Meta’s new Muse Wonder models promise “Active Memory” across conversations.
Developers: what’s ONE instruction you’re tired of repeating to your coding assistant? 👀
Looking to connect with more developers, SaaS founders, and people building with AI 🤝
I build mobile apps and would love to meet others to swap ideas, share experiences, and learn from.
If we haven’t spoken yet, say hi 👋
X’s algorithm is open source. Here’s what its ranking weights reveal about growing here 👇
POSITIVE SIGNALS
Like: +0.66
Repost: +1
Reply: +5
Quote: +5
Share via DM: +5
Share by copying the post link: +20
Follow the author: +4
Click an external link: +0.2
Time spent reading also contributes through separate attention signals.
NEGATIVE SIGNALS
“Not interested”: −52.272
Mute the author: −58.8
Block the author: −31.2
Report: −234
HOW THE SCORING WORKS
These aren’t points earned per engagement.
X predicts how likely a particular viewer is to take each action, then multiplies that probability by its weight.
Example: a 10% predicted chance of replying contributes 5 × 0.10 = 0.5 before other adjustments.
The same post can score differently for different viewers.
MY GROWTH TAKEAWAYS
① Ask specific questions people have something to say about.
② Share useful examples, resources and lessons people would send to a friend.
③ Deliver on your opening. Give readers a reason to spend time with the post.
④ Keep promotion relevant. Predicted negative feedback can hurt ranking.
⑤ Build around a recognizable audience. Recommendations use each viewer’s engagement history.
These are practical interpretations of the public code. Defaults and live experiments can change.
Which is hardest for you: getting replies, getting shares, or keeping people reading?