I think this paper harkens us back to earlier points made in this thread about the danger of making inferences in the tails of statistical distributions from statistics derived about the sample mean. The highest median "IQ" for a profession is 128, and ranges for all the professions included capture the ~140+ spectrum at the upper end.
Aquinas spitballing...
If I may, can I please restate what I think your underlying (unspoken) thought process is? I come from an economics and business background, so human capital theory is dear to my heart.
I get where you're coming from. From your posts, it sounds like you're attributing an aggregate theory of diminishing marginal returns to human capital to a specific subset of the population, namely gifties. But, the correlations and concavity of earnings in ability (loosely proxies by educational attainment and/or profession) from that literature are derived from the broad calibration sample. They carry no meaning for particular population subsets because they weren't derived *in entirety* from those subsets.
ETA: Per ultramarina's point, I've had average job evaluations by an old boss who didn't understand my work. I wouldn't say perceived productivity is a good proxy for actual productivity, especially when the group of interest is one whose expected ability outstrips the average ability of managers.
