When Elon open sourced X's algorithm, he said X was "actively seeking criticism to improve fairness and quality of algorithmic recommendations."
Here is some.
Every reaction has a price:
Like +0.5
Reply +5
Block -31.2
Mute -58.8
Report -234
But the code also shows the abuse system is not applied equally. X's AI safety pipeline tags targeted insults with FOSNR_ABUSE_INSULTS, which hides the post from recommendations. Accounts labeled High PageRank are skipped. Their posts are not even scanned. They are automatically immune to X's own detection model.
I checked 7,992 of Eitan Fischberger's archived posts. 406 attack named people, "sick freak", "evil witch", "piece of human garbage". No sign any were hidden. They do BETTER than his normal posts.
Are accounts like Fischberger classified as High PageRank? Because the code is clear about everyone else: post these exact insults as a regular account, or reply to him in kind, and the model can scan you, label you, and limit your reach. He may never be scanned at all. One rule for some, another for everyone else. Why?
💬 34🔁 71
♥ 260👁 21K








































