The Library of Congress recently posted a webcast: Library of Congress New Bibliographic Framework Initiative: Update Forum with Eric Miller. This is LC's initiative to develop a replacement for the MARC format.
The new bibliographic framework transition initiative will be focused on the Web environment, Linked Data principles and mechanisms, and the Resource Description Framework (RDF) as a basic data model. Zepheira, led by Eric Miller, has been active in the development of Semantic Web and library standards as well as open source tools to support linked data technologies and library applications. Zepheira will provide both a model (or models) to serve as a starting point for discussion and an analysis of related initiatives. Miller spoke to Library of Congress staff with an overview of the project and a report of progress to date.
In this webcast, Eric Miller describes the transition away from MARC as an evolution, not a revolution. It's a very interesting talk on how Zepheira is developing a linked data model that "deconstructs" and repurposes the valuable entities, subjects, etc. that are already contained in MARC records.
My persistent question, which I haven't seen an answer to yet is: Why at such an early stage of data modeling are we narrowing in on Linked Data? Shouldn't we be looking at a variety of models for a MARC replacement?