Tuesday, April 24, 2018

I am ok

Yesterday a van was driven into crowds of people on a street here in Toronto.  Ten people died, many more were injured.  The first I heard of this was Facebook prompting me to tell everyone that I am okay.  It was strange to be asked to confirm that I am not in fact dead or injured when I have no idea what it is exactly that might have injured me, but I ended up clicking to tell Facebook and my friends there that I am fine.

I won't be doing that again.

It made me feel weird and uncomfortable.  Facebook 'helpfully' listed all of the people I knew who live anywhere near this area who had clicked that they were safe, but of course I know more than one hundred more who had not clicked.  I don't know why they didn't click, but presumably most of them simply didn't log onto Facebook.  This ambiguity didn't bother me at all because even if they did click it doesn't tell me they are safe *now* - in the time since they clicked they could have died from something else entirely.  In fact if the only thing I knew was that ten people had died it is still more probably that someone I knew died from something else entirely than that they were hit by the van.  Toronto is a big place.

This system is basically just exploiting people's inability to intuitively grasp extremely low probabilities to inveigle them into staring at Facebook, worried about an unlikely but anomalous event, instead of worrying about that which is actually threatening.

Facebook is just trying to leverage tragedy for personal gain.  They pitch it as trying to help people cope with disaster but their system is mostly worthless.  It is clearly designed to get people to log in to Facebook every time they find out about a disaster both to click 'safe' and to obsessively check the list of other people to see who else has clicked it.

We know that desperately checking social media to see if people have clicked on a thing does not make people happier.  All it does is feed our collective addiction.  I don't want my friends wasting time trying to find out if a problem that is far removed from me randomly hurt or killed me by extreme random chance - you can't do anything about it anyway, so just move on with your life.

Tragedies happen.  It sucks.  But nothing will be improved by sitting staring at Facebook refreshing the page.  Go out and do things that bring you joy, and accept that disaster sometimes strikes but sitting in terror of unlikely events will bring nothing but sadness.

5 comments:

  1. I feel like you've entirely missed the point.

    I'm at a hockey game with 22,000 people and there's an incident. Everyone in the world knows about it within an hour. Dozens of people want to know if I'm okay. Do I mass email everyone I know? Do I respond individually to each person? Text? Call? Send a letter?

    The best way to communicate the information is updating a central location where everyone knows to go look. That's the intent of the check-in software. It's the same as the manual process of a centralized bulletin board when you're searching for survivors.

    Is it a bit ridiculous to use it for an incident where 25 people were injured? Absolutely. Unless, of course, you're one of the hundreds of people who work there and might have been out at lunch. Then maybe there is value.

    I think it's a lot more useful for large events. But I'm sure you already know this because you researched its history before ranting, right? :-)

    https://en.wikipedia.org/wiki/Facebook_Safety_Check

    ReplyDelete
    Replies
    1. A database for this sort of thing is potentially useful, if the data is actually good. This data is horrible, absolutely chock full of false negatives. In fact the *majority* of your data is false negatives. That is worthless as far as I am concerned. You didn't call everyone you know after an incident before facebook came around, and you shouldn't change your strategy now. Ignore it until it matters to you.

      Delete
    2. Did you have a lot of friends and family in Nepal during the earthquake? Chennai during the floods? I suspect the data was better for those events in those areas of the world.

      I agree, it's not helpful for small scale events.

      Delete
    3. But still you have the problem of massive false negatives. And even then all it means is that you know, not that you can do anything useful. Data that is massively unreliable and which you can't act on (but which causes other undesirable behaviour) is worse than no data at all.

      Delete
  2. Your estimate of the likelihood of a person dying since they checked in vs. the likelihood of them having died in the event is off. Applying Ontario's mortality rate to Toronto I get 34.4 deaths per day. But, of course, mortality rates are greatly skewed by age. If you are checking on someone around your age they are considerably less likely to die in a few hours after checking in than they are to have died in the attack (assuming the attack was randomly distributed, which it wasn't, obviously the attack targeted women and targeted people who would be at Finch in the afternoon).

    ReplyDelete