Chris Dent shared a link with me that discusses unique vs. descriptive identifiers:

Information-Free Identifiers: A Key to Flexible Information Systems
Part I, Part II

Despite the importance of getting it right and the consequences of getting it wrong, identifiers are, in fact, got wrong most of the time. The requirement for uniqueness in identifiers is so basic, and so universally appreciated, that it is usually taken as a sufficient as well as a necessary condition. The general idea seems to be that as long as the basic requirement for uniqueness is satisfied; an identifier is a convenient vehicle for conveying other information about the thing identified. This destabilizes the identifier and gives rise to the problems discussed in this article. Part 1 examines in detail the problems associated with altering the properties of a thing?s identifier and describes the basic principle of identifier stabilization through the use of information-free identifiers. Part 2 focuses on implementation considerations.