The purpose is to identify the most popular books on the planet.
A person takes a photo(s) of their book shelf and uploads it into an online database (anonymously if so desired). Text recognition software reads the book titles on the spines and enters them into a database.
EDIT: Apparently the rest of this is already fully baked... so my invention ends with the Scanning of the Bookshelf.
A person could search for the most popular books. Or, the most popular books containing a keyword ("auto", "cooking", "zebra" etc). One could also use this for recommended books - 150832 people who read "the black stallion" also read "black beauty".
Cross referencing with other databases can add meta-tags to books ("fiction", "non-fiction", "space travel" etc.) - see link for halfbaked database idea (minus bookshelf component)
In recent years, similar information can be gathered from book store sales data - but computerized information of older book sales does not exist. Also, discarded books (firewood) will be eliminated from this database.-- knowtion, Mar 15 2009 Another Database Idea Internet_20Book_20Database [knowtion, Mar 15 2009] Delicious Library http://www.delicious-monster.com/Book/Media collection management software, with webcam-based identification [placid_turmoil, Mar 16 2009] (?) LibraryThing www.librarything.comList and review your books [colorclocks, Mar 16 2009] I like it - I've a feeling there's may have been something discussed here before to to with analysing the contents of a photo of someone's bookshelves - and we must be on the cusp of it being an available web-utility (upload your bookcase, and we'll tell you what you've got type deal) which would attract millions.-- zen_tom, Mar 16 2009 ...my bookcases always have a lot of dust...-- xandram, Mar 16 2009 Baked. LibraryThing doesn't have the text recognition from photo thing, but it has everything else.-- colorclocks, Mar 16 2009 random, halfbakery