A general rule for critical analysis of any research that purports to model human thinking: generate a random dataset. If the effect is stronger in the random dataset than in the research, the research is bunk.

The belief in D-K persists because expert humans - think Neil Degrasse Tyson - fail spectacularly when just outside their area of expertise, because they can’t see the infelicities in their models (or hubris). That’s not a general effect on all humanity.

Nope, worse than that. The refined analysis is that asked to score themselves on a *percentile* basis against an unknown cohort will substantially outperform against random weighting. Which is strong evidence that, even when the system is rigged, humans are really good at general self-assessment.

It's this year's Kindle candle table bundle

Haven’t listened to the podcast, but you do know that the research that D-K relied on is sloppy, and, if anything, disproves the D-K effect? The study itself is the best argument for the effect … because many social science researches are incompetent at basic data analysis.