If the books you want to discuss are still under active copyright (Stephen King, Cormac McCarthy, etc) instead of the public domain (e.g. Mark Twain, Shakespeare, etc), how do you envision the shared annotations feature to work?
For example, google books doesn't show pages 43-44 (and many other pages are missing) in Blood Meridian:
It seems that to create a website to fulfill your idea, we would need to have a blanket license to not only store 100% of the text digitally on the servers' harddrives but to also display any part of the complete text to all members so they can annotate it. A giant like Google Inc. was not able to get such terms from publishers.
I don't think you need copyright-free access to the whole book. Interesting quotes/passages are few and should be OK to provide under fair use (e.g. seehttp://onlinebooks.library.upenn.edu/fair use-explain.html).
Well, I wasn't limiting it to meme-friendly fragments such as "To be or not to be" from Hamlet. To go back to the premise mentioned by theswan, it was "good college-level literature course".
That means most of the text like Hamlet is discussed and annotated front to back. A literary guide such as Norton Critical edition of Hamlet will have annotations for every single line of the play. Hamlet is easy to digitize into RapGenius because it's 400 years old and public domain.[1],[2]
For a recent book still under copyright such as Twilight or Harry Potter, the rabid fans could conceivably want to discuss every page of the book. Therefore, a thousand fans "sharing annotations" leads to reconstructing the entire book. If the entire book isn't presented by the website to annotate, what exactly would they be annotating?
For literary and difficult books such as Ulysses by James Joyce, the entire book begs to be annotated. If a permissive license doesn't exist to present 100% of book for thousands of professors and students to share annotations, I'm not sure what the value is.
What do those stored annotations point to if members are providing the books?
This isn't RapGenius where all of the lyrics and annotations are shown on the same screen.
If the data structure of the stored annotations includes "pointers" to a specific book, what does it "point" to? A page number? Many epub/mobi books don't have absolute page numbers. For dead tree books, even the page numbers can change between the printing of the 1st hardback cover to the 2nd paperback edition.
When I think of "shared annotations", I'm thinking of virtual comments written in the margins of a book that anyone else can see. How would members "provide" books for that scenario? Upload epub & mobi files?
I guess it would be easier to quote (copy&paste) a particular passage a book and then follow up with some commentary but that type of thing can already be done today in any book discussion forum. To me, that's just quoting/citing and not annotating.
For epubs from the same source, epubcfi[1] can be used to point to a passage. It is somewhat resilient to editing, as long as the gross structure of the document remains the same. (Parent tags, file name, ids.) iBooks uses this standard for annotations, but doesn't expose it anywhere.[2] I don't know if any other book readers use epubcfi for annotations.
I think mobi files loose enough structure in translation that it wouldn't be possible to locate an epubcfi pointer within them. However, it might work with the newer "format 8" files, if they are converted from an epub. (I haven't investigated how much structure is lost.)
[2]: some records in my database don't have an epubcfi, but I haven't investigated whether they're user-created annotations or internal ones. (e.g. current location)
This could presumably be handled similar to how patch does it. When you make an annotation, you store an approximate location and say a hundred words of context. When someone loads your annotation, even if they have a slightly different edition, you can look for almost-the-same text, searching outwards from an appropriate point.
For example, google books doesn't show pages 43-44 (and many other pages are missing) in Blood Meridian:
http://books.google.com/books?id=s-QzccStux4C&lpg=PP1&dq=blo...
It seems that to create a website to fulfill your idea, we would need to have a blanket license to not only store 100% of the text digitally on the servers' harddrives but to also display any part of the complete text to all members so they can annotate it. A giant like Google Inc. was not able to get such terms from publishers.