> Does it really correlate with protected group memberships?
Yes: "the vast majority of families (73%) who buy organic describe themselves as white" [1]
Now, you might say, "Okay then. Let's prohibit ad targeting by food preference." But the trouble is that there are hundreds of factors that correlate with protected group membership. Are you going to ban them all?
> But the trouble is that there are hundreds of factors that correlate with protected group membership. Are you going to ban them all?
It's a tough question to answer. I believe you were debating me about the use of ZIP code as a proxy in another comment thread so I'll touch on that also.
If it's found that organic food preference is systemically being used as a proxy for a protected class, as ZIP code has been in the past, then it might end up being banned. By following this process could you end up with a list of hundreds of banned factors? In theory yes, but in practice it seems unlikely, just because of how long it would take to study each factor and go through the process of adding it to a banned list. It took a long time for redlining to be made illegal. The article you provided says that organic food preference is gradually starting to mirror the general population, so if current trends continue it eventually won't be useful as a signal for race-based discrimination. Furthermore unlike redlining, there's no vicious circle of "I eat non-organic food -> I can't get a loan to buy organic food -> I have to continue eating non-organic food". By the time you had enough data to decide whether discrimination based on a particular factor constitutes de facto discrimination against a protected class, that correlation might have ceased to exist.
