Hall Lane, in Newtown, CT, is similar to many other “country roads” in rural New England, flanked by stone walls and tidy homes built in faux colonial or ranch style.
It’s also a typical country road in New England in that it’s a “one-and-a-half” lane road that may or may not allow two smallish vehicles going in opposite directions to pass each other.
Newtown Chief of Police David Kullgren said on a typical day, Hall Lane gets maybe 125 to 150 vehicles on it, mostly local traffic, with no issues.
However, as the use of navigational applications on cellphones has become more ubiquitous, and vehicle traffic on nearby Interstate 84 has become heavier and heavier, Hall Lane became a poster child for how the apps can turn a bad situation worse. Instead of 125 cars and light trucks a day, every time there was an accident or other delay on the Interstate, the apps would send 600 drivers looking for a way around the holdup down the narrow lane—including cars, heavy trucks towing 50-foot long trailers, and full-sized buses and motor coaches. At least three of these heavy vehicles bottomed out on the small street and had to be towed away, causing even more delay.
It’s a worldwide problem that has been going on for nearly as long as navigation apps have been the de facto tool drivers use to get around. It’s been studied in academic labs and tackled by local jurisdictions such as Newtown.
Kullgren said Newtown’s solution, after studying traffic logistics for about a year, was to turn Hall Lane into a one-way street. So far, that has helped eliminate the “Waze effect” on Hall Lane; he said neighboring small streets have not been affected yet.
Yet there doesn’t seem to be any sort of consistent progress being made technologically to improve awkward rerouting.
‘They can’t do a whole lot of thinking’
Jane Macfarlane, director of the Smart Cities Research Center at the University of California, Berkeley and Lawrence Berkeley National Laboratory, has been tackling the unintended consequences of navigation apps for some time. In 2019, her article published in IEEE Spectrum mentioned small streets in Los Angeles, San Francisco, London, Boston, and suburban Tel Aviv that have suffered similar bottlenecks as Hall Lane. She said that not much has been done, as far as she can see, to improve things.
“Nothing has been done that I’m aware of inside of the machines that do the routing,” Macfarlane said. “There aren’t many players, obviously, but I don’t think a lot has been done on this topic with those players.”
In western Connecticut, one state legislator is proposing a regulatory approach to compel app developers to eliminate small local roads like Hall Lane from their algorithms. Mitch Bolinsky, who represents Newtown in that state’s house of representatives, has introduced HB 5071, which would “prohibit persons, firms, or corporations that operate a global positioning system or mobile application used for navigation purposes in the state from proposing routes of travel that detour heavy motor vehicle traffic from commuter routes onto local roads.”
The bill is still in committee, with no guarantee to advance, but Bolinsky said the obvious ramifications of the status quo—traffic jams that are transferred from a main arterial to smaller roads, damage to the roads that are not built to handle the weight of heavy vehicles, and potential danger to local residents and pedestrians from massive increases in the number of vehicles on those roads—make the effort worth it.
“Whether it’s a state highway or what I would call a local arterial, those roads are part of our commuter mix every day,” Bolinsky said. “They are adequate bypasses. They don’t contain parked cars and they have better foundations so they don’t get torn apart. We are trying to keep unfamiliar traffic from being re-routed onto roads that are just completely not appropriate for them.”
Macfarlane says implementing such a scheme is easier planned than done.
“It sounds like a good idea, but when you’re looking at a routing algorithm, you have to remember it becomes much more difficult; these algorithms are running on the fly as you are driving. They can’t do a whole lot of thinking, if you will. They are very intent on identifying their next route. That is a function of the network you have given it.”
Ideally, according to Macfarlane’s University of California, Berkeley colleague and co-author Ismaeel Babur, the navigation apps could be able to analyze much more than the simple “origin-destination” information drivers put into them before leaving home. But that has yet to happen.
“Because they are approaching it from a computer science perspective, they want to route thousands of people on the fly,” Babur said. “What you really want to know is if somebody has a purpose to travel a given street, whether they live there or are delivering food or working there. And routing them by purpose is a much harder problem.”
Much as Newtown opted to create a one-way street to eliminate bottlenecks on Hall Lane, Macfarlane said the city of Berkeley, CA, tried another non-computational approach: placing giant flowerpots in intersections, to slow traffic down and make roads less appealing to the apps’ consideration of faster bypasses.
“In the urban environment, that’s what you are seeing,” Macfarlane said, “control of the traffic that slows it down. Once you slow it down, that data goes into the navigation app, and those roads are then not as interesting to the app because they are going slow. That’s been the current solution to this problem.”
Google, which owns the Waze navigation app in addition to Google Maps, did not respond to requests for comment.
Existing plans in place
While getting local traffic diversion plans into navigation apps might seem difficult, it’s not as if those plans do not exist. One of the factors that led Kullgren to suggest local roads be taken off the apps was a plan approved by the Western Connecticut Council of Governments.
In 2011, he said, the council, working in concert with the Connecticut Dept. of Emergency Management and other agencies, created a diversionary plan for the region’s two main arterials, I-84 and U.S. Route 7, that eliminated the use of local roads for arterial traffic between the arterials’ exits in the region.
“Working with local municipalities, they were able to develop plans to keep traffic on state roads if something happened between exits 9 and 10,” he said. “The instructions and maps are there, you can find that online.
“The state transportation department uses the same protocol if there’s an accident or a huge construction project. Their policy is to use only state roads in their diversionary plan. That information can be supplied to the travel app companies. You don’t have to reinvent the wheel, it’s all there.”
Bolinsky said the state Department of Transportation (DoT) would seem to be a natural mediator between local councils and the app developers.
“I think the inconsistency of having, in Connecticut’s case, 169 municipalities funneling their personal preferences into any GPS host would be a really unreasonable burden. It’s already in existence within the DoT, so they are, without a doubt, the most appropriate partner.”
Macfarlane and Babur believe state and local transportation and planning agencies, plus other entities including real estate developers, logistics and delivery companies, and insurance carriers, might also be natural customers of a platform able to model highly complex traffic patterns that can work in concert with navigation apps. One such model the researchers at Berkeley Lab and the university have developed uses a parallel discrete event simulator, and can model 19 million daily trips over 1 million road links in a 6-to-8 minute span. Such a tool would be useful to app developers for everyday traffic, but could also be lifesaving in planning traffic flow in the face of fast-moving disasters such as wildfires. (During wildfires in the Los Angeles area in 2017, some navigation apps, sensing empty roads, routed drivers into danger zones).
“Instead of trying to model where the fire is going, let’s model an example fire and say, ‘What is the best way to get people out of here?’” Macfarlane said. “How long will it take? When do I have to tell people to leave before they open the door and see the fire in their backyard? With this tool, we have the capability to evaluate all those ‘what if’ scenarios.”
Should apps develop ‘social sense?’
In a recent paper published on PLOS One, researchers from the University of St. Gallen in Switzerland and the University of Michigan also examined existing simulations. They also looked at news stories about app-created bottlenecks and countermeasures employed by both public agencies and private individuals affected by what the study’s authors called “negative social externalities” (NSEs) the apps can cause.
The Swiss researchers suggested app developers could create a more holistic and empathetic version of their platforms, just as they have already built in options for preferences such as the most fuel-efficient route. “For instance, users could be informed about potential consequences of their route, such as disrupting school dismissal times or causing undue noise during late-night hours,” they wrote. “This context could allow drivers to make more informed decisions, and they might willingly choose to stay on a congested highway for a few extra minutes to minimize negative effects on local communities. Moreover, these broader criteria can be formally integrated into the design of routing algorithms.”
For everyone involved in such situations, the question of addressing the negative effects of routing technology seems to come down to the situation-dependent ability or willingness to invest substantially in a mitigating factor, whether that is physical infrastructure or adding computational complexity—and those investments seem hard to come by. In western Connecticut, for instance, Interstate 84 is still on its original 60-year-old roadbed, with two lanes of traffic in each direction. Adding an additional lane each way would help alleviate congestion in theory, but Bolinsky said the state DoT does not have such a project in its capital plan over the next 10 years, “and we are very close to, if not at, critical mass right now.”
In more urban areas, Macfarlane said, there’s simply no room for more traffic lanes and the solution needs to be something else.
“Our population is increasing and we are urbanizing,” she said. “It’s not going to get better. There used to be capacity on the roads, and there just isn’t much left.”
Gregory Goth is an Oakville, CT-based writer who specializes in science and technology.
Join the Discussion (0)
Become a Member or Sign In to Post a Comment