@ILA_NewsX
The recent monetization change has destroyed every small and midsize creator on this platform. Save citizen journalism. #FireNikitaBier https://t.co/joylJH8cQe
X open-sourced its Grok 'For You' feed algorithm. Modular and transparent, it predicts multiple user actions from behavior and will be updated on GitHub monthly.
Real-time analysis of public opinion and engagement
Community concerns and opposing viewpoints
Users distrust the Grok/transformer approach and resent automated curation — many complain “I don’t want my news from an autobot,” call Grok “puke” or “ass,” and say predictive ranking floods their feed with repetitive content shortly after a single engagement.
The new monetization changes are blamed for crippling small and midsize creators — multiple posts say earnings and impressions plunged, making it “impossible” to earn or grow.
” Low impressions and poor relevance are recurring complaints.
open-sourcing the model is accused of turning content into score-chasing formats that force creators to adapt or vanish.
Calls for concrete action appear repeatedly — users ask to “roll it back,” demand better support, and some plead to “Save citizen journalism” (with targeted criticism like “#FireNikitaBier”).
several say the feed over-promotes certain political content or personalities and misrepresents their interests.
A mixture of resignation and anger runs through replies — some users are ready to quit or stop using For You, while others beg for improvements and clearer explanations from the platform.
The recent monetization change has destroyed every small and midsize creator on this platform. Save citizen journalism. #FireNikitaBier https://t.co/joylJH8cQe
Cool but everyone knows to stay off the For You tab already. :-/
I don’t want something deciding what it thinks I will like. The only one who knows that is me,let me decide for myself.
Community members who agree with this perspective
Many replies call the move “huge,” “legendary,” and a trust-builder, praising X/Elon for letting the public inspect the production feed instead of hiding it.
Commenters highlight the model’s shift to predicting likes, replies, clicks, watches and more at once, and note the architecture is modular and cacheable.
Threads emphasize that knowledge removes guesswork—people ask how to increase impressions, expect to tweak content to favor signals the model rewards, and claim small creators could benefit.
dwell time (watch/stop), bookmarks, video watch time, and clicks. Several replies also mention author-frequency penalties, limited external links, and premium-user weighting as things that materially affect reach.
users repeatedly ask “where’s the GitHub link?”, want plain-English or 10‑year‑old explanations, and seek simple guides and SDK support to experiment with the repo.
Concerns about feedback loops and reduced diversity are raised; some worry high-ranked content can self-reinforce and narrow what people see, while others point out heavy negative-signal weights (not interested, block, report) that can quickly hurt reach.
a few replies call the algorithm “dumb” or uncertain it will work perfectly, yet most accept that showing the code is preferable to secrecy and will invite faster fixes and scrutiny.
With mentions of four‑week updates, people anticipate rapid evolution, community audits, new tools/SDKs, and experiments that could reshape how creators and platforms interact.
Transparency always breeds confidence. This is a solid step forward.
I need it
Bullish.