Why does mojibake happen at all?
Text is stored as numbers, and an encoding is the agreement about which number means which character. Write with one agreement and read with another, and the same bytes come out as different characters.
Can "ã‚ãªãŸ" style garbling be fixed?
Almost always. That shape is UTF-8 bytes read as Windows-1252, where every byte simply became a different character - no information was lost, so pasting it into Repair brings the original back.
Why can't the Shift_JIS style be fixed?
Byte sequences that do not exist in Shift_JIS are replaced with U+FFFD as they are decoded, and that replacement throws the original numbers away. No tool can recover them; you have to reopen the original file with the right encoding.
Why are several candidates shown?
Garbled text alone does not say which encoding was used to read it, so every candidate is tried and ranked by how naturally the result reads. The top one is usually right, though short strings can fool it.
Is my text uploaded anywhere?
No. All conversion happens in the browser. Only the share button sends anything, and only the URL you asked it to shorten.
Can I garble emoji and half-width katakana?
Yes, but many encodings cannot represent those characters at all, so they come out as "?" or U+FFFD. Emoji exist in almost no encoding other than UTF-8, so garbling loses them.