Dying to Know Uber's Secrets, Data-Hungry Cities Get Creative

Researchers want to know how ride-hailing companies are affecting their streets, but don't have much information to help them.
Uber Lyft and other ridehailing companies are clearly changing how people move and work but don't much like sharing data...
Uber, Lyft, and other ride-hailing companies are clearly changing how people move and work, but don't much like sharing data with researchers eager to know the details.HOTLITTLEPOTATO

Less than a decade after a startup called UberCab launched in San Francisco, there’s a growing sense in American cities that things have changed. Ride-hailing services like Lyft, Juno, Gett, Via, and of course Uber have upended how people travel around the places where they live. A ride is faster, cheaper, easier to find, and maybe even safer than ever before.

Still, the spread of these services leaves plenty of questions about what, exactly, has shifted—and whether it’s all for the better. Who is using ride-hailing apps? How often? Have these services put extra cars on city streets, or exacerbated congestion and air pollution? And are the companies’ “driver-partners” making enough to live?

The answers inform how cities make all sorts of decisions, about traffic patterns, public transit, even the social safety net. But finding them requires data from the companies in question: How many cars are they using? How much are they charging? Where are their riders going, and when?

So it's too bad Uber et al.’s love for sharing doesn’t extend to its data. Turning over intel raises questions about rider and driver privacy. It could endanger proprietary info, the secret software sauce that gives them an edge over competitors. Worst of all, it could make them look bad—justifiably or not.

Fair enough, but frustrating for the researchers and cities eager to understand what’s happening on their streets. Especially frustrating given how much ride-hailing companies talk about cutting down car ownership and emissions, and giving workers the flexibility they need.

“There’s a lot of talk about wanting to be collaborative, but when it comes time to actually walk the walk, a lot times no data is forthcoming, or the data that's forthcoming isn't useful in any way,” says Joseph Castiglione, the deputy director for technology, data and analysis at the San Francisco County Transportation Authority.

Not to be denied, many researchers get creative. To determine how many and where ride-hail vehicles were operating in San Francisco in late 2016, savvy Northeastern University data scientists tracked drivers looking for riders by systematically pinging Uber and Lyft APIs every five seconds. From there, the agency extrapolated the vehicles’ movements around the city. Et voila—a comprehensive (if conservative) map of the pickups and drop-offs around town, which Castiglione’s agency can use to shape traffic management policies. (The agency is also in talks with the companies about data sharing pilot projects.)

Other scientists have collected their own ride-hail numbers by sending students to catch rides en masse, or signing up to drive themselves.

Sometimes, the inventive approach yields mixed results. Last week, researchers funded by MIT’s Center for Energy and Environmental Policy Research published a troubling research brief. The non-peer reviewed working paper found that 74 percent of Lyft and Uber drivers across the US earn less than the minimum wage in their state, with a median profit of just $3.37 an hour after accounting for driving expenses—before taxes. Without access to official Uber data on driver salaries, the team, headed by Stephen Zoepf of Stanford’s Center for Automotive Research, used numbers from a 2017 survey of driver-partners conducted by the driver news hub The Rideshare Guy.

Uber’s response was swift and harsh. “MIT = Mathematically Incompetent Theories (at least as it pertains to ride-sharing),” Uber CEO Dara Khosrowshahi tweeted, with a link to Uber chief economist Jonathan Hall’s response to the work.

Hall argued the MIT team misinterpreted the survey results and skewed its calculations. He pointed to another recent paper he wrote with Stanford researchers, which found gross hourly earnings between $21.07 and $15.80 per hour. On Twitter, academics who have collaborated with Uber’s research and policy team on their own work criticized the paper, too.

On Monday, Zoepf acknowledged the problems with his work and re-crunched his numbers, finding a median profit somewhere between $8.55 and $10, still below the minimum wage for many drivers. He says he will further revise the study. Zoepf also motioned toward the root of the problem, calling on Uber to “help make an open, honest and public assessment of the range of ride-hailing driver profit after the cost of acquiring, operating, and maintaining a vehicle.”

Other researchers sympathize. “I ask [Uber and Lyft] for data all the time,” says Bruce Schaller, a former New York City transportation official turned consultant, but notes he hasn’t had luck receiving numbers directly from the companies. Today, Schaller gets much of his info from New York’s Taxi and Limousine Commission, which began forcing private for-hire vehicles to turn over data on operations last year. But other, smaller cities don’t have that clout.

Uber, for its part, runs a policy and economics shop, staffed by data scientists outside researchers regard as sharp and rigorous. Its Movement platform also provides data on traffic conditions, though it does not furnish the granular numbers researchers need to answer questions about where riders are coming from and going. The company also says it’s looking for new ways to portion out data. “We’re actively looking in a bunch of different directions to be able to share data in different ways and for different purposes,” says Andrew Salzberg, who heads up transportation policy at the company. (Lyft did not respond to requests for comment on its data-sharing policies.)

Indeed, accessing data from any private company can be a drawn-out hassle, involving complex data sharing agreements. “These things take time,” says Susan Shaheen, a UC Berkeley engineer who studies transportation. “These relationships take time.” Her Berkeley team is in the midst of a multi-year collaboration with Uber, Lyft, and the Natural Resources Defense Council on ride-hailing and emissions. She expects to publish a study by the summer.

But that “slow down and think a bit” approach is at odds with the speed of transformation in American cities, and the American labor landscape. It does not help that ride-hail companies have been less than truthful about driver pay in the past. (Last year, Uber agreed to pay $20 million to settle a Federal Trade Commission lawsuit alleging it misled drivers about how much they could earn driving for the service.) Some officials and researchers want to see Lyft and Uber offer up specific data to those who do not agree to co-author papers with their in-house scientists.

“How are we going to move forward given that things are moving so quickly now?” says Shaheen. “How are we going to get past this data impasse and develop those really rich, robust, relationships to respect and trust around these very sensitive questions?”

One possibility: Some kind of third party, who mediates sharing between private and public. Perhaps the government? “It is very important that accurate, objective data is available on workers’ earnings, expenses, and other aspects of work in new and existing industries,” says Alan Krueger, a Princeton University labor economist who has worked with Uber. He has called on the Commerce and Labor Departments to collect data from platform companies to give everyone a better understanding of how the “gig economy” works.

Zoepf, who authored that controversial MIT research, says he’s hopeful, too. “This has been a tough episode on all sides, but it is my sincere hope that it leads to a healthier relationship in the long run,” he said in a statement to WIRED. “It is in everyone’s interest to allow people to make fully informed decisions about when and how to participate in the gig economy. There are lots of smart, good people inside these companies, that want to not only make a profit but also generate the best information to ensure that good decisions are made.”


Street Smarts