Joel's article confuses characters and code points and can be very misleading.

>In Unicode, a letter maps to something called a code point which is still just a theoretical concept.

This is incorrect. Letter or character (user perceived character) can be multiple code points. Code points don't have intrinsic meaning across languages. If you want to count number of letters in a text or edit text, you must work at grapheme cluster level. Absolute minimum about Unicode must at least mention grapheme clusters and user perceived characters.