Facebook’s formula prioritized anger and ended up spreading misinformation

The Hill 3 min read

10 Aug '26, 13:56current

Summary (TL;DR)

Facebook's algorithm ranked angry emoji reactions five times more valuable than likes, prioritizing emotional content that increased engagement but also spread misinformation and low-quality news. Internal documents from Frances Haugen revealed that data scientists warned about this in 2019. Facebook gradually reduced the angry emoji's value, cutting it entirely by September 2021, which reduced misinformation and violent content exposure.

Story at a glance

Internal documents reveal Facebook’s algorithm prioritized angry reactions, which were disproportionately likely to push out misinformation to users.
Data scientists raised red flags internally at the company to warn of the dangers associated with the angry emoji reaction.
Facebook has now demoted the angry emoji reaction within its algorithm.

New details about the inner workings of Facebook’s algorithm reveal its reaction emojis were ranked five times more valuable than likes. Emoji reactions created more engagement, which is key to the social network’s business model.

Facebook introduced new reactions to its platform five years ago, with “love,” “haha,” “wow,” “sad,” and “angry,” alongside its “like” button.

These reaction buttons informed how the company’s algorithm pushed content to its users, as revealed by documents turned over by Frances Haugen, former Facebook product manager, to the Securities and Exchange Commission and to Congress.

The more reaction emojis were used, the more Facebook’s algorithm would prioritize emotional and provocative content. This opened up the opportunity for controversial posts to be highly valued, which Facebook employees recognized. In internal documents reviewed by The Washington Post, one employee said that favoring these types of posts could open, “the door to more spam/abuse/clickbait inadvertently.”

America is changing faster than ever! Add Changing America to your Facebook or Twitter feed to stay on top of the news.

In 2019, those fears were confirmed when data scientists at Facebook said that posts that received angry emojis were disproportionately likely to host misinformation, toxicity and low-quality news.

The strategy behind how Facebook viewed and used its angry emoji reaction, now confirmed through Haugen’s release of internal company documents, means that for three years the company’s algorithm was in fact prioritizing low-quality content. According to The Post, Haugen told the British Parliament on Monday, “anger and hate is the easiest way to grow on Facebook.”

CEO Mark Zuckerberg himself took to encouraging Facebook users to utilize the angry emoji button, replying to a user’s comment that wanted to use a dislike feature.

Eventually, the company adapted its algorithm to move away from prioritizing the angry emoji in 2018, downgrading the value to four times that of a like reaction. Then again in 2019, Facebook created a mechanism to “demote” content that was disproportionately receiving angry emoji reactions.

Finally, in 2020 the company cut the value of all of its emoji reactions to one and half times that of a like, after research found that users didn’t like when their posts received angry reactions. By September, Facebook cut the value of angry reactions to zero, completely eliminating it from its equation, but not yet removing it from its algorithm.

According to The Post, after changes were made to zero out the angry reactions, users received less misinformation and violent content.

READ MORE STORIES FROM CHANGING AMERICA

‘PARKS AND RECREATION’ ACTOR CHARGED WITH VANDALIZING GEORGE FLOYD STATUE

CRIMINAL CHARGES NOT OUT OF THE QUESTION IN FATAL SHOOTING ON MOVIE SET, DA SAYS

DAVE CHAPPELLE RELEASES NEW STANDUP VIDEO REFUSING TO APOLOGIZE FOR TRANS CONTROVERSY

NEW EGYPTIAN MUMMY DISCOVERY COULD REWRITE HISTORY

BIPARTISAN LEGISLATORS DEMAND ANSWERS FROM FAUCI ON ‘CRUEL’ PUPPY EXPERIMENTS