A ship is seen stranded on the shore after an earthquake and tsunami hit Donggala, Central Sulawesi, in Indonesia.Muhammad Adimaja / Antara Foto Agency / Reuters

On Friday, an earthquake and tsunami struck the island of Sulawesi in Indonesia, flattening hotels, shopping malls, and hundreds of homes, and killing at least 1,200 people. The government expects the death toll to rise.

In the days after the quake, outsiders have focused on the failure of the local tsunami early-warning system. Many of the deaths occurred in Palu, a medium-sized city at the end of a long, skinny bay, nestled between mountains. “Tens to hundreds” of people were at the edge of that bay, enjoying themselves at a beach festival, when the water started rising. This scenario—beachgoers utterly ignorant of a tsunami at the very moment it struck—is exactly what early-warning systems are supposed to prevent. Why didn’t they work this time?

The tragic answer has to do both with broken buoys and the horrifying and irrepressible power of tsunamis. “A tsunami will come faster than any early-warning system can possibly work,” says Mika McKinnon, a geophysicist and disaster researcher based in San Francisco. In some disasters, early-warning systems will never be fast enough.

Not that Indonesia didn’t try to create one. In 2004, a magnitude-9.3 earthquake erupted off the coast of Sumatra, spawning a catastrophic tsunami that rippled across the Indian Ocean and killed more than 230,000 people in 14 countries. In the years following that disaster, international organizations and the Indonesian government installed 22 floating tsunami-detection buoys between the islands of its archipelago. But this system has since faltered.

“Buoys have a lot of parts and electricity, so they attract fish. Fish attract fishermen. And the fishermen abuse the buoys,” says Louise Comfort, a professor of international relations at the University of Pittsburgh. By 2016, when another large earthquake struck Sumatra, every single one of the 22 stations had been destroyed, vandalized, or raided for parts. A team of American and Indonesian researchers—led by Comfort—have proposed and tested a lower-cost network of tsunami sensors, but they do not yet have funding to install it.

Even without that network of sophisticated buoys, Indonesian authorities still managed to use a more rudimentary set of GPS sensors and ordinary buoys to predict last Friday’s tsunami ahead of time. But this, too, was for naught. The earthquake that spawned the tsunami also took out the local power and communications grid, muzzling tsunami sirens throughout Palu. As the dreadful wave approached, beachfront alarms did not sound.

In fact, that power outage also broke the rudimentary network of tidal gauges used to predict the tsunami in the first place. Tragically, the Indonesian government took this lack of data as an all-clear: “They canceled the tsunami warning at the exact same time that the tsunami was coming ashore,” McKinnon told me.

Yet it’s not clear whether—even if the buoys did work—they would have saved many lives in Palu. The U.S. government, for instance, runs the world’s most sophisticated tsunami early-warning system. But it’s less miraculous than most people think, says Chris Goldfinger, a professor of geophysics at Oregon State who studies what tsunamis would do to the coast of the Pacific Northwest.

“There’s this idea that the [early-warning] system will provide some useful warning to you, if you’re on the beach, directly adjacent to the fault line, and that’s not really true,” he told me. “In most cases, if you’re on a coastline that has a big subduction zone, the earthquake is the warning.”

Take, for instance, the big subduction fault that happens to be right near an American coast: the Cascadia subduction zone—the subject of that terrifying 2015 New Yorker story. If that subduction zone slips, it will probably loose a devastating earthquake on the Pacific Northwest, originating somewhere near the coast. “It’s just going to go without any preamble or any warning,” Goldfinger said. “And then that starts the stopwatch.”

The stopwatch, that is, for the tsunami. If the Cascadia quake lasts for two minutes, it could spawn a 15-foot tsunami. If it lasts for longer than that—and the ground could shake for up to six minutes—it could unleash a wave “on the order of 90 feet,” Goldfinger told me.

But how would we know this wave is coming? Most of the U.S. early-warning sensors sit relatively offshore. If Cascadia triggered a tsunami, its point of origin would likely sit between the coastline and that array of tsunami sensors. So any wave would have to travel out to sea before it could be detected by a U.S. government sensor. Even once it’s detected, news of the wave must still be transmitted to a satellite, sent back down to Earth, and reviewed by a person in the U.S. tsunami center before, finally, resulting in an official early-warning alert.

This whole process could take three to five minutes, Goldfinger told me. Meanwhile, the inbound wave would be approaching the coast. He warned that beachgoers near the epicenter of a future quake may only have 15 minutes before the tsunami strikes. So if you feel violent shaking, but don’t evacuate until you get a tsunami early-warning alert, you may have burned up a third of your time.

“We love technological solutions to problems. But the tsunami-warning system is really not a technological solution to the problem of a near-field tsunami,” Goldfinger said. “For the people on the coast here [in Oregon], it’s like a broken coffee machine. You put your money in, it pours all the coffee out, and then the cup comes down afterwards. Being off by a few minutes isn’t going to cut it.”

“That’s why some of us, to be a little tongue-in-cheek, call it the Japanese tsunami-warning system,” he added. “That’s its real function. It will warn Hawaii or Japan of an outgoing wave.”

The daunting lesson from this scenario is that education and preparation will save far more lives than a sophisticated tsunami warning system. That’s why everybody needs to know one big lesson about tsunamis, McKinnon told me: “If you’re at a coast, and you feel severe shaking, you run for high ground as fast as you can. If you don’t see high ground, then you get on the roof of a building—preferably one with concrete pillars.”

It helps, she said, that the same type of sign denotes tsunami-evacuation routes around the world.

As for the Palu quake, locals had other forces working against them. Last Friday’s tsunami was not triggered by a big subduction-zone quake. “It was generated by an underwater mountain moving sideways or by an underwater landslide,” McKinnon said. “That means it’s not going to travel a very big distance.” It also would have made it hard to detect.

That same kind of “small” tsunami is a risk on the Pacific Northwest and down the Pacific coast, Goldfinger said. (The Port of Los Angeles would be devastated by just a 1.5-foot tsunami, he told me.) So he preaches a kind of low-key but alert preparedness for anyone who lives near the Pacific Ocean. “I don’t obsess about it,” he said, “but I don’t like to stay overnight in parts of the coast that are in these low-lying, difficult-to-evacuate places. And when I go to the beach, I tend to gravitate to places with good evacuation routes—just as a matter of everyday life, the same way you’d skirt around an icy patch on the road.”

That kind of risk management, he said, could do a lot more than fancy technology. “We often think,” he told me, “that we’re protected from something that we’re really not protected from.”

We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.