The entire point of this exercise is to address zaidf's use case, which is trying to attach importance to messages that are outside the realm of Facebook's "important" data points. If they aren't your friend, and they aren't a friend of your friends', but they still have an important message -- how can we identify this?
The obvious hope here is that they can identify important messages by attaching a small (but real) cost to them, while limiting abuse by way of rates and maybe tweaking the cost.
You may not want these messages, but the fact remains that there are important messages that go missed because they can't smartly be deemed "important" in the Facebook ecosystem. Maybe there are sinister intentions beneath the surface, but this is still a legitimate attempt to solve a legitimate problem.