Facebook must understand how weird that sounds. How can they not know why people are recommended to each other?
They really do need to dig into the issue, if in fact they don't know. Because something seriously need to be excluded from their recommendation algorithm if the article is true.
She can't reveal the patients to them, so Facebook wouldn't be in a position to give a specific reason -- only generically how the algorithm is calculated.
In fact, they can expect very serious investigations throughout Europe if someone complains about it to their local information commissioners or equivalent here.
If they are in fact sucking in contact details from users phones and using them for matching and recommendations, that would seem to be something that would be serious enough to likely require express consent (in other words: users taking an explicit action, rather than being "opted in" implicitly by agreeing to a TOS or similar) under EU data protection regulations.
Not a lawyer, but I'd be surprised if there isn't one or more data protection violations lurking in there somewhere.
not surprising really, I have often written software, returned a decade later and had to unravel why it does what it does, even though I am one person and it's just a few thousand lines of code, I can well imagine the FB system is too complex for any individual to inspect, and if the record keeping and documentation on the whys as well as the hows is incomplete then we are we are...
Technical question: does anyone know if FB can easily answer why X was recommended to Y? That is, do they have to manually check or can they query this easily and get the precise answer why a recommendation was presented(for example: the reason why X was recommended to Y on date d, cannot be that X is connected to Z and and Z to Y, if X connect to Z on d+1; I assume bad inferences like this will probably happen often if such questions are answered manually).
Such a terrible excuse. FB you only have one job! Fail.