I think the original poster's email also raises some questions about what a library catalog is for. To my mind, it's meant to be an inventory of what a particular library holds. The holding is the crucial element, and the extensive bibliographic data serves a discovery purpose. I think the web enables libraries to let go of the discovery aspect: users find out if they want to read a particular book (article, DVD, manuscript, whatever) from sources other than the catalog, and the catalog exists to tell them if their library has it. What I think linked data can enable for libraries is to move away from extensive description of our resources (once that has been done by someone else, perhaps in a centralized linked data store, ahem LoC/OCLC), and to focus on linking our holding, our unique copy, to other discovery sources and other descriptions of the resources. I think it can enable an even more efficient version of copy cataloging.
Because let's be honest: If I'm looking for information, I'm going to look on the web, not on the library website. And once I've found the thing I want to read, all I need to know is how I can get it. I imagine that using linked data to link our holdings to other representations of a work on the open web could get the information users really care about--whether we have the thing they want--into the search engine representation of the thing.
I also think this would enable us to spend more of our resources describing and providing access to our unique collections, which is another whole ball of wax, with different uses for and types of linked data creation.