When spam is posted, it’s almost always linking to some other place. The idea is to reject any edit that includes a link that has already been identified as a spam destination.
(can anyone think of any?)
Once a problem domain is banned, all links to it are banned. If a problem domain becomes converted to a “good” domain after being banned, it’s still impossible to link to it. It’s tricky to even *talk* about the fact that it is now “good”.
(I’ve seen many “good” domains turn into problem domains. I’m speculating: Perhaps the reverse happens. Maybe that’s just wishful thinking.)
Work-around: use some sort of euphemism / name-mangling / phonetic alphabet to talk about the domain name.
Say a link to a domain name already exists on some page. Then later that domain name is banned. The next time a person edits that page, the wiki will reject those edits, unless he finds all the banned domain page links in that page and deletes them. That could be confusing if that person is simply trying to fix a tiny typo, several scrolls away from the problem URI.
Work-around:
Wiki that are friends with each other could share their link ban lists, for example all MoinMoin wikis that have the antispam subsystem enabled automatically use the blacklist provided on the project’s page.
It could be also possible to re-use blacklists used by software like adblock.
A wiki could grant the impression of a successful edit. Just replay the preview page as if it were a successful edit.
(See UserTargettedParallelUniverse, IdeasToPlace #99)