There’s a distinction between what these companies can predict about you vs what they can roll up into products that other entities want to pay for. For exampl

There’s a distinction between what these companies can predict about you vs what they can roll up into products that other entities want to pay for. For example, a ton of advertisers will predetermine some targeting criteria based on simple demographics, brand loyalty or rewards program data, etc. and then commit to it for a whole ad campaign, even if it means leaving money on the table by not electing to dynamically shift into more precise targeting. For these clients, no amount of fancy predictive capability or algorithmic targeting will matter, they just don’t care. So Facebook or whomever just offers them big, sloppy and easy-to-conceptualize segments or buckets of users. They just want aggregate intelligence anyway, so nobody in the transaction cares much if it’s wrong about your age bracket or general TV interests. This doesn’t mean Facebook is unable to produce far more alarming forecasts of your behavior, or assign you to categories based on things like political activism, privacy conscientiousness, or detect personal life details like your location trail, purchases, etc. The really scary stuff just doesn’t end up being surfaced in connection to lowest common denominator adtech products.