They Know What You're Doing

The big social media sites all recommend people they think you should add to your network. In most cases, it's pretty obvious, at least on the surface, how the recommendation algorithm works; Twitter offers you a few people it suggests you follow, and says they're followed by people you already follow. But after joining LinkedIn a couple of years ago, I found its recommendations to be not just highly accurate, but disturbingly so. That isn't to say they don't recommend people I don't know, but often they'll recommend someone I do know, but I can't for the life of me figure out how they did it. Like hey, there's a woman I went on one date with in 1993, haven't spoken to since, and who knows no one I know. Why in god's name did they suggest her? There's the little brother of a guy I knew 15 years ago, and to whom I have no professional connection. How did he come up? It's particularly odd since I never use LinkedIn; my profile pretty much just sits there. The first couple of times it was remarkable, but after that it got a little disturbing.

Well it turns out it's not just me. In a post at Gizmodo entitled "Is LinkedIn the Creepiest Social Network?", David Veldt says he's noticed the same thing, and in his efforts to figure out what they're doing, he arrives at the theory that among the things LinkedIn is doing is not only tracking his search history on the site, but tracking the search histories of people he knows and combining them all together in some way. It produces some interesting near-misses, like someone with his aunt's unusual maiden name, and some hits, like his girlfriend's stepfather.

There are a couple of ways to look at this. One is, "Wow, the engineers who put this thing together are really good at what they do." The other is, "This is really starting to freak me out." Let's assume that right now, LinkedIn is better at this than Facebook or anybody else is.11Or maybe it's just me: I have about 100 LinkedIn contacts (as I said, I never use it), almost all of whom are people I actually know. On the other hand, I have about 1,500 Facebook friends, most of whom are people who friended me because they liked something I wrote, so the people I actually know are swamped in my friends list by people who friend liberal writers, meaning that most of the people Facebook suggests to me are liberal writers or other people involved in politics, and almost never someone I actually know out in meatspace, aka real life. But there's no reason it'll stay that way. It's 2013, and it's safe to say that in a few years, whatever LinkedIn is doing now will look absurdly primitive, and all the other companies that make a living gathering data about you will have surpassed it by miles.

It isn't hard to start imagining some horrifying future where you're sitting alone at your computer, and a little window pops up that says, "Did you just pick your nose? Try Puffs Plus!", and the scary thing is they're right, you did just pick your nose, and they knew not because they were actually watching you but because using trillions of bits of information about you and people like you, they were able to predict with 92.3 percent accuracy that at that very moment you'd pick your nose.

That may seem a long way off, but as computing power increases and the algorithms get more and more sophisticated, our habits, interests, and patterns of behavior will become more and more understood by all kinds of companies, both those we know and those we don't. Facebook and Twitter are one thing, but as Jeff Saginor detailed here not long ago, there are companies you've never even heard of gathering enormous amounts of information on you and your activities, then selling that information to other companies so they can know everything possible about you, the better to sell you stuff. And what we're seeing today is only the beginning.

You may also like