LibraryThing is an online service to help people catalog their books easily. Because everyone catalogs together, you can also use LibraryThing to find people with similar libraries, get suggestions from people with your tastes and so forth.
It's these guys "I'm gonna make a lot of money" scheme that runs on the idea that lots of people will use it to catalog their books (whatever they mean by 'catalog') and, in the social-tagging free-love world, eventually there will be this omnicatalog like wikipedia.
Of course, it doesn't really rely on all those book lovers for metadata. It harvests Amazon.com metadata for current and poplular works and for older, more obscure writings it takes full advantage of all the millions of staff hours the Library of Congress has spent over the years cataloging everything that moves. By the way JH Libraries catalog is also listed as a source for data.
The interface is maddenly clunky. (Who would have thought Horizon could look so good in comparison!) There is a lot of back and forth and screen re-freshing. Importing is whacky--I tried to import my wishlist from Amazon and got a bunch of books I would never read. Maybe I got part of someone else's wishlist. Exporting doesn't follow any bibliographic formatting standards (e.g., RIS). So how am I going to compose a references cited list for my term paper?
That said, it's an interesting idea for non-scholarly work. Has the shared tagging thing going on that creates something resembling a 'people who like what you like also like' world and there is an interesting API called thingISBN (like OCLCs xISBN) which is a kind of FRBR-izing thing -- see the 14 June entry for http://www.librarything.com/thingology/.
I'd be interested in hearing what Scrapplers think of it.