Three former Facebook contractors say Trending news algorithm was not ready to run autonomously, user testing was overwhelmingly negative
How Facebook's foray into automated news went from messy to disastrous. — It seems Facebook's human news editors weren't quite as expendable as the company thought.
Context & Ripple Effects
This closes a loop that opened in May, when former Trending curators alleged in a Gizmodo account that the team suppressed conservative outlets and leaned on personal news judgment — reporting later echoed by workers describing the group as ill-managed in New York Times coverage. Under that political pressure, Facebook cut the humans: sources reported in late August that it was letting go of its entire 15-18 person contracted editorial staff (Quartz).
Now three of those same contractors say the algorithm that replaced them failed user testing overwhelmingly before launch and was never ready to run autonomously. The significance is direct: the bias controversy pushed Facebook to automate an editorial function its own pre-launch data said wasn't ready.
First-order effects
- Facebook's Trending module runs without any editorial check on a system its own testers rated negatively, meaning the product live now is one the company had evidence against shipping.
- The 15-18 laid-off contractors become the public witnesses to the decision, with their testimony directly contradicting the framing that editors were expendable.
Second-order effects
- Publishers and outlets that benefited from curated trending placement lose a human gatekeeper who understood context, shifting distribution further toward whatever the algorithm surfaces.
- The move hands ammunition to critics of platform editorial discretion: having been accused of human bias, Facebook now owns the failures of machine curation instead.
Third-order effects
- If the pattern holds, platforms facing accusations of editorial bias will reach for automation as a neutrality fix even when internal testing says otherwise — trading visible human judgment errors for less accountable systemic ones.
- It foreshadows the structural question that recurs across Facebook's history: whether news distribution on social platforms can be governed by engineering alone or requires an editorial function the companies keep dismantling.
The trend: Social platforms are replacing human editorial judgment with algorithms as a response to bias controversies, often ahead of what the technology can reliably deliver.