False Tsunami Warning Sent To the East Coast, Gulf of Mexico and Caribbean (nbcnews.com)
An anonymous reader shares a report: Some residents along the East Coast received a false tsunami warning on Tuesday morning after a private company sent out an alert following a monthly test by the National Weather Service. A tweet from the National Weather Service (NWS) in Charleston, South Carolina, said the alert was sent around 8:30 a.m. ET. "We have been notified that some users received this test message as an actual Tsunami Warning," the NWS tweeted, adding that a tsunami warning was "not in effect." In a statement to NBC News, the NWS said that a routine test was sent out and that the agency is investigating why it was communicated as an actual tsunami.
The NWS must not have checked his references. At least they are consistent about the ~30 minute later "Oopsy, we made a bad"
Why would Accuweather (the company who sent this) want the liability and burden of being responsible for (or even touching anything to do with) sending a life-or-death tsunami warning? Would you, as a company say "ok, the National Weather Service sends these things out, let's let them handle this the whole way"? Why would you think it good to take on that role?
Step one: disconnect the outside from the inside. Big toggle switch on the wall, itty bitty network jack gets torn out. We call this: network isolation. You can easily build this into your testing software. If google.com is reachable, don't send tests.
Step two: announce it in-advance. Tomorrow, at noon, precisely sixteen hours from now, we'll be running a test. It'll say "missile carried by tsunami". There will only be one test. It will specifically not say "seriously really". We call this: planning ahead. Again, you can easily build this into your testing software. If there's no record of an announcement, don't send tests.
Step three: non-accidental confirmation. Tests can be sent with the push of a button. Real-world warnings require a human being to type the words "send real warning to EVERYONE". It's not case-sensitive, and three type-o's are permitted. Again again, you can easily build this into your testing software. "Click OK" is replaced by "type this phrase" -- and specifically not for tests.
Step four: two heads are better than one. Tests can be sent by one human. Real-world warnings require a second human to do the same things as the first human, within the same few minutes. Again, easily built into your software. A single command is ignored, two commands are executed.
Two in four weeks is not only intolerable, but it's actually more painful than the actual events would have been. I don't know how many death-bed confessions occurred in hawaii, and I can't imagine the health results of that kind of stress on an entire population. But I do know one thing very well: notification fatigue completely destroys the future. How many Hawaiians will simply ignore it next time, and die as a result?