Research notes on the X algorithm and tweet quality
Build-in-public posts on the engine, the math behind the score, and what we're learning from real engagement data. No listicles. No "10 secrets the algorithm doesn't want you to know."
- ·9 min read
Six tweets, six scores: a walkthrough of what actually moves the number
Real scores from the live engine on six tweet patterns — contrarian diagnosis, build-in-public receipt, tactical checklist, AI hype, engagement bait, and the em-dash trap. What each pattern signals to the algorithm, why scores cluster where they do, and what to change if you write one of these.
Read → - ·8 min read
Why your tweets don't break out: the velocity-gating window
X's For-You algorithm gates out-of-network distribution on early-engagement velocity. The first 30 minutes decide whether your tweet leaves your follower graph. Here's the math, the workaround, and why the headline metric most tools show you (likes after 24h) is the wrong number to optimize.
Read → - ·7 min read
Phoenix-style scoring vs likes-prediction: what we actually measure
Predicting like count is the wrong question. Twitter's algorithm weighs 18 distinct user actions with different coefficients — and four of those actions actively penalize your tweet. Here's what each one is worth, and what changes when you stop optimizing for likes.
Read → - ·5 min read
TweetSim v1 roadmap: what ships, what's deferred, what we won't build
Honest roadmap: paid tiers, Stripe billing, X OAuth, persona ingest, trained view-curve model. Plus the things we deliberately won't build (chatbot vibe-checks, faked social proof, magical 'algorithm hacks').
Read →