I think that social networks are not meant to be moderated at scale. We are meant to have what I call 'overlay networks': we occupy the same infrastructure but see content filtered to the style that befits us. Most social networks have the notion of friend symmetry, but I think that read-time filtering needn't be like that.

To that end, I made a trivial Chrome extension and an equivalent CRUD backend that just helps me store lists of users I like and dislike. The former are highlighted, and the latter are simply removed from comments.

As an example, the user I'm responding to is someone whose comments I like so I have had them in my highlight list for two months now and not regretted it https://overmod.org/lists/view?pk=ELpqNsanTYP9_wZXNjdF-FcEOc...

My personal tool is particularly idiosyncratic but I think information sieving is particularly important these days, so I recommend everyone build something like this for themselves. One thing I've found it particularly helpful with is the usual outrage bait. But I also killfile users who I think particularly misunderstand the comments they respond to, and I also killfile users who express what I think are low-information views.