The Woman in the Crosswalk & the Ghost in the Machine

Self-Driving Cars, Pedestrian Safety, and the Future of City Streets

Moments before 11:35 am on July 14, 2023, my grandmother was waiting to cross the three-way intersection where Bishop Street meets State Street in New Haven, Connecticut, on her way to a hair appointment. A morning thunderstorm had left the street damp and the air muggy, but otherwise conditions were fair when a driver rushing to complete a food delivery ran down my grandmother in a Toyota Corolla. The impact of hitting an 89-year-old woman was enough to damage the car’s passenger-side windshield, and the police report noted that the victim sustained a “suspected serious injury.” She had been rushed by ambulance to Yale New Haven Hospital, so the officer wouldn’t know until later that she had survived and was in stable condition. But the crash marked the beginning of the end for my grandmother, and she spent the remaining four months of her life in and out of hospitals, dealing with and inevitably succumbing to complications.

I was thinking about my grandmother as I took my first ride in a self-driving car last week.

Waymo “robotaxis” are now available in Austin via the Uber app to those who opt in. Having seen them driving around my neighborhood for the better part of a year, I was curious—eager, even—to experience the technology firsthand. From a user experience perspective, the ride was seamless. The car arrived with my initials flashing on the rooftop LiDAR sensor, and I unlocked it using the app. Inside, two screens welcomed me by name, and I started the ride when I was ready. The cabin was comfortable and clean, with controls for music and temperature.

But after a few minutes, the “Wow, everything’s computer” novelty wore off, and my attention shifted to the other drivers on the road—the ones who were not computers. The Waymo obeyed the speed limit and traffic signs—already better than most human drivers—but it also demonstrated situational awareness: at a stop sign downtown, it waited until pedestrians had cleared the intersection before moving.

This, to me, is the promise of self-driving vehicles: every year, more than 2 million people are injured and over 40,000 die in crashes. Waymos don’t drink and drive. They don’t speed. And the company’s data shows they’re already significantly safer than human drivers. The value proposition of self-driving technology is the possibility of saving 40,000 lives, reducing life-altering injuries, and eliminating the billions spent on crash-related medical costs and property damage.

If a Waymo had been driving in New Haven in July 2023, my grandmother likely would have made it to her hair appointment—and probably to a good number more after that.

Would that have been worth it?

A couple months ago, I argued that “Cities Need ‘Waymo’ Autonomous Vehicles.” I wrote that piece not to be a Pollyanna, but to underscore the life-saving potential of this technology—too often dismissed by some Cassandras and doomsayers—and to point out that it’s no longer science fiction. Self-driving cars are already operating on the streets of Phoenix, San Francisco, and Austin, and they are only going to become more ubiquitous. We can disagree about how fast the robotaxi revolution is coming, but it is coming.¹

Self-driving cars will save lives—but many urbanists rightly fear that cities will repeat the midcentury mistake of planning themselves around cars, compounding sprawl and congestion at the expense of walkability and transit. I argued that cities should adopt sensible policies that embrace this life-saving technology without killing the urban environment—and that they should do it now, while self-driving vehicles are only operating in a few cities.

Because the opportunity for robotaxis is that they can make streets safer for everyone—not only vehicle occupants, but also for the man (and the 89-year-old woman) on the street.

Coincidentally, I’ve been reading Movement by Nicole Gelinas. The book’s subtitle, New York’s Long War to Take Back Its Streets from the Car, might sound like anti-car hysteria to some, but Gelinas is a policy analyst at the conservative Manhattan Institute. Perhaps surprisingly, she argues that “the automobile [has] failed urbanism as an engine of growth.”

Of course, while New York may be America’s most urban city, it too has struggled with the same challenges that others have faced since the car arrived en masse on city streets. Historically, streets were shared spaces: pedestrians and cyclists negotiated with horse-drawn carriages, pushcart vendors, and later omnibuses and streetcars. It was often chaotic, often dangerous—but everyone, regardless of transportation mode, understood that they had to look out for one another.

Car-centric planning was supposed to improve safety, but the opposite became true: by the 1920s, “Cars were already beginning to pose a far greater danger to pedestrians than trains had, and car drivers showed an early, alarming tendency to fail not just to look out for walkers but for themselves.” Between 1910 and 1929, New York’s population grew by 44%, but cars increased by 1000%. The number of pedestrians killed in car crashes rose from 232 to 952 per year.

But it was the pedestrians who were told to watch out.

The mass enthusiasm for automobiles also led to a massive sense of entitlement. Drivers demanded sole use of the once-shared streets, road infrastructure through existing neighborhoods and city parks, and free parking, especially in residential neighborhoods. The more space the city gave, the worse things got. By 1960, the New York City Planning Commission warned that congestion was “slowly but surely strangling the economic life of [Lower Manhattan] and unless drastic remedial action is taken soon the condition will become unsupportable.”

New Yorkers largely agreed, but that didn’t mean they were ready to give the city back to people on foot. Over the next 60 years, neighbors, activists, and politicians fought for better transit, bike lanes, pedestrian zones, wider sidewalks, more crosswalks, slower speeds, and an end to the “crime” of jaywalking.

By 1990, New York’s traffic death rate had dropped to 9.6 per 100,000—still high, but far below the national rate of 17.9. New York’s outperformance was due in part to “the affordability, convenience, and ubiquity of public transportation as well as densely built housing, stores, and offices” that reduced the need to drive. It also had something most other cities lacked: “pedestrian mass.” A distracted driver is more likely to notice a crowd in Manhattan than a lone pedestrian in the suburbs.

The numbers kept falling. By 2001, while traffic deaths dropped nationally by 5%, New York’s had fallen by half. By 2019, traffic deaths were down to 220, including 124 pedestrians. The city’s traffic death rate that year was 2.6 per 100,000 people, compared to 11 nationally. New York’s great secret was to recognize, as police commissioner Bill Bratton did, that in the vast majority of traffic deaths, it was the vehicle driver who was at fault.

New York stopped blaming the victim, and rightly so: walking in a city should be a pedestrian activity—mundane and not life-threatening.

Despite the progress it had made in reclaiming its streets, congestion in New York got much worse with the introduction of ride-hailing apps in 2011, and then again as the pandemic scared many former transit riders into private cars. While New York eventually capped the number of for-hire vehicles allowed in the city, the introduction of congestion pricing—which was on “indefinite pause” at the time of the book’s publication—seems to have finally made a difference. Ten weeks in, congestion pricing has been something of a “policy miracle,” according to

Sam Deutsch

, reducing driving times, increasing transit usage and safety, boosting business, raising millions for transit—and winning public support. Assuming the Traffic King in Washington DC doesn’t kill it.

Reflecting on New York’s Uber experience, Gelinas is not bullish about self-driving cars, calling the technology “unproven.” That claim didn’t survive her publication window, but she is right to be concerned that cities might “reshape themselves around technology, not the other way around” and undo the progress they’ve made in reducing traffic deaths by repeating the mistakes of the past. Of course, New York, with its vast transit network, is not a helpful comparison to other American cities in this respect, but even in New York’s outer boroughs, to say nothing of Austin or Phoenix, it’s hard to get around without a car. Banning self-driving cars won’t induce people there to ride a bus that doesn’t exist; they’ll just drive instead—and that won’t make cities safer. But Gelinas’s broader conclusions about what made New York safer can.

We now have a century of experience with which to understand how cars and people interact in dense urban environments. Cities don’t have another century to prepare for self-driving vehicles—but they have enough time to act before the robotaxi revolution rolls into town. And lots of the things they might do to prepare for self-driving vehicles are things they should be doing today to reduce traffic congestion and improve the experience of pedestrians.

Here’s a short list: Charge for parking. Build dedicated drop-off and pick-up zones. Install pedestrian-friendly infrastructure like crosswalks and curb extensions to improve visibility. Repurpose traffic lanes for sidewalks, bicycles, and scooters. Plant trees. Expand outdoor dining and amenity spaces. Put a price on vehicle access to downtown. These are not robotaxi-specific reforms, nor are they exhaustive—they’re good policy, full stop.

Most importantly, such policies can work to change urban car culture from one of entitlement and victim-blaming to accountability and awareness.

Any amount of free urban space given to cars will only induce demand and generate congestion. When the revolution arrives, robotaxis should instead have to share that space—and pay their fair share for it. Self-driving cars will help reduce the staggering loss of life on America’s roads, and that alone marks a massive improvement over the status quo. With good policy, the robotaxi revolution can be a boon to urban life—so long as we remember that cities are for people first, especially the woman trying to cross the street.

1. For two different takes, see

Tomas Pueyo

here and

Andrew Miller

’s rebuttal.