Base32
Base64 without the confusing letters.
Where it came from
Base32 was standardized in RFC 4648 (2006) to fix Base64's weakness for humans: Base64 uses 0/O, 1/l/I which get confused when read aloud or copied. Base32 restricts itself to A–Z and 2–7, avoiding those glyphs entirely.
That made it the choice for things typed or spoken: TOTP authenticator secrets, Bitcoin/Tor onion addresses, and DNS labels that must survive case-insensitive systems.
How it works
Like Base64, it groups bits — here 5 bits per character, 8 characters per 40-bit block. Output is ~20% bigger than Base64 but safe to read.
A padding character "=" fills the tail. Crockford's variant drops padding and ambiguous chars (I,L,O,U) entirely. UWC encodes/decodes Base32 in your browser, UTF-8 safe.