The encoding method specified is theoretically practical. As far as I can tell, the ONLY drawback to this scheme is mentioned in the very last section before the references: the encoded address tends to look like gibberish. (One ramification not mentioned is that if you sit on a keyboard, and happen to enter exactly 20 characters this way, chances are excellent that the result would be a legal (but hopefully unused) IPv6 address.

Whether we would WANT to use this scheme is another issue... but it seems to me that we could.

(Note: If this ever goes to a print edition, please do NOT credit me...)