Hacker News Reader

The data don't lie on a line, but at least they lie on a plane.

Thanks! I agree. The R-squared of these numbers is 0.11, which is much closer to 0 (no relation) than to 1 (perfect correlation).

That’s a very misleading framing though. An R-Squared if 0.11 would be pretty damn important if rigorously demonstrated. This is not rigorous though.

Optical regression score: "All over the place worse than my darts game after a few pints of gin"

Though, often a rank vs. rank scatter plot is more enlightening than a scatter plot of the raw data.

pints of gin?!

That's part of the joke, stolen from Critical Drinker's YouTube commentary on film narrative.