I've often wondered if this is because, from the HR perspective, false positives are much more costly. I.e., Those who would be helped by limiting the false negatives are project managers who are too downstream in the process for HR to care.
If you reject someone who would have been a good fit but still hire someone good eventually, that’s ok.
A bad hire is not fun, not for HR, not for the hiring manager, not for the colleagues and ultimately not for the person who got hired. Especially if they had to move for the job, possibly with family.
A bad hire has massive negative consequences for many people. Of course you want to avoid that, if that comes at the cost of not hiring someone who would have been great occasionally, that’s unfortunate but acceptable.
If you're hiring people at the very top or with rare skillsets, like building browser engines, building databases or people whose skill is at author/contributor level to relevant technology (for the job in question), you're probably not going to go with the usual process anyway as there is less uncertainty.
What virtually every attempt shows is that it is better to hire people who are roughly qualified and see what happens, managing the results instead of filtering. Having an HR strategy for hiring leads to creating biases (i.e. requirements based on your current work force) that gets you a group of overly similar and therefore collectively incompetent work force.