> But the new wisdom is that any policy which disproportionately affects a certain group is racist.
That doesn't sound like an accurate interpretation of that document. Here's the non-strawman version:
> Policies that are purported to be “color-blind” or “race-neutral” have all too frequently been designed to benefit White children and families, and harm children and families of color directly and indirectly.
…
> In the past, public policy has too often been siloed, and policies intended to support children have failed to address the larger struggles their families face. Policy has even actively undermined families in the name of protecting children—as we see with the child welfare system, which disproportionately threatens and separates Black and Indigenous children from their loved ones.
What that's getting at is the idea that there are many cases where policies did not incorporate an explicit racial check but still produce that outcome. For example, if your city has a magnet school which doesn't have good bus service, it could act as a racial filter if a higher percentage of white families can afford to own cars, even if nobody who set the policy had that goal. That policy would be bad because the filtering effect would be real and isn't relevant to the stated goals of the program — whereas having, say, a height threshold for a basketball team is pretty reasonable because that's directly linked to performance, whereas it wouldn't make much sense for, say, a biathlon team.
In the context of that article, this is a notorious problem in the child welfare system where you have major events based on subjective judgements about the quality and safety of the home environment or the ability of families to provide for a child. It doesn't need to have a “Is this person black?” question on a form to still have negative outcomes if there are subjective factors like predominantly white middle-class caseworkers feeling less safe in a poor majority-black neighborhood and that influencing their assessments. That's what they're talking about when they say that polices must recognize how these different factors intersect — it's awareness, not an absolute, and it's an attempt to correct for the way data-driven decisions can still be skewed by societal biases.
Going back to this discussion, as I explained the difference is that the policy is clearly linked to a simple, neutral objective (everyone gets sick, nobody wants to) and the way to comply with it is easily available to everyone at no cost. That's not racist because it's not based on fixed attributes which people cannot change and there's no significant barrier to compliance. In contrast, a policy which, say, said people who are at elevated risk of sickle cell anemia couldn't eat a restaurant would be racist because there's no plausible safety justification for such a policy, everyone would know why that one “objective” factor was selected, and there's nothing which an affected person could do change to comply with the policy.