Perhaps what it comes down to is a confidence interval, and the broad overconfidence of tech companies when their product meets the real world. There's nothing

Perhaps what it comes down to is a confidence interval, and the broad overconfidence of tech companies when their product meets the real world. There's nothing inherently wrong with saying "oh, you like X, maybe you should try Y". The problem comes in when you're too focused on providing the Y suggestions, in my opinion. There's a general overconfidence in the tech industry, particularly in advertising, and even more so in social media targeted advertising, that the suggestions that they make are actually useful to society as well as individuals. I'm not saying that they're wrong about it, but I am saying that just because the previous status quo was "no measurement, trust us, it works" in the TV/radio/billboards era, and now you're bringing in "look, we measured it!" doesn't inherently mean that you are correct about that measurement, or that it is the right thing to measure in the first place. Perhaps the old status quo factored in innumerable intangible social factors that could then somehow translate into tangible results that you're not capturing in your new metrics. The economics of the social advertising industry are wholly untied from providing genuine value, and this manifests itself in myriad ways: you're only seeing a symptom of it here.