The XML Paradox

I have been working on my tutorial for the O'Reilly Tools of Change conference. I'm presenting PDF as a cost-effective option to create revenue from the the backlist as an alternative to XML. As a dedicated markup advocate from the days of SGML, and someone who helped simplify SGML down to XML, I still find it odd to be talking about other kinds of solutions, but I think I learned something from my custom web site customers... The XML Paradox is that XML is a high-quality archival medium, and obviously then, books and scholarly content would make the jump first. It just makes sense that everyone would use the high-value format for the longest-lived, highest value content. Wrong! The economics of publishing have played out the opposite way. The more ephemeral the content, the faster production methods can change. So newspapers were doing full-text databases from very early on. In the scholarly markets, journals are now almost all electronic. Books, however, are only starting to move fitfully in the XML direction, and are mostly not digital at all. So the least archivable stuff, moves to the best archival format fastest — because serial content does not have a legacy that needs conversion to make a new channel profitable, so the payoff from a production change can be pretty fast. A publisher with a rich backfile has items that can earn for 20 years or more — as long as costs can be controlled. So any change to the book production process has to pay off immediately on new books. And for any large-scale change across a publisher's line to be successful, it must be very cheap for old books. And that's where e-books stand, revenue unearned because there's not a clear path to get it. XML is great, and enables the production of an optimized presentation for a new media format, but it's not cheap at all. It's an expensive and tricky management challenge to change editorial production processes for new content, and data-conversion costs for old content are very high. Once the data is in hand, the development cost to create a new output format (print, web, handheld, or whatever) is not cheap either. Problems like typesetting, layout and display all have to be solved anew for each output format. It takes work to optimize presentation, especially from the level of abstraction gives good XML that power. So page images (and especially PDF) get a big boost from the XML paradox because they capture a lot of the production value of the existing process and they're the cheapest searchable format to produce from paper. So here I am, a guy who courted his wife over conversations about markup, working with page images. We are managing them with very rich metadata at a fine level, to capture much of the commercial benefit of XML, but still, I'm enabling something I used to rail against. And it's not easy to make page images work over the web, let publishers control the presentation, and still be good to readers. In this discussion I am leaving out the small number of crown-jewel properties that earn large amounts quickly in a new channel, and thus merit technology investment — Projects like that are important, but don't shift the business as a whole. And their emphasis on frequent updates makes them similar to serials in the need for continuous editorial management. Coming soon: I used to think that page scanning projects were a waste of money in terms of long-term investment, and I hope to post soon about why I no longer believe that either.