cognitive science
and more
Why Facebook's safety check ("Marked Safe") is dumb and dangerous

If you live in a region where terrorist attacks happen with some regularity, you are probably familiar with Facebook's Safety Check, or Marked Safe, feature. This works as follows: When a terrorist attack happens (or another disaster, but this post is purely about terrorist attacks), Facebook determines which of its users were nearby when it happened. Facebook does so based on three sources: your phone or tablet's location, your home city, and the location of your last connection to Facebook (i.e. they check the location of your last IP address). It then asks those nearby users to confirm that they are safe, which is posted on Facebook, thus re-assuring friends and family.

Surely there is nothing wrong with letting friends and family know that you are safe! So why is this—in the format of a Facebook-wide safety check—a bad idea?

First, more than anything else, systematically marking people as safe in this way accentuates the possibility that they may not have been safe, that they were in serious mortal danger and barely survived. It increases the subjective probability that loved ones may get hurt in terrorist attacks. And it increases the feeling that we are all in danger all of the time.

People who are affected by terrorist attacks deserve unconditional sympathy and support. But empathy doesn't have to, and shouldn't, turn into unreasonable fear.

Because fear is exactly what organizations like Islamic State (IS) want to accomplish: They want to look far more dangerous to the Western world than they really are, and the Western media, including Facebook, are doing an admirable job as their public-relations department. (On a related note, I wonder how many Western teenagers were inspired to join IS after the media kept feeding them Rambo-like images of Jihadi John and other IS supervillains. Imagine that you're a disturbed and angry teenager living the anti-American dream in La Castellane, in the poor northern quarters of Marseille. Wouldn't you want to be a famous bad-ass like them?) In reality, the risk of dying in a terrorist attack is tiny compared to many other causes of death in the Western world. For example, around 100,000 Europeans die in traffic accidents every year, of which more than 3,000 in France. So, even in terror-stricken France, traffic casualties alone outnumber terrorism casualties by roughly twenty to one. But we don't worry about dying in traffic—we worry about dying in terrorist attacks, because the media remind us of this possibility over and over again. And Facebook's Safety Check is one particularly insidious example of this.

Second, I wonder what exactly Facebook is trying to accomplish with their Safety Check, and whether they actually accomplish this. In part, they probably want to score good-guy points by showing how engaged they are with their audience, and they may succeed in this goal. But at the same time, Facebook is run by human beings, and they may genuinely be trying to help their users by informing them of their loved ones' safety. In this case their goal is admirable, but their solution is not.

To illustrate, say that one of your friends is not marked safe, what should you do? According to Facebook:

If you or someone you know is not safe, please contact your local emergency response team immediately.

Wait … what?! This statement is ambiguous. It could mean that you should contact the authorities if you know for a fact that someone is not safe, for example because you saw that person get injured. That would be reasonable advice, although somewhat out of place on the Facebook help site. But given the context, it suggests that you should contact the authorities if someone is not marked as safe on Facebook. That would be very bad advice, as you can easily demonstrate with a back-of-the-envelope calculation.

Take the Bataclan attacks in Paris of November last year. With 130 casualties, this was the deadliest-ever terrorist attack in France. How many people were nearby according to Facebook? This is difficult to say, but I think that about 10 million is a decent guesstimate. (Only about half of all people are on Facebook. But that's not relevant for this calculation.) So the probability that someone was hurt, given that she was nearby, is about 0.001%. Phrased differently, 99.999% of nearby people were fine. Now let's say that one in four nearby people, for whatever reason, don't bother to mark themselves as safe on Facebook. This means that 2.5 million people are not marked as safe because they didn't bother, versus 130 people who are not marked as safe because they actually died. Therefore, even if someone is not marked as safe, the probability of her actually being hurt is only around 0.005%.

Phrased differently: About 99.995% of people who are not marked as safe are actually fine. Of course, the numbers that I fed into this calculation are guesses. But I dare to say that any realistic set of numbers will lead to the same conclusion: The Safety Check is useless, because it doesn't indicate whether someone is actually safe or not.

But now imagine that it's two days ago (July 14), and you're looking at the Facebook profile of a friend who is visiting Nice—and she is not marked safe. How would you feel? Would you have the clarity of mind to do the calculation that I did above? Or would you be very worried, contact other equally worried friends, and perhaps even, as Facebook may or may not recommend in their ambiguous statement, contact the French authorities? Only to find out later that your friend's phone died because she forgot to bring her UK-to-European power adapter.

This is why I think that Facebook's Safety Check feature is dumb and dangerous. It is dumb, because it is a poor check. If anything, it checks how likely someone is to grab their phone first thing after a terrorist attack. But as a safety check after terrorist attacks, it is useless because almost everyone is safe anyway and marking yourself as safe (or not) adds little information. And it is dangerous because it creates fear by boosting the subjective feeling that we are all in danger. And fear is bad—it causes people to do dangerous things, and to vote for dangerous politicians (Trump, Front National, FPÖ, PVV) who will happily do even more dangerous things in our name.