David Gill asked what we might want for Web 3.0. For my part, it's more stability, meaning widespread implementation and adoption of stable references that we can all use in the ongoing conversation that we are having on our blogs and in our digitally-implemented scholarship.
On this blog, I have already made reference to Worldcat.org as one kind of stable reference that we can all point to when citing a book. In doing so I praised librarians. A task that librarians have not taken on via their catalogs is the indexing of the contents of their holdings. Not being a trained librarian, I don't want to misspeak but it is my understanding that the discipline has had an historic focus on the cataloging of physical objects - books, journal volumes, etc. - rather than on cataloging the discrete bibliographic entities within those physical objects. I.e., librarians don't separately describe the articles in a particular journal volume.
In a networked world, it's the individual article that we care about, rather than its appearance in a particular physical volume. The Handle.net system is becoming a widely adopted standard for making unambiguous and stable references to individual digital objects, such as published journal articles. For example, http://hdl.handle.net/10.2972/hesp.75.4.453 is a stable reference to an article in Hesperia 75.4 (2006). Of course, you have to have an affiliation with a subscribing institution or pay money as an individual to read this article, but that observation is part of the conversation initiated on this blog by Chuck Jones' earlier post. Charles Watkinson, listed as an AWBG Contributor, has shepherded Hesperia into this world of stable references and that is a good thing.
It can be a hassle to find Handles for particular articles but I have had reasonable luck using Crossref.org's DOI lookup form. Unfortunately, it is still the case that the vast majority of published articles have no widely used and computationally actionable identifiers.
Another form of stable reference is the geographic entities being defined by the Pleiades Project. Here, the link http://pleiades.stoa.org/places/639166 is a stable reference to the record for the site of Xanthos in Lycia. Pleiades has a sophisticated data model which I encourage readers to explore, but let me just say for now that the project has established concisely composed URLs, such as that for Xanthos, that stand as usable infrastructure in a network-based historical geography of the ancient world. Tom Elliot, again an AWBG Contributor, has posted about the form and longevity of Pleiades URLs at his Horothesia blog.
For texts, I look forward to the widespread adoption of the Canonical Text Services (CTS) protocol, which is the work of Neel Smith and Chris Blackwell. CTS defines itself as "a network service for identifying texts and for retrieving fragments of texts by canonical reference". In this scheme, "urn:cts:greekLit:tlg0012.tlg001" is an unambiguous reference to the Iliad that leverages the existing Thesaurus Linguae Graecae Canon of Greek authors.
Why does reference stability matter? I suspect that for many readers of this blog this is not a necessary question. We're interested in substantive conversations about the ancient world and the analogy between linking and source citation is well understood. I am interested in the extent to which stable references will make conversations more discoverable. The use of links into Worldcat.org should make it more likely that I can find your reference to a particular book. Right now, Google does not do a good job of enabling such discovery. The link http://www.google.com/search?q=link%3Ahttp%3A%2F%2Fworldcat.org%2Foclc%2F144767610 does not return the blog post that made reference to this book. I do hope that in the near future the use of stable references will help bind together discrete contributions into a continuous conversation about the ancient world.