Because there aren't sufficient repercussions. You can see this everywhere: lack of social accountability leads to more openly incivil behaviour. This isn't just online, but offline it works the same: if the cost of being incivil is low enough, there is a lot gain from it. See for example: backroom deals, mob behaviour, road rage.
A few years ago someone pointed me to https://ncase.me/trust/ (I think I learned about it here), which captures the problem very well for me: if social cohesion goes down (i.e. fewer interactions but with more different people), the penalty for inciviliy goes down but the rewards stay the same.
Also: there is really no central, balanced standard for justice/justness scale, but only an illusion of it(probably generated through said cohesion).
I learned it being an non-American with some English literacy on internet, and am being reminded of it every week.
Justice and/or “being nice” perceived scale is just an AI model of events around you that one uses to predict certain outcomes, and most of logic you see in it are purely coincidental correlations to aid in footprint reduction.
I would imagine it's more for things like illegal content (i,e. child sexual exploitation) as opposed to people being "mean". But I'm sure abusive behavior is a component.
I do not regret downvoting the comment since it is an unproductive take in 2020, and there's been more than enough incidents in the past decades to show why content moderation is necessary and inherent human virtue is not sufficient. I've deleted the tweet (the intent wasn't public shaming, it was more highlighting an unusual take).
That said, creating a throwaway for snitch-tagging isn't a moral highground.
