The breadth of predictability is why it’s such an effective measure. Most tasks involve many different skills, so it’s helpful to have a single measure that’s correlated with a bunch of different competencies. That’s why we use what are essentially IQ tests in everything from assigning jobs in the military (ASVAB) to selecting lawyers (LSAT). There’s tremendous social value in a single test that can scaleably sort through millions of people even if it’s not the most predicative test for a specific problem domain or a specific individual.

Also, IQ predicts chess performance as well: https://www.sciencedaily.com/releases/2016/09/160913124722.h...

> Also, IQ predicts chess performance as well: https://www.sciencedaily.com/releases/2016/09/160913124722.h...

"The study found that intelligence was linked to chess skill for the overall sample, but particularly among young chess players and those at lower levels of skill."

We might have different definitions of the word well.

Edit: Upon reading the linked study in the article, it's even worse than I thought:

https://www.sciencedirect.com/science/article/abs/pii/S01602...

"Effect sizes were small-to-medium in magnitude; variance in chess skill explained by cognitive ability was similar in magnitude for Gf (6%), Gsm (6%), Gs (6%), and Gc (5%), with an average of 6%. Full-scale IQ explained < 1% of the variance in chess skill."