I still don't really understand why we had to use punycode for this instead of plain UTF-8?

Unicode domains are nice, because it's always frustrating to have to misspell words just to make them fit within ASCII for no good reason (other than technical) but it really should have been handled in a better way, both on the technical side (punycode is ugly and scary for everyone) and the administrative side (it's actually not that difficult to handle phishing or to semi-automatically attribute domains with accents to the existing owners of the equivalent domains without accents, for example, at least for most European languages).