Open Office has a far richer format than HTML allows (e.g. "this is a footnote", "this footer on every page, 2cm from the bottom", ...). Of course, one could extend HTML to handle all that, but just starting with XML is cleaner.

HTML made for continuous display and re-flowing for different sizes.

ODF is for publishing, i.e. where you can have separate pages, with footers, margins, absolute position inside page (versus whole document), corrections tracking, authorship, figures.

HTML _can_ give you close result, but this is not HTML domain, hence separate standard.

Really, what we need to adopt en-mass is Tex.

All sorts of things Tex doesn't handle.

Embedded editable files, for instance.

Or markup for tracking changes.

Tex might make a good starting point for building a system that _did_ do all of that, but nobody seems to be working on that.

> Or markup for tracking changes.

I've often wondered why Microsoft chose to embed a weak implementation of version control into their mutable document format instead of doing something like passing around a signed chain of immutable documents.

The people that made change tracking probably wanted the feature to work and be usable, rather than create another crypto product and have no one figure it out.

Also, if you don't trust the other parties, you can use the compare feature an get a diff of the documents. Keeping your own copy is a easy, simple, way to get a correct view of changes to a document.

Right, so, "signed" in the same sense that a chain of git commits are signed. In other words, resiliency not trust, and full history instead of whatever Track Changes provides.

Sure, and Microsoft provides document signing/rights management, as well, if you can get the supporting infrastructure. Office apps also have built in change revision history. Track Changes is sorta orthogonal here, I think.

Because I don't want a list of separate changes - I want a series of notes _inside_ the document, colour coded for the editors, marked as accepted/rejected, etc.

You can do a mediocre job of inferring the latter from the former, but for 100% support you need to actually put the notes into the file itself.

That seems like a bad idea. TeX doesn't allow separation between markup and presentation very well. Something like DocBook[1] with a presentation engine like XSLT might be better, if it wasn't for the verbosity of the XML source.

[1] http://www.docbook.org/whatis

+1.

Any representation that makes people think "this is a paragraph", "this is a title", "this is a list" and then apply styles to each is an improvement over the current "lets vomit text on a page, then tweak individual sections independently" method endorsed by most word processors.

Tools that do this like (Pandoc being the one I like) do a great job of getting people to think of documents as text that gets styled later.

I won't even acknowledge a document unless it's hand-written postscript.