There are many dimensions to digital preservation. Typically we think about the individual objects, the digital bits that form the object and our ability to render, present or perform the object for a human to view or for another machine to use.
Another aspect is locating and retrieving the object. In file and archival systems a file address system is maintained. On the Web, URLs are the equivalent addresses. Here is a problem: within a managed file system it is possible to maintain consistency of addresses through changes and upgrades. The Web is a big open system where URLs are published but not always maintained, as we know through broken links. In such an open system those managing the object cannot manage the proliferation or the publication of its URL. What they can do is manage the connection to the original URL whatever changes may take place at the server end.
One of our colleagues on the KeepIt project, Chris Gutteridge, has strong views on maintaining, or preserving, URLs. Chris is lead developer of the EPrints repository software, so his views translate to practical application and affect systems running EPrints. In his inimitable style, Chris has recorded a video (6 May 2009) on the importance of preserving URLs.