Photo by Stevesworldofphotos on Flickr.

Most of us only see ghosts when people dress in costume on Halloween, but bus riders deal with “ghost buses” on a regular basis. These are not spirits haunting your ride to work, but Metrobuses that mysteriously disappear, or never appear, on the NextBus real-time prediction system. The system now has fewer errors, but riders still encounter problems.

Ben Ball, a representative on WMATA’s Riders’ Advisory Council (RAC), offered an example of the “ghost bus” problem:

Every day at around 4:40 or so, I check the Metro website to see when the next N2 or N4 bus will approach Ward Circle heading eastbound to Farragut Square. There’s usually an N2 in around 8 minutes, and an N4 at around 10 minutes. This week, both are displaying 24 minutes, and there is no bus displayed on the arrivals map. But that doesn’t mean that there’s no bus.

In fact, when I went toward the Red Line on Monday based on the fact that the next N2 wasn’t supposed to come for 24 minutes, an N2 streaked right past me as I was walking up Nebraska Avenue. And then there was another one at Tenleytown — two ghost buses in a row.

When a bus disappears from the system, the most likely culprits are a lost GPS signal, a bus logging off the system, heavy traffic, a detour, or a bus breakdown, said WMATA chief spokesman Dan Stessel. A bus will disappear from the NextBus system if it stands still for more than 2 minutes or deviates too far from the assigned route.

A bus arriving without ever appearing on NextBus can happen when a driver doesn’t log on, equipment fails, or the data feed breaks, Stessel added. This problem has receded to some degree. Before 2011, Stessel said, some 300 buses ran daily without active GPS signals and radios. This year, there are just 40-50 on an average weekday.

WMATA re-launched NextBus in 2009, after discontinuing service in fall 2007 due to accuracy problems. NextBus receives data from GPS locators on the buses and uses them to estimate when a bus will arrive at a given stop. Users can access bus predictions from the WMATA website or through a mobile app.

WMATA uses two metrics to measure NextBus’ performance. “Predictability” is how well the system locates each bus and determines when it will reach a stop. This depends on factors Metro is responsible for, such as bus drivers logging on, working equipment and accurate schedules and stop locations. Predictability has improved over the past 2 years, Stessel said. WMATA gave itself a predictability score of 87% in April of this year, up from 85% in April 2011 and 77% in April 2010.

The second metric, “accuracy,” depends largely on the NextBus software. A prediction counts as “accurate” if, when the system predicts a bus to arrive within 5 minutes, the bus actually arrives within 3 to 7 minutes. On this metric, NextBus scores above 90%.

What has WMATA done to make the predictions more realiable? Stessel explained that Metro set up an education program and uses performance center monitoring to ensure that bus drivers remember to log into the system and stay on throughout their shift.

Metro has also upgraded onboard radio and communications equipment, and has a project underway to replace existing systems with more modern technology. The newer systems will “poll” buses’ GPS locations every 30 seconds or less; the older systems only “poll” every 120 seconds, meaning that buses can travel a fair distance before the NextBus system knows about it.

Still, Stessel said, there is no way to completely resolve prediction issues. “While the new technology will greatly increase reliability and data availability, factors like detours, traffic and weather” will always play a role.

In the meantime, Ball laments that there is no easy-to-use system for reporting bus outages to WMATA. Former RAC chair Dennis Jaffe noted in 2010 that the generic feedback form is complicated and hard to use. There is also no easy way to report NextBus errors from the WMATA bus prediction interface or the mobile apps.

What have your experiences been with NextBus?

Jeremy Barr is a graduate journalism student at the University of Maryland. He previously worked in non-profit communications and has interned in politics on several occasions. In the last year and a half, he has lived in Adams Morgan, Logan Circle and Mount Vernon Square. Email him at