:D I get your point. The specific reason I said it "wouldn't take much at all" to convince me otherwise is that under a negative result, we do not know:
-> Was Facebook listening under previous versions?
-> Did the researchers adequately trigger ads?
-> Are subjects able to notice all direct ways in which listening data is being used?
Instead, I would simply accept a negative as showing that no, Facebook isn't listening. (I'd happily accept false negatives.) Instead, what is really interesting is that numerous people decided to do an informal experiment such as put their phone next to their TV set to Spanish-language programming and then, to their shock and horror, Facebook began showing them Spanish-language ads.
-> Was Facebook listening under previous versions?
-> Did the researchers adequately trigger ads?
-> Are subjects able to notice all direct ways in which listening data is being used?
Instead, I would simply accept a negative as showing that no, Facebook isn't listening. (I'd happily accept false negatives.) Instead, what is really interesting is that numerous people decided to do an informal experiment such as put their phone next to their TV set to Spanish-language programming and then, to their shock and horror, Facebook began showing them Spanish-language ads.