I like the options to control the size of the 'IDs' and the charset used.

I wrote something like this a few years ago called 'acts_as_hashed' that as you might guess, uses a hash function to come up with a unique non-sequential identifier. https://github.com/briandoll/acts_as_hashed