The problems not so much UTF-8 vs UTF-16 as the fact that there is so much software out there that can't handle unicode codepoints outside the BMP. For example

The problems not so much UTF-8 vs UTF-16 as the fact that there is so much software out there that can't handle unicode codepoints outside the BMP. For example: https://github.com/rails/rails/issues/3727 https://github.com/rails/rails/issues/3727 Another example is MySQL. utf8 won't work; you've got to use utf8mb4 instead. (OT, but won't this be a problem again if and when we get unicode characters that are 5 bytes in UTF-8?)