SSC Journal Club: Expert Prediction Of Experiments

SSC Journal Club on DellaVigna & Pope, who tested not which incentive scheme works but WHO can best predict behavioral-econ results. Findings: knowledgeable academics beat randos only slightly; prestige, citations, and field don't matter; the small expert edge vanishes under wisdom-of-crowds (5 averaged undergrads beat 1 expert) and is screened off by a simple common-sense pre-test (i.e. 'superforecaster'-type individuals match experts). The killer: top-business-school MBAs -- the highest-IQ sample -- do WORSE than undergrads, which is fatal to a pure-IQ explanation. Scott infers a separate 'rationality' skill (predicted by, but not identical to, IQ/training), and carefully distinguishes punditry (predicting outside a tested theory) from genuine expertise (engineers, doctors), warning against over-generalizing the result to discredit real experts.

Quality 70 · Strong. Strong, upper (70). A clear, insightful study writeup with a genuinely sharp observation (the MBA paradox killing the IQ story), the rationality-skill conclusion (echoing Superforecasting), and a careful punditry-vs-expertise caveat that resists the lazy anti-expert takeaway. Capped as a single-study analysis.

Claude’s paradigm shift 48 · Moderate. Moderate (48). Analyzing others' study; the fresh bits are the MBA-paradox reading and the punditry/expertise distinction.

Real-world impact 2 · Minor. 2 -- minor/within-blog. A forecasting/expertise journal-club post; within-discourse.