Written by
Facebook released the Like button in 2009 and it changed the way people shared content.
The idea wasn’t new—once-popular, now marginal, sites like digg.com and del.icio.us had been letting people “like” articles for years before that. But at these companies, the content was the star. Facebook laid curation over an already robust social network and, for the content creators, made it simple for anyone to attach that iconic little thumbs-up to their work.
They created a new universal microcurrency—I might not pay you for your writing, music, or whatever, but I’ll give you a fillip of approval and share what you’ve done with my friends.
Facebook said in May 2013 that it was recording 4.5 billion likes a day and in September of that year reported that 1.13 trillion had been submitted all-time. Researchers started using that macro-level data to make predictions; students from MIT developed a gaydar algorithm that was pretty good at guessing a man’s sexuality.
Since then, the power of predictive software has advanced rapidly; these types of programs only get smarter and faster as more data becomes available. A group from the UK discovered that from a person’s likes alone they could figure out the following, with these degrees of accuracy: Whether someone is . . .
- Caucasian or African American: 95%
- A man or a woman: 93%
- Gay or straight: 88%
- Democrat or Republican: 85%
- Lesbian or straight: 75%
- A drug user: 65%
- The child of parents who got divorced before he or she turned: 60%
Again, this is not from looking at status updates or comments or shares or anything that the users typed. Just their likes.
You know the science is headed to undiscovered places when someone can hear your parents fighting in the click-click-click of a mouse. A person’s “like” pattern even makes a decent proxy for intelligence—this model could reliably predict someone’s score on a standard (separately administered) IQ test, without the person answering a single direct question.
This stuff was computed from three years of data collected from people who joined Facebook after decades of being on Earth without it. What will be possible when someone’s been using these services since she was a child? That’s the darker side of the longitudinal data that I’m otherwise so excited about.
Tests like Myers-Briggs and Stanford-Binet have long been used by employers, schools, the military. You sit down, do your best, and they sort you. For the most part, you’ve opted in. But it’s increasingly the case that you’re taking these tests just by living your life. And the results are there for anyone to read and judge. It’s one thing to see that someone’s Klout score is 51 or whatever in advance of a job interview. It’s another to know his IQ.
If employers begin to use algorithms to infer how intelligent you are or whether you use drugs, then your only choice will be to game the system—or, to borrow wording from the corporate world, “manage your brand.” To beat the machine, you must act like a machine, which means you’ve lost to the machine. And that’s all assuming you can guess at what you’re supposed to do in the first place.
Apparently, one of the strongest correlates to intelligence in the research was liking “curly fries.” Who could reverse-engineer that? But while Facebook does know a lot about you, it’s more like a “work friend”— for all the time you spend together, there are clear limits to your relationship.
Facebook only knows what you do on Facebook. There are many places with much deeper reach. If you have an iPhone, Apple could have your address book, your calendar, your photos, your texts, all the music you listen to, all the places you go—and even how many steps it took to get there, since phones have a little gyroscope in them. Don’t have an iPhone? Then replace “Apple” with Google or Samsung or Verizon. Wear a FuelBand? Nike knows how well you sleep. An Xbox? Microsoft knows your heart rate. A credit card? Buy something at a retailer, and your PII (personally identifiable information) attaches the UPC to your Guest ID in the CRM (customer relations management) software, which then starts working on what you’ll want next.
This is just a sliver of the corporate data state, the full description of which could take pages. For the government picture, a sliver is all I have, because that’s all we’ve been able to see of it. We do know that the UK has 5.9 million security cameras, one for every eleven citizens. In Manhattan, just below Fourteenth Street, there are 4,176. Satellites and drones complete the picture beyond the asphalt.
Though there’s no telling what each one sees, it’s safe to say: if the government is interested in your whereabouts, one sees you. And besides, as Edward Snowden revealed, much of what they can’t put a lens on they can monitor at leisure from the screen of an NSANet terminal, location undisclosed. Because so much happens with so little public notice, the lay understanding of data is inevitably many steps behind the reality.
I have to say, just pausing to write my book, I’m sure I’ve lost ground. Analytics has in many ways surpassed the information itself as the real lever to pry. Cookies in your web browser and guys hacking for credit card numbers get most of the press and are certainly the most acutely annoying of the data collectors. But they’ve also taken hold of a small fraction of your life, and for that small piece they had to put in all kinds of work.
No matter how crafty the JavaScript, they’re villains in the silent-film vein, all mustachios and top hats. Or, a more contemporary reference: they’re like so many pasty Dr. Evils—underworld relics holding the world hostage for one . . .million . . . dollars . . . while the billions fly by to the real masterminds, like Acxiom.
These corporate data marketers, with reach into bank and credit card records, retail histories, and government filings like tax records, know stuff about human behavior that no academic researcher, fishing for patterns on some website, ever could.
Meanwhile, the resources and expertise the national security apparatus brings to bear makes enterprise-level datamining software look like Minesweeper. This data, despite the “mining” metaphor, isn’t a naturally occurring resource; it comes from somewhere—and that somewhere is you. The companies and the government are collecting disparate pieces of your private life and trying to fashion them back into an image they can master.
The more privacy you lose, the more effective they are. The fundamental question in any discussion of privacy is the trade-off—what you get for losing it. We make calculated trades all the time. Public figures sell their personal lives to advance their careers. Anyone who’s booked a hostel in Europe or bought a train ticket in India has had to decide if the private room is worth the extra money.
And not to confuse the issue here, but many people, men and women, trade on privacy when they walk out the door in the evening, giving it away, via a hemline or a snug fit, for attention.
So the exchange isn’t new. But our trading partners, and their terms, are.
