Laura Bliss is a staff writer at CityLab, covering transportation and technology. She also authors MapLab, a biweekly newsletter about maps (subscribe here). Her work has appeared in the New York Times, The Atlantic, Los Angeles magazine, and beyond.
It takes a volunteer army of student mapmakers to build a better transit tracker.
On a a glorious 75-degree afternoon in early May, students at the University of Maryland, College Park sprawled in tank-tops and bathing suits on the quad, hatching post-finals plans. But not all students: In a windowless computer lab in the geography building, a handful of industrious undergrads were glued to banks of monitors, mapping buses.
Stop by stop, pick up by drop off, they traced the trajectories and timetables of a Loudon County, Virginia commuter bus service into the backend of Moovit, a partially crowdsourced transit navigation app. Or at least they were trying to. “I was in the wrong Virginia!” shouted Kurt Willson, a freshman GIS major, as he clicked through OpenStreetMap for the location of a stop on the online platform.
Soon enough, he found it: a church on the corner of 33rd and Main St. in Purcellville. He hovered his cursor over the spot on the map and clicked to add the bus stop icon. Later he’d enter in all of the headways for the Washington, D.C.-headed route he’d been assigned. “One down”—Willson cocked his head to scan the Loudon County website for the timetable he’d pulled up—“29 to go.” He groaned goodheartedly.
Moovit relies on these kinds of mapathons—often student-led—to help power their transit app, which company representatives say is available in 2,000 global cities. (I also recently wrote about this effort in the MapLab newsletter.) Like competitors such as Transit and CityMapper (which cover 125 and 39 cities, respectively), Moovit uses transit schedule data to display information and suggest the best routes across several non-driving modes: buses, trains, bikeshare, walking, and ride-hailing. It also tries to predict bus arrival times more accurately, thanks to algorithms that pull in from information from GPS trackers onboard buses, as well as variables like the historic performance of a route, the weather that day, ridership patterns, and crowdsourced incident updates.
To do that accurately, Moovit requires meticulously accurate maps for its estimated 4.7 million bus stops. That’s where the UMD kids come in. Many of them are members of the UMD chapter of Gamma Theta Upsilon, a geography honors society. In 2017, as one mapathon among many throughout the school year, the club had entered the university’s own extensive bus network onto Moovit; this year, the students targeted Loudon County coaches after Moovit conveyed requests it received from commuters there.
Although none of them had ever used these particular buses, many understood the importance of a decent transit app.“The D.C. Metro app is really, really bad,” Joseph Cunningham, a senior studying GIS, told me. “They never update it, and it’s always crashing on my phone.” Cunningham was painstakingly plotting each stop along the route he’d been assigned. “Just looking at this already, I feel like it makes things more user-friendly,” he said.
In that sense, these students were one block in a long chain of human and computer labor marshaled in the pursuit of a still-elusive goal: a transit app that really works for bus riders, everywhere. Anyone who has ever stood at a bus stop knows how interminable the anticipation can feel when there’s no indication of when the vehicle is coming. Even systems that have “real-time” bus information, thanks to onboard GPS trackers, can still be highly unreliable. Server crashes, traffic incidents, slow passenger boardings, and the weather regularly mess up arrival estimates. Research shows that bus waits actually feel longer when riders lack reliable information at hand.
The reverse is also true. Kari Watkins, an assistant professor at Georgia Tech who has researched transit rider behavior, led a Seattle-based study that found that simply adding real-time information decreased riders’ perceived wait time by 0.7 minutes. That information came in the form of OneBusAway, an open-source software that Watkins helped develop, which combines location data from the buses and a predictive algorithm. A number of cities have adopted OneBusAway as the basis for their own apps, and have seen positive results: For example, New York City saw an increase in weekday ridership of nearly 2 percent, attributable to real-time mobile updates, in 2011 to 2013. “When you’re gaining a sense of control over your trip, it helps you feel more willing to take it,” Watkins told me.
That might count for a lot right now. Public transportation systems, especially buses, are in trouble. Ridership has dropped to its lowest point in 30 years on the nation’s rubber-tired workhorses. Several factors seem to be pushing people off, including cheap gas prices, the popularity of ride-hailing, and service cuts along important routes that haven’t been restored. And in cities where traffic is getting worse, buses are getting slower and less reliable, which is feeding a vicious cycle.
Meanwhile, an expanding universe of multimodal navigation apps and predictive software is trying to address that uncertainty, both for consumers and transit agencies themselves. Optibus, for example, is a firm that works with cities to plan bus systems using artificial intelligence that can catch inefficiencies and redundancies. Part of the benefit, they say, is that the oft-unreliable “real-time” data that is given to the public becomes more accurate. Moovit also has several products that it sells to transit agencies, one of which is a similar “real-time data” solution.
It’s hard to say which transit apps and prediction services are the most reliable. “To my knowledge, no one has done a scientific comparison of times shown within apps,” said Sean Barbeau, the principal mobile software architect at the Center for Urban Transportation Research at the University of South Florida. What’s more, with the exception of OneBusAway, which uses open-source software, they’re all “black boxes”—their algorithms are proprietary and haven’t been assessed by researchers. Moovit likes to call itself the “Wikipedia of transit maps” because, according to representatives, it crowdsources about 60 percent of its data from millions of volunteer transit mappers around the world. (Its competitors do not.) But unlike Wikipedia or, for that matter, OpenStreetMap (which Moovit uses as one of the layers for its editing platform), it’s not actually open-source.
At UMD, however, none of the students seemed fazed that they were donating their time to build the underpinnings of what is ultimately a private service. For many of them, the motivation was simply to help transit users, or “to give back,” as Willson put it.
That, and a sheer love of maps. “Everything has a place and a reason why it’s there,” said Andrew Lazara, a junior majoring in geography who helped coordinate the event, as he quickly mapped his route. “Connecting each of these points, you’re sort of making history. And you’re creating the ability for other people to form future connections, too.”