Across millions of Substack posts, the strongest predictor of a post’s likes is the average likes of the author’s previous 10 posts, explaining roughly 86% of variance. Posting more often beats writing longer, and a first‑post 'boost' plus pricing and category choices further tilt outcomes. This implies path dependence: once an audience is built, its inertia dominates performance.
— If platform metrics mostly reflect prior audience momentum, not per‑post merit, media economics and public debate are steered by reinforcement dynamics that entrench incumbents and muddy quality signals.
Uncorrelated
2025.04.05
100% relevant
The 'Post Likes Model' reports the moving average of the prior 10 posts’ likes explains ~86% of variance in a post’s likes.
← Back to All Ideas