In fact this requires special language-specific knowledge anyway (which unicode provides in some tables and algorithms actually). In some languages ä should sort exactly as if it were 'a'. "aa", "äb", "ac". In others it should sort as a distinct letter (but not necessarily between 'a' and 'b'). Different Latin languages sort differently, I'm not sure if exact UTF-8 (or UTF 16 or UTF 32) byte ordering is actually appropriate collation in any latin-alphabet language.

But I do suspect it had something to do with ascii compatibility, I don't recall what. Very little of unicode is accidental, there's usually some reason for whatever in it.

Except not really, since in Swedish, it's sorted xyzåäö, not aåäbc. Also, it used to be that w and v were equivalent sorting-wise and you'd mix them together.