Not such a bad thing. Each rejection results in a narrowing of the scope and claims of the patent. The resulting claims that are granted will be far narrower than scope of the original specification.
In Facebook's case, though—see above—there were two final rejections. A final rejection happens when all the claims are invalidated, so your submission is now vacuous.
Obligatory warning: I'm not a patent lawyer but I've responded to both final and non-final office actions, eventually getting a patent
It's called an IPR, an inter partes review [0] in the USPTO. Most potential challengers, though, are likely to save their ammunition for if and when they get sued on the patent. (And they might quietly let Facebook know they have prior art, so as to discourage FB from getting aggressive — FB might want its patent to remain officially valid, so as to be able to continue to rattle its saber at others, instead of having invalidity officially confirmed.)
[0] https://www.wikiwand.com/en/Inter_partes_review
It's called an IPR
This sounds like it's not exactly shadow banning; your direct friends can still see your comment. I'm vaguely aware that they have special comment sections on "public" pages, articles, etc, that filter the giant thread down to stuff your friends actually wrote. I'm guessing that at some point, if you were banned from the page/etc, then your friends also couldn't see what you wrote. If that's true, then this is a little nicer to the banned user than a shadowban.
More importantly, it's more effective because it's less likely to be detected.
It still shouldn't be patentable because there's prior art involving a bunch of systems with various forms of audience selection such that how to implement this specific set of audience selection rules is obvious to most software developers. Patents cover how, not what, and they do not cover inventions whose implementation is obvious to most experts in the field.
This is (ahem) patently not true; there are plenty of patents whose implementation is incredibly obvious. Insultingly obvious. As in, "Did that company try to patent string comparison?"
Look up the "standard network byte order" patent, for instance. It's now expired and was never enforced. The friend-of-a-friend who got that patent was heard to exclaim "I can't believe they gave that patent to me!", which should tell you a lot about the quality of the patent industry then -- IMHO it hasn't changed for the better.
I'm not entirely sure why, but there appear to be perverse or at least badly aligned incentives for patent examiners, and it would appear they only look for prior art in specific locations, and these aren't the ones that the industry uses.
So you get situations like someone getting a patent for putting structures on multiple lists at a time (e.g., a lookup list, and an LRU replacement list, and maybe a list for keeping track of locks) and this is something that people have been doing since rocks were young, but some nincompoop at OraGooSoftBook got a patent on it because these structures appear in OS textbooks or maybe Knuth and not anything the examiner encountered in law school, or any legal publications. And now FaceGooSoft has a legal lock on code you wrote (hey, you did diligently read all the patents relevant to your industry before you wrote a single line, just like everybody else does right?).
They should not be, and somebody should probably be lobbying congress to address the patent office not being sufficiently strict.