

Is it only an idea? I can think of automatic tagging of all posts. If you have access to a post and all its comments, probably you can programmatically assign a tag to it. Based on “words cloud” or something like that. Annoying posts usually have a lot of comments which simplifies automatic tagging. It can make it possible to filter out specific topics, or, contrary, browse them specifically.
If computing these tags is not expensive, they can be computed and stored internally in the app at client side. If this will work and will be useful, it can be moved to server-side in one of lemmy’s updates. Each post will have have probable tags in metadata with % of how sure an algorithm was about assigning this tag. Personally, I think affecting your feed by picking appropriate instance doesn’t work, and I do hope other instance-independent ways to browse lemmy will become available. But right now I haven’t found a time even to check Lemmy’s api to see what’s already available.