The problem with reality is there is no mathematical metric function where you can plug as input a human and you get as output a number and then you can sort this humans out so you can hire the best. Even for small domains like programming with language X and framework Y is not possible to create a good such metric function, someone could be good at memorizing APIs, someone could be fast at creating good solution but others might be slow but their solutions are even better and others are good at debugging issues.
If you can meassure skils then you should be capable of hiring people with this simple 2 steps
1 Person A administers the tests 2 Person B checks the results but he has no idea who the candidates are
No more problems with gender,race,culture,political affiliation
Someone explain me how you always have to meet the candidate because you have a skill to read they competence and cultural fit from their face and body proportions and you are special and this proven physical biases never affect you the special one. I mean "you" in a general term not the comment author.
But outside of a perfect mathematical universe competency is multi dimensional and difficult to measure. Bias does creep in. Treating it as a scientific process seems just as reductive as reducing a person to just their gender.