Mapping Hacks

I set off for Where 2.0 with some trepidation, but it was largely unjustified; and it was chance to see many old friends, make a few new friends, and to see many of them doing well out of the conference. I left with my head spinning, and these notes are what came out the top of the mental centrifuge…

Gnat’s opening remarks stressed the importance of open access to data in the emerging geospatial web. “At its heart,” Gnat said, going straight to my heart, “it’s all about geodata… we need open standards for creating data so it is not locked into a particular vendor.” Open source geospatial software is beginning to be taken for granted; open data is the next step, currently suffering from a lot of the same FUD as free software once did (and was, until recently, still suffering in geospatial).

The buzz around GeoRSS continues so promisingly - Mikel bounced up at the first break to exclaim, “5 for 5 talks so far!”. GeoRSS has become the defacto standard of choice for getting structured data with a geographic component out onto the web, without the overhead and constraints of full GML. GeoRSS is where the geospatial web and the semantic web join together at the edges. It’s more or less perfect as it is; I’ve heard some talk about adding addressing and calendaring to it; this honestly needs to go into a different namespace. As more GeoRSS data starts to appear, the idea of what complex things can be built semiautomatically.

I recall hearing a lot of GeoRSS-potential-excitement from Chris after the last location intelligence conference. The use of “mash-up” is starting to grate a little, I admit; especially in contexts where it’s barely justified or just a trendy re-description of a practise that has been going on for a long time (if i overlay a WMS layer with a WFS layer, can that really be called a mashup, Arnulf? ). As a semweb-head, I resist the idea that humans should be spending their time doing things that bots can achieve both more creatively and efficiently. The more stuctured data that GeoRSS passes around, the more we’ll need intelligible, reusable descriptions; I’m fired up to pick back up on pieces of transportation modelling vocabulary, in particular, with an eye to GeoRSS syndication.

The release of MetaCarta’s GeoParser API was a surprise big hit; I hear so much about it at home, that I forget how clicky the idea is for others. GeoNames have a GeoRSS reparser service which I was pleased to hear about in Adena’s recent article on Fun with GeoRSS; I’ll be interested to see both how GeoNames’ brute-force approach compares to what the GeoParser can do, and how or if people will come up with interesting post-processing hacks based on the GeoParser API. One niggle I have with gutenkarte is that it reliably attempts to geocode peoples’ names which happen to be placenames. It may have wonderfully sophisticated inference about spatial things, but doesn’t complement that with inference about non-spatial things - inferring proper names through proximity to speech acts, repetition in context, etc.

The key here is another kind of convergence - the complementary use of natural language analysis techniques alongside bodies of structured information (such as a gazetteer, set of GPS tracks etc) to complement and enrich each other. Arguments about metadata often overfocus on “human error” and our limited inclination, or capacity, to expend our semantic energy on describing things for unknown, future others. But here are glimpses of a solid “middle way” in automated suggestion that allow humans to focus energy on what they’re actually good at.

About the most thought-provoking talk for me was Steve Morris on “The Disappearing Data Problem”; digital archiving of geodata and metadata, and the work being done at NCGDAP. I’m not sure how seriously to take the assertion that we may be living in a “digital dark age”, to be looked back on with incredulous startle in 20 years time. The thrust seems to be that we’re archiving the data substrate, but not the presentation effects; an archive of geodata+metadata won’t allow us to produce “a map” as it would have been seen. We’re nonchalant about what we throw away, chasing the present, and viewing updates as a moving truth, not a reflection of a truth that has passed yet is still valid. Without formal archiving processes in place, we risk losing track of ‘obsoleted’ data collected both by states and by corporations.

The other big resonance of the conference for me was as the North American “coming out party” for OSGeo; the first public event since its inception where the tribes have really been out in force, visible both in the throng of remarkably tasteful t-shirts, and the repetitive presence of OSGeo members on the stage. Gary Lang’s talk offered more insight into the roots of Autodesk’s decision to open source MapGuide, and how it’s working out for them. Sean S.mentioned in the OSGeo BOF that their users are getting a faster response cycle to their needs; soon the open source package is going to outstrip the “Enterprise” version in what it can do. As government in North America begins to actually mandate Open Source in their software contracts, “a rising tide raises all boats”, we hope.

The “Future of Data Industry” panel puzzled me. Steve Coast had given a well-received and high-profile presentation about openstreetmap the previous morning and at this point had worn his voice out to a husk. He was set up on a panel with three commercial data providers, keywords “value chain”. These people deal with the built environment, with human artifacts and human movements. The focus is street and addressing data, as that’s where the obvious “solutions” are - in routing and distribution logistics; in spatially targeted marketing and land/building use brokering.

The word “geodata” travels in my mind with “public” and “state-collected”, and that’s where the real booty is; in remote sensing and natural resource management, and in civic information. Some of the whole potential body of geodata will be replicated on commercial terms, or in open-source, open-data basis. But much of it either requires really specialised equipment and arcane technique, or doesn’t occupy a place in the “value chain”, for the narrow definition of proprietary value. Public/private sector working together, both reaching out to user base for ‘ground truthing’, production of collective works with open interfaces and re-use rights taken for granted; a good precedent to be able to establish in a domain that can come to effect so many other domains, not just with technologies but with operating ideologies.

Mike Liebhold talked in his excellent keynote about “mining geodata, but missing metadata”. Without common data, open to all, free to re-use and recombine, we won’t get past the point of static, purpose-specific mashups into the serendipitous terrain where we are finding out things we never expected to find. Perhaps i’m projecting, but it feels like ‘geospatial’ (hey, if ‘local’ is now a noun, who can constrain me?) is about to hit a new phase. Mike’s picture of the classic hype/antihype graph entertained me; “locative media” is just emerging from the trough of antihype now, on the way to becoming something new. The prospect of making new things whose uses can’t predict or control always excites me.

Oddly enough, Jack Dangermond put it best; “it’s a great time to be in the geospatial community”. The infrastructure is built, stabilising and starting really to break into popular consciousness; driven in the ‘mainstream’ mind, we like to think, by Google Maps and Earth; now hackers want to reach down into the guts and fix and remake them, and the OSGeo toolsets are right in place for that. Not wanting to take the edge off self-congratulation, but there’s a lot of work to do - particularly in data packaging and search, in documentation and in educational toolkits - and I’d like to help make some of that happen, too.

Posted in events, community, osgeo |

You can follow any responses to this entry through the RSS 2.0 feed. Trackback from your own site.