Hacker News new | past | comments | ask | show | jobs | submit login

It's true that there's a correlation / causation problem here, but consider the problem from Google's POV: they rank a certain set of results by hand, then feed those into various learning algorithms to find good predictors for quality.

If an analysis such as this finds such correlations, it's likely that Google's algorithms have also picked up on them. In other words: maybe people diligent about meta tags and H1s also write better content and should therefore rank higher (correlation). But "good writing" is hard to judge algorithmically, and Google may therefore well be turning some correlations into causations.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: