X's open-source algorithm adds labels that strip scores from engagement-bait posts
ThePeterMick · x · 2026-11-22
X added three new labels to its open-source algorithm on Nov 19 — "promotes replies", "promotes shares" and "promotes clicks" — targeting posts that beg for engagement. Reading the code shows each label makes one score component ineligible: "promotes shares" posts lose share/DM/copy-link points (copy link alone is worth 20), while "promotes replies" posts lose reply points including the mutuals bonus (20 between mutuals). The feature is switched off for now, but signals where the platform is heading: earn replies, don't ask for them.
More from Apps
- Signal restricts mobile screenshots as it confronts 'existential threat' of AI surveillance — TechCrunch AI · 2026-11-22
- Solo consultant's 11-tool AI stack: Claude Code builds, Codex reviews, bots run marketing — jacob_posel · 2026-11-22
- Litmus: an open-source local AI-text detector that dares to say "Uncertain" — l_hallee · 2026-11-22
- Use AI to turn voice-dictated conference notes into personalized follow-up emails — annetgriffin · 2026-11-22
- New tool lets you visualize and search your X mutuals — nutlope · 2026-11-22
- Design canvas tool mainframe rolls out more invites and launches two new features — vinvan · 2026-11-22