It seems worse statistically than hate speech, because it widely prevents and distorts communication, and keeps people in separate "bubble" reality tunnels. As Jaron Lanier put it, “This is an epochal development. The version of the world you are seeing is invisible to the people who misunderstand you, and vice versa”.
A bottomless rabbit-hole of paranoia, in which sincere attempts to understand the views of others are seen as insincere or suspect; in which attempts to be courteous and accommodating are mistaken for conspiratorial deviousness or other petty nefarious behaviour. In which people routinely and reflexively attribute the worst possible motives to others, when innocence fits as easily. I've seen friends fall out because of it.
Again, to quote Lanier, “society has darkened a few shades as a result”.
It has a number of causes, some accidental, some engineered. I stopped using social media because of it, but I wonder if it applies to other online social spaces. I've never been a technophobe, I've always enthusiastically promoted technology, so it's difficult to change my mind on these things.
Anyway, back to the real world of coffee, sunshine and birdsong...
Decades ago, I remember innocently asking a female colleague a work-related question, and she took it the wrong way (made some assumptions that weren't true), but she immediately saw my confused reaction, and realised she'd got it wrong. That's the cliche about facial expressions, etc, aiding communication. You don't get it online.
Lanier's point is that it's worse with algorithm-delivered news and social media feeds. Not only are you being fed different, personalised bubble realities, dependent on your online inclinations, but you have no way of knowing what others are seeing. Yet because of the traditional model of media in our heads (everyone accessing the same narrow range of TV news and newspapers, etc) we still have this anachronistic notion of sharing roughly the same reality (outside our physical space) - and it becomes an increasingly false notion.
You can see how that leads to paranoia when people not only don't have the same reference points, but don't even know it. In that environment, if I asked the innocent question that was taken the wrong way, there's probably no way back for that relationship if you don't have the shared reference points to mend the paranoia, the poison - the wrongly imputed slight or deviousness or whatever.
It might not even be that overt. It might be more subtle, but measured statistically across hundreds of millions of communications, it entails a kind of "darkening" of the communications realm, a) if we're not very careful personally, and b) if the business model for social media doesn't change fundamentally. Read Jaron Lanier's book, Ten Arguments..!