No time for an exegesis. Just some rough notes to further tease out some details first presented in Dumb eBooks Must Die, Smart eBooks Must Live
Static ePub is dead. Accept it or embrace the death of current print publishing. There is no other choice other than smart eBooks.
Levels of metadata
– Level 1: Can be incorporated quickly into an eBook
— general standardization
– Level 2: Requires copyeditor-like specialty to extract from manuscript
— manual, helps to create metadata categories
— at some point semi-automated, “drag & drop”-like
– Level 3: Queries to writer(s)
— manual, extract enhanced meaning from author(s)
– Level 4: Deeper metadata via metadata specialists
— manual, with some automation (query of metadata databases)
– Level 5: Coarse connections to outside metadata
— automated with manual refinement; semantic web meshing
– Level 6: wikimetadata, incorporation of metadata from readers/outside specialists
— manual, creates an additional and participatory experience (WikiPedia)
At least 3 levels of metadata can be included in published books at the start.
Levels 5 and 6 transform the static object of “book” into a ticket to further experience. This fundamental point would justify higher eBook pricing.
Since not all metadata is included — and really can’t all be included, ever — the book remains open-ended, living, and a portal to further, growing, and dynamic information. The value of the book actually increases over the course of its ownership.
People would understand they are not buying a “thing” any longer. Instead of regarding the thickness or thinness of a paper object to judge its monetary value/price, people will judge the amount of intelligence in the book.
An illustration of some of the connectedness that’s possible between books will be shown when I finally get around to doing a series of posts about several books I’ve read recently.
Final thought: Which would people prefer to pay for today? A printed map? Or a live, connected Google Map with dynamic/customized points of interest overlaid as well as intelligence?
Each contingency branched at several places; she learned them all until she could close her eyes and see the entire great structure, decision tree after decision tree branching and rebranching, dozens of them. As new data came to her from hard-copies or from Sandaleros, she mentally redrew every affected branch. For each decision point she assigned a text from the Quran or, if there were conflicting possible applications, more than one text. When she could see the enormous balanced whole spread out behind her closed lids, she opened her eyes and taught herself to see it in three dimensions within the cell, filing the space, palpable growing branches like the tree of life itself.
— Beggars from Spain by Nancy Kress
Links:
Metadata will Rule the World
Rushdie’s Stream library & Borges’ Print library
The Book of Sand
Experimental DML over digital repositories in Japan
The eBook Test 
Since not all metadata is included — and really can't all be included, ever — the book remains open-ended, living, and a portal to further, growing, and dynamic information. The value of the book actually increases over the course of its ownership.
In lit courses, I can see assignments given out to extract/parse/create new metadata from each student's reading experience and adding his/her name to the metadata.
Mmmm, reader-experience-as-metadata.
This is all fine but has two drawbacks for fiction.
1. I write novels. In doing so I generate a lot of metadata but I don't want to include it in the book – for aesthetic reasons and because it would be tedious. As for the semantic web, life is way too short to structure a novel like that! If third parties want to add value, or this could be crowdsourced, as you suggest, fine, but
2. Fiction with separable metadata always seems really crappy to me. Like those post-modern novels that are 25% footnotes. Yeuk. Mostly, you read the metadata and think 'so what?' The reason being that, if the information is integral to the enjoyment of the novel, the author has already included it in the text.
>>>This is all fine but has two drawbacks for fiction.
Including metadata would be a function of the publisher (which is now why most self-publishing is doomed to FAIL).
You also need to broaden your view of metadata and seriously consider how it would enhance the value of your book.
It almost sounds like what you're saying is that the "book" format as we know it is evolving into an entirely different and semantic web medium, which sounds somewhat plausible.
Although take into consideration that ultimately it's human beings and our biological interface — not technology that limits what's possible with any given medium.