I've been working a lot with binary data and doing conversion in JavaScript and recently converted a byte array from 8 bytes to suddenly 20 bytes.
Surma and Jake were kind enough to deep dive it in their 203 podcast, but it boils down to this:
[Remy] had a byte array, he decoded it he re-encoded it and suddenly the string got longer and that's because the original byte sequence that he put in contained invalid UTF-8, so each invalid byte was turned into a question mark at a box, which in turn when re-encoded into UTF-8 turns into three bytes and so that's how the byte array got longer.
So, the question mark in a box, � is 3 bytes specifically 0xEF 0xBF 0xBD and JavaScript is going to encode into this character if it encounters a non-UTF8 character, like my original string.