i spill out of the station with my bike, a three-lane motorway ringed with concrete blocks all of my prospect. i ask the station guard if there's a map here, or if he has any idea where i'm going. "ironically enough, i'm visiting the ordnance survey." "sorry, i'm not from round here," he replies. i set off down the carriageway and buy a geographers A-Z streetmap from a garage.
it's an interesting meeting. the charming and obliging jon bryant, who represented OS R'n'D so valiantly at the collaborative mapping show'n'tell, couldn't make it; an interesting mixture of semantic web interested parties; a chap jon goodwin looking at OWL, a spatial cognition researcher working with random users and professionals, a quiet goateed bloke 'looking at the grid and agents', who said, i recall, nothing; another quiet chap working on their mastermap SVG project; a sage and hauntingly familiar looking interested consultant type; a gimlet-eyed bridge between research and implementation, running a strong commercial line, and a shy researcher, "doing a PhD in ontologies". (sorry for not remembering anyone's names at the OS, hoping no-one who reads this offended on that score)
i was there with nick gibbins, who i'd met only through irc before, another member of the #foaf ivory-tower, secret-handshake clique. he ran through a mix of exec summary style info about his project at iam, the euro-funded aktors; and some nice demo material, researcher discovery and implicit trust in authority on topic in the academic funded scene; and the demo on collaborating agents working on flood relief, which i'd heard many good things about post-www2003.
the usual suspicious questions; scalability of triplestorage and query, expressiveness of RDF, complexity of implementing OWL-Full; semi-automatic generation and exchange of ontologies, the overhead of making sameAs-type statements manually. they're curious as to what the semantic web will buy for them as an organisation; reluctant to commit to projects in a commercial space until the benefits, in a commercial sense, are proven. they've not assembled their feature classification work into a taxonomy, a controlled vocabulary, an ontology; this seems like on of the most interesting and useful, give-a-little take-a-little, things the OS could do for the semweb.
my angle on this stuff, i declaim, is somewhere between art and technology for social change. "this doesn't have to be about standard upper ontologies, inference engines for KM; it can be about open vocabularies for infosharing, decentralisation of storage, user ownership of data. i rush through, rather quickly, explanations of my mudlondon bot, and the utility, even necessity of open map data to the community wireless networking; compare and contrast maps.nocat with the consume nodedb. and openguides; publishing metadata from a selection of vocabularies in RDF as well as HTML, spatially located content, the localisation of the web and the services available from it; the "with wearables, we can annotate space wiki-style" dream. as a brief coda, i haul up a large image from the waag's realtime GPS project; "these are the sorts of means to which we have to resort to build free images of our world".
retrospectively; i wish i'd asked for a proper demo of this mastermap. i had the subtle and unexpected impression that it was mostly demo-ware at this point. it would have been nice to take away more contact; i was too predictably confrontational in the presentation of my concerns. but those concerns are strong; the lack of open maps, open geodata in the UK is hamstringing community efforts, withholding community resources, hamstringing small startups on some next tech wave.
they're waiting to see if we fail (or by some chance succeed). i ham up my indignance, even directing it at nmg at one point; "the webont wg can talk til they turn purple..." "(and they do)", emphasising pragmatism and empiricism in this work, throwing data at the world and seeing what sticks; finding our own errors by trial, a modelling peer review process. john goodwin points out that the OS have guaranteed unique identifiers for features; one of the more useful short-term and non-loss-inducing things they might do is to release their unique-feature-identifiers as a uri scheme, similar in intent to the TAP KB; i assume openguides would be interested in reusing those in its RDF serialisations of spatial metadata.
the subjects of provenance, and of trust, layered on top of instance data, seem to be of strong interest; an interesting negotiation, between statements "that the OS has been imposing its view on england, its government and citizens, for most of the last 200 years... for the last 20, it's been listening to what the customer wants, and telling them regretfully they can't have it." but now perhaps there is a technological solution to their core problem; maintaining their IPR and status as trusted source, while allowing user contribution - but very much 'using' the users.
so they're looking for a killer RDF app, like everyone else. i maintain that the solution and the problem are the same thing; not a killer app, but a thousand interlocking sub-apps, enhancing each other; the swarm of bots, propelling themselves about a fluid continuum.