It's interesting. What would be an "acceptable" mix of demographics for political ads recipients? That is to say, what is the answer to that question where ther

It's interesting. What would be an "acceptable" mix of demographics for political ads recipients? That is to say, what is the answer to that question where there is no PR story, where Facebook doesn't look bad? If the answer is, none at all... I'm sure we can see where they are coming from. In my opinion you don't really need to do the NYU study. Intellectually honestly, many political ads will disproportionately appear in front of users with different demographics than their census tracts, regardless of their targeting parameters. In my experience many of the demographics of users in many software products are arbitrary, telling you nothing about the content and much more about acquisition channels and technology usage patterns at a particular point in time. As far as I know, Facebook allows some targeting parameters for political ads. So they should publish how often those targeting parameters are selected. Great, advocate for that. Intellectually honestly, that will conclusively show that ad buyers have a wide diversity of targeting parameters that, in aggregate, represent a complex mix of objectives oftentimes only adjacent to a specific election. Almost certainly Facebook already looked at this and found that geography, gender, age and proxies for user's race (like "multicultural affinity") are among the top choices, and that looks bad, even though it may be an important part of all ads targeted anywhere. Is NYU's study going to have enough power to measure targeting in an intellectually honest way? They can certainly write something descriptive. That descriptive, "Well here are some ads we looked at, and some of them disproportionately appeared in front of users with e.g. this ethnicity more often than others, which we editorially chose" - I can see how that is a lose-lose for Facebook.