Hypothesis: A good paradigm for proofing and marking up a book is an outline.
Several assumptions that help this to work.
1. Without any exceptions I can think of, any comprehensible printed
text can be completely, unambiguously outlined.
We know from experience it works. Any XML document, including
an XHTMLdocument, complies by definition. It's not just a good idea,
it's the law.
2. An outline is easy to define and easy to understand. Conceptually,
it's simply a regular hierarchical structure, with every syntactic
element completely embedded within another below a simple
sequential list of top-level elements.
3. Any syntactic element can be structurally identified as one
of three types.
a.) A section.
b.) A sequence of characters.
c.) A position offset from the start (of the text, and/or of an element.)
We know from experience that this works. Any HTML element can
be bound by one of only two types: a <div> or a <span>. What we
need to do is to associate logical divs and spans with syntax.
==================================================
Benefits:
A book that has been outlined is probably simultaneously
easier to build, to read, to comprehend, to verify visually, to verify
grammatically with software, and to transform into ebook markups
than any other format. And structurally, it's self-validating.
Low barrier to entry. Anyone can proof with confidence from the
start, with a brief introduction and a list of syntax elements.
==================================================
Proofing interface.
Notice that the proofing representation can be entirely separate from
the serialized representation - i.e. how it's stored in a file for instance.
What might it look like? We have lots of history for this - there are
not many ways to represent language that are more universal than
an outline. Almost all of us come pre-trained.
Say the convention is to start an element with a newline, a plus
sign, and a syntax tag, on a line by themselves. Paragraphs are so
common that they can just start with, say, two blank lines. An
element's content continues with indented content. An element
ends with the start of another element at the same indentation
level, two blank lines (another paragraph), or outdented content.
+chapter
+chapter-heading
Chapter The First
It was a dark and windy ...
I think I'll play with this a bit and see how far it goes. Is anyone
familiar with other attempts in this line?