Schneier on Security
A blog covering security and security technology.
« Backdoor Found (Maybe) in Chinese-Made Military Silicon Chips |
| The Psychology of Immoral (and Illegal) Behavior »
May 30, 2012
The Problem of False Alarms
The context is tornado warnings:
The basic problem, Smith says, it that sirens are sounded too often in most places. Sometimes they sound in an entire county for a warning that covers just a sliver of it; sometimes for other thunderstorm phenomena like large hail and/or strong straight-line winds; and sometimes for false alarm warnings warnings for tornadoes that were incorrectly detected.
The residents of Joplin, Smith contends, were numbed by the too frequent blaring of sirens. As a result of too many past false alarms, he writes: "The citizens of Joplin were unwittingly being trained to NOT act when the sirens sounded."
Posted on May 30, 2012 at 6:44 AM
• 42 Comments
To receive these entries once a month by e-mail, sign up for the Crypto-Gram Newsletter.
"False Positives"... like clicking "OK" on Windows...
I learned to ignore storm and tornado warnings as a kid - 25 years ago. Even today - thanks to monthly tests and "emergency awareness" weeks - I don't give them a second thought when they go off even when I know a storm is forecast.
Of course - this all assumes you even hear it in the first place.
Reminds me of the months following last year's earthquake/tsunamis here in Japan. Everyone's mobile phone earthquake alarm would sound about three times a day or more for aftershocks until everyone turned them off or set them to alert only higher-strength quakes.
Still now, they tend to go off without being followed by anything at all, and these days we just ignore them until the ground actually moves... which kind of defeats the purpose.
I Boulder we had the additional problem that it wasn't always clear whether you were getting a tornado warning ("Head for the Basement!") or a flash-flood warning for Boulder Creek ("Head for the Hills!").
I would be cautious in accepting the premise of this article. The book in question is written by a VP at a private company which competes with the National Weather Service (and which has lobbied for the NWS to be privatized). That doesn't mean the criticisms are wrong, but they should be taken with a grain of salt.
I don't know very much about how the local tornado warning siren system interacts with National Weather Service warnings. But the NWS issued an accurate tornado warning for the Joplin tornado (separate from the 5:10 warning for the weaker one mentioned in the article) at 5:17 PM local time, a full 24 minutes ahead of touchdown. This is about double the average lead time, and probably reduced the death toll substantially.
But as I said, I don't know much about the warning siren system. The NWS warnings do include information about the location and heading of the tornadic cell, and this should make it possible to determine which specific sirens to turn on and when. Maybe someone who knows more about how the siren system works can give some thoughts. I just want to defend the NWS warnings, which were quite good that day, and remind people that the book in question is written by a competitor to the NWS trying to encourage more adoption of his company's warning systems.
The same phenomenon happens with car alarms. Nobody is thinking that a car is being stolen anymore when it goes off.
False alarms may be the least of the problems with sirens:
"James Spann, a longtime television meteorologist at Birmingham's ABC affiliate, says the reliance on sirens has led to dozens of deaths over the years. "In the siren mentality, it's the idea that you're always going to hear a tornado siren before a tornado strikes. And I believe it's a farce." Sirens are decades-old air-raid technology from World War II that were designed, principally, to warn people who are outdoors of threats. Today, homes are built and insulated so well that outdoor warnings rarely make it inside."
What If Dorothy Had A Smartphone?
The boy keeps crying wolf, only now it's in a modern setting.
I think Randall Monroe makes an interesting counterpoint. We live in a hurricane belt. We get more advanced warnings than folks in a tornado belt but many people avoid evacuation until *everyone* evacuates. Folks who've considered the warnings more thoughtfully - or who have endured 18 hours idling on I-95 - take a measured response and leave just before everyone leaves. This is a different "learned" response than "ignoring the warning". It is still a learned response but it is better suited to mitigating risk because it applies data to refine the response.
We have to be careful not to dismiss warnings too quickly because the false positive rate *appears* high. Instead, we ought to help people learn to consider a variety of factors before dismissing a warning as a false positive.
I wholeheartedly agree. Or my preferred response, "Well...duh!"
I want to know when my *life* is in danger! I can see a "bad" storm. I want to be warned when a full fledged tornado is headed my way. I understand that some communities don't have the capability to turn on only some of their sirens, but you would expect a better system where tornadoes are prevalent.
I'm thinking that a reverse-911 system makes the most sense, since the radars can track to the street level.
We see it all the time where I work. "Tornado's headed right for us" Everyone goes downstairs to shelter in front of the big glass doors to be sucked out when it hits...and we never see anything.
I started staying home monday mornings just to miss the tornado warnings.
I'm a certified storm spotter and this is a topic that's frequently talked about. It's a real catch 22. Tornadoes are inherently unpredictable. There is no way to know if one is going to touch down or not until you actually see it happen (either with the old mark-1 eyeball, or a radar). Once the Tornado exists, there is no way to predict it's exact path, or how long it will last.
If the weather service waits to send out alerts until it's on the ground killing people, there is little point in sending out an alert. As a result a "better safe than sorry" approach is taken. Once all the conditions are met that allow for a tornado to form the alert is sent out. Unfortunately that still makes for a lot of false alarms.
Randy: Weather radars do not track at the street level, in fact they don't go down below 4000 feet, and that number goes up the further you are from the radar.
An air raid siren is not a false alarm just because a bomb doesn't land smack dab on your house.
The activation of a severe thunderstorm or tornado warning when there is an observed hazard in or approaching the area advised is not a false alarm.
Drills are not false alarms.
Public sirens are intended to alert persons outdoors that there is a potentially dangerous weather approaching and that they should seek shelter and additional information. The public sirens here in the Twin Cities are activated only in the areas within the storm track, not for the entire county as it once was.
Sirens are only one element of the diverse efforts of NOAA and NWS. Their data are much more granular these days and are available by radio, television, online, and by smartphone. Anyone who cares about hazardous weather can track it for hours before it arrives.
It's all a question of trade-offs. You may need to watch NOAA radar instead of The View and listen to the weather radio instead of your favorite drive time shock jock.
@Dylan, Re: radars resolve to 4000 feet.
Hmmmm...I'll have to pay closer attention the next time there's a severe storm in my area and the local weather man is broadcasting the details. *He* sure makes it seem that he knows exactly where the "eye" of the storm is and where it's headed and when it will get there.
Same as fire alarms. Hardly anyone even pays attention when they go off because we've all been through so many tests and false alarms.
@Randy - Tornado heading for town A
1 - local government decided not to sound the alarm in community B and the tornado changed path and partly hit B as well - got sued.
2 - Same government dept sounded the alarm in A,B and C and got congratulated on it's comprehensive proactive response to public safety.
The problem with sirens in the weather context is that small scale weather remains unpredictable. People in this thread keep talking about the false positive but neglect to mention the entire history of weather is based upon the sheer number of false negatives (Galveston, I'm looking at you.)
The NWS can't win because if they sound the warning and nothing happens people will complain and if they fail to sound the warning and something happens people will complain. In essence, a weather warning is a "best guess" and often a poor one. We've gotten much better with hurricanes but with tornadoes and other much smaller scale weather phenomenon the remote data sensing equipment just isn't in the field and honestly, if coldly, is cost prohibitive.
Dylan is referring to the altitude of the radar image, not the resolution of the image.
The resolution has gotten better and better with more powerful radar and processing technology. However, this doesn't mean the image you see is more accurate, only more precise. It measures the rain/wind/etc that appears 4000 ft above the ground, not at ground level.
And with tornadoes, the 'eye' of the twister may be in a significantly different location 4000 ft up in the air than it is where it contacts the ground.
There is even a bigger problem with most installations of sirens I've seen:
They are awesome during the test and totally inaudible during a real weather event. I have lived/worked within a block or 2 of 4 different installations. During extreme weather, you cannot hear them. Worthless.
Trust me, if you live in Dixie Alley and have seen what multiple EF4 outbreaks, some with tracks more than 100 miles on the ground can do, a simple warning will suffice, and you will keep one foot in the safe area and monitor events until all clear is given.
When I was growing up in the Twin Cities (Minnesota), we got about one tornado warning per month in the spring and summer, and I never saw a tornado.
Last year we got only one warning, and a tornado formed within hearing range of our siren. The National Weather Service's upgrade has made a big difference, and I'm sure that people are going to start paying close attention to those sirens.
Julien: When I hear a car alarm go off, I am *hoping* the car will be stolen. That way I don't have to listen to the freaking alarm anymore.
Re: weather sirens. I used to live in the Midwest. Whenever I heard one, I would go outside and wait for the fireworks. Never did see anything more than lightning and hail ...
Ahhh....that makes sense now. I feel *slightly* vindicated.
Re: location of the 'eye' of the twister
I'll bite for the obvious question. How much different can it be? 1000...2000...4000...10,000 feet? I'm betting 2000...which is less than 1/2 mile...which is also skinnier than some large tornadoes.
I think I'm gonna stop here though, since I'm obviously *not* a meteorologist.
Randy -- luckilynevermetatornado
Coincidentally, Ars Technica has an article today on a radar originally tasked to track shuttle debris which is now being used to study weather. Resolution is 1mm at a distance of 2km - it can actually track individual raindrops.
One Indiana county (I forget which) just announced that it is going back to only sounding the sirens when there is a Tornado Warning. Not for every other damn thing the weather might do. GOOD FOR THEM! I'm tired of them going off every time there is a storm warning here in Marion County. If it's not a tornado, I don't want to know.
I live in western OH and we dont have a tornado siren. We have a NOtornado siren.
In the 24 years Ive lived here it has gone off ~60 times (that I am personally aware of) and it is always either a test, an accident or, in about 5% of the incidents, a tornado had passed thru approx 15 minutes EARLIER but NEVER has it gone off BEFORE a tornado.
The tests/accidents speak for (and ill of) themselves, but I suspect the "there was a tornado here a while ago" issues are due to the fact that the NWS district which is responsible for my area is way east #ie downwind# of here in Wilmington OH and we are at the very beginning of their area of responsibility and they dont see enough info to predict one until it is too late and no one who does have the info cares about THIS area.
Old problem. The boy who cried wolf.
Think about car alarms... I don't think that I can even recall the last time that I have heard a car alarm go off and my first thought was "Maybe a car is being stolen!"
I was a student at the Rochester Institute of Technology in the 1970s. RIT also contained the National Technical Institute for the Deaf. The new, 12-story dormitory for the 900+ deaf students featured strobing lights instead of audio alarms. At night, the strobing lights were an impressive sight. Students who wanted the light show and/or hated the deaf students often pulled the alarms. Sometimes alarms were pulled more than four times in one night. The evacuation time went from 7 minutes to 45 minutes, and some deaf students hid in closets or under beds when the resident advisors used their master keys to check the rooms.
There was a real fire in the winter trimester, and some students failed to evacuate and were trapped by the smoke and flames. Fortunately, they were on the fourth floor and were evacuated via hook-and-ladder trucks. If the fire had been on the top three floors, students would have died because the ladders could only reach the ninth floor.
Basic human psychology, pretty much like "porn creep", which is a condition whereby a person no longer reacts to ordinary sexual stimuli as a result from the constant or excessive watching of pornographic material.
It's all a question of trade-offs. You may need to watch NOAA radar instead of The View and listen to the weather radio instead of your favorite drive time shock jock.
Excellent, funny quote. In my area, there's a test every Friday at 11:00am sharp. I guess by doing that, obviously there's some desensitization if there was an actual storm at that precise time. First off, unless you live in a bunker and can't look outside and see the dark clouds and put 2 & 2 together, I guess that's the least of your concerns. But, with weather forecasting these days, it's not like these storms just creep up on you out of nowhere, again if you can drag yourself away from "Dancing with the Stars".
Another funny thing about my area is that tornados and other extreme weather seem to divulge around like water hitting a big rock in a river. Odd..guess it's just random chance, or luck.
We still have those sirens where I live, but I've never understood why. In the '50s and '60s they were supposed to tell us we were about to get nuked, but the fallout shelters have all disappeared (or if any still exist, they might as well be gone since nobody knows about them).
Even without frequent false alarms, the tests they conduct at 11 am on the first of the month are by far the most common time to hear them (& I wonder if our next war opponent will take advantage of this by attacking at 11 am).
Since if the alarm were real, the first thing you or I really should do is turn on a TV or AM radio to hear instructions, maybe it would make more sense just to give the instructions by PA speaker after an initial horn blast. That way the large number of people who now ignore the alarms will still hear if there's an actual need for them.
Actually, in Poland the most common use of sirens is to signify mourning – during state holidays, for example. Technically we learn at school what are the sounds for air raids or chemical emergencies, but nowadays what you think when you hear a siren at a random time is "god, another presidential plane".
Our best 1st level warning is the cable TV provider running an NWS banner across the bottom of the screen listing every country supposedly affected. We then go to the NWS site forecast.weather.gov for details.
The best use of our local siren is the 9:45 PM 2 or 3 second Do-you-know-[where-your-children-are? / what--the-*%&-time-it-is?] bleep.
The NWS did their own study about the 2011 tornado oubreak down in the south (which I believe includes the Tuscaloosa tornado).
The forecasting worked very well those days, but not everyone listened. They had a lot to say about making warnings more informational and consistent, and better communicating to the public.
It's amazing that some people are still shocked by this. Aesop warned us about this over a millenium ago.
> The basic problem, Smith says, it that
> sirens are sounded too often in most places.
Much, MUCH too often, at least where I live. They sound it for tornado "watches" (i.e., when there is in fact no tornado). They sound it for thunderstorms. Once they sounded it for a thunderstorm warning (i.e., when there was not in fact a thunderstorm, but the weather people thought it was possible one might develop). Sometimes they sound the thing in obviously clear weather, when the sky is bright blue from horizon to horizon. The most egregious case of all was the time they sounded it because there was a tornado warning in Crawford County, Pennsylvania. (I live in Crawford County, Ohio, three hours' drive from the Pennsylvania state line.) Basically, as near as I can tell, they just like turning the thing on.
I've been ignoring it for years now. It goes off a dozen times a year or more. I'm not going to drop everything a dozen times a year on the off chance that some day it *might* actually be for real. Given the general level of confidence displayed, if we ever do have a tornado come through the area, the siren probably *won't* go off then.
> thunderstorm warning (i.e., when there was not in fact a thunderstorm...)
I meant "watch" there, not "warning".
The smoke alarm in my kitchen goes off incessantly. I imagine that the manufacturers of smoke alarms are encouraged--perhaps forced--to make them ridiculously oversensitive; opting towards endless false positives to avoid the risk of a single miss (and, no doubt, a lawsuit).
Every time the headache-inducing squeal goes off, I have to convince myself not to disconnect the thing permanently. (Worse, you can't cancel the alarm completely; if you silence the alarm, it keeps making loud, high-pitched squeaks at intervals for the next ten minutes. I assume this is designed to ruin my dinner.)
I wonder just how many fire deaths are caused by people disconnecting smoke alarms because they got fed up with false positives.
I did some digging on this a couple years ago and was surprised to learn the following:
1) There is no federal standard -- what the sirens mean is decided by the states (e.g.: in California, they may be sounded for wildfires).
2) In Ohio, the siren's meaning is that its citizenry should turn on television or radio and listen for information.
Once I realized the siren meant "turn on your radio" and not "run for your lives," I made my peace with it.
Schneier.com is a personal website. Opinions expressed are not necessarily those of BT.