It is widely used by libraries, data centers, archives, museums, publishers, and government agencies to provide reliable references to scholarly, scientific, and cultural objects.

Throughout the 1990s, the Internet Engineering Task Force and other organizations developed standards for persistent identifiers for web resources, including URN, PURL, Handle, and DOI.

[2] In 2001, John Kunze of the University of California and R. P. Channing Rodgers of the United States National Library of Medicine released the first draft of “The ARK Persistent Identifier Scheme,” designed in response to the needs of their two organizations, as an IETF working document.

[3] In explaining their motivations for creating a new system, Kunze later wrote that “each [persistent identifier] system had specific problems.” In contrast to the decentralized structure of the web, with many independent publishers, Handle and DOI were related centralized systems which charged for inclusion; they were “antithetical,” according to Kunze, “to an implicit principle that Internet standards must not endorse control by any one entity, over access to the networked resources of another entity.” URNs were free, but lacked a resolver discovery services, and, wrote Kunze, “it seemed to me that the IETF community lost interest in creating a whole new Internet indirection infrastructure that would add little to existing web and DNS mechanisms, especially in light of the small part that indirection plays in keeping links from breaking.”[2] In contrast to these other systems, the ARK scheme proposed that “persistence is purely a matter of service,… neither inherent in an object nor conferred on it by a particular naming syntax.” The most an identifier could do to solve the problem of persistence, then, was to indicate an organization’s commitment.

If a web server was queried with an ARK, it should return the resource itself or some surrogate for it, such as “a table of contents instead of a large complex document.” If a question mark was appended to the ARK, though, it should return a description—metadata—instead, which “must at minimum answer the who, what, when, and why questions concern an expression of the object.” (The scheme also included a guide to Electronic Resource Citations, a simple format for structuring this metadata.)

[2] A complete NAAN registry[6] is maintained by the ARK Alliance and replicated at the Bibliothèque Nationale de France and the US National Library of Medicine.