Hyperlocal

I've been speculating for some time that the toolsets to create hyperlocal geospatial data and services will soon be within reach of the people who have an acute interest in that sort of very place-specific knowledge. That is to say, me. And perhaps my neighbors. First, the adoption of building information models (BIM), and CityGML have made it possible to drive geospatial referencing into the architectural and engineering world. Secondly, inexpensive (relatively) GPS receivers have made it possible for the average Jo/e to digitize geographic features of intrinsic importance and thus create geospatial databases of very local features: street trees, neighborhood landmarks, signage, street lights, and so on. With Google as a platform, nowadays literally anyone with a little bit of technical gumption can become a geographer.

I can picture a time not far off when our communities will be tagged, mapped and annotated in any number of different by self-motivated individuals who take an active interest in documenting their communities. I imagine most local governments would love to have data at this level of granularity. But I imagine that it will be way too expensive to collect for the forseeable future. Moreover, most government GIS shops place great emphasis on data quality standards and will need to have some amount of reliable metadata in order make use of these data sets. It may be that if it's an emergency and it's the only data, then the data are going to be better than nothing.

So today Yourstreet announced their launch (or is it a re-launch). Given my interest in the subject, I headed over to the site to check it out. I typed in my zipcode and was given a Google map with a scattering of markers denoting events culled from local news sources using the MetaCarta text-to-gazetteer tools. (I noted that several stories were culled from another town in Texas with the same name.) I can add a news story, start a 'conversation' linked to place (address, city name, or zipcode), and display 'neighbors' -that is, other members in my locality. It's a pretty basic toolset at this point, but perhaps that misses the real point that the average Jo/e now has a place to blog about very place-specific things that matter quite acutely to a very small set of people. The question then becomes: does Yourstreet offer enough inherent value to me that I can begin to posting and create the network effects that will be necessary to sustain Yourstreet (and other similar services)?

Congratulations to Al Gore

Hooray for Al Gore. What a vindication after having the presidency stolen out from under him. Congratulations!

Does 80% of government data really contain a locational component?

If you've read any amount of the GIS literature, you've no doubt come across the assertion that:
"80% of government data contains a locational, or geospatial component..." or, some words to that effect.

Have you ever stopped to wonder where this assertion comes from? Have you ever seen a citation for it? I haven't. How many of the results of this query cite the original source? (Exactly none, in case you were wondering.) Furthermore, how was the 80% figure arrived at? And, what is the origin for this assertion? I have always thought that the assertion came from the 1987 report, from Lord Chorley & co., Handling Geographic Information, but I'm not certain.

I should add that I suspect that most government data does indeed contain some locational component, - a place name, an address, the name of a landmark, or even geographic coordinates. So, I it's completely plausible to me. But, why do we as a community continue to repeat the assertion, taking it at face value? We take it as an act of faith really, don't we? Seems plausible, after all.

Here's the citation for the Chorley report, if anyone is able to get a copy and verify:
Handling of Geographic Information. Report to the Secretary of State for the Environment of the Committee of Enquiry into the Handling of Geographic Information. Chairman: Lord Chorley. London: HMSO, 1987. ISBN 0 11 752015 2

Lowering the costs of metadata

Recently I've been thinking a lot about the role metadata plays in discovering geospatial data and making judgements about fitness of use. It's becoming increasingly clear that the costs of generating standards-compliant metadata (e.g. FGDC, ISO 19115) are comparatively high. Organizations that have no clear mandate to create metadata, create either minimal metadata or worse, none. Even those who are mandated to create metadata (like federal and state governments) sometimes fail to provide metadata. I suspect that some managers of GIS shops go through some cost-benefit calculation that goes something like this:

I produce metadata at my expense.
You, external user, use that metadata to access my data.
Therefore, I incur all costs, you get all the benefit.

So, how do we lower the costs of creating meaningful metadata? I recently picked up the most recent version of Information Architecture by Morville & Rosenfeld. Part of my motivation was to gain a better understanding of the bigger issue of how to structure web-based content to make it finadble and accessible. But more importantly, I want to understand how practices of tagging and collaborative classification schemes assist in finding geospatial data and linking it to other content, which may or may not be geospatial in nature.

Certainly tagging has problems and limitations that have been pointed out elsewhere. One big disadvantage is the lack of support for formal metadata classifications outside of the 'keyword' category. Another problem is how to implement? Data producers will tag their data with certain keywords, but how do we allow the broader community to tag data in ways that are meaningful to them? The Geodata Commons whitepaper acknowledges this difficulty. By temprament, I favor a solution that is open and collaborative and thus, takes advantage of network effects and improves with use. But practically speaking, I'm not so sure this would preserve the value inherent in a controlled vocabulary.

FOSS4G 2007

Spent last week at the FOSS4G 2007 conference in beautiful Victoria, BC. An excellent conference overall. The quality of the sessions I attended was quite high, almost universally. In the coming days, I'll offer some of my observations from the conference.