The model already found titles that score higher. Try one.
I scraped/analyzed about 150k Hacker News submissions and trained a model to estimate how a new title + URL + text might perform. The tool returns predicted score, uncertainty, front-page probability, similar historical posts, and feature contributions. It is meant as a launch/title debugging tool, not a guarantee machine.
ForesynWanna keep in touch?
Built this solo over a weekend. Soft-launching before the HN post on Monday. If you scored a draft and the prediction either nailed it or whiffed, I want to know.