It’s one thing to tell riders when the next bus is supposed to arrive – and include a symbol claiming the prediction is live and up-to-date. But it’s another thing to offer accurate information, as many “real-time” arrival predictions tend to be off by a few minutes.
A few minutes might seem like a small change to drivers. But to people riding transit, it makes a huge difference.
Mobility Lab’s recent report (pictured to the right) found that inaccurate real-time information (RTTI) sparked frustration with transit among riders in Arlington, Va. Some riders switched to other modes, while others stopped seeking out RTTI altogether.
This is a shame, because accurate real-time information has the potential to increase transit ridership and improve customer experience. Our literature review found that the availability of accurate real-time information can give transit users more control over their trip-making decisions, shorten wait times, and reduce overall travel time. Inaccurate information, on the other hand, can mislead people into waiting longer for the bus or even missing it.
While this study’s findings are specific to Arlington County, the idea that RTTI can influence travel behavior got us thinking: when does the accuracy of real-time transit information matter most? And how do we improve it?
Real-time accuracy can help make infrequent service more palatable
When I’m traveling to work in the morning, I almost never look up real-time information for my bus. Why? Because even though there’s no bus shelter at my bus stop, and even if the weather is brutal, I know that a bus will most likely be there within five minutes.
For bus routes with frequent, reliable service, accurate information may not be necessary. In Mexico City, for example, Mobility Lab contributor Andy Furillo found that the lack of RTTI was never a problem. The service was so frequent, with buses or trains arriving every three to five minutes, that he never felt the need to check when the next bus was coming, even in the evenings and on weekends.
But what about when transit service is not frequent or reliable? If a bus is scheduled to come only every 20 or 30 minutes and does not reliably arrive at its scheduled time, a rider wishing to wait less and make more informed travel decisions might want to know when it’s really coming (as our literature review within the report found). Riders can access this information through various means, including their computer (perhaps before leaving the office), a dynamic messaging screen (like Redmon screen), or a smartphone app.
When service is frequent, inaccurate real-time information doesn’t make much of a difference in travel plans. Even if one bus shows up two minutes earlier or later than the real-time prediction indicates, the wait time will probably be more or less the same.
But on infrequent bus routes, inaccuracies can cause missed buses and long waits. Over time, repeated inaccurate predictions might cause riders to distrust the technology or the transit system, negating the benefits of providing the arrival information in the first place.
When transit systems can’t increase service levels to the point that transit users don’t need real-time information, improving RTTI accuracy can still make public transit more usable and attractive as a transportation option.
If transit agencies can’t keep up with changing technology, the accuracy of real-time information may suffer
Given how advanced and ubiquitous technology has become, to many people inaccurate real-time transit information must seem like something out of the dark ages..
So, what’s the deal?
Many transit agencies use automatic vehicle location (AVL) and computer-aided dispatch (CAD) systems to collect real-time information data via GPS systems on buses. These systems originally reported locations every three to five minutes over the radio and can now report locations every 30 seconds via LTE or 3G networks. Some transit systems (such as the DC circulator) now use smartphones installed in each vehicle to transmit location information, which can be updated as frequently as every three seconds.
As our report’s literature review and previous Mobility Lab reporting show, there are a variety of methods transit systems can use to provide real-time information. Some of these methods are more accurate than others.
Technological changes don’t necessarily solve all glitches that can affect real-time arrival information. For example, in 2016, WMATA updated its Metrobus real-time system to increase accuracy, switching from NextBus to BusETA and joining the OneBusAway project. This new system combines on-vehicle GPS responders with computer modeling to predict bus arrival times. Greater Greater Washington, reporting on this new system, found that BusETA seems to be more accurate, but the issue of “ghost” buses (buses that don’t appear in real-time due to communication issues) and buses showing up a minute or two early or late persists.
But the inability of a transit agency to keep up might be the greatest barrier to real-time accuracy. In San Francisco’s MUNI system, the failure to update real-time technology was particularly disastrous, according to Andy Bosselman of StreetsBlog SF. At the beginning of 2017, NextBus (which MUNI still uses) did not have any location data for more than half of MUNI buses. Why? In 2002, MUNI installed GPS systems that used AT&T’s 2G network on each of the system’s vehicles. AT&T announced in 2012 that it would shut down 2G services in 2017, but despite the five-year warning, the San Francisco Municipal Transportation Agency (SFMTA) failed to update these GPS systems on about two-thirds of its bus fleet.
Given MUNI’s 57 percent on-time performance, the lack of real-time information exacerbated public frustration with the system. As Bosselman wrote, “The one thing that makes our public transportation system usable to thousands of people no longer works.”
Riders suffer from inaccurate information. But they could be part of the solution
Crowdsourced data from riders might be a solution.
One trip-planning smartphone app, Transit, now offers a crowd-sourced data option that its developers hope will improve the accuracy of the real-time information it provides its users. In the app, when a rider maps out a trip using public transit and clicks “Go,” the app uses their location to supply real-time information on the transit routes they take to other app users. Transit also tells you how many people you are helping by providing your own data.
Just one Transit user using the “Go” feature improves RTTI accuracy, but it has a network effect—the more people using this feature, the better. Transit partners with several transit agencies across the U.S. and Canada, such as Boston’s MBTA and Baltimore’s MTA, and serves as the official real-time and mapping application for these systems.
However, like other forms of RTTI technology, crowdsourcing faces limitations. In my experience, the Transit app generally does a good job offering accurate real-time information. But I’ve noticed major issues with how it tracks my location using the “Go” feature that I fear could lead to inaccurate data.
For instance, when I recently used the “Go” feature to track a morning trip on WMATA’s 38B towards the Ballston neighborhood of Arlington, the app showed that I was on the bus before it even got to my stop. And the other day, when I rode the 31 towards DC’s Friendship Heights, the app asked me which bus I had taken (I could have also taken the 30S or 33). The app initially bugged out and wouldn’t let me select an option, then automatically selected the 33 for me. If these issues are widespread, they may contribute to inaccurate RTTI on the app.
In addition, people without access to smartphones – who comprise 23 percent of the U.S. population, including 33 percent of people with an annual income under $30,000 and 56 percent of people over age 65 – can’t contribute to or benefit from this data. Transit agencies could help address this issue by partnering with crowdsourcing apps to provide this real-time information in other ways, such as on public display screens.
So while good real-time information can certainly make transit easier to use and improve the rider experience, in the end, technology and operations limit the extent to which it can mitigate the downsides of infrequent service.
Photo by Jeffrey/Flickr.