You'd have people queuing up to insult you so they could watch the reaction.

The real world "say it to my face" really means "say it somewhere I can fight you if you're rude enough".

Yes. The underlying threat of violence, or at the very least social exclusion, if one goes too far is what keeps most in person conversations far more civilized than their online counterparts.

Even assuming the correctness of your argument, the threat of violence only applies where it is credible. Strong bullies aren't intimidated by their victims, men generally aren't intimidated by women, and so on.

There is in fact an orthogonal effect where people will fear expression in the face of violence. Is "civilized" expression dominated by the strong better than the rudeness of online communication, where there are different power dynamics? I don't know, but I wouldn't be so quick to dismiss it.

Perahps at least a portion of the "uncivilized" speech you see online is the outcry of those who are not allowed to express their opinions elsewhere.

You assume one-on-one conversations. In a group setting even a strong bully can be subdued by the group. In such a group, a man isn't going to be allowed to insult a woman without other men stepping in. But you bring a good point. Should insulting, disrupting voices be tolerated in a conversation in the name of freedom of speech? I would say no, because they make further speech impossible, but I see the problem in deciding what constitutes disruption/insults.

Social exclusion works online -- and did work on usenet, until platform owners disabled it in order to promote engagement/enragement for ad impressions.

Not really - it certainly didn't work against spammers, and the whole thing was entirely unauthenticated.

A little bit. It was a lot easier to escape exclusion by changing your identity in Usenet compared with the real world so motivated trolls would often do that.

That's the real-world implications of an in-person conversation.

Obviously extending this to online, two people could be far from each other. But it still would require one person to see the emotional response from the other, which may or may not be pleasant.

Not exactly, the real world implications could go well beyond observing your targets emotional response.

And how would that work exactly? Strap the other person into a chair and force them to watch you "Clockwork Orange" style?

As an adult your emotional responses to communication are entirely your own affair.