Traffic

by Tom Vanderbilt

  Indeed, there is a strong argument against the idea that we should emulate the actions of people like race-car drivers in everyday life. In a well-known (but not since repeated) study conducted in the 1970s, researchers from the Insurance Institute for Highway Safety looked into the off-course driving records of a pool of stock-car drivers. These drivers were no doubt capable of handling themselves around tight turns, no doubt superior at anticipating their moves ahead of time, no doubt possessed faster reaction times than ordinary people. How had they actually performed on the road, off the track? They’d not only gotten more traffic tickets (which we would expect given their penchant for risk) but they’d also had more crashes than the average driver. Racers possess superior control of a car, to be sure, but control alone does not win races. They also need that ineffable something within that tells them to push just slightly beyond their limits, and the limits of every other driver, to win. As Mario Andretti put it, “If everything seems under control, you’re just not going fast enough.” They had, one might argue, put themselves into positions in which their skills were not always enough to keep them out of trouble.

  In everyday traffic, “good driving” has little to do with cornering ability or navigating between tight packs of high-speed vehicles. It’s more a matter of just following the rules, staying awake, and not hitting anyone. This is not to say that racing cannot teach us things about everyday driving. Racers, Betchner said, sit erect and close, alert for feedback signals that can be felt in the pedals and steering wheel. The typical driver’s posture, however, is terrible. “Most of us sit back, the ‘Detroit lean,’” he said. “The car communication is horrible.” Some drivers, he lamented, sit so far back they cannot reliably depress the brake pedal far enough to activate the antilock system. Or consider vision, the sense that is supposed to account for 90 percent of our driving activity. The racer’s dictum that you should always be looking ahead to where you want to go next, which helps them speed through turns, is just as apt for something as prosaic as navigating an intersection. One reason for the high numbers of pedestrians struck in the crosswalk by vehicles turning is that drivers are simply not looking in the right place; they may be concentrating on making the corner itself as they turn (particularly if they are on a cell phone or otherwise distracted), rather than on what the result of their turn will be. In racing, this slows you down. In real life, it means you might hit someone.

  Everyday driving also presents those moments for which nothing in our previous experience can have adequately prepared us: the oncoming car crossing the line, the sudden obstacle in the headlights. At Bondurant, I went through repeated drills—for instance, driving a car as fast as I could toward a set of cones, hitting the brakes hard enough to activate the antilock system (something that actually took me several tries), and then steering off into a small lane marked by different cones. I was struck by just how much control of the vehicle I had under full braking. The ABS did not help me stop any more quickly; indeed, another exercise, one that involved steering into one of three lanes at the last moment at the command of a signal, drummed home the idea that certain crashes, inevitable if I had braked, could be rather easily avoided by simply steering. It did, however, open my eyes to the ability one has, with ABS, to stop and steer at the same time.

  That may seem, like the other lessons at Bondurant, rather common knowledge, but the wealth of evidence derived from studies of what drivers actually do in the critical moments of emergency situations suggests otherwise. First, drivers are actually quite reluctant to steer when an obstacle suddenly looms in front of them. The majority of drivers brake first and steer last, if at all, even in tests where steering is physically the only way to avoid a crash. This may be because steering might seem to put the driver in an even more precarious position, or it may be because the driver is unaware of the way the car is capable of handling, or it may simply be a form of “operant conditioning”—pressing our brakes, like staying in our lane, has so often been the right thing to do in everyday driving, it begins to seem the only thing to do. But research has also shown that drivers rarely activate the brakes to their full power. Other studies have demonstrated that when steering is attempted, the maneuver tends to be in the same direction the obstacle is moving, which hints that drivers are not “looking where they want to go” (and moving in that direction) but are focused instead on the obstacle to be avoided.

  Whether or not the “muscle memory” of my evasive actions on the test course can be sustained over years of uneventful driving is an open question. The major problem is that so many things can go wrong in traffic that it would be impossible to teach, much less remember, appropriate responses for each scenario. Add to this the problem that because these events are unexpected, our reaction times are slowed; the emotional duress of a potential crash might even further slow our reactions—sometimes, studies have shown, to the point where we do nothing.

  Then there is the shifting, dynamic nature of traffic itself. It is sometimes impossible to say what a “correct” evasive maneuver would be in the moment of trying to avoid another driver, as it could be canceled out by an unexpected countermove by that other driver. In one trial, forty-nine drivers were put in a driving simulator at Daimler-Benz. As they approached an intersection, a car that had been stopped on the crossroad suddenly accelerated into the intersection, then halted in the drivers’ lane. The reaction time of every driver was sufficient, in theory, to avoid a crash. But only ten of forty-nine did. Part of the problem is that they had only time enough to react to the presence of the approaching car, and not enough time to fully discern what the intruding car was going to do. It was less about a correct maneuver than a roll of the dice.

  Whether advanced driver training helps drivers in the long term is one of those controversial and unresolved mysteries of the road, but my eye-opening experience at Bondurant raises the curious idea that we buy cars—for most people one of the most costly things they will ever own—with an underdeveloped sense of how to use them. This is true for many things, arguably, but not knowing what the F9 key does in Microsoft Word is less life-threatening than not knowing how to properly operate antilock brakes.

  This uneasy idea is one of the many unresolved tensions and contradictions found in driving and the traffic it spawns. There is the contradiction of the car itself: With its DNA steeped in racing, today it’s often just part of a loosely organized, greatly inefficient mass-transit system, a “living room on wheels.” To drive safely is often to become rather bored, which may lead us to become distracted and thus less safe. On the other hand, if we drove like racers, we would have little problem becoming distracted or falling asleep, but we would inherently be driving less safely. (Even the most skilled drivers cannot overcome the fundamental physics of things like stopping distance.) We all think we’re better than the average driver. We think cars are the risk when on foot; we think pedestrians act dangerously when we’re behind the wheel. We want safer cars so we can drive more dangerously. Driving, with its exhilarating speed and the boundless personal mobility it grants us, is strangely life-affirming but also, for most of us, the most deadly presence in our lives. We all want to be individuals on the road, but smooth-flowing traffic requires conformity. We want all the lights to be green, unless we are on the intersecting road, in which case we want those lights to be green. We want little traffic on our own street but a convenient ten-lane highway blazing just nearby. We all wish the other person would not drive, so that our trip would be faster. What’s best for us on the road is often not best for everyone else, and vice versa.

  The reason I have avoided talking about the negative environmental consequences of the car is that I believe, as was once said, that it will be easier to remove the internal-combustion engine from the car than it will to remove the driver. With fuel economy liberated by some renewable, sustainable fuel source of the future, all the dynamics of traffic I have described will only become more amplified. As Larry Burns, vice president of R&D and strate
gic planning at General Motors, put it to me, “Of all the externalities of an auto that I worry about—energy, environment, equality of access, safety, and congestion—the one that I think is toughest to solve is congestion.”

  Even if the driver is still in the car, whether he or she will be driving in the future is another question. Virtually all of the perceptual limitations we have in driving—blind spots, overdriving our headlights, problems in detecting the rate of closure—are being addressed by scientists and car manufacturers. High-end cars already bristle with these features. An ad for BMW’s xDrive system, which “uses sensors to monitor the road ahead,” puts it succinctly. It says, “xDrive reaction time: 100 milliseconds. Human reaction time: unnecessary.” Technologies like “gaze detection,” in which the car will tell the driver that he or she is not paying attention (by tracking eye movements), are on the horizon.

  The future of driving will probably look a lot less like the track at Bondurant and much more like the 200,000-square-foot parking lot at AT&T Park (ordinarily home of the San Francisco Giants) during the World Congress on Intelligent Transport Systems. The parking lot had been converted into a “Innovative Mobility Showcase” for any number of high-tech traffic devices. It looked like a kind of strange carnival of human limitations. There were “Intelligent Intersections” that could alert drivers when an approaching driver did not seem, as calculated by sensors and algorithms, intent on stopping and “Dynamic Parking” demonstrations that promised to end, through real-time sensors, the search for open parking spots.

  I was riding in a Cadillac CTS with C. Christopher Kellum and Priyantha Mudalige, two researchers with General Motors. The car, via GPS technology and receivers, was communicating with the other cars, also equipped with the technology. GM calls its technology “vehicle to vehicle,” and the idea is that by connecting all the cars in a kind of mobile network, this shared intelligence can help you “watch for the other guy,” as Mudalige put it. A screen displayed the fact that we were connected to two other vehicles. The researchers are aware that any system released into the real world would have to contend with hundreds more at a time. “We do lots of simulations to understand what happens when there’s two thousand vehicles in the same spot,” said Kellum. “We need an intelligent way to parse out what information is important and what’s not important. If there’s an accident a mile ahead, you want that information. If it’s just some guy driving a mile ahead, you don’t really care.”

  If this sounds familiar, it’s because it is: This kind of incident detection and evaluation was one of the key tasks the Stanford team had targeted in getting their robotic car Junior to drive successfully in simulated urban traffic. I was, I realized, sitting in Junior’s cousin. Kellum asked me to change lanes, even though I knew, in this case, that a neighboring car had crept into my blind spot. As I put the signal on, I felt a small, Magic Fingers–style vibration in my back. This is known as a haptic warning, and it is used so that the driver will not be overwhelmed with visual or auditory information, or to underscore warnings he or she might disregard. (As you will feel when your car has drifted off the road into gravel, haptic warnings can be crudely effective.) One of the issues that haunts driver-assist technologies like “lane-departure warnings” is that these warnings can become ever more prescient, ever more sophisticated, but drivers still have to pay attention to the warning and be able to react accordingly.

  Or perhaps not. Next, Kellum asked me to drive at a steady clip toward a parked car far in the distance. “Whatever feels comfortable,” he said. He then asked me not to press the brakes. “We’re going to go up there and our car’s going to brake automatically,” he said. “In real time we’re constantly assessing how far away we are, the closing speeds, and when to start braking. I’ve done this at seventy-five miles per hour.” This was essentially the same exercise as at Bondurant, but instead of being asked to lock up the ABS, I was being asked to sit back and do nothing. I was in Junior, and I was riding shotgun. The stopped car quickly loomed into view. Time seemed to slow for a moment. (In reality, as studies have suggested, it probably sped up and this was just my memory playing tricks.) A chill shot through my body; the hairs on my neck tingled. Images of blooming air bags and the buckling necks of crash-test dummies ran through my head like a fleeting nightmare. The car came to a perfect stop.

  Somewhere down the road, in some distant future, humans may evolve to become perfect drivers, with highly adapted vision and reflexes for moving seamlessly at high speeds. Perhaps, like the ants, we will turn the highways into blissfully cooperative, ultraefficient streams of movement, with no merging or tailgating or finger flipping. Long before that happens, however, a sooner future seems likely: cars driving themselves, at smoothly synchronized speeds to ensure maximum traffic flow and safe following distances, equipped with merging algorithms set for highest throughput, all overseen by network routers that guide cars down the most efficient paths on these information superhighways. Maybe this will be the traffic nirvana for which we have been searching. We would do well, though, to remember the warning from the mid-twentieth-century traffic engineer Henry Barnes: “As time goes on the technical problems become more automatic, while the people problems become more surrealistic.” Even if drivers are taken away from the wheel, can we ever take the mere fact of being human out of traffic?

  Despite possessing the small diploma known as a driver’s license, I was, throughout the course of this endeavor, a novice in a complex field. I relied on the help of many people in many places, without whom this book would have been impossible.

  In no logical order, then, and with any omissions purely unintended, allow me to unravel the roster of gratitude, beginning geographically with the American Middle West. At the University of Iowa and at its National Advanced Driving Simulator, Daniel McGehee, John Lee, Omar Ahmad, and Tara Smyser patiently explained their findings and looked the other way as I skidded out of control in Virtual Iowa on the world’s most advanced driving simulator. At the University of Michigan, Michael Flannagan and Daniel Blower at the Transportation Research Institute, and Barry Kantowitz in the Department of Engineering, walked me through ergonomics, vision, and other topics. Over in Warren, Michigan, and in Detroit, Richard A. Young, Larry Burns, and Linda S. Angell of General Motors popped open the hood on the automaker’s research. In Chicago, Howard Hayes and Larry Peterson of Navteq walked me through the company’s traffic monitoring operations, while Jean Gornicki took me on a Navteq mapping drive of the suburbs. At the University of North Dakota, Mark Nawrot taught me Motion Parallax 101, among other things.

  In Los Angeles, special thanks are due to John E. Fisher, Assistant General Manager of the Los Angeles Department of Transportation, and Frank Quon, Deputy District Director of Operations for District 7, for sharing their extensive knowledge and insight into how traffic in L.A. functions. Thanks also to Marco Ruano, Dawn Helou, Afsaneh M. Razavi, and Jeanne Bonfilio of Caltrans, and James Okazaki, Kartik Patel, and Verej Janoyan of LA DOT. Thanks to Chris Hughes, Claire Sigman, and Shane Novicki at Clear Channel’s Airwatch in Orange County, as well as Vera Jimenez at CBS2 in Los Angeles, for dishing on L.A. traffic in all its infinite varieties. Sergeant Joseph Zizi of the California Highway Patrol gave me an intimate view into patroling the highway and answered any number of statistical queries. At UCLA, a number of people across different departments shared their expertise: Donald Shoup, Jay Phelan, Brian D. Taylor, Randall Crane, and Jack Katz. At Stanford University, thanks to Sebastian Thrun and Michael Montemerlo.

  In the New York region, thanks are due to Kay Sarlin, Ryan Russo, and Michael Primeggia of the New York City Department of Transportation. Sam “Gridlock Sam” Schwartz of Sam Schwartz PLLC and Michael King at Nelson/Nygaard provided invaluable insight and commentary on New York traffic. Aaron Naparstek was a constant source of traffic inspiration, and under his editorship, streetsblog.org remains the world’s single best source of transportation news and opinion. At the New Jersey Department of Trans
portation in Trenton, Gary Toth and Yosry Bekhiet gave me a tour of the city’s highway overhaul and patiently explained “Jersey jughandles” and other exotic traffic creatures of the Garden State (where this book began). In the Washington, D.C.–Beltway area, special thanks to Nancy McGuckin and Alan Pisarski; and, at the Federal Highway Administration, thanks to Tom Granda, Carl Anderson, Doug Hecox, John McCracken, Michael Trentacoste, Bill Prosser, and Ray Krammes for the tour of the Turner-Fairbank Lab, the lively roundtable discussion, and subsequent conversations. At the National Highway Safety Administration, thanks to Charles Kahane and Patricia Ellison-Potter.

  In Canada, Gerry Wilde offered his theories on risk homeostasis (and top-drawer espresso). Baher Abdulhai, founder and head of the Intelligent Transportation Systems Centre and Testbed at the University of Toronto, explained the “fundamental diagrams” and other traffic intricacies to me. In Mexico City, Mario González-Román took me driving on the monumental Segundo Piso and helped in countless other ways. Thanks also to Agustín Barrios Gómez and Alan Skinner. Alfredo Hernández García, executive director of traffic control and engineering at the Secretaría de Seguridad Pública of the Gobierno del Distrito Federal, opened up the city’s Traffic Management Center in the Colonia Obrera. Thanks also to Claudia Adeath at Muévete por tu Ciudad, which deserves kudos for trying to calm Mexico City’s often hostile traffic.

  In England, thanks to Malcolm Murray-Clark, Director of Congestion Charging in London, and Phil Davis, at Transport for London’s London Traffic Control Centre. Peter Weeden of the Royal Kensington Borough Council graciously offered his time and expertise. John Adams, professor emeritus at University College London, offered his always trenchant thoughts on risk. At the Transport Research Laboratory in Wokingham, Janet Kennedy shared her expertise and the lab’s driving simulator. Thanks also to John Groeger at the University of Surrey, Jake Desyllas at Intelligent Space, and Bill Hillier and Alain Chiaradia at Space Syntax. In Germany, Michael Schreckenberg at the University of Duisburg-Essen’s Physics of Transport and Traffic department held a wide-ranging and illuminating symposium for me on the personal and system-wide physics of traffic. At the Bundesanstalt für Straßenwesen (Federal Highway Research Unit) in Bergisch Gladbach, Germany, Karl-Josef Höhnscheid and Kerstin Lemke answered my questions about the autobahn and other topics. Thanks also to Juergen Berlitz at the ADAC (Allgemeiner Deutscher Automobil-Club). In Copenhagen, thanks are due to the esteemed traffic guru Jan Gehl, at Jan Gehl Associates; and Steffen Rasmussen, of the city’s Traffic and Planning Office. In Italy, many thanks to Paolo Borgognone and Giuseppe Cesaro of the Automobile Club d’Italia for the traffic knowledge and the excellent cacio e pepe. Thanks also to Andrea del Martino at the Laboratory of Complex Systems at “La Sapienza,” and Max Hall, physics teacher and Roman Vespa rider.