August markup conferences always leave my brain feeling a bit bubbly. Or maybe it's just melted. As expected, Balisage offered amazing food for thought that should last until next year's event.
One of the fun things about this conference and its predecessors is that the themes that emerge over the course of the conference are rarely the ones I expected when I first explored the program. Threads from individual sessions mingle into a richer conversation. Despite the incredible variety of tasks people try to solve with markup, we can still talk to each other amidst that diversity. The speaker's podium, conversations over meals and drinks, and conversations in the hallways all mingle into complicated but exciting story.
Hugh Cayless explains
applying SVG tools to papyrus
Teasing out some of the threads, there are some common themes. I can't say all of them promise bright sunny futures, but nonetheless they point to useful work.
Humans matter. One of my greatest concerns about this conference is that a lot of past talks have emphasized automating knowledge processing to pull humans as far out of the loop as possible, while including footnotes that this is, you know, difficult. This year there were a lot more talks emphasizing how human decisions and interactions affect this work - and not just in the design stage. Jerome McDonough's "sociotechnical" look at how humans behave (or don't) with markup in the library world was especially enlightening in this regard, something I wish everyone creating an XML vocabulary would get to see.
Human interfaces matter. I enjoyed a set of talks in the first two days which tooks a variety of different approaches to bringing people into describing and managing data.
We're not nearly done writing tools, even after a decade. There was an explosion of tools in the early XML days, and then a retreat to a smaller set of tools. However, we're still missing a lot of parts, even some basic ones. Norman Smith explored the many good reasons for writing your own parser, and what's involved. Mohamed Zergaoui explored the state of streaming processing technology and while what he presented was great, it made me wonder about just how long it's taken us to get here. Also in the "I wish I'd had that long ago" category was David Lee's work on creating an XML-oriented shell, xmlsh. If only I'd had that available before.... wow! At the end of the show, Liam Quin's talk on his nearly 20-year-old lq-text package reminded me of how much had come before - but also how there are many more pieces needed.
After years in which the W3C XML Schema community pitched its work as not that painful, the talks which used it either were amazing rocket science aimed at dealing with its difficult aspects, or amazing rocket science integrating it with other technologies to go far beyond its bounds. Balisage definitely pitches itself as high-end challenging stuff, but these two talks were remarkable and brain-melting. On a slightly lighter note, Syd Bauman explored how different approaches to schemas can fit different situations, something that often gets lost in the rush.
The other traditional mind-melting subject, knowledge representation, mostly in Topic Maps and RDF, was here in force. Pierre Levy gave a mind-expanding talk on a "Semantic Space" abstraction that goes beyond either Topic Maps or RDF in its scope, though I'll admit that I'm even more skeptical of that than the rest of what was shown. Christo Dichev explored possibilities for translating between Topic Maps and RDF, while Peter Brown asked hard questions about context in both human and machine contexts.
Sadly, this year I missed the overlap talks, but it's good that the conversations on that obscure but necessary subject continues.
Standards work is sometimes insane and often broken, but somehow the participants still find enough value to continue working on them.