the ratings data could support Farrar’s case that Uber performance-manages its drivers, he says. Though the Uber spokeswoman said that “[d]rivers are contacted directly if their rating falls below the average for their city so they have a chance to adjust their behavior BEFORE a deactivation,” several drivers have reported being deactivating from the platform without clear warning or explanation, or believe they’ve been sent fewer ride-requests as a result of perceived poor service. If proven, Farrar thinks it could be evidence that drivers are actually controlled by the platform via “algorithmic bosses,” as tech ethnographer Alex Rosenblat writes in her book Uberland.
“Uber is trying to avoid any sign that they might be managing people or this might be an employment relationship,” said Farrar. “They want to avoid saying they’re dismissing you because you’re not doing the job right.”
Uber says they’ve released all the data they can and are mandated to. “Our responses have been within the guidelines published by ICO (UK data authority) and we’ve provided millions of lines of data in response to these requests as well as detailed explanations on why we cannot provide some of the specific data he requested,” an Uber spokeswoman told CityLab in an email. “[E].g. some of it never existed in the first place and some of it can’t be shared without infringing on the rights of other individuals.”
(There’s disagreement about whether gig workers have the same kind of data-access rights that other users of a digital platform would under GDPR, because the lines between personal information and proprietary business intel are blurred, as The Economist reported this May.)
Even millions of individual data points won’t be enough to truly articulate how Uber controls its drivers’ movements, says Uberland’s Rosenblat, because the logic of the algorithms that nudge them is always changing and recalibrating.
But for drivers, Farrar says getting a peek under the hood of the Black Box could spark a U.K. labor revolution. What happens when drivers are confronted with how much they’re owed in back holiday pay, for example? And for policymakers, Rosenblat says it could give them a clearer understanding of the outcomes of those automated decisions, however dizzying—and a chance to legislate accordingly.
“Right now, so much of this discourse happens in a data vacuum,” Rosenblat said of the international debate over gig worker treatment. While Uber and Lyft can selectively deploy their own in-house data to fit narratives they like, she says, politicians and organizers are left clinging to what seem like feelings, or individual’s anecdotes.
“Almost everything we get from companies … [has] been shaped with lawyer’s inputs, to make sure that it tells only half the story or a quarter of a story, or only the story an employer wants to tell,” said Bill Sokol, a labor lawyer that practices in the Bay Area.
But harder, and more complete, information does exist, as efforts like Farrar’s show. It’s just locked up.
From the driver’s seat
In the U.S., the question of how to classify gig workers has also preoccupied lawmakers. California is currently debating Assembly Bill 5, a controversial piece of legislation sitting in California Senate committees that could redefine ride-hail drivers as full employees. Uber and Lyft, along with some drivers and union leaders, have raised concerns about what the bill might mean for driver flexibility; now AB5 is stalled pending more research.
But establishing stronger labor protections is not contingent on transforming employment status: After Uber and Lyft went public this spring, and ride-hail drivers organized local and international strikes in response, increasing wages—and wage transparency—has become an urgent priority.
What’s different in the U.S. are the mechanisms by which people can gain access to their data. Because the country isn’t covered by GDPR, individuals can’t make piecemeal data requests like Farrar’s. In cases where data has been unearthed, it’s generally been in bulk, under pressure from cities like New York.
Other times, companies have proactively released limited amounts of data, both to justify their expansion and to assist with local initiatives. Uber Movement, a data-sharing partnership between Uber and dozens of cities around the world, launched in 2017; it features visualizations of travel times and traffic speeds, and little else.
But there are also a few U.S. efforts to enhance drivers’—and cities’—ability to collect that information themselves, without having to rely on company disclosures. This year, a consortium of cities banded together to launch an open-source data project called Mobility Data Solutions (MDS), which allows cities to collect their own real-time info on scooter and dockless bike trips. Uber, Lyft, and Bird are supporting a bill to block it, citing fears that the data would not be stored securely. Many transportation experts believe those privacy concerns are valid, as my colleague Laura Bliss reported; others think the companies are acting in their own interest, which is to keep cities in the dark.
The Driver’s Seat Cooperative offers another, more personal, solution. The app, which is still in its beta version, allows ride-hail drivers to turn on a Waze-like GPS system that runs in the background while they work, recording their location. That data will then be aggregated, anonymized, and shared with a cooperative made up of participating drivers. Co-founders Matt Schumwinger, a former data consultant, and Hays Witt, a former organizer with the Partnership for Working Families, promise that drivers in the co-op will be able to be compensated for their industry insights, as well: Cities will be able to buy that data, and drivers will share in the revenue.
Witt, like Farrar, wants to use the information to help drivers earn more stable incomes, and cities craft more well-informed policies. “I heard from drivers that one of the key issues was they didn’t have good information that they could use to plan their day or make business decisions,” he said. “And at the same time, we were hearing from cities and municipalities that they were interested in improving ride-hail both for drivers and in communities they’re accountable to, in terms of providing quality efficient equitable transportation, and that they were unable to get the data they needed to do that.”
San Francisco is the Driver’s Seat Cooperative’s first local partner. Data from the group (accessed for free in its beta mode) will help San Francisco’s Local Agency Formation Commission (LAFCO) conduct a study on the state of the city’s gig economy. The effort is led by Bryan Goebel, LAFCO’s executive officer, who spent a stint working for on-demand delivery apps after a career in transportation journalism. “Having struggled with this kind of work over the past year, I was like, well, we don’t really have data,” he said. “I’m trying to figure out how can we work with companies, how can we amplify the voices of on-demand workers and find some solutions. Data is part of that.”
Goebel isn’t working towards a specific policy outcome like a city-wide ride-hail minimum wage, though he won’t rule it out. He says the goal is to “better understand the geography of on-demand workers … and provide some additional data on what their days are like.” How many hours do they work? How much money do they make? Do they travel by car or by foot or by bike, as Goebel once did, sweating as he crested the San Francisco hills? How has their experience changed over time? Once the study is complete, he’ll deliver it, along with a set of recommendations, to the San Francisco Board of Supervisors, who will decide what to do next.
The price of data
Uber and Lyft drivers often talk about ride-hailing work as a game; each day, they chase surges and pursue rewards. Harry Campbell—the blogger, YouTuber, podcaster, and former driver known as The Rideshare Guy—has built a media empire guiding drivers how to work smarter, not harder.
In a way, the gamification of the app makes the tasks feel self-directed, as a recent paper argues. Katie Wells, a postdoctoral research fellow at Georgetown’s Kalmanovitz Initiative for Labor and the Working Poor, and co-authors Kafui Attoh of the CUNY School of Labor and Urban Studies and geography professor Declan Cullen interviewed 40 Uber drivers in D.C. for a series of reports on how the on-demand economy affects workers, concluding, among other things, that the nature of the work breeds isolation. (Uber has taken issue with some of their findings for drawing on interviews conducted in 2016, before the company rolled out in-app tipping and an Uber Pro rewards system.)
As gig economy worker and researcher Sarah Mason argues in her Guardian essay, “High score, low pay: why the gig economy loves gamification,” Uber also deftly translates that “desire towards the production of profit for the employer.” Swap data in for profit—it’s easy to do, because one is easily converted into the other—and you have a more comprehensive picture of what driver labor really looks like, according to Wells. Idling, re-routing, and cancelling rides may not make drivers any money under the current system, but it gives Uber valuable info on what routes and times of day drivers prefer.
“[W]e’re their guinea pigs now,” said Joe, a driver Wells and her co-authors interviewed in 2016, whose name was changed to protect his livelihood. “We’re building that data for them, and they have it.”
It’s challenging to determine how companies should compensate for that extra labor, Wells says, or to say if they even should. What’s important is to start recognizing the processes—and the people—behind data collection, especially as the promise of the ultra-optimized Smart City gets closer to reality. “How could it be bad to know more?” said Wells. “The question is, how do we come to know that thing?”
Witt, too, says he’s wary of commodifying individual data points, instead stressing the more meta potential of data ownership and production. He wants to compensate drivers for the data they collect through the Driver’s Seat Cooperative, but his ultimate goal is to use it in a way that benefits the profession as a whole. “The value in the aggregate is much greater than the sum of its parts,” says Witt. “What’s really important is that people are thinking about how drivers can share in that value.”
Having the ability to shape city policy is one value proposition; gaining the same level of control as a true entrepreneur would is another. Drivers might not feel like guinea pigs if they, too, got to wear the white coats every once in a while.
It can also correct for the ways Uber and Lyft already deploy data, which is more strategic. “Platform companies give platform workers views of their own data that are designed to incentivize and spur specific behavior,” said Witt. They’ll tell drivers how many more rides they’ll have to complete before earning a bonus, for example.
These limited disclosures encourage them to stay on the platform longer, and direct them “to work in ways that are inherently isolating and that function to undermine anything approaching collective (whether that action involves a collective appeal for higher wages, or safer working conditions),” as Attoh, Cullen, and Wells write.
But in different hands, data could also break down those silos. For one thing, organizers could use it to strategize about which drivers to bring into their meetings, says Sokol, the labor lawyer. “If you can create an app that actually generates what I would call good evidentiary proof of those who are really employees doing the work on a regular full-time basis, versus those who are simply casual or part-time employees … you would basically know who is it you’re trying to organize,” he said.
It could also push both full and part-time drivers to feel engaged enough to act. “The most important power workers have isn’t always through the court, but through direct action—through strike and protest organizing,” says Farrar. “The data helps unlock that energy.”