The biggest problem with Unicode parsing is that you don’t know how long a Unicode string is without parsing it. Which often leads to double parsing it. But we’d have this problem even with Unicode fixed to 2 or 4 bytes (like utf-16 and utf-32) because not all glyphs are going to be printable characters (eg accents).

…or you allow for every possible combination of characters, accents, etc and fixed at specific number of bytes and then suddenly you have a 6 or 8 byte character set that breaks backwards compatibility with ASCII and increases global network throughput by six times despite that mostly being empty space, while increase a greater burden on font developers to cater for every subtle variation of glyphs.

Sure things seem messy now, but I honestly think utf-8 is the least worst solution to the problem.