This only works if it's semi-automated on the sender side. The app should add tags based on keywords, which the author can delete before hitting send if they're wrong. That way, as a follower, I can have some confidence of not seeing the stuff I opted out of, without having to build such a filter myself.

[Insert AI buzzwords]

The more manual alternative is to have multiple identities (subkeys?) and have the app keep an eye out for when you post baby photos with your Bitcoin persona. But this is inflexible and requires the sender to decide on how to split things.

Some reader-side filtering based on other heuristics is probably still needed though.

[Insert AI buzzwords]

Or we just reinvent social norms about which topics are taboo in public spaces. Meh.