If a book predates computers, I can understand OCR, but not for something that's been written in the past ~20 years and is popular enough to not be "lost".

Wrong.

Prior to 2000, the MS may never have been submitted electronically; or might be in an obsolete format (Word Perfect, anyone?).Then, until circa 2008, the book was almost certainly copy-edited on paper -- that is, the editor printed out a paper copy and the copy-editor corrected typos by hand, which where then manually transcribed on a DTP system to produce the final output. The author therefore doesn't have an as-published electronic copy.

To make matters worse, all the big publishers outsourced copy-editing, typesetting, and printing many years ago. The typesetting was probably executed using Quark Publishing System on MacOS by an external agency, who then burned any backups on floppy disk or CDROM. The publishers do not own these, and in fact cannot acquire them without paying the typesetting bureau a three (or four) digit copying fee. So neither the authors nor the publishers own an as-published copy.

These days stuff is typeset on Adobe InDesign, which is a whole lot more ePub-friendly, and which can import Quark files ... except that Quark's format is notoriously idiosyncratic and import ops commonly lose some formatting info. (Such as italics, font changes, etc.)

TL:DR is that any book published before 2007 or thereabouts may be impossible to republish without either OCR or re-typesetting from scratch.

(TeX is pretty much unheard of among authors outside the science fields, and thanks to M$ churning the MS Word file format repeatedly between 1990 and 2008 it may be difficult to do anything with the original manuscript.)

Going forward the picture is brighter: I have as-published ebook editions of all my books and know how to crack the DRM on them to pull that text out in a legible form, with formatting intact. (Yes, I appreciate the irony of this. Why, just last week I emailed three DRM-cracked novels I downloaded off BitTorrent to an editor. I wrote 'em, the editor has a license to publish them in the UK, so it's legal, but ... the irony! It burns!)

Publishers are as incompetent as anyone else. At ICON this year I got to go to a reading by David Weber (of _Honor Harrington_ etc), and he talked a little about dealing with a publisher and gave 2 examples of global search-and-replaces gone terribly wrong - Barons -> Earls, and Lords -> Knights, which resulted in him proofreading the entire books just to fix the S&R. Turns out publishers apparently don't use revision control either.

(Well, most of them. I think Tim O'Reilly might argue the toss :)

Authors of novels are even worse. They are in a career that typically doesn't get started until they're in their mid-thirties. So unless they got their start in IT/CS, you're looking at J. Random Middle-Aged Non-Techie.

Some of us know what a revision control system is. Most haven't got a clue, and aren't interested.

I'd bet a lot of them have seen the RCS-like abilities of Time Machine or Dropbox and been delighted or would react with delight if it those were demonstrated to them. A lot of them still wouldn't have a clue nor be interested, though.

If people have built content management systems for publishers like newspapers, someone must have thought of CMS for other publishers with roots in the dead-tree business.

Newspapers and magazines need CMSs because they may need to pull or revise stories. Novels ... not so much. Yes, it's good to fix typos. In the world of paper publishing the author gets just one shot at that: the book goes back to be re-flowed for a different page size when it goes from hardcover into A-format paperback, and there's an opportunity to fix snags.

This is, obviously, going to change. In fact, I expect the bigger publishers to wake up real soon now and realize that not only do they have to do e-pub related things like drop DRM and fix the broken territorial rights problem, they also need to do internal process-related things -- like set up a CMS, in-source their typesetting, and modify their workflow to allow updates to existing-and-out-there books, rather than viewing the books as a terminal state at the end of a production process.

(Reasons why it should change? Well, for one thing it makes possible the issue of "director's cut" versions of a book. Yes, crass commercialism. But it also makes it possible to retarget books on a different audience. It makes it possible to rebundle in-series books in an omnibus edition. It makes it easier for the publisher to directly issue extracts from the book via their website, for promo purposes. Once layout/typesetting stops being a single step and becomes an ongoing process, there's every reason for in-housing it. And that, in turn, will require revision control and archival functions that publishers are currently unfamiliar with.)

they also need to do internal process-related things -- like set up a CMS, in-source their typesetting, and modify their workflow to allow updates to existing-and-out-there books, rather than viewing the books as a terminal state at the end of a production process.

I think there's a 3-way non zero-sum opportunity (win-win-win) lurking here if one can get away from the idea that the unit of production is a book as a lump of dead-tree or even as a particular string of bits or even as a stream of versions of a particular string of bits. Instead of focusing on curating individual books, wouldn't everyone win by developing CMS for curating collections of things, like author bibliographies?

In a way, this is what disparate parts of the industry already do. (Like this book by C. Stross? Why not try this other book by C. Stross? Like this book that's supposed to be "Hard Sci-Fi?" Why not read this other book...?) They're already trying to get customers to think of their brand as a good source of [some classification]. They're also motivated to get customers to think of an author as a brand as a good source of [some classification]. This is also how publishers and editors approach the brand represented by anthology series. Really, publishing at its core is the curation of a bibliographies. Customers and authors would benefit from such directed curation. (On the other hand, this might also perpetuate restrictive terms between authors and publishers.)

(As I get older, I am becoming more sympathetic to this: paper stays put when you take your eyes off it, while bits do not.)

Besides, how difficult can it be? Find a competent English major, fuel him/her with some red bull, pizza and bit of cash and they'll fix it. Hell, if they had a mechanism to do so the fans of these books (Dune etc) would probably fix it for free.

This happens all the time. DVDs cost less to produce than VHS but sold for more because customers were willing to pay. TicketMaster can charge its service fee because customers will pay, even though it's certainly lower cost for TM to email PDFs than to staff a physical box office and buy ticket printers. A smart capitalist captures the consumer value he creates, regardless of his own costs.

This might be true if people were rational utility-maximisers. They're not, and people are quite capable of choosing not to buy something if they feel it's a rip off.

Ticketmaster is an effective monopoly, publishers are an effective oligopoly. Neither are exactly capitalism's finest examples.

I was talking more in terms of a bigger picture view: In a competitive market with high net profit margins more competitors should enter the market & undercut the existing prices. Mean that in the long run, the sales price of an item should approach the cost price. Now producing & delivering an additional ebook costs perhaps 1 cent, while producing an additional paper book costs perhaps a dollar. So ebooks should be sold at 1 cent + minor contribution to fixed cost + near zero profit, while paper books should be sold at 1 dollar + minor contribution to fixed cost + near zero profit. i.e. ebooks should be dirty cheap compared to paper books.

Yet in practice ebooks are somehow maintaining their prices way beyond what one would expect per economic theory. Little surprise then that various publishers are currently in trouble for price-fixing of ebooks.

Courts work slowly though & in the mean time I'm still pissed about the price I'm paying for those (badly OCR'd) bits & bytes.

I see there is a recently published set I didn't see before containing all 5 books:

http://www.amazon.com/Song-Ice-Fire-Continues-ebook/dp/B007B...

Is that the one that is fine?

I didn't buy them as a set - I bought them one by one.

Hope that helps. Fun books.