On the afternoon of Jan. 9, nearly 10 million people across Los Angeles County received a warning that they were in danger.
“NEW: This is an emergency message from the Los Angeles County Fire Department. An EVACUATION WARNING has been issued in your area,” the alert began, going on to list suggested actions and resources.
The message met a population already on edge. It was the third day of dry, gusty Red Flag weather conditions, which were stoking the spread of wildfires across the county. The two largest and most devastating blazes, the Palisades Fire and the Eaton Fire, had already sent more than 180,000 people away from their homes and destroyed thousands of structures, and both were still burning.
A correction went out about 20 minutes later, clarifying that the warning had only been intended for those in the area around the Kenneth Fire, one of the new starts threatening communities near West Hills. But for many, the false evacuation alerts continued, buzzing their phones overnight and into the next day.
“There is an extreme amount of frustration, anger, fear with regards to the erroneous messages,” Kevin McGowan, Director of the Los Angeles Office of Emergency Management (OEM), said in a press conference the following morning. He promised this was an “all-hands-on deck” situation and that his office was working with the Federal Emergency Management Agency to identify and fix the issue. In the meantime, McGowan said L.A. County would use the state’s system to send its alerts and told residents to cross-check any evacuation messages with the county’s maps.
For many in L.A., evacuation warnings — sent to alert residents that they should prepare to evacuate — offered a life-saving opportunity to escape the fast-moving fires before the area was officially under an evacuation order, which means the threat is imminent. But the false alarm and delayed response, as well as the repeating incorrect warnings, opened the door for uncertainty and even harm if recipients had tried to evacuate en masse and blocked roads needed for those who were actually fleeing danger. The errors also threatened the integrity of these alerting system, with the county acknowledging in a statement they were a “serious breach of public trust” and experts questioning how many people might have chosen to opt-out of alerts because of them.
One week later, authorities are still investigating what went wrong with L.A. County’s alerts. What we know so far suggests multiple points of failure within the complex system, including one issue experts say they have been drawing attention to for years.
Untangling What Went Wrong
At the press conference on Jan. 10, McGowan said the incorrect alerts were a technical issue rather than a human error. “There is no one sitting at a desk right now initiating emergency alerts,” he said.
Determining where an issue lies within the alerting process can be complicated, because the system itself is layered. While emergency alerts often resemble text messages, they actually stem from a unique partnership of government offices and agencies with wireless providers and private companies. L.A. County uses a third-party software system called Genasys to enter in its alerts, which are then dispatched through FEMA’s Integrated Public Alert & Warning System, known as IPAWS. From there, the message is distributed as a Wireless Emergency Alert, which most major cell phone providers in the United States carry as a courtesy. And what that alert actually looks like when it reaches you depends on even more variables, like the model and age of your phone.
A FEMA spokesperson tells Rolling Stone that the agency is working with the state, county, and other local safety offices across L.A. to troubleshoot the issue, as well as wireless carriers and the Federal Communications Commission. Zach Stanford, a consultant for Bent Ear Solutions LLC and former Oklahoma emergency manager, said the technical issues they’ve identified so far can be broken down into three categories: Alerts being received by residents after they expired or were cancelled, alerts being received multiple times, and alerts being received outside of their intended target area.
Going down the list, the first issue is one FEMA has seen before during other incidents that involved power interruptions. When emergency alerting authorities enter messages into the system, they choose a length of time for them to remain active. That doesn’t mean you’re supposed to get it more than once, but it does allow for new people entering the at-risk area to receive the message as they enter the intended warning perimeter.
While just how long an alert goes up differs depending on the threat, Stanford tells me, “I’ve never known a Wireless Emergency Alert to be valid for less than 15 minutes, and they can’t be valid longer than 24 hours.” But if the wireless infrastructure goes offline while an alert was scheduled, messages that had been set up might go out when those sources come back on, even if they had technically cancelled within the system, according to FEMA.
The second issue — repeating alerts, or “echoes” as the L.A. County OEM has been referring to them — also points to problems with cell carriers.
“Our preliminary assessment is that these recurring erroneous notifications are due to issues with telecommunications systems, likely due to the fires’ impacts on cellular towers,” a statement from the L.A. OEM explained on January 10. The FCC did not respond to Rolling Stone’s request for comment on the investigation.
The last issue — the accidental county-wide send — was the one Stanford says is the most complicated to pin down. The L.A. OEM said their investigation shows an “accurate, correctly targeted alert” was sent from the Emergency Operations Center on Jan. 9, but “after it left the EOC,” it was erroneously sent way beyond its intended boundaries. The county said its software provider, Genasys, has been conducting tests to see how this happened.
In a statement to Rolling Stone, Genasys did not expand on what might have caused the problem but said the company had instituted “additional safeguards” to prevent it from happening again, including a new confirmation message that will appear if an alert is being sent without a saved target area, meaning it would be sent to the entire county.
Stanford says it’s difficult to know from outside of the investigation whether this issue was an operator error or a system malfunction.
“These are very different issues we need to better understand,” he says.
Repairing the Trust
So far, the parties involved in the investigation remain focused on figuring out what happened and preventing it from occurring again in the near-term rather than proposing any significant fixes. But there is one simple way officials can ease some of the confusion associated with these alerts moving forward — including more details.
While WEA do have character limitations, Jeannette Sutton, director of the Emergency and Risk Communication Message Testing Laboratory at the University of Albany, has spent years testing the most effective language to include in these counts. Through her research, she’s found the most effective alerts include six components: The name of the source sending the information, a description of the hazard, the location of the hazard, the consequences of the threat, the correct protective action to take, and when to take it.
The county-wide message sent to Angelenos included many of these details but lacked the location specificity, allowing it to be easily misinterpreted by the broader audience. In the case of a real wildfire threat, it would have left residents confused where “your area” was, requiring them to seek additional information instead of starting to figure out how they were going to keep themselves safe. Additional details on timing could have also helped those who received the repeating alerts recognize that they were false alarms, rather than causing further confusion and fear.
“In some cases, the more of those messages you get that are really undefined, you might decide, ‘Well, this isn’t for me,’ and then you just stop paying attention — which is the last thing you want people to do,” Sutton says.
Another consequence might be people choosing to opt-out from receiving any further alerts from the county. But there’s no way to track whether this happened, as Sutton says alerting officials have “zero insight” into those numbers, which are not provided to them by cell carriers.
“So, we don’t know what that broader impact is on that particular message that went out,” she says, adding, “But the repeating messages, my gosh, you can imagine people are turning them off.”