In mid-March of 2006, NISO convened a roundtable of experts and thought leaders in digital resources, at the National Library of Medicine in Bethesda, Maryland. The goal of this meeting was to establish some consensus around the use of identifiers for text, video, music, and other media in the digital realm. In breakout discussions, three characteristics of an identifier were ultimately defined: granularity, semantic opacity, and persistence.
The granularity of an identifier refers to precisely what it identifies. An ISBN, for example, identifies a stand-alone, trade-able publication (a book or a chapter). It does not identify an illustration, a diagram, a bibliography. The publication is the extent of the ISBN’s granularity. Other identifiers (such as the DOI) can identify components of publications.
Semantic opacity refers to the degree to which the identifier is a “dumb number” – a random string of numbers that carries no intelligence. The ISBN is only partly a dumb number – it begins with 978 or 979, which indicate that the thing being identified is in the book supply chain; it then has a publisher prefix. The string following the publisher prefix is semantically opaque, and the ISBN ends in a check-digit that validates the number.
Persistence refers to how long the relationship between the identifier and the object will last. Identifiers on shipping containers, for example, do not need to be persistent after the container has been unloaded and its contents dispersed. Identifiers on books need to be persistent for a much longer period of time, as information about a book can be created long after the book itself has gone out of distribution.
Essentially, all an identifier does is say, “This thing is not that thing.” It doesn’t say what the thing is, or offer any insight about any of the thing’s characteristics. An identifier expresses uniqueness. And that’s all it expresses.