When it gets hot enough, as it has across the South in recent weeks, barefoot toddlers suffer second-degree burns from stepping onto concrete. People who fall on the blistering pavement wind up with skin grafts. Kids stay inside all day, “trying to survive.” Windshield wipers glue themselves in place, and the ocean transfers heat back into your body. One electric blackout could bake thousands to death inside their homes.

You would think people would flee such a hellscape expeditiously. But as record-breaking heat fries the Sun Belt, the region’s popularity only grows. The numbers, laid out recently in The Economist, are striking: 12 of the 15 fastest-growing cities in the U.S. are in the Sun Belt. Of the top 50 zip codes that saw the largest increases in new residents since the start of the pandemic, 86 percent were in blazing-hot Texas, Florida, and Arizona.

To be sure, during the pandemic people also moved to a few relatively cool cities in Idaho, Utah, and Colorado. But hot places overwhelmingly dominate nearly every ranking of population growth and migration: The 50 counties with the greatest extreme-heat risk grew by nearly 5 percent from 2016 to 2020 due to migration, according to Redfin data. Meanwhile, the 50 counties with the lowest heat risk saw their population decrease from migration by 1.4 percent in the same time period. Those hot counties were led by Texas’s Williamson County, near Austin, whose population grew by 16 percent from inbound migration, and where 100 percent of homes have a “high heat risk.”

[Read: The problem with ‘Why do people still live in Phoenix?’]

The South may be approaching the approximate ambient temperature of Venus, but that’s no deterrent. People keep wanting to move there. (I count myself among these people, as someone who has dedicated the past year of my life to finding a house in Florida.) This unstoppable appeal of Sun Belt cities rests on three factors: These places tend to have less expensive housing, lots of jobs, and warm winters. None of these is sufficient to attract people in large numbers, but together they seem to generate an irresistible force, sucking up disaffected northerners and Californians like a fiery tornado.

Cheap housing

These days, you don’t have to wonder how the other half lives. You can open up Redfin and see how much house you can get in Dallas for less than your New York rent. The median home price in Los Angeles is $975,000. The median home price in the Phoenix suburb of Chandler is $520,000. Once you have this knowledge, it can be hard to evict it from your mind. What would you do with an extra half a million dollars?

The one thing every sunny, growing city has in common is affordable housing. This explains why Los Angeles, with its unimpeachable weather, is losing residents (including to Phoenix). It’s “vastly easier to mass produce housing in the suburbs of Phoenix or the suburbs of Houston than it is anyplace in coastal California or the Northeast,” says Edward Glaeser, an economist at Harvard. Cities in the Northeast and West tend to make it harder to get construction permits, and they have zoning requirements that make building affordable housing in desirable areas difficult. Plus, it’s just easier to build on an immense, unending desert than around the mountains of California or in old cities like Boston. In a study in 2007, Glaeser found that in the 1970s and ’80s, housing supply increased by 20 percent more in the South than elsewhere in the country.

In fact, many people seem to end up in the South because they aim for the perfect climate of California, quickly realize they can’t afford it, and settle for a similarly warm, cheaper place, like Phoenix or Austin. Just ask Elon Musk.

A “business-friendly” environment

Not all hot, affordable places are created equal. Austin became a pandemic boom town, but Midland, a West Texas city that’s just as warm and even less expensive, did not. This is where a complex mix of economic growth, human capital, and a certain yuppie je ne sais quoi come into play.

The Sun Belt cities that have soared are mostly in states with low taxes, which helps attract businesses. But many are also home to prominent universities that churn out highly educated workers. They’ve successfully created “agglomeration economies” of lots of similar types of companies in close proximity. Austin has the University of Texas, an Apple campus, and throngs of upwardly mobile Californians and New Yorkers who have fled high house prices. Midland, well, does not.

Austin began its strategy of luring tech workers as early as the 1970s and ’80s, when UT’s then business-school dean, George Kozmetsky, recruited computing companies to the area and launched incubators to nurture local talent (one of his mentees was a UT student named Michael Dell). Companies tend to cluster near other, similar companies—a phenomenon that explains Silicon Valley in California, Kendall Square in Cambridge, and Research Triangle Park in North Carolina. Knowledge workers like to be near people who can provide them with mentorship and job leads. You might not want to stay at your job forever, so it’s nice to have other companies to jump to. Businesses and educated workers tend to attract each other, and attract more businesses, creating a virtuous circle. In the first year of the pandemic, Austin had the highest inflow of tech workers of any major city.

[Read: When will the Southwest become unlivable?]

Many of the booming Sun Belt cities also possess the seeds of a hip Millennial lifestyle: Live music, outdoor recreation, and interesting bars and restaurants. The newcomers demand even more microbreweries and tapas places, which then sprout up and attract more newcomers. “When the people come, they in turn change the place,” says Cullum Clark, the director of the Bush Institute-SMU Economic Growth Initiative. “The place becomes bigger; it becomes richer; it becomes more cosmopolitan.” And expat Californians tend to like that.

Warm winters

Walter Bimson, the chair of Valley National Bank and a mid-century booster of Phoenix, once explained that people would surely move to the desert city, because they “want to flee from shoveling coal and from shoveling snow.”

His hunch—that people love sun—has persisted as a lay explanation for Sun Belt migration, but polling on the significance of weather to people’s moving decisions is sparse. Weather often gets wrapped into a nebulous factor called “amenities” or “quality of life,” which can also include the local schools and crime rates. When asked, many people name better weather as a reason for their move, but not the reason. In 2018 survey data shared with me by the Bureau of Economic and Business Research at the University of Florida, the top reasons people gave for moving to Florida were to be closer to family (37 percent), to start a new job (22 percent), and then the climate or weather (14 percent.) A 2014 Gallup poll found that weather was a prominent reason for those seeking to leave Illinois, Maryland, and Idaho, but it wasn’t the top reason for movers from any state. People who moved during the pandemic were likely to cite financial reasons or COVID risk as their motivations.

Still, it would be weird to ignore the sun in Sun Belt. This is something that all the experts I spoke with eventually conceded—that weather is hard to find in the data behind the Sun Belt’s rise, but it’s also hard to explain away. “No variable better predicts metropolitan-area growth over the last 120 years than January temperature,” Glaeser says. “Everybody likes playing golf in the winter,” says Enrico Moretti, an economist at UC Berkeley. “People really don’t like cold winters,” says Jenny Schuetz, a senior fellow at the Brookings Institution. People might not always admit it, but they appear to like warm weather.

Warm winters seem to act as an accelerant on cheap housing and plentiful jobs. People will vaguely consider a place with lots of new businesses and $300,000 homes, but once they see a few hundred Instagram posts of 70-degree February days, they call the moving company. “I think it’s word of mouth. It’s Instagram. A place gets buzz,” says the University of Toronto professor (and Atlantic contributor) Richard Florida. “‘My friends are there. It’s fun. They’re going out to restaurants; they’re going to the beach; winter doesn’t look cold.’” If you can work remotely, why not?

Chambers of commerce, real-estate agents, and industries that attract workers to warm-weather states tend to play up the “golf in the winter” element and play down the “lava-hot July.” “January, February, and March are the three reasons why people were attracted” to the Sun Belt, says Andrew Ross, a professor at NYU who has written several books about Sun Belt cities. “Housing-industry developers, real-estate brokers, they don’t talk about the summers. They talk about January, February, and March.”

A Sun Belt tipping point?

That said, all of these new Arizonans and Texans might leave if their cities continue heating and also escape the realm of affordability. There’s already an exodus afoot from Miami, for instance. The conservative governors of some Sun Belt states likely don’t appeal to the creative, liberal types who are drawn to cities. “Let’s you’re a gay designer who moved to Miami. Now you look up and you see Ron DeSantis,” Richard Florida says. “And you go, What the fuck am I doing here?”

[Read: Summer in the South is becoming unbearable]

Florida thinks the next great migration, for both climate and affordability reasons, will be north. Midwestern towns that can offer good amenities without scorching summers, such as Madison or Pittsburgh, are poised to offer a Sun Belt alternative. The Midwest currently has the most worker-relocation incentive programs, which pay remote workers to move to an underpopulated city. (Indiana alone has 16.) Or people might migrate to slightly cooler parts of warm states, choosing Flagstaff over Phoenix. (This is already happening, to some extent.)

For now, though, climatologists’ dire predictions don’t seem to be fazing people. Sure, Texans would prefer for it to be cooler, but the heat is apparently just this side of tolerable. Many residents of Phoenix and Dallas spend their summer days rotating between air-conditioned houses, air-conditioned cars, and air-conditioned offices, minimizing the felt impact of the triple-digit heat. “The fact that people are moving more into climate-risky places, and away from lower-risk places, suggests either that climate risk isn’t a primary factor that’s driving this, or that something else, like the cost of housing, is weighing out more than the climate risks,” Schuetz says. Knowledge of flood risk can nudge people toward lower-risk homes, according to a study by Redfin, but it’s unclear if this is true for more widespread climate risks, like heat. What’s the “safer” house if 100 percent of a county has a high heat risk?

It would be unfair to write off people moving to the Sun Belt as irrational or ignorant. We would all love a cheap house and a good job in a city that’s just warm enough. But life involves compromise. There aren’t enough houses in California. There aren’t enough jobs in Cleveland. Sometimes the best you can afford is Phoenix. “We can’t tell them not to move there,” Schuetz says, “unless we make it feasible for them to live in other places that are lower risk.”

The man who has been hailed as “the best state chair in the country” is not a national household name. He’s not even a household name in his own state. But on a recent afternoon in the small village of Grafton, Wisconsin, Ben Wikler might as well have been Bono.

Two dozen middle-aged and retired volunteers stood in line to clutch the hand of the chair of the Wisconsin Democrats. “Thank you for everything you do,” they said, beaming at Wikler as he took a lap through the Ozaukee County party headquarters. “We’re so happy you’re here.” Like proud children before an admiring parent, the volunteers told him how much money they’d raised and how many doors they’d knocked on this summer.

“This is Connie,” someone said, patting a woman’s shoulder. “She just won the school-board race.” “Yay, school board!” Wikler cheered.

He was there to kick off the last day of door knocking for a Wisconsin state-assembly candidate who had very little chance of winning in solid-red Ozaukee County, an exurban district on the shore of Lake Michigan north of Milwaukee. But the point was not to win, it was to lose by less. That afternoon, Wikler managed to deliver a speech with almost the same inspirational zeal as Aragorn at the Black Gate. “This election is a demonstration to ourselves as Democrats and to the country that there is change happening right now,” he told the volunteers—and a reminder to Republicans “that Democrats have not given up on democracy.”

Since becoming chair in 2019, Wikler has brought his party back from virtual irrelevance in Wisconsin. Four years after Donald Trump had demolished the so-called blue wall in the upper Midwest, Wikler’s leadership helped tip Wisconsin—and the entire presidential election—to the Democrats in 2020. Then, earlier this year, the millions of dollars Wikler had raised helped a progressive candidate prevail in the off-cycle state-supreme-court race, which will likely lead to a reworking of Wisconsin’s extremely gerrymandered maps.

Wikler’s talent is getting people to show up. He does this by framing every race as the election of a lifetime. “Resources tend to flow toward the places where they can make a difference or their imagination has been captured,” he told me.

[Read: It ain’t over in Wisconsin]

Resources is something of a euphemism; he really means dollars. Thanks to legislation passed by Republicans a few years ago, Wisconsin is one of the few states in which individuals can donate unlimited amounts to political parties, which can, in turn, transfer unlimited funds to candidates. It is Wikler’s particular genius to have turned that weapon of fundraising against the party that made it law.

In the run-up to next year’s presidential election, American eyeballs will once again be on Wikler’s home. “If we could have a Ben Wikler in all 50 states, the Democratic Party would be in better shape,” Jon Favreau, the podcaster and former Obama speechwriter, told me. But people may be getting tired of elections with existential stakes, however much the party spends persuading them to go out and vote. Capturing imaginations once again, especially on behalf of an elderly incumbent with less-than-great approval ratings, could be Wikler’s most formidable challenge yet.

I hitched a ride to the Ozaukee County event with Wikler’s posse in their rented minivan. When I slid open the back door, I found the state party chair buckled into a seat in the middle row, his head grazing the ceiling. The 42-year-old Wikler, who is goateed and tall (6 foot 4), was wearing clear-framed glasses and a denim shirt over denim jeans. He looked like a Brooklyn dad—but Wikler is a dad from Madison, a fact he is very proud of.

I’d hardly sat down before Wikler launched into a 30-minute refresher course, for my benefit, on Wisconsin’s idiosyncratic past. Robert La Follette and the state’s socialist roots. Senator Joe McCarthy. Governor Tommy Thompson’s welfare reform. Then more recent history: Scott Walker’s ascension to the governor’s mansion in 2011, and Republicans’ success in flipping both chambers of the state legislature. Walker’s Act 10 legislation, which eroded the power of public unions. The GOP’s controversial and secretive redistricting project.

“How many times have you delivered that spiel?” I asked when he was done.

He smiled. “There’s actually an extended version.”

Today, Wikler lives in his childhood home on Madison’s west side with his wife, his three kids, and their enormous, excitable Bernese mountain dog. But before moving back to the upper Midwest, Wikler was the Washington, D.C., director of the progressive organization MoveOn, for which he led protests against Republican attempts to overturn the Affordable Care Act. Prior to that, Wikler hosted a politics podcast called The Good Fight after a spell as a researcher and producer for Al Franken. The former senator from Minnesota remains a close friend. “He’s just brilliant—really funny and a really good writer,” Franken told me of Wikler last month, over the phone. “He has the full package, and that’s hard to get in a state chairman.” (The title of Franken’s 2003 book, Lies and the Lying Liars Who Tell Them, was Wikler’s idea, Franken said.)

Then, in 2016, Trump hurtled through the blue wall, winning Wisconsin’s Electoral College votes for the Republicans for the first time since Ronald Reagan in 1984. Which is why Wikler ultimately decided to move back home and help revive his party’s fortunes.

As chair, Wikler is known for posting climactic Twitter threads about Wisconsin elections that go viral. He’s constantly giving interviews to convey the urgency of races up- and down-ballot. The central strategy of his chairmanship, Wikler told me, “has been to buy a bigger siren, and put it as high up as we possibly can.”

Most state parties in America have somewhere around half a dozen full-time paid staff members, but Wikler has expanded his staff from 30 to 70. He has a comprehensive digital operation, an in-house research group, and a full-time staff of youth organizers.

Since 2019, Wikler has used his connections in national politics to raise more than $110 million, an astoundingly high amount for a state party. His team’s most successful money-gathering endeavor was getting celebrities such as Robin Wright and Julia Louis-Dreyfus to care about the Badger State: In September 2020, the Wisconsin Democrats hosted a Zoom table reading of the 1987 film The Princess Bride that reunited most of the original cast. The event attracted more than 100,000 viewers and raised $4.25 million. So they did it twice more, with the casts of The West Wing and Veep.

Wisconsin could have gone the way of neighboring Iowa, which has turned sharply to the right in these past six years. In the Badger State, the trend toward Democrats began in 2018, when many voters revolted against Trump. But thanks in large part to the machine that Wikler has built, the party has continued to win by bigger and bigger margins in the state’s metropolitan areas in the past few cycles, and it’s losing by smaller margins in the Republican-leaning suburbs of Milwaukee. Although Democrats nationally have been hemorrhaging voters in rural areas, they’ve managed to at least stop the bleeding in rural Wisconsin, Craig Gilbert, the retired Washington bureau chief for the Milwaukee Journal Sentinel, told me.

[Read: The most dangerous Democrat in Iowa]

Statewide elections have proved to be the most rewarding battlegrounds for Democrats. In Wisconsin, Biden beat Trump in 2020 by 20,000 votes, and last year Democratic Governor Tony Evers narrowly won reelection. The only major disappointment was Mandela Barnes’s loss to the incumbent Republican senator, Ron Johnson. But just this past spring, Wisconsinites elected Janet Protasiewicz to the state supreme court in a race that broke turnout records and attracted donations from George Soros, Steven Spielberg, and Illinois Governor J. B. Pritzker.

Wikler’s legacy as a Democratic leader will be the nationalization of the state party’s donor base—something he’s achieved by arguing that Wisconsin is at the epicenter of America’s political battle. Whether that’s good for democracy is another matter.

The wealthy Democrats from California or Illinois who’ve done much of the donating are not ideal stand-ins for regular Wisconsinites. “Elections shouldn’t be a tug-of-war between a handful of billionaires on the right and a handful of billionaires on the left,” Matthew Rothschild, the former executive director of the Wisconsin Democracy Campaign, told me. “But Ben didn’t make the playing field. Republicans in Wisconsin made the playing field. The U.S. Supreme Court made the playing field.”

If Wikler’s strategy is to make politics in Wisconsin national, he is also committed to hyperlocal campaigning: Democrats should have a presence everywhere, Wikler believes. Which is why the van drove another two hours west from Grafton to Baraboo for an annual agricultural-equipment expo.

The state party’s Rural Caucus had set up a tent between the crop-spraying-drone display and a demonstration area for grinding forest products. Wikler gave a pep talk to some of his members before striding over to the Sauk County Republicans’ tent. “Hi, I’m the Democratic Party chair,” he said, extending his hand toward a trio of 60-something men chatting in the shade. For a few minutes, the four men went back and forth, a little awkwardly, about the successes and failures of the former Governor Walker and whether any of them were particularly excited about a second nomination of Trump. (They weren’t.) It was all pleasant enough.

Then, as Wikler turned to leave, one of the men took him aside. “I gotta tell you something,” he said, in a low voice. “I spoke with a gentleman over at your tent this morning, and I have never met a finer man or had a more reasonable conversation.” Wikler beamed. “As a party chair, that’s a delight to hear,” he replied.

We left Baraboo in the late afternoon for a volunteer picnic in Middleton, a leafy Madison suburb along Lake Mendota. The gathering was held in a lush backyard, full of unruly flowering shrubs and the kind of wacky animal lawn ornaments that seem to announce, A Democrat lives here!

The yard was full of gray-haired volunteers from different neighborhood door-knocking teams. “I don’t think we could have done anything without Ben,” JoAnna Richard, the host of the event, told me. “His leadership has been key: his connections, and how we fundraise and organize year-round.” A few minutes later, Wikler was giving his third and final motivational speech of the day, thanking people for their work over the past few years. We’re “building something bigger than any of us,” he told them. “You’re at the heart of that project, in a place that is the most key furnace for democracy—the key engine, the center of the web.”

Republicans are working hard for a rebound in Wisconsin. Later this month, they’ll host the first debate of the GOP presidential primary in Milwaukee, and the Republican National Convention will be held in the same city next summer. That national attention will be good for the state party, which has recently under-raised Democrats.

“They’ve been very good at getting Hollywood money,” Brian Schimming, the state GOP chair, told me by phone, with what sounded like a mix of shade and envy. “It’s hard to compete with” the Democrats’ celebrities and wealthy out-of-state donors, he said. “I need to nationalize Wisconsin a bit more.”

This time around, Republicans are certainly going to be more focused on fundraising. “Ben would be kidding himself if he thinks he or his successor can always win the money race,” Rothschild told me. But money is not the race that ultimately matters.

“I’d rather have my problem than the problem Ben has, which is an extraordinarily unpopular sitting incumbent,” Schimming told me. “Our folks are really fired up about this race.”

Wikler, in fact, does seem a little nervous. He worries about a low-turnout election—and that people aren’t taking seriously enough the very real possibility of a second Trump presidency. “In 2020, people were ready to do anything to beat Trump. I had people retiring early and moving to Wisconsin to volunteer,” he told me in the car. “None of that’s happening right now.”

[Read: The long-shot candidate who has the White House worried]

Every recent presidential election in Wisconsin has been decided on a razor-thin margin, and Wikler’s job is to engage more than just the highly educated, high-income activist types. He’ll need to stitch together a delicate coalition and get them all to fill out a ballot: young people in Dane County; Black voters in Milwaukee; moderates in the suburbs and the small cities around Green Bay. The hurdles are already high, and Biden doesn’t exactly get many people’s blood pumping. “I’ve been concerned about that since 2020,” Favreau said. “It’s easy to see a scenario where a couple people say, ‘[Biden’s] too old. I’m going back to Trump.’” It’s even easier to see a situation in which some Wisconsinites, weary of it all, simply don’t vote.

In JoAnna Richard’s backyard in Middleton, Wikler was winding up his pep talk, a little breathlessly. They’d be working “throughout this year, and into next spring in the local elections, and into next fall in 2024,” he said. “And then we’ll continue six months after that in the 2025 local elections! And the next state-supreme-court race—”

A few people audibly sighed at this point, likely in anticipation of another two exhausting years door knocking and phone banking and envelope licking in defense of democracy. A man near me shouted, “We’re tired!” But that moment of wavering enthusiasm lasted only a fraction of a second before the whole group began to laugh.

Sure, they’re tired. But for Wikler, they’ll show up.

Will everyone else?

A few weeks ago, Dan Afergan, a software engineer, met a few friends at 540 Rogues, a bar in San Francisco’s Inner Richmond neighborhood. As Afergan and his companions nursed their drinks, someone walked in with some unusual news: “There’s a Cruise out there with a cone stuck on it.”

Afergan stepped outside to check it out. Sure enough, a self-driving cab from the company Cruise, which is majority-owned by General Motors, stood frozen in the middle of the street, its hazard lights blinking. A bright-orange cone was perched on the robotaxi’s hood.

“At the time, I thought it was a dumb prank,” Afergan told me later. “But one friend said, ‘No, I’ve heard about this.’ Until then I didn’t know that there are a bunch of people who are anti–autonomous vehicles.”

[Read: Seven arguments against the autonomous-vehicle utopia]

The “coning” that Afergan witnessed was part of a campaign launched by Safe Street Rebel, a local activist group previously known for organizing protests in support of bike-lane construction and public-transit funding. Now its members have turned their attention to robotaxis. According to government data reported by the news site Mission Local, Cruise and its rival Waymo—a subsidiary of Google’s parent, Alphabet—together operate 571 self-driving cabs in California. Users can hail them via an app. Service is concentrated in San Francisco, where the companies have been subject to a variety of limits imposed by the California Public Utilities Commission. The two companies now want the CPUC to remove those restrictions, despite objections from San Francisco’s police union and transportation and fire departments about robotaxis’ troubling habit of blocking traffic and obstructing emergency vehicles. The commission has postponed a decision twice but is expected to vote tomorrow.

After realizing that placing a simple orange cone on the hood seemed to paralyze a state-of-the-art autonomous vehicle, Safe Street Rebel posted a TikTok video encouraging San Francisco residents to try it for themselves. Hell no. We do not consent to this, a caption declares over a clip of a robotaxi on a city street. As the video ricocheted across social media, Cruise and Waymo were unamused, threatening to call the cops on anyone who placed a cone on their cars.

One might dismiss the guerrilla-style coning of robotaxis as one more sign of an anti-tech backlash, or just of San Francisco being San Francisco. But those narratives understate the significance of the current uproar. For the first time, urban residents, tech companies, and public officials are debating whether and how self-driving cars fit into a dense city. This is a conversation that needs to happen now, while autonomous-vehicle technology is still under development—and before it reshapes life in San Francisco and throughout urban America. A century ago, the U.S. began rearranging its cities to accommodate the most futuristic vehicles of the era, privately owned automobiles—making decisions that have undermined urban life ever since. Robotaxis could prove equally transformative, which makes proceeding with caution all the more necessary.

In the utopian version of the robotaxi story, trips in autonomous electric vehicles become so affordable, easy, and pleasant that many people decide to forgo owning their own car. Because San Francisco is close to Silicon Valley and is home to so many investors and tech journalists, the city is a high-profile, high-stakes testing ground for the emergent technology. It’s also a more challenging environment for autonomous cars than sprawling places such as Phoenix and Las Vegas, which have fewer pedestrians, cyclists, and transit riders.

The novelty of self-driving cars is a key part of their appeal. “Tourists take pictures of me, so I get to feel like a small celebrity,” a San Francisco resident named David Anderson, who says he requests a Waymo ride multiple times a week, told me. Beyond the wow factor, these companies offer a service akin to ride-hailing, which researchers have found worsens urban congestion and pulls riders away from transit.

Perhaps the most appealing argument for robotaxis and other self-driving vehicles is that human drivers are so fallible. Last month, a Waymo co-CEO published an op-ed in the San Francisco Chronicle extolling the safety of the company’s vehicles, and Cruise ran a full-page ad in The New York Times and other newspapers presenting its technology as a solution for the 42,795 road deaths last year in the United States. (The companies have been reluctant to share data on their operations, hampering unaffiliated researchers who might weigh in objectively.) For now, though, robotaxis are creating a slew of headaches for San Francisco officials. Even with limited deployment, the vehicles have blocked traffic lanes, obstructed buses and streetcars, driven over a fire hose, and entered an active construction zone. Since June 2022, San Francisco logged almost 600 instances in which robotaxis made unplanned stops—some lasting hours—on public streets. That count was limited to incidents reported to city officials, suggesting that the actual number could be far higher.

[Read: Finally, the self-driving car]

“These vehicles perform very well in basic suburban driving conditions, but they face challenges when cities have greater levels of complexity, and particularly when they are in unexpected situations,” Jeffrey Tumlin, the head of San Francisco’s Municipal Transportation Agency, told me. “In a city like San Francisco, the unexpected is ubiquitous.” Jeanine Nicholson, the city’s fire chief, offered a blunter robotaxi assessment: “They’re not ready for prime time.” Steven Shladover, a research engineer at UC Berkeley who has advised California officials about autonomous vehicles, told me that Cruise and Waymo vehicles sometimes show admirable sophistication. But, he said, “you’ll almost inevitably encounter a situation where the vehicle will act like an inexperienced driver. It’s an adolescent, not yet an adult.”

Cruise and Waymo have responded to critics by touting their vehicles’ overall safety record, which they argue is far superior to human drivers’. “We should be doing everything possible to quickly and safely scale this technology and combat a horrific status quo,” Cruise declared in a statement last month, after CPUC postponed its decision on easing limits on robotaxis. “Every single day of delay in deploying this [life]-saving autonomous driving technology has critical impacts on road safety,” Waymo asserted.

Robotaxi companies are under pressure to scale up quickly. Having invested billions of dollars, their backers want to see growth (the demise last year of Argo AI, a prominent robotaxi competitor backed by Ford and Volkswagen, undermined investor confidence in the industry). Cruise is aiming to put 1 million robotaxis on U.S. streets by around 2030, and CEO Kyle Vogt said during an earnings call last month that “certainly there is the capacity to absorb several thousand [robotaxis] per city at a minimum.” (When I requested comment from Cruise by email, the company did not respond to my questions about its expansion goals.) Any delay from the CPUC—one of whose five members is Cruise’s former managing counsel—makes the company’s objectives harder to achieve.

Despite its current challenges, self-driving technology is steadily improving, inferring new lessons from reams of data collected from vehicles plying public streets. Eventually, robotaxis might avoid the kinds of traffic and safety hazards that have afflicted San Francisco. But even if robotaxis operate perfectly, what would life be like in a city where they are ubiquitous?

Before gas-powered automobiles arrived en masse, American streets bustled with activity. Pedestrians, horse-drawn carriages, and bicyclists jostled for space, and children played stickball, marbles, and other games on the pavement. Streetcars carried millions of passengers on 45,000 miles of track; in the 1920s, most of Chicago’s nearly 100 streetcar lines operated 24 hours a day, with some providing service at eight-to-10-minute intervals in the dead of night. Photographs of urban thoroughfares at the dawn of the 20th century may appear chaotic, but the danger was limited, because no one traveled much faster than 15 miles per hour.

Early on, cars were too pricey for all but the most affluent urban residents. But after the introduction of the Ford Model T, U.S. car sales surged, rising from 181,000 in 1910 to 4.5 million in 1929. Traveling faster than anything else on the street, these vehicles soon presented a mortal threat to pedestrians and children. Some 25,800 people died in crashes in 1927, a per-capita fatality rate substantially higher than today’s despite Americans owning far fewer cars at the time. “The dead were city people, they were not in motor vehicles, and they were young,” the University of Virginia historian Peter Norton wrote in Fighting Traffic: The Dawn of the Motor Age in the American City.

[Read: Everyone has ‘car brain’]

In the early 1920s, Norton recounts in his book, St. Louis and Pittsburgh residents erected immense memorials to those killed in car crashes. In Cincinnati, a 1923 ballot initiative proposed a mandate that all motor vehicles within the city be outfitted with speed governors set to 25 miles per hour. “Forty-two thousand people put their names on petitions, just in that city,” Norton told me. “That’s a sign that there were a lot of people troubled by car domination.” Alarmed, the auto industry rushed to mobilize against the Cincinnati measure, which was defeated.

Seeking to avoid debating whether fast vehicles could coexist with urban neighborhoods, the car industry worked with friendly government officials to reframe road safety as the responsibility of the individuals at risk of being struck. Car groups funded school curricula instructing children to stay out of streets and worked to establish jaywalking as a crime. Meanwhile, city sidewalks and public spaces were torn up to expand traffic lanes.

Urban cars proved devastating for streetcars unable to navigate around a motor vehicle blocking their tracks. “The arrival of private automobiles quickly gummed up streetcar efficiency and made them much less competitive and comfortable,” Nicholas Bloom, a Hunter College urban historian and the author of The Great American Transit Disaster: A Century of Austerity, Auto-Centric Planning, and White Flight, told me in an email. “Streetcars lacked exclusive rights of way, so exploding auto traffic dramatically slowed streetcar service.” Automobiles also enabled many city residents to relocate to suburbs unreachable by transit. By the 1950s, American streetcar service had collapsed. In 1960, just 12 percent of commutes to work occurred on transit; by 2019, the share had tumbled to 5 percent.

The aftermath of these early auto-centric decisions still reverberates today, causing cities to become dirtier, more dangerous, and less fun. More than half of the land in many downtowns is used to move and store motor vehicles, occupying space that could otherwise accommodate housing, retail, playgrounds, and parks.

Many cities are now taking steps to correct past mistakes. Last year, Denver voters passed a referendum that will allocate millions of dollars to improve sidewalks. Striving to make public transportation competitive with car trips, Phoenix and Madison, Wisconsin, are planning their first bus-rapid-transit lines. (Such moves could have aided streetcars a century ago.) In recent years, California, Nevada, and Virginia have moved to decriminalize jaywalking. Progress is gradual, but it is real.

Autonomous vehicles threaten that momentum, for the simple reason that self-driving cars are still cars. Whether operated by a human or software, automobiles generate pollution, require traffic lanes, and endanger pedestrians and cyclists.

One member of Safe Street Rebel told me he agreed with AV boosters that self-driving cars could make car trips easier than ever—which is exactly the problem (he asked to remain anonymous because of the dubious legality of the group’s activities). “We have these two competing visions for the future of transportation,” he said. “We’re now talking about tearing down sections of freeways in San Francisco, but AVs go completely against that, because they need that road space to go quickly. If we have more AVs, do we have to keep those freeways? Or can we invest in better transit, so we don’t need those freeways?”

[Joshua Sharpe: We should all be more afraid of driving]

Norton, the University of Virginia historian, thinks the San Francisco activist’s concerns are valid. “Once we have streets with robotaxis, there is definitely a risk that the city feels that it doesn’t have to supply basic public transportation,” he told me. In fact, such views have already been shared. “Don’t build a light rail system now. Please, please, please, please don’t,” Frank Chen, a partner at the venture-capital firm Andreessen Horowitz, told The New York Times in 2018. “We don’t understand the economics of self-driving cars because we haven’t experienced them yet. Let’s see how it plays out.” The year before, officials in Miami-Dade County, Florida, cited autonomous vehicles as a reason to refrain from expanding public transportation.

But any suggestion that the vehicles will significantly improve mobility on their own seems fanciful. A few years ago, researchers provided 13 Bay Area volunteers with a personal chauffeur who would bring them wherever they liked, mimicking the experience of accessing a self-driving car. During their week with the chauffeur, the test subjects traveled a whopping 83 percent more car miles than they did previously. Autonomous vehicles would be an environmental disaster if they induced anywhere near that much extra driving. They could also create unprecedented gridlock on highways and streets.

Shladover, the Berkeley research engineer, thinks such fears are overblown. “It depends on how the self-driving cars are used,” he said. “If they are deployed in ways that are complementary to transit, such as serving parts of the city that have minimal transit access, that is a significant plus.” But would for-profit companies focus on so-called transit deserts, or would they cater to the needs of a smaller subset of wealthier customers? For AVs to complement transit lines, urban residents must be willing to hop into robotaxis with strangers. That assumption is built into Cruise’s small, podlike Origin shuttle vehicle, but the troubles of Uber Pool and Lyft Line cast doubt on the idea.

Even if AVs do live up to their hype technologically, their long-term effect on cities is hard to predict. “While there’s a lot of data indicating that AVs can contribute to overall safer streets, the reality is that they exist in a transportation ecosystem and are not a panacea,” Drew Pusateri, a Cruise spokesperson, told me in an email. “We need a much broader approach to road safety that includes more investments in mass transit, wider, slower roads and a variety of other solutions.” But the past century suggests that when a transportation system is built primarily for cars, people using other modes get short shrift. That may remain true even when the cars are driving themselves.

In San Francisco, public officials and activists are raising fundamental questions about the desirability of autonomous vehicles within cities—questions that have seldom been aired in public. Instead, elected officials seeking an aura of innovation in states such as Texas and Arizona have actively pursued early AV deployments, minimizing regulations and shunning hard questions. Only now, in California, are robotaxi companies finding themselves in the unfamiliar position of playing defense in the public arena.

Regulators at the CPUC and elsewhere should encourage a vibrant public dialogue around autonomous vehicles, and learn from it. The worst thing they can do is rush decisions to scale an alluring new technology whose downsides could be catastrophic.