Is this clunky? Yes. Is there a better way of doing it? Maybe. Is it an explicitly biased policy? Yes.
However, correcting a biased system requires that there is a period with enforced change bias to correct the existing bias. This is because there are many ways previous explicitly biased policies entrench that bias even if the current environment is no longer explicitly biased.
Even if we were to assume that everyone is suddenly 0% biased on anything, a lack of explicit corrective actions will necessarily be long and dragged out continuing to harm the classes that are no longer explicitly discriminated against.
So given we know that hiring biases do still exist, and we know that many groups are underrepresented due to historical biases we end up stuck trying to fix a complex problem with very clunky binary tools.
As a very simple example: imagine you have a job where the starting rate for one group is $2 and for the other group it’s $1. Everything else is fair people get raises at the same rate (eg 10% or something). Then one day the law says “no, that’s bullshit”, and now everyone’s starting rate is $2. At this point all the people who were already employed are still earning half as much as their colleagues, despite the policies now being completely “fair”. If you just leave things as they are, eventually things will end up equal across both groups, but you have a potentially multiple generation latency. You could tried to correct the incomes, but that looks like discrimination: “we’re doubling the income of everyone in group B, group A is explicitly excluded”. The thing is in this case, even with this biased 2x income, the disadvantaged group is still worse off, due to the years of unfair compensation. If you try to correct for that you end up with another super biased behaviour.
What we’re seeing is an attempt to correct historical biases, and any such correction is necessarily biased, because otherwise even a perfectly unbiased continues to harm historical victims.
However, correcting a biased system requires that there is a period with enforced change bias to correct the existing bias. This is because there are many ways previous explicitly biased policies entrench that bias even if the current environment is no longer explicitly biased.
Even if we were to assume that everyone is suddenly 0% biased on anything, a lack of explicit corrective actions will necessarily be long and dragged out continuing to harm the classes that are no longer explicitly discriminated against.
So given we know that hiring biases do still exist, and we know that many groups are underrepresented due to historical biases we end up stuck trying to fix a complex problem with very clunky binary tools.
As a very simple example: imagine you have a job where the starting rate for one group is $2 and for the other group it’s $1. Everything else is fair people get raises at the same rate (eg 10% or something). Then one day the law says “no, that’s bullshit”, and now everyone’s starting rate is $2. At this point all the people who were already employed are still earning half as much as their colleagues, despite the policies now being completely “fair”. If you just leave things as they are, eventually things will end up equal across both groups, but you have a potentially multiple generation latency. You could tried to correct the incomes, but that looks like discrimination: “we’re doubling the income of everyone in group B, group A is explicitly excluded”. The thing is in this case, even with this biased 2x income, the disadvantaged group is still worse off, due to the years of unfair compensation. If you try to correct for that you end up with another super biased behaviour.
What we’re seeing is an attempt to correct historical biases, and any such correction is necessarily biased, because otherwise even a perfectly unbiased continues to harm historical victims.