This highlights a problem with the Web: the fusion of document identifiers and the channel through which the (or "an") information/service provider makes it available.

Suppose stakeholder S wants to mirror a document D (published in an HTML-based format) originally distributed by authority A. This will probably involve having to touch D in order to rewrite links to point at S's copies of auxiliary resources instead of A's originals. Simply copying the representation's bytestream is not sufficient.