Hi Jon,
I agree, the weakest point on that list is the regions. I just finished the post that pertains to that bullet (http://www.pbs.org/idealab/2008/01/tagging-to-physically-defined.html) but, like you said, doing an “geographic image map” could well be overcomplicated and not worth the trouble. I have some ideas as to how it could be less complicated (use a program to automatically generate those regions using geopolitical maps, then if an item is geotagged by a human in the traditional “by point” way the program can ask “Is this story relevant to Philadelphia, Cheltenham, etc. ?” based on which of those communities contain that point).

Anyway, I like the idea of pulling in “radius of impact” as a simpler although possibly incomplete solution (the idea had actually occurred to me when I was brainstorming for that most recent post, but after reading this I decided it was worth mentioning as well). Is that what you were intending when you mentioned affecting greater areas, or do have you another metric in mind.

The problems with pre-defined areas (tagged, or automated) are mainly in how loosely they are defined – for example I’m currently working on a project for Arts Council England that covers the West Midlands – and we’re working with the BBC and Travel West Midlands (a local bus/tram company) – what is tricky is that ACE’s West Midlands, the BBC’s West Midlands and TWM’s West Midlands are all different – to the extent of featuring or not featuring whole counties. That the county the “West Midlands” is a tiny subset of all of these further complicates matters.

In blog posts on by Birmingham blog, I represent city-wide or council decisions by a geo-point on the council house – which is not an ideal solution either.

It’s a difficult, and complex job to map the infulence – your tax example is a good one – where strict defined boundaries come into play, rather than declining importance as you are farther from the epic-centre of the event. I think that this is where attention data has to come into play, for it will define your level of interest not only in subject areas (tax in this case) but interest in locations – where the user themsleves could have a “region of interest” (or more than one) compiled from geotagged news that they express interest in. Where the “attention regions” overlap with the “radius of impact” (and of course the subject inerest) is our user’s sweet spot.