Laura Bliss is CityLab’s West Coast bureau chief. She also writes MapLab, a biweekly newsletter about maps (subscribe here). Her work has appeared in The New York Times, The Atlantic, Sierra, GOOD, Los Angeles, and elsewhere, including in the book The Future of Transportation.
In an excerpt from the new book The Future of Transportation, CityLab’s Laura Bliss adds up the “price of anarchy” when it comes to traffic navigation apps.
This essay is adapted from SOM Thinkers: The Future of Transportation, published by Metropolis Books.
There might not be much “weather” to speak of in Los Angeles, but there is traffic. It’s the de facto small talk upon arrival at meetings or cocktail parties, comparing journeys through the proverbial storm. And in certain ways, traffic does resemble the daily expressions of climate. It follows diurnal and seasonal patterns; it shapes, and is shaped, by local conditions. There are unexpected downpours: accidents, parades, sports events, concerts.
Once upon a time, if you were really savvy, you could steer around the thunderheads—that is, evade congestion almost entirely.
Now, everyone can do that, thanks to navigation apps like Waze, which launched in 2009 by a startup based in suburban Tel Aviv with the aspiration to save drivers five minutes on every trip by outsmarting traffic jams. Ten years later, the navigation app’s current motto is to “eliminate traffic”—to untie the knots of urban congestion once and for all. Like Google Maps, Apple Maps, Inrix, and other smartphone-based navigation tools, its routing algorithm weaves user locations with other sources of traffic data, quickly identifying the fastest routes available at any given moment.
Waze often describes itself in terms of the social goods it promotes. It likes to highlight the dedication of its active participants, who pay it forward to less-informed drivers behind them, as well as its willingness to share incident reports with city governments so that, for example, traffic engineers can rejigger stop lights or crack down on double parking. “Over the last 10 years, we’ve operated from a sense of civic responsibility within our means,” wrote Waze’s CEO and founder Noam Bardin in April 2018.
But Waze is a business, not a government agency. The goal is to be an indispensable service for its customers, and to profit from that. And it isn’t clear that those objectives align with a solution for urban congestion as a whole. This gets to the heart of the problem with any navigation app—or, for that matter, any traffic fix that prioritizes the needs of independent drivers over what’s best for the broader system. Managing traffic requires us to work together. Apps tap into our selfish desires.
Everyone wants to feel in control. In L.A., nothing concerns locals more than traffic—not personal safety, the cost of living, or even the housing market—according to a 2016 poll by the Los Angeles Times. Drivers there spend an average of 80 hours in gridlock every year, according to a report by Texas A&M University. That’s a full three-and-a-half days.
So virtually everyone is following a routing app, in L.A. and beyond. A 2015 Pew survey found that 90 percent of Americans who own smartphones get their driving directions from them, at least some of the time. Our troubled psyches are soothed by the constant movement the apps encourage us to be in. By isolating the source of jams in closures and crashes, the apps teach us why congestion even exists, a question that can be as aggravating as the thing itself. “Why is traffic slow? How can I get around it? Waze tells me,” the columnist Gary Richards wrote in the Mercury News in 2016. Brian Roberts, a writer and producer who authored a book about L.A. street shortcuts in the 1980s, declared in a 2015 L.A. Times op-ed that Waze feels like a “a quantum leap forward for those of us who want to keep our blood pressure down during the daily commute.”
But researchers have known for decades that driver-first traffic “fixes,” even with the best of intentions, have deleterious effects on transportation networks overall. The principle of “induced demand” is one expression of this. When engineers widen roads in order to accommodate growing traffic volumes, the outcome in the long run will rarely be a faster journey for drivers. That’s because “the traffic we see does not represent the full demand for peak travel at the prevailing cost, since congestion itself causes many potential rush-hour trips to be canceled, diverted, or rescheduled,” the urban economists Richard Arnott and Kenneth Small wrote in a 1994 issue of American Scientist. Like liquid, traffic expands to the space available.
Another expression of the counterintuitive relationship between person and society is a theory developed by John Nash, the famed mathematician best known from the 2001 film A Beautiful Mind. The “Nash equilibrium” describes a situation where, in a system of decision-making actors, nobody is motivated to make a different decision than the one they are making, because a different decision would leave that individual worse off, even if it would improve overall group welfare. In other words, it’s when a bunch of individual self-interests work against the good of all.
Road traffic is a great example: absent other incentives, I’m always going to choose the fastest route home that is available to me, even though taking a longer, more circuitous route would help spread out traffic and ease congestion for other drivers across my city. Traffic engineers have long assumed that the Nash equilibrium describes real-world rush hours pretty well.
In fact, mathematical studies and behavioral experiments dating back to the 1960s have shown that the collective delay is almost always worse in the Nash equilibrium, a.k.a. “user-optimized” driving scenario, compared to a world where drivers worked as a team for smoother traffic overall. Imagine a centralized transportation planner who assigned commuters their routes based on what was going to most benefit everybody. That god-like figure could impel some drivers to protract their journeys in order to improve the overall flow and decrease the cumulative time spent in traffic.
In this “system-optimized” equilibrium, our trips would be less harried on average: One widely cited 2001 paper by computer scientists at Cornell found that a network of “user-optimized” drivers can experience travel times equivalent to what a network of “system-optimized” drivers would experience with twice as many cars. Transport engineers call the difference between selfish and social equilibria the “price of anarchy.”
By and large, current GPS routing systems help drivers double-down on our self-centered tendencies, and the price of anarchy may be growing as a result. It’s true that Waze and other systems divert drivers to emptier pockets of road space, which may improve travel times across the network. But Alexandre Bayen, the director of UC Berkeley’s Institute of Transportation Studies whose current research is focused on modeling the effects of mapping apps, believes there is a limit to that systemic benefit as the share of drivers plugged into apps grows.
In a 2017 talk at the Cal Future Forum, Bayen walked through a computer simulation that showed how drivers reacted to a car crash on the highway with and without the help of a navigation system. His model demonstrated that when just 20 percent of drivers were using apps, the total time that all drivers spent in traffic actually increased. Prompted by their smartphones to use “faster routes” on surface streets, the simulated commuters clogged exit ramps as they veered off the crash-afflicted highway. The back-up they created sent ripples into the travel lanes behind them, creating delays for highway drivers. Meanwhile, they added vehicles on neighborhood roads that weren’t designed to handle the through-traffic.
There hasn’t been enough research or available data to say for sure, but with adoption rates as high as they are, Bayen believes that navigation apps are creating more problems than they’re solving in many crowded cities. “If these companies were ever subpoenaed by public agencies, we strongly believe their data would clearly reveal that,” Bayen told me. Other researchers agree that it’s possible, though not certain, that these tools are making traffic worse, not better.
Drivers have always made self-serving choices, of course—but often with imperfect information. Ioannis Paschalidis, a traffic engineering scholar at Boston University, explained his reasoning in this light. “The selfish decisions that people are making without an app might not actually be accurate estimates of congestion; therefore, their independent selections might not always be minimizing their actual travel time,” he told me. “So it is plausible that congestion gets worse with apps, because it means more users are using accurate information.” The more we know, in other words, the more we might be dooming the system.
Waze insists otherwise. “With more and more drivers on the road every day, Waze works to spread congestion evenly across public roads to make the driving and commuting experience better for everyone,” a spokesperson stated in response to questions about the app’s possible negative impacts. She also pointed to Waze’s efforts to share data with city governments through its “Connected Citizens Program,” in which it passes along accident reports and road closures to traffic officials.
In unusual circumstances—including hurricanes, fires, and other emergencies—this data has proven invaluable to public agencies. But from the perspective of “eliminating traffic,” the researchers told me, it’s not the most salient. As long as officials and academics are unable to measure the volume of app-reliant drivers passing through city streets at a given time, it’s hard to say for sure whether apps are “defeating” jams or exacerbating them.
Clearly, residential neighborhoods are bearing a heavier traffic burden. Sherman Oaks, a wealthy residential neighborhood in L.A.’s San Fernando Valley, is a good example of this collateral damage. The enclave is less than a mile to the northeast of I-405, one of the busiest commuter arteries in the county, which pulls drivers up and over the Sepulveda Pass to West L.A.’s many commuter destinations. As such, the narrow, windy, unstriped streets have become popular alternate routes, and the city’s own studies have found that traffic has become measurably worse in recent years, in part due to the apps. I happen to have grown up there, and I’ve seen it myself: On weekdays by 8 a.m. the roads are clotted, noisy, lurching with the penetrative anxious movements of drivers in a rush.
City officials have responded by installing signs to slow speeds and restrict turns, but these have been difficult to enforce. Sherman Oaks homeowners are not alone in demanding a more meaningful response. In the Bay Area, homeowners have taken to reporting false car crashes and even physically walking in the street with the app running on their phones in hopes of screwing with the algorithm. In the suburbs of Washington, D.C., Boston, Toronto, Paris, and London, homeowners are slapping up DIY detour signs and staging protests. Some resistors would like to see their neighborhood taken off the map entirely, as in Leonia, New Jersey. There, pissed-off locals have attempted to ticket through-travelers from using their town as shortcut to the George Washington Bridge. Currently, the Los Angeles Department of Transportation is trying to convince Waze, Google Maps, and Apple Maps to effectively remove parts of Sherman Oaks from their service.
Would all communities get away with such hijinks? The Waze backlash raises the question of who gets to politicize traffic, and to what end. After all, routing apps don’t exist in a vacuum. They disrupt a system that has long funneled traffic, noise, emissions, and fatalities into poorer neighborhoods of color, with serious consequences. In the U.S., blacks and Latinos are disproportionately killed in car crashes compared to whites, in part because they are more likely to live near high-speed, high-capacity roads. Living near highways is associated with higher rates of asthma, cancer, and heart disease. “The poor disproportionately bear the costs of automobility,” wrote the historian of technology Paul Josephson in his 2017 book Traffic.
Bear in mind, too, that as new interstate highways plowed through dense urban centers while under construction in the 1960s, some “15,000 families and 1,500 business [were] being displaced each year,” as President John F. Kennedy stated in a 1962 address to Congress. This pattern frequently played out in lower-income black neighborhoods. During that era, in many cities, highway protests led by white, wealthy activists got their way more often than those led by activist of color. Today, poorer neighborhoods of color tend to have less political power to demand design improvements such as sidewalks, bike lanes, crosswalks, stop signs, and speed bumps. In the near future, algorithm adjustments could be included in that list too.
Yet traffic apps have transformed the politics of traffic in addition, perhaps, to its physics. They have created new constituencies of people outraged by what cars are doing to their neighborhoods. Perhaps their anti-app energy could be channeled into a solution that produces smoother traffic flows for everybody.
After all, there are known and viable solutions to congestion. Economists and urban planners insist that tolling the use of city streets, with prices rising at peak travel times, is the only meaningful way to reduce congestion. The results of “congestion pricing” policies in the cities that have implemented them, from London to Stockholm to Singapore, are promising. And there may be real societal gains to be had from the apps themselves. With adoption rates as high as they are, most of us have already bought into the idea of following an omniscient traffic planner; now it might just be a matter of working towards “system-optimized equilibrium” instead of one designed around selfishness.
Paschalidis, the Boston University traffic expert, has also imagined a system where app companies adjust their algorithms to factor in the cost of increased congestion and send users on routes that cut down on system-wide travel times. (As Rosenblat found, they are already adjusting the algorithms for more expansive purposes than we realize.) Bayen envisions something similar, but accompanied by financial incentives—drivers could be rewarded with a couple of dollars every time they opt for a slightly longer route.
Such a scheme sounds like the most perfect of all navigation-app worlds. But it’s antithetical to their business models for companies like Google or Waze to even fathom giving away their most meaningful data, the lifeblood of their products, and it’s hard to imagine that drivers would accept slower routes, at least knowingly. What’s more, even in cities where congestion pricing exists, the utopian vision of fast-flowing streams of traffic does not. Tolls help mitigate delays, but traffic eventually expands to meet the financial and physical space that’s available to it. It’s a byproduct of a healthy economy, after all.
Maybe it’s worth asking what exactly we want traffic to look like. Clearly, the era of highway engineers nudging drivers onto the straight wide roads they wanted us on is over. It isn’t clear that cities like L.A. are preparing themselves for what should come next, let alone fully addressing what’s happening now. What kinds of vehicles should be prioritized? What kinds of neighborhoods? Given that bad traffic affects non-drivers, too, what kinds of people?
Perhaps the best approach to the future of traffic starts with redefining the problem we’re trying to solve. For most of the 20th century, the principal concern of transportation economists was to reduce travel time across a given distance. But getting as many people as possible to and from work, school, and shopping—period—might be the more important task. If the goal is reframed as increasing access, rather than increasing speed, then the answer involves more than traffic apps and vehicle transponders. Land use patterns would have to be rethought so that people can live closer to the destinations they care about. People should be able to walk as much as they want to, and use bikes, scooters, buses, and trains. Autonomous vehicles, if they come, ought to be carefully folded into the mix with care so as not to double down on congestion and carbon emissions.
The apps aren’t going away any time soon. Their data is too precious, and we’re too seduced by their aura of authority. It’s easy to believe we’re being guided on the smartest routes when we’re using them, even when the streets are as familiar to us as the scuffed leather of the steering wheel. But no single tool is going to “solve” congestion, despite the claims of tech companies. Nor are autonomous cars, new highways, or any one policy.
Perhaps the best political outcome of the Waze resistance would be for some savvy leader to channel the outrage of aggrieved residents into support for a future of transportation that’s less about making way for traffic and more about mobility. Instead of censoring apps and blockading public roads, we could change the map itself. Would we have to set aside the delicious human satisfaction of outsmarting the system in place? If we’re all riding in autonomous vehicles, certainly. Otherwise, I’d say not. Grab your bike—I know a good shortcut.