« Social Good versus Economic Value? | Main | A Few Things Heard on the Way Out to Lunch »

March 10, 2006

User-defined book slices?

Tim O'Reilly's mention of building a digital economy and user-enhanced networked services got me wondering whether too many discussions of e-books are tied to the idea of exchanging book-sized objects. Once digitized, there's no reason a user should receive book-content they don't desire in order to use the portion that they do desire. Robust searching and retrieving of slices of books would require some great tagging effort, but why not let enthusiastic readers contribute those tags? A user interested enough to list literary references to "Kilroy was here", for instance, could also mark the spans of text which make those references.

Posted by jgmorse at March 10, 2006 01:20 PM

Comments

Tagging significant spans of text is exactly the approach used at the Early 19th-Century Russian Readership & Culture project ( http://www.library.uiuc.edu/spx/rusread/ , with the category search available at http://media.library.uiuc.edu/cgi/t/text/rusreadtc?c=rusreadtc;page=cat ). While small, the project is really the only I know of that attempts to provide access points to a level below the bibliographic item (using catalogers' jargon).

Posted by: kshawkin at March 10, 2006 01:31 PM

"... there's no reason a user should receive book-content they don't desire in order to use the portion that they do desire ..."

I think one possible candidate for such a reason is that the user may not realize that the portion of the book they don't desire is, in fact, desirable to them. Which I don't mean to be facetious, just to emphasize that while an entire 'book' is likely an overly large context, it's not trivial to define size of the spans of text in the example:

"A user interested enough to list literary references to "Kilroy was here", for instance, could also mark the spans of text which make those references."

One desired functionality of any tool making use of such tags would be a way to easily 'zoom-out' to some incrementally wider view than the tagged span, going on up to the full text of the book. It then becomes an interesting question: what size should the 'default' context be (easy answer, whatever the user wants it to be)?

What kshawkin is suggesting is, I think, the right direction. Rather than try to create a 'system' which has 'intelligence' only to the extent that the algorithms designed run it have 'intelligence' we let the intelligence of the whole system accumulate via input from its users. This is the spirit of the various Wiki intiatives (which kshawkin's example links to) as well as enterprises like Craigslist. It'd be nice, for example, if a text, once tagged, could be re-tagged in a different way by a user if that user found the tagging to include too short or too long a span, and you could, Wiki-style, revert to a previous system state in the event that vandals messed about with it. I think this basic approach to book tagging contrasts rather sharply with Google's algorithmic approach, which has certain obvious merits as applied to web content.


Posted by: homelessdave@homelessdave.com at March 10, 2006 02:12 PM

Login to leave a comment. Create a new account.