If you look at the Like button, and their other endeavors, I don't think Facebook's ambitions are limited to just getting visits and clicks. They want to become a utility as indispensable as HTTP.
And Facebook are really smart. Keep in mind that if there is a drop, they've probably known about it for months and months, and we're just finding out it about now indirectly.
That said, I hope they don't succeed and a distributed system does succeed.
I'm really hoping for decentalized peer to peer communities that are elastically created from locally computed interest profiles. Not a distributed facebook per se, but more like a user-centroid HN that aggregates similar people and interests... that would really make my academic life soar.
In the end, I don't think Facebook would fight for the long tail of user demographics. What they have is quite profitable enough. Why expend 90% more effort to convince all of the tinfoil hat wearers to join?
Facebook private groups
https://www.facebook.com/groups
are working quite well among my circle of friends to provide a focused view of one or another niche community that forms a natural subset of my friends. There is a particular Facebook secret group that I visit absolutely every time I'm on Facebook, which is several times a day. That's a practical place for me to ask focused informational questions, and also a place where I can share a lot of camaraderie with dear friends who have had similar--and yet very diverse--life experiences. Because of the stronger filtering abilities of Facebook, I can indeed see how it could supplant many niche sites. (Thus far, HN gets the second share of my attention after Facebook, among all websites, and it will be interesting to see what the consensus here is about the overall community experience in the next year or so.)
Since none of the groups I joined have the same level of activity I'm used to on specialized forums for those subjects, I never bother to check up on them - which of course lends itself to a vicious circle of inactivity.
What do you imagine they would do otherwise? More to the point, what's the business case for doing what you describe? They could require FGroups to self-categorize in order to narrow and aid their advertising sales, but that's not foolproof and lots of groups would just as soon choose "Other."