Granted it's not groundbreaking but it's attempting to make UUIDs "better". For computers it doesn't really matter what an identifier looks like (minus memory to store long strings, maybe). For users, it does matter.
Long UUIDs are just an artifact we've grown accustomed too. By no means are they the ideal solution.
The type of namespacing OpenPhoto does doesn't apply to everyone either. Sometimes you do need a UUID that isn't namespaced. Sometimes you don't.
While UUIDs are not as "nice" to look at, they contain the same information (origin, unique ID) as mentioned by the OpenPhoto guys.
Except v5 UUIDs which solely rely on random numbers.