E.g. looks and facial symmetry both correlate with fertility and IQ, probably through the mechanism of mutational load. So you could say, "good looking people make more money" as an example here, but the effect could be from IQ because they actually do go together.
The same could be happening for height. This may be the reason people instinctively are attracted to tall people.
It's very believable that intelligence and height correlate, but the question then becomes "Do we overweight height in decision making, relative to its predictive power for success/intelligence?" If height very weakly predicts increased intelligence and we weight it strongly, that's still far from optimal decision making.
The correlation with intelligence isn't even necessary to ask that question. Hiring taller people also has upside in that other people (eg, your customers!) are biased toward them as well. However, and I admit this is conjecture, I still suspect allowing your own biases to influence decision making is a net negative.
Too many acceptable candidates? Nobody I've ever worked for had this problem. If you meet the bar, you're getting an offer, even if Chuck Norris and Jeff Dean carpooled to the next session.
At the level of phone screens and first interviews, there should be many potentially acceptable candidates. If there aren't it's indicative of a flaw at the very top of the funnel (ie, the opening needs to be better publicized, or the org. may have a public image problem).
I wholeheartedly agree, elevating people above the level of responsibility/work they can handle is a bad idea, as is taking away responsibility from more qualified people.
Where you run into trouble is the people judging qualifications. Human judgment is an insanely noisy tool that heavily weights a bunch of stuff with no actual effect. For a stupid example, look at the earning difference between tall men and taller men -- it's about $800/yr/inch [1]. The effects of gender, race, and apparent country of origin (eg "foreign sounding" names) are all larger and equally useless in actually predicting job performance.
I'm not saying hire dim women over bright men. As I see it, the claim is that there are often many candidates who are equally "qualified" within our margin of error, and in those situations it's wise to pick the disadvantaged candidate because A) your flawed human judgement's probably lying to you ie they're more qualified than you think and B) if they are truly equal, elevating them will help encourage the next generation of candidates with their background, as TFA suggests. We can argue whether such a strategy will discourage the advantaged group's next generation in an equal and opposite manner, but I suspect the effect is strongly net positive on the size / quality of the labor pool.
Lastly, I realize I've made this comment quite narrowly focused on a hypothetical "hiring" scenario with a single human arbiter, but I hope HN can extrapolate the ideas to other choices we make as a society.
1: https://www.theatlantic.com/business/archive/2015/05/the-fin...