l5-fooled-by-randomness_inverse-skills-problem

The inverse skills problem is Taleb's observation that evidence of skill decreases as organizational rank increases. Lower-level workers generate large decision samples, while CEOs make only a few consequential choices, making their results statistically indistinguishable from luck.

CEOs make perhaps five to ten truly consequential decisions per year, a sample too small for statistical differentiation. With such small samples, variance swamps any signal of skill, so outcomes reflect randomness more than ability.

Compensation is highest exactly where the randomness-to-skill ratio is highest. Corporate reward systems pay for outcomes rather than process, so executives presiding over bull markets are rewarded as if they caused them.

Outcome-based evaluation fails when decision sample sizes are small and variance is large, such as executive roles. In these cases, process quality and decision reasoning are more reliable indicators than realized results.