As for why OCR is used, just look at the range of books that are now available electronically. There are a lot of books from the 1980s and earlier that are now available as EPUBs. Do you really think that it would have been easier to use the source files off floppies? (Keep in mind that the floppies are, for the most part, long gone.)
How are reprints of such books achieved?
How about effectively dealing with ligatures?
I know that people like to harp on and on about how great LaTeX or InDesign are, and when it comes to complex layouts (particularly LaTeX if there are lots of referenced figures and InDesign for text wrapping, etc.), I agree. If you were writing a scientific article, I'd tell you to use LaTeX. If you were creating a magazine, I'd tell you to use InDesign.
But for your typical, text-only, fiction novel, the kind you're most likely to read on a Kindle? I don't think it really makes a difference.
I guess they could generate PDFs from Word files with Acrobat and shoehorn them into the process, but any publisher that is actually satisfied with doing that is bonkers.
(And even if they weren't, I would appreciate such examples.)
People make a big deal about how using Word will lock you into a certain format when they're storing their data on floppies or, even worse, rasterizing it and keeping just the PS. Seems like the latter is a lot worse than the former.
So they're not willing to provide the original source files, or what?
> edits late in the process are made to the typeset InDesign files, after the structure has been lost.
Wow, they do this so haphazardly that they don't even bother to go back to the original Word file and add the same edits?
Is InDesign used for all books? I don't see why you would need it for something without any figures/graphs/tables. You could just use Word, and things would work out fine.