Facebook’s news feed algorithm has been blamed for fanning sectarian hatred, steering users toward extremism and conspiracy theories, and incentivizing politicians to take more divisive stands. It’s in the spotlight thanks to waves of revelations from the Facebook Papers and testimony from whistleblower Frances Haugen, who argues it’s at the core of the company’s problems.
But how exactly does it work, and what makes it so influential?
While the phrase “the algorithm” has taken on sinister, even mythical overtones, it is, at its most basic level, a system that decides a post’s position on the news feed based on predictions about each user’s preferences and tendencies. The details of its design determine what sorts of content thrive on the world’s largest social network, and what types languish — which in turn shapes the posts we all create, and the ways we interact on its platform.
Facebook doesn’t release comprehensive data on the actual proportions of posts in any given user’s feed, or on Facebook as a whole. And each user’s feed is highly personalized to their behaviors. But a combination of internal Facebook documents, publicly available information and conversations with Facebook insiders offers a glimpse into how different approaches to the algorithm can dramatically alter the categories of content that tend to flourish.
The top post on a Facebook user’s news feed, shown as the biggest box, is a prized position based on thousands of data points related to the user and post itself, such as the poster, reactions and comments.
As users scroll farther down the feed, the smaller boxes here, the algorithm dictates each post’s position. The algorithm is precisely tailored to each user but also reflects Facebook’s strategy to favor certain content or behavior, illustrated in the following feeds.
Since 2018, the algorithm has elevated posts that encourage interaction, such as ones popular with friends. This broadly prioritizes posts by friends and family and viral memes, but also divisive content.
This was a departure from Facebook’s previous strategy in the mid-2010s, which optimized for time spent on the site and notably gave greater prominence to clickbait articles and professionally produced videos.
Each user’s feed reflects their expressed interests. For a subset of extremely partisan users, today’s algorithm can turn their feeds into echo chambers of divisive content and news, of varying reputability, that support their outlook.
Some critics argue a news feed that orders posts from newest to oldest is better for society. This wouldn’t prioritize divisive content, but could give greater space to more frequent low-engagement posters, such as that one distant friend with a new baby.
When Facebook launched the News Feed, in 2006, it was pretty simple. It showed a personalized list of activity updates from friends, like “Athalie updated her profile picture” and “James joined the San Francisco, CA network.” Most were automatically generated; there was no such thing as a “post,” just third-person status updates, like “Ezra is feeling fine.” Starting in 2009, a relatively straightforward ranking algorithm determined the order of stories for each user, making sure that the juicy stuff — like the news that a friend was “no longer in a relationship” — appeared near the top.
Over the past 12 years, almost everything about the news feed algorithm has changed. But the principle of putting the juicy stuff at the top — or at least the stuff most likely to interest a given user — has remained. The algorithm has simply grown ever more sophisticated to the point that today it can take in more than 10,000 different signals to make its predictions about a user’s likelihood of engaging with a single post, according to Jason Hirsch, the company’s head of integrity policy.
Yet the news feed ranking system is not a total mystery. Two crucial elements are entirely within the control of Facebook’s human employees, and depend on their ingenuity, their intuition and ultimately their value judgments. Facebook employees decide what data sources the software can draw on in making its predictions. And they decide what its goals should be — that is, what measurable outcomes to maximize for, and the relative importance of each.
Troves of internal documents have offered new insight into how Facebook makes those critical decisions, and how it thinks about and studies the trade-offs involved. The documents — disclosures made to the U.S. Securities and Exchange Commission and provided to Congress in redacted form by Haugen’s legal counsel — were obtained and reviewed by a consortium of news organizations, including The Washington Post. They have focused lawmakers’ attention on Facebook’s algorithm and whether it, and similar recommendation algorithms on other platforms, should be regulated.
Defending Facebook’s algorithm, the company’s global affairs chief, Nick Clegg, told ABC’s “This Week” earlier this month that it’s largely a force for good, and that removing algorithmic rankings would result in “more, not less” hate speech and misinformation in people’s feeds.
In its early years, Facebook’s algorithm prioritized signals such as likes, clicks and comments to decide which posts to amplify. Publishers, brands and individual users soon learned how to craft posts and headlines designed to induce likes and clicks, giving rise to what came to be known as “clickbait.” By 2013, upstart publishers such as Upworthy and ViralNova were amassing tens of millions of readers with articles designed specifically to game Facebook’s news feed algorithm.
Facebook realized that users were growing wary of misleading teaser headlines, and the company recalibrated its algorithm in 2014 and 2015 to downgrade clickbait and focus on new metrics, such as the amount of time a user spent reading a story or watching a video, and incorporating surveys on what content users found most valuable. Around the same time, its executives identified video as a business priority, and used the algorithm to boost “native” videos shared directly to Facebook. By the mid-2010s, the news feed had tilted toward slick, professionally produced content, especially videos that would hold people’s attention.
In 2016, however, Facebook executives grew worried about a decline in “original sharing.” Users were spending so much time passively watching and reading that they weren’t interacting with each other as much. Young people in particular shifted their personal conversations to rivals such as Snapchat that offered more intimacy.
Once again, Facebook found its answer in the algorithm: It developed a new set of goal metrics that it called “meaningful social interactions,” designed to show users more posts from friends and family, and fewer from big publishers and brands. In particular, the algorithm began to give outsize weight to posts that sparked lots of comments and replies.
The downside of this approach was that the posts that sparked the most comments tended to be the ones that made people angry or offended them, the documents show. Facebook became an angrier, more polarizing place. It didn’t help that, starting in 2017, the algorithm had assigned reaction emoji — including the angry emoji — five times the weight of a simple “like,” according to company documents.
“The goal of the Meaningful Social Interactions ranking change is in the name: improve people’s experience by prioritizing posts that inspire interactions, particularly conversations, between family and friends,” Facebook spokesman Adam Isserlis said. “We’re continuing to make changes consistent with this goal, like new tests to reduce political content on Facebook based on research and feedback.”
While the choices behind Facebook’s news feed algorithm can broadly elevate certain types of content, the same algorithm will produce different results for every user, because it is built to learn from their individual behaviors. If you rarely click on videos in your feed, you’ll be far less likely to see a viral video than your friend who loves videos. If you spend most of your time interacting with Facebook Groups, posts from those groups will figure especially prominently in your feed.
Internal documents show Facebook researchers found that, for the most politically oriented 1 million American users, nearly 90 percent of the content that Facebook shows them is about politics and social issues. Those groups also received the most misinformation, especially a set of users associated with mostly right-leaning content, who were shown one misinformation post out of every 40, according to a document from June 2020.
One takeaway is that Facebook’s algorithm isn’t a runaway train. The company may not directly control what any given user posts, but by choosing which types of posts will be seen, it sculpts the information landscape according to its business priorities. Some within the company would like to see Facebook use the algorithm to explicitly promote certain values, such as democracy and civil discourse. Others have suggested that it develop and prioritize new metrics that align with users’ values, as with a 2020 experiment in which the algorithm was trained to predict what posts they would find “good for the world” and “bad for the world,” and optimize for the former.
Still others, including Haugen, would like to see Facebook’s power over the algorithm taken away altogether: They argue we’d all be better off with social media feeds that simply showed us all of our friends’ posts in reverse-chronological order. But even that would come with trade-offs: The users and institutions that post most frequently, with the largest existing audiences, would dominate our feeds, while worthy ideas and clever videos from those with smaller followings would have less of a chance of reaching people who might be interested.
Kate Rabinowitz contributed to this report.