These days, it should be fairly easy to combat this. Everyone subscribes their messages. Client sw allows whitelisting the signatures. Whitelisted signatures can vouch for new signatures to be added. User can easily silence any signature. Messages that are hidden (because the signature is not whitelisted) are shown if a whitelisted message replies to them.
I'm sure I didn't catch all the edge cases, but the main idea is that the system is distributed, built on reputation, and self-managed. Everyone is responsible for the content they receive.
That's just bozo filters with a side of Web of Trust. It's been done. Bozo filters help with spam on the client side but don't do anything about spam on the server side. Servers still have to host everything because they can't know a priori what clients will want or not.
Not that your idea is necessarily bad but it breaks a lot of the utility of Usenet to have whitelist based filtering. I can't know ahead of time I like your posts and want to whitelist you. I also don't want to have to "know a guy" to get involved in a group (via Web of Trust).
Is spam message a problem if noone reads it? Storage is cheap.
As for knowing a guy... Maybe someone will browse spam and vouch for you for this message and if more people like it, you are accepted to their whitelists. But to be honest, if people don't want to read my messages, that is their prerogative.
I'm sure I didn't catch all the edge cases, but the main idea is that the system is distributed, built on reputation, and self-managed. Everyone is responsible for the content they receive.