Yes, the author is assuming for the sake of argument that salary is dictated by talent. But if talent is distributed on a normal distribution and compensation i

Yes, the author is assuming for the sake of argument that salary is dictated by talent. But if talent is distributed on a normal distribution and compensation is fat-tailed, that might mean that there is a non-linear relation between talent and compensation, or it might mean that other factors entirely are at play. My guess is that the compensation graph shown is skewed by the presence of executive salaries, which have been shown to be uncorrelated to corporate financial performance, suggesting talent is not a significant factor at the fat end of the distribution; more likely, the fat tail is a social network effect. To be fair, the author concedes this possibility in a footnote.