I rather suspect that the “Consider” mechanism ought to be sufficient to (continue to ...) handle this, whenever it (fortunately, rarely) appears. PM is, by and large, remarkably free of spam, and I think that the “self-policing” mechanisms, working as they should, are responsible for that. If it ain’t broke . . .
Also: I don’t personally believe that author-based heuristics will be useful, or successful: “user-ids” are just too easy (for a bot) to come by. When a “bot” comes calling, its authors generally prove to be very ingenious. (Bah...)