People have a tendency to talk about their city’s terrible drivers with a weird sense of pride. In Boston, they’re called Massholes. In Los Angeles, even the 18-wheelers go bumper-to-bumper with cars on the freeway. In Philly, being aggressive and rude behind the wheel is just a given. (The particularly bad drivers, however, are the ones with Jersey plates.)

Wherever you go, it seems, traffic is terrible. Everybody’s tailgating. And nobody’s using blinkers or paying attention. The automobile, the sociologist Henri Lefebvre once said, was “the last refuge of chance and risk in an increasingly controlled and managed society.” And although different cities have their own cultural quirks—in Honolulu, honking your horn is a faux pas unless it’s really an emergency—most bad drivers are bad for the same reasons: People are inconsistent, easily distracted, and generally pretty bad at risk assessment. All this is why self-driving cars promise to save so many lives.

That is, if they ever gain widespread adoption. Google has logged nearly 1.5 million miles of test driving on public roads in autonomous mode, and in all that time its cars have only caused one minor accident. But critics have pointed out that one fatal accident occurs for approximately every 100 million miles that Americans drive, a figure that makes Google’s test driving record seem puny. Academics who focus on robotics and self-driving cars have been quick to note that even as Google expands testing on public roads to other cities beyond Mountain View, California—including Austin and Kirkland, Washington—it still hasn’t taken to the roads in a region that gets a lot of snow and ice.

“Driving a million miles in Southern California? Big deal,” said Missy Cummings, the head of robotics at Duke University. “I would take 10,000 miles in North Dakota in the wintertime over one million miles in San Francisco. I think we need to start being more targeted and more smart about how we’re doing tests. Let’s stick that thing in Boston for a year. There’s not a white line on the road, the weather is shitty, and the traffic is worse.”

It’s fair criticism. If self-driving cars are to transform society the way so many technologists claim they will, they will eventually have to prove they know what to do in a blizzard—even if knowing what to do means refusing to drive through it. But critics who are pushing Google to begin testing in icy conditions sooner rather than later are unlikely to convince the company to do so before it’s ready.

“For now, if it’s particularly stormy, our cars automatically pull over and wait until conditions improve (and of course, our test drivers are always available to take over),” wrote Google in a December 2015 report, referring to rainy roadways. “To explore even more challenging environments, we’re beginning to collect data in all sorts of rainy and snowy conditions as we work toward the goal of a self-driving car that will be able to drive come rain, hail, snow or shine!”

There are at least some companies already testing autonomous driving technology in a region known for its inclement weather—though not necessarily on public roads. In Ann Arbor, Michigan, there’s a testing site called Mcity that’s used by Ford, General Motors, Honda, Toyota, Nissan, and others, according to The New York Times.

As the leading and most public-facing company in the driverless car space, Google is under enormous pressure to both get the technology right and to win a massive battle for public opinion. To be successful, Google has to select locations for test driving very, very carefully. “When we think about the technology rolling out over time, we imagine we’re going to find places where the weather is good, where the roads are easy to drive, and the technology might come there first,” said Chris Urmson, the director of Google’s self-driving cars project, at a conference in March.

Which brings us back to Honolulu. It never snows there. It rains—more on the windward side of the island than the leeward side—but not substantially more than in Kirkland, Washington, where Google is already testing its vehicles. And there’s ample room for relatively uncomplicated test routes—long highways like H1, and the mostly flat roads flanked by pineapple fields en route to the country. Even downtown Honolulu isn’t as congested or dense as other metro areas. (Navigating tourist-choked Waikiki would probably pose more of a challenge for any driver, computer or human.) And despite perennial complaints about potholes, the upkeep of roads—with mostly clear signage and bright-enough paint on main roadways for a computer sensor to read—is mostly decent.

There are other reasons driverless cars would be transformative for Honolulu, in particular. The city, on the island of Oahu, routinely tops national “worst traffic” lists. Most commuters leave their homes before 7 a.m. in an attempt to avoid traffic, according to Honolulu magazine, and people who live on the west side of the island routinely spend two hours stuck in traffic just getting to (or home from) work. Public transportation options are mediocre at best, and pedestrian safety is abysmal. Hawaii has one of the highest rates of pedestrian fatalities in the country, and a startling portion—about one-fifth—of overall traffic deaths include pedestrians.

So, to sum it up: Honolulu desperately needs to address its problems with gridlock and pedestrian safety, and it has consistently good weather. If driverless cars are going to take off anywhere, Oahu seems like a strong candidate for early adoption. That’s still no guarantee.

“I believe that driverless taxis are going to induce a large-scale abandonment of car-ownership in urban areas over the next two to three decades,” said Shem Lawlor, the director of Clean Transportation at the Blue Planet Foundation in Honolulu. “However, since the pricing will still not be as cheap as walking, biking, or transit—and since it’s logistically impossible for driverless taxi services to ever move a sizable percentage of peak-hour traffic volume, I believe we are going to see a tremendous increase in demand for public transit, biking infrastructure and walkable neighborhoods.”

Then, there’s the question of whether Hawaii residents would actually welcome such a test project—and it’s not clear that they would. Debate over expanding Oahu’s public transportation has raged for decades. Only now is a controversial elevated rail line, first proposed in the 1960s, finally under construction. (Whether this is a good thing is still very much a point of contention in the Islands.) Taxi companies in Honolulu have fought ride-sharing services like Uber and Lyft, arguing they ought to be outfitted and regulated like taxis. And for most people living on Oahu, car ownership is considered essential.

Whether exposure to alternatives like driverless cars could change people’s attitudes—in Honolulu, or anywhere else—remains to be seen.