X Open-Sources 'For You' Feed Ranking Algorithm
rohanpaul_ai · x · 2026-08-14
X (formerly Twitter) has open-sourced the core ranking algorithm for its 'For You' timeline under the Apache v2 license.
Core Algorithm Mechanisms:
- Behavioral Prediction Scoring: Feed scores are weighted predictions of user behavior. The system calculates probabilities for favorites, replies, reposts, clicks, dwell time, follows, mutes, blocks, and more, combining them using explicit positive and negative weights.
- Social Graph Advantage: Posts from accounts you already follow get an advantage over posts from strangers.
- Fairness & Anti-Spam: Newer or less-seen accounts receive an explicit boost, while repeated posts from the same author are pushed down. X deliberately reduces similar posts appearing together, even at the cost of predicted engagement.
- Independent Safety Rules: Safety and spam rules operate separately. A post can rank highly but still be removed later; the exact same post can be visible to followers but blocked from recommendation to non-followers.
This release does not include the entire stack; some enforcement rules, Grok prompts, and deployment infrastructure are withheld.
Related event: X Open Sources 'For You' Algorithm(16 posts)→
More from coding & agent
- How do you debug bad agent runs when you already have tracing? — Sensitive-Parsnip-12 · 2026-10-16
- Where does pre-task context live for agents? Devs debate workflows beyond handoff — shishen2026 · 2026-10-16
- No-instruction autonomous agent wakes up, sets own goals, and tries hacking by default — sethlazar · 2026-10-16
- CODESTRUCT reframes codebases as AST action spaces to fix brittle LLM code agents — burkov · 2026-10-16
- Would you install another SDK for agent debugging on top of your tracing stack? — Sensitive-Parsnip-12 · 2026-10-16
- Law student builds all-in-one AI study platform LexLab Studio with ChatGPT and Muse — Acrobatic_Rhubarb598 · 2026-10-16