Publishing often seems to be a mysterious process. Content goes in, books and a whole lot more come out later. What happens in the middle is a complex brew of human intervention and automated processing - generally opaque to readers, and even authors.
O'Reilly is cranking open the doors a bit again, re-opening http://labs.oreilly.com with a mix of components that should excite publishing geeks. The three initial projects come from wildly different aspects of publishing, with one focused on reading ebooks, another on producing ebooks, and yet another taking on the complicated headache of presenting metadata for books in many formats.
I'm not sure how many people really want to play with book metadata, though I'll be putting this to immediate use in another application I'm writing. My broader hope is that it will demonstrate RDF and more specifically Dublin Core, hopefully inspiring more people to try those technologies. (I may even access the data as RDF instead of squashing it into XML, as I've done before.)
I'm guessing that the DocBook and XSL communities will enjoy the DocBook to ePub generator. DocBook may not solve every publishing problem, but it makes a lot of our problems simpler.
And Bookworm? That's for all of us to enjoy, I think.
Hopefully there will be a lot more coming here. I might even have to come up with something to contribute.