Publishing and the Open Web Platform

W3C OWP Workflow workshop. Fascinating. Depressing. Inspiring. Far too much to talk about. Got to talk to loads of interesting people, finally met some people I've been discussing things with online (finally meeting Dave Cramer from Hachette and Romain Deltour was good). Thanks so much to Liam Quin.

Adam Witwer talked about O'Reilly's migration through XML first to HTML first and the open sourcing of HTMLBook. Fascinating stuff but I'm not so convinced that the lessons that O'Reilly have learned apply outside of their very special author/publisher community. Having said that some of it makes sense to me — for simple content (anything that flows without needing spreads or any other specialised layout), the O'Reilly approach should work. He previewed Atlas, O'Reilly's integrated, web based authoring and compositing tool. Adam suggested later that you could pull authors away from Word by giving them more — additional features like collaboration).

Atlas is an interesting approach. There are some things that are only going to work with the O'Reilly author community — github diffs and comments can't replace revisions because, in the general case, these are different concepts. As Liam pointed out the author wants to know that the editor dropped the heading level:

<h1>Publishing with HTML</h1>
<h2>Publishing with HTML</h2>

github's diffs aren't going tell her that the heading level changed; they're going to tell her that two '1's were changed to '2's. Perhaps the W3C Change Tracking Markup Community Group could help here?


The keynotes were followed by three panel discussions. First, a discussion of using HTML5 as it is for publishing. Lots of interesting information. It's pretty clear that the official future is CSS print. Also pretty clear that so much is missing from CSS print that it's not usable without serious vendor extensions. Most of the W3C folk there made it clear that CSS for print is in most cases utterly underspecified. There is a desperate need for users (publishers, that's you) to join in.

I was there giving my spiel about authoring and workflow (editing in this case). Much Word mocking happened over the two days and the fundamental disconnect seems to be that web people and techies hate Word for perfectly understandable reasons and the publishing folk like Word for perfectly understandable reasons (these are often the same reasons). You can't take away someone's toys (let alone the tools of their profession) just because you don't like them. We need to have something better to offer. I'm just not convinced that Atlas is it for the average author, editor or publisher.

Dave Cramer and Phil Madans discussed the Hachette HTML/CSS driven publishing workflow. This sounds absolutely amazing although pretty well limited (right now) to the standard black & white text (with funky extra bits). They've not yet really pushed the envelope with it. Just like O'Reilly they're heavily dependent on vendor extensions to CSS to get something close to good typography.

Gerry Leonidas pointed out that we are expecting programmers to be able to do what a typographer does after a five year apprenticeship. I think that's slightly wrong — we are expecting programmers to provide the tools that the typographer needs.

We came up with the a few things that needed to be done. I think the most important was — reach out to the publishing industry. Todd Carpenter pointed out (rather forcefully) that this is bloody difficult. He suggested various publishing conferences. That will work fine for the Americans (perhaps) but not for others (like UK publishers).

The W3C needs input from the publishing industry if the OWP is to support professional publishing fully. How does the W3C get the publishing industry to become involved when much of the industry doesn't even realise that becoming involved is possible let alone desirable? Hachette's workflow (and O'Reilly's) are likely to become the model that the W3C work from if no one else says anything. Right now, I don't see how to even convince people that they might want to say something. How do we solve this? Answers on a postcard.