There is a philosophical problem with all identifiers, particularly obvious in an international context. Identifiers are not only used by computers, but also by humans. Humans have a strong preference for using mnemonic names: you call the address field in a form address and not x12 . Such identifiers are easier to create, easier to remember, easier to understand, easier to guess, easier to transcribe, and easier to identify with. The prime example is URIs: they are machine-readable, but you also find them printed in advertisements and user manuals, and their creators definitely make every attempt to make them easy to remember.

Should identifiers therefore always use English-like words, ASCII letters only? This may make things easier for some people, but harder for others. Is it worth to make things easier for the majority of actual users, while making it a bit more difficult for accidental outsiders? And what if they have to contain information that is originally not in ASCII? URIs created as the result of a form submission are a prime example.