Mastodon isn't a Twitter clone but it was made to fill that exact niche. Short posts with mentions, hashtags, and URLs. Such details give the feel of a platform. The velocity, the atmosphere. On a personal level that's just what I like. And I think it would be reasonable to assume many of those who signed up do, too.
A social network is also a kind of common playground. Individuals have the capacity to spoil it for others. For example, when there was a bug in Mastodon that allowed you to make links spin, many made use of it for fun, which spoiled the experience for people with motion sickness. The more vectors there are for how information is transmitted, the more pronounced this is.
Right now, there is not a whole lot of ways that individual posts can escape their boundaries to affect the whole experience. They can use lots of line breaks or a wall of text, but it's limited by a maximum height. They can use caps, but they don't stand out that much. Custom emoji are one of the weaker spots when people use them as fonts, and that's annoying.
But adding more elements to the mix? More control over visuals? People WILL make it look ugly, that's a guarantee.
There's another angle to this. The fediverse is large, and its participants who are not Mastodon have different ideas and design philosophies. Some of them are publishing all sorts of lists and quotes and strike-through items. And when those posts show up on Mastodon, those visuals are lost.
That is not ideal, when it happens! But interoperability between systems with different goals and capabilities can't ever be full, it's a spectrum.
What does not necessarily make sense to me is the urgency and fervour with which this issue has been discussed. Although it is not uncommon for Mastodon--for whatever reason--to frame every feature request as a question of life and death, when Mastodon is ultimately a place where we make a meme out of someone's name for a day, and then someone posts a picture of What Must Not Be Named
Mastodon, rich text
@Gargron βAnd I think it would be reasonable to assume many of those who signed up do, too.β the polls seem to tell that assumption was incorrect anyway.
Wrt. potential for abuse, I don't think there is significantly more. If you are afraid of headers for instance, we could have them be displayed with the same font size and weight than regular bold text, and it wouldn't be more disruptive. Things like emoji and unicode abuse is more disruptive.
You are right that perfect compatibility with every possible other software is not reasonable, as AP allows full HTML. However, I don't think this means we should draw a line and never move it. We can reassess the situation when it changes. Currently, software such as pleroma makes reasonable use of some HTML elements, which, once stripped, change the meaning of the statuses. We have an opportunity to change that.