> to the extent that it affects productivity
I believe that's the key. In a similar fashion, we (society) don't want kids to interfere with productivity, so we stuff them in kindergartens and schools so they are out of our way until they reach their 20s and can be regarded as fully productive. And we put old people in nursing homes for the same reason.
I wonder if the current practice of everyone hiring only "the best of the best" is suboptimal for society as a whole, since it leaves a lot of potential productivity on the society table: Someone with a disability might be (say) only 5-10% as productive as the average healthy adult, but add up everyone with some sort of disability and it sums to a nontrivial amount of underutilized capacity.
But I'm not sure. It's a game of numbers: There are administrative costs to hiring and managing (say) 10-20 people to do one person's job. If the "less productive" part of the population is small enough and the productivity difference between "less productive" and "more productive" is large enough, it could be a net productivity win to pay the "less productive" some small amount of money to go away and stop bothering the "more productive".
Which is essentially what we do today.