You might learn more from the first person you looked at from a random employee at a random company. The second person, you might not. The 100th, you wouldn't.

You might learn more from the first person you looked at from a random employee at a random company. The second person, you might not. The 100th, you wouldn't. There is only so much you can learn from failure, when it is the rare, immense success you care about. It might be rare, but in terms of outcome averaged sampling, you'd learn much more from examining the few, significant strings of immense success. It is not a series of independent random events, summing up. Wealth distribution is not normally distributed. It is not a series of independent multiplicative random events. Wealth distribution isn't even log-normal. It's a severe power law. There is some ordering principle that is not captured by simplistic luck based models that people put forward. Luck may still dominate, but it is not clear how, or why. And strict wealth based "Matthew effect" models don't work either, since throughout most of history the advantages of a socioeconomic advantage decays, rather than compounds, over 15 generations. Skill hypotheses become much more plausible when you look at the statistics.