There is no "most" correct, since the "length" of UTF encoded text is ambiguous. The point of the post is to highlight which semantics are the most useful and the tradeoffs.