The amusing Black Mirror video above, though artificial, is arguably a prescient comment on the implicit risks inherent in a tendency for "oversharing" and/or "data leakage" in the IT-enabled SNM (Social Network Marketplace) and other personal-data-related databases (e.g., health, insurance, banking, education), where private data that one might have previously perceived as being peculiar and useful/relevant in one context only is subsequently seen to be useful/relevant in another, or maybe many other contexts. These are typically the data connections and interconnections that the SNM operators and data miners would tend to seek/exploit, for financial gain.
When I was contracted in 2003 to get a failed data analysis project back on the rails, I learned quite a lot. It was a complex IT project to implement a BI (Business Intelligence) system and we had to train the users in the application of BI (it's actually quite powerful and hairy stuff) to meet the growing and complex business needs of the power (energy) company who had contracted me into the project recovery role. I learned that what the Defence sector had always taught was true - that all data/information can be interconnected at some stage - and that, for BI, the world could be simply envisaged as one or more universes of dynamic data - each having its own peculiar descriptive and dynamic data model and that, as in the popular SF concept of parallel universes, there was the potential to interlink these data universes (mass aggregations of dynamic data sets), constantly combining/recombining and drawing data from one to the other, enabling the BI analyst to discover data relationships in a way that would probably not have previously been feasible on such a mass scale, using the then prevailing technologies.
Fast forward to 2018, where we can perhaps now better understand why we might have the apparent privacy shambles that we see around us. It was a gold-rush, opportunistic, every man for himself. Presumably the Google/Facebook founders (and others) would have seen it coming. There were little/no regulations to limit or constrain the progress of BI and its application in the field of mass demographics. Now that some regulations have belatedly been or are being implemented, it arguably may be too late anyway - locking the stable door after the horse has bolted; Pandora's box has already been opened.