There was bred a while back where anon simply split notables into two categories:

[Anon notes]

…

[Non-anon notes]

…

It was strikingly effective, as all the news spam gets chunked together. And it's an easy binary decision so low overhead, "Was this note generated by anon or not?"