X's Open-Source Algorithm Adds Labels to Penalize Reply, Share and Click Baiting
ThePeterMick · x · 2026-11-19
Tracking X's open-source algorithm, @ThePeterMick reports the Nov 19 release added three new labels — "promotes replies", "promotes shares", "promotes clicks" — for posts that fish for engagement.
- Each label marks its own slice of the post's score as ineligible: a reply-bait post would earn no reply points (replies are worth 5 points, 20 between mutuals).
- The feature is switched off for now; no detector exists in the code yet, and it's unclear whether it replaces or stacks on the existing engagement-bait rule that hides posts from non-followers.
- Takeaway: write posts people genuinely want to reply to, and don't ask for replies.
More from Companies & People
- Researcher Tells Max Tegmark: Boycott Anthropic Entirely or Stop Signing Open Letters — AlexTensor · 2026-11-19
- Netflix's first fully in-house animated film Steps mocked as AI slop after 5 years — aakashgupta · 2026-11-19
- Jensen Huang resigns after 33 years: Nvidia built to survive anything, including him — r0ck3t23 · 2026-11-19
- Jeremy Howard's savage take on lab motives: 'That's why I want Google to win' — jeremyphoward · 2026-11-19
- Nobel laureate Susumu Kitagawa joins CuspAI advisory board, its second alongside Hinton — jonfildes · 2026-11-19
- Sarvam criticized as not founder-led: 'can't match Sam/Dario' as sovereign AI bet — cneuralnetwork · 2026-11-19