A view of an Anchorage neighborhood after the 1964 quake.
A view of an Anchorage neighborhood after a 9.2 quake in 1964. Everett Historical/Shutterstock

In 1964, an oceanic plate suddenly drove north beneath the southern Alaskan coast, thrusting under the North American Plate like a spatula under a stuck pancake. The earthquake splayed the coastline south, and tsunamis caused by displacement on the ocean floor rippled as far away as California. One-hundred and thirty-one people died.

Fifty years later, in March of 2014, a panel of earthquake experts sat in front of a House subcommittee session dedicated to the Great Alaska Earthquake’s half-centennial memory. This time, though, discussion pivoted to another fault line ripe for a megathrust quake: the Cascadia subduction zone.

Panelists made a plea for an earthquake early warning system.

"For example, along the coast of Washington, Oregon, and California, subduction zone earthquakes have occurred every few centuries, and the last one was way back in 1700," Lisa Grant Ludwig, a geologist and a professor at the University of California-Irvine, told lawmakers. "The old saying, 'Those who ignore history are condemned to repeat it,' should not apply to earthquakes in the US, because we know as a direct result of federally-supported research and seismic monitoring we have an earthquake problem."

Not long after Ludwig and her colleagues finished presenting, Rep. Peter DeFazio (D-Oregon) lost his shit on C-SPAN. The fact that the US Geological Survey had only spent $10 million since 1999 on an early warning system rattled him.

"We've spent $10 million—wow!—since 1999!" DeFazio said. "We're looking at a $100 billion problem in the Pacific Northwest. We've spent $10 million."

John Vidale, director of the Pacific Northwest Seismic Network, told DeFazio: "I think we should build the early warning system. The physics is simple. We know how to do it. It's a bit of a challenge to make sure we don't get false alarms, and we have to adapt the emergency broadcasting system to react in seconds rather than minutes…"

DeFazio cut him off and threw up his hands. "But, I mean, if Romania can do it, can't the United States of America?! I don’t know. I guess we've fallen so far, and the Republicans are disinvesting so much we can't, but this is crazy."

Today, John Vidale has a beta-version of an earthquake early warning system on his phone. But on February 18 of this year, when a 4.3-magnitude earthquake shook central Washington, the prototype delivered the news late.

Vidale and his colleagues at the Pacific Northwest Seismic Network have been working over the last several years to fine-tune an earthquake early warning system for the West Coast. (Japan's, the most advanced in the world, rolled out in 2007.) But in order to make an early warning system sing, a network of hardware sensors have to act like one hyper-accurate, knee-jerk reflex. The sensors—either buried several feet underground or shielded from the elements above it—have to be delicate enough to pick up the first, weak "P-wave" signals from a fault, sturdy enough to send those signals back to researchers without distortion, and agile enough to do all of that within a second.

"And if we do it quick enough, we can warn people before the stronger waves hit," Vidale says. If the system's successful, it could alert people in Seattle to a quake by as much as several minutes ahead of local impact, though that would depend on where along the Cascadia subduction zone the earthquake strikes.

But it doesn't help that a lot of the earthquake instrumentation embedded in Washington State dates back to the 1960s. Vidale estimates that as much as two-thirds of the state's 400 sensors are pretty much useless for an early warning system. A lot of the instrumentation takes a full minute to report back. And while the hardware from the '60s is pretty sensitive, it also tends to clip—or distort—when sending the signal.

Each year, seismologists like Vidale have asked the USGS for funding between $16 and $17 million to build out the early warning system. The seismologists' final network would consist of about the same amount of hardware currently in place, but two-thirds would be seismometers, and a third GPS. Vidale says that much of the total ask would go to paying the salaries of 30 to 40 people split between California and the Pacific Northwest to process the data.

This past year, the USGS kicked in $6.5 million for the effort, more than it usually budgets, but less than half of what the researchers would need on an annual basis. Those same researchers are hoping the USGS will ramp up to the full amount in the coming years.

Or, the USGS could fund an additional $40 million upfront and possibly get an early warning system running in the Pacific Northwest within two-to-three years. California could complete one in an even shorter timeline, Vidale says, maybe even within 12 months.

Replacing all the seismology instrumentation would take a full decade, but with a $40 million investment right now researchers could get a "good set of instrumentation in the ground" in five years, Vidale says. In that time, researchers could also figure out algorithms that wouldn't trigger false alarms and help bring a super-fast emergency broadcasting system online. After all, an earthquake early warning system also needs a way to disseminate that information through cellphones, building alarm software, TVs, and radios. And that's where state funding comes in.

But Vidale's not apoplectic about the pace of progress. (Even though the New Yorker piece cited a one-in-three chance of a Really Big One in the next 50 years, Vidale says the chance of a 9.0 is closer to one-in-six.) Plus, implementing a new system gets easier as the technology gets faster and more precise. He's gunning for an early warning system, but doesn’t want to unnecessarily freak people out. (What's the point?)

"I sort of think we're on a good path," he says. "We could have set it up 10 years ago, but it would have been a lot harder."

Still—and obviously—having an functional early warning system as soon as possible is better than, you know, not.