By turning a series of lonely events into something like a story—by combining all your friends’ actions into a community, or even a conversation, on your home page—news feed gave Facebook a soul. Thanks to news feed, people started finding one another and working together in ways that had never happened on the site before. “Before news feed, you could join groups, but discovering them was not super easy,” Cox explained at a recent Facebook press event. “Within a period of two weeks after the news feed launch we had the first group with over a million people—which means a million people had seen the group and taken an action to join it.” There’s a punch line to this story that’s a testament to our enduring ambivalence about news feed. The first Facebook group that news feed propelled to million-plus membership was a protest against News Feed.
In the years since its launch, Facebook has constantly changed both the appearance and the mechanics behind news feed, and the current version is the product of one of the most complex computational processes you deal with on a regular basis. Facebook has always tried to present only a subset of the stuff people are posting—the posts it thinks you’re most likely to like. In its earliest incarnation, engineers tuned news feed manually, tweaking the frequency of certain kinds of posts—more pictures, fewer news stories—in response to user engagement. Later on, they developed a more formal algorithm, sometimes called EdgeRank, which took into account broad factors about people’s relationships in deciding what showed up on your feed. (For example, EdgeRank might determine that a photo from your mom is more important than a news story from your friend, partly based on information that Facebook’s engineers had hard-coded into the system.)
But EdgeRank wasn’t sophisticated enough. Ideally, everyone’s ranking algorithm should be personalized, and news feed should recognize those preferences and tweak our stories accordingly. “A few years ago, we stopped working on EdgeRank, and started working on a machine-learning approach,” says Serkan Piantino, the engineer who worked on some of news feed’s earliest ranking systems. Now, every time you load up news feed, the new system takes thousands of factors into account to present a feed that is personalized to your tastes. The machine-learned algorithm constantly tweaks itself based on how you interact with it: If you click like on a lot of memes, you’re going to see more of those. For every user, the system has to instantly analyze and rank an average of 1,500 posts every time the site is reloaded.
And engineers keep making the system more complex. Lars Backstrom, the engineering manager for news feed ranking, says that one of his team’s current goals is to get news feed to present information that hasn’t been explicitly shared by your friends. For instance, say you love Ricky Gervais, but none of your friends care for him. They aren’t likely to post anything about his new Netflix show—but given what news feed’s ranking system knows about your interests, it should determine that you might like Willa Paskin’s review of Derek more than you might like, say, another post about your mom’s friend’s visit to the Hamptons. “If something really interesting happens in the world, we should know enough about you to pull that in even if you haven’t explicitly connected to it,” Backstrom says.
He adds that news feed doesn’t really do this yet, “because we aren’t very good at it.” But the fact that Facebook is working on this problem illustrates the scale of its ambitions for news feed. Facebook doesn’t just want to be the front page for your social life; ultimately, it wants to be the one place online you check for everything you care about.
You can question whether this is good for society. One persistent worry about the news feed approach to information is the Filter Bubble critique—the idea that by engaging only with stuff that’s been algorithmically determined to appeal to us, we’re all tunneling into echo chambers of our own preferences. I explored that critique in my own book, though research into the question has since shown that the bubble is, thankfully, more porous than we might fear.
Another critique of news feed is that it has turned us all into narcissists, and worse, that it’s making us depressed about how much better everyone else’s life is. The trouble with that critique is that News Feed is only a reflection of your own interaction with it: If it’s serving up stuff that makes you sad, it’s only because that’s the stuff you’re most engaged with. If it’s true that news feed drives us crazy, we theoretically have the power to fix it. The news feed we have is the news feed we deserve.