on the semweb, the geoweb, and the media lab

I enjoyed giving a braindump yesterday to the placemap workshop that Matthew Hockenberry is running at the MIT media lab. I decided to talk without slides or notes, but wing it using a del.icio.us linkdump made for the placemap workshop; this turned out to be a good decision, as my laptop hard drive died horribly an hour before setting off to give the presentation. I'm back in flatland again, on a third-hand old Thinkpad.

I wasn't really in visionary mode yesterday, but that wasn't such a bad thing either; i think the media lab has no shortage of visionaries with interesting ideas, but could perhaps benefit from a bit more connection to open source and open standards best practise. I emphasised these themes as i've been thinking about them a lot recently: the night before i'd written a draft attempt to explain the issues behind building a Spatial Data Infrastructure in terms that a bright 10-year-old could understand, for inclusion in an upcoming Wsfii.London publication and handbook.

Matthew's email roundup of the workshop/talk themes entertained me: There seems to be quite a bit of debate between those of us who believe in the Semantic Web and those who don't, but this sort of academic controversy isn't a bad thing. It seems terribly pertinent today, because Zack Rosen's rather harsh words about how the semantic web research community is failing the big vision have been doing the rounds, and they cut rather deep for me.

I met RDF for the first time in summer 2002, and it instantly grabbed hold of my brain and distorted its shape. The fit between RDF concepts and structures, spatial applications, and my previous experiments with messaging bots and interest in wireless networking, was a new and startlingly complete shape.

Between then and now, i've spent most of my time and code energy chasing that vision, wrestling with impedance mismatches in the tools, rewriting the same ur-application about 10 times over as the tools mutated and got better, other applications and data sources appeared. RDF made everything a lot easier for me for a long time; on each iteration, there would be a tiny web service interface left over from the last one, which would save me another loop of code: just point at this URL, learn the graph, and keep going; no new formats to think about, no parsers to modify. I haven't been able to bring myself to use a SQL store other than as a parallel stream to do spatial queries over an RDF store, since; rigid tables and relationships just don't fit my mental model.

At the core of my motivation was the feeling: here is an incredibly potent new set of tools and techniques for associating and making inferences from different kinds of information about people, things, places, interactions over time: it seems like those tools are going to inevitably exist; and i want to do what i can to make sure they are put into everyone's hands who would like to use them.

I put a lot of energy into infecting London's media arts scene with the "gonzo collaborative mapping on the semantic web" meme. The endlessly mutable and wonderful UO text on The Freemasons of the Future was influenced by my enthusiastic campaign to bootstrap London's semantic web, and in a sense to ensure that We can all become the freemasons of the future, if we want.

nodel, its companion software bbox, and the Soft Architecture which they were developed to help support, was my last attempt at clinching a solution to this problem: a semantic web spatial application toolkit which had translator gateways to talk to many different web-addressable APIs on many different services that do one simple thing well. I made nodel as a potential answer to some of the questions that Zack Rosen is asking of the academic RDF hardcore now.

RDF is used in nodel as a bridge and as a brain; some projects like OpenGuides output it directly; others like EVNT use custom formats which can be very easily translated into and out of an RDF graph model, which then is used to multiplex a lot of different data sources together. The dullest part of the work in dealing with, for example, the myriad slight format variations of RSS feeds, can be obviated by translating them all into the same underlying model, bypassing the need to think about how differently formatted data structures are stored differently; they're all reducible into a common concept graph.

I've found it can be very hard to communicate the small core of a big vision. I stalled development work on nodel after it reached a usable alpha because of the biggest impedance mismatch of all; my own lack of skill in thinking and working in a visually effective way on the applications i build; and no resource to spend on working the vision into the brain of a good UI designer. (UI designers who are prepared to do open source, freetime work seem impossibly hard to come by). I've taken a break from it for the last few months, and spent most of that time very enjoyably reading urban planning classics and making soft furnishings, but i hope to find energy to go back and file off its sharp or broken edges, in another couple of months. (and shift it from using the cheetah templating system to the kid one... but that's a rather dull story).

The RDF-driven semantic web gave me the opportunity to collaborate-without-really-trying with a lot of brilliant people. It led me to meet Schuyler and Rich and co-create Mapping Hacks with contributions and inspirations from a lot of those brilliant people. So many of the techniques and stances that are now taken for granted by those people trying to transcend the commercial view of 'Web 2.0' - information mashups, navigation through social networks, wiki-nature geoannotation - came directly from experimental work on RDF and the first couple of waves of interest in the Semantic Web Thing.

I wish technology wasn't so full of "religious wars". I wish the funding structures of arts and academic research projects weren't so biased towards moving on to the "next thing", to pushing the envelope rather than posting the envelope. But mostly, i wish these ad-hominem attack stances towards different sets of 'innovators' with different strategies for building, didn't feel so bloody personal.

-- Jo
[]