Facebook 'Meaningful Social Interactions' algorithm

Occurred: October 2021

Can you improve this page?
Share your insights with us

Leaked documents reveal Facebook deliberately pushed users into seeing more emotional, provocative, and negative content in order to deepen engagement and drive more revenue.

Between 2017 and 2020, an update to Facebook's 'Meaningful Social Interactions' (MSI) algorithm saw the weighting of angry emoji substantially increased.

The update was made despite Facebook researchers having warned senior leadership at the company, including Mark Zuckerberg, that it would likely lead to higher levels of spam, misinformation and disinformation, and abuse.

The update also had the effect of undermining efforts by the company's content moderators and integrity teams to manage toxic and harmful content.

The leaked documents were shared with the SEC and the US Congress by whistleblower Frances Haugen. Speaking in the UK parliament, Haugen argued 'anger and hate is the easiest way to grow on Facebook.'

Operator: Meta/Facebook
Developer: Meta/Facebook
Country: USA; Global
Sector: Technology
Purpose: Increase engagement, revenue
Technology: Content ranking system
Issue: Ethics; Hate speech/violence; Mis/disinformation
Transparency: Governance; Black box; Marketing

Page info
Type: Incident
Published: October 2021
Last updated: January 2022