However, postcodes are not useful when *browsing* data: We do need standardized, well understood, regional geographic descriptors.

For example: When doing an on-line search for jobs, how would I describe my home location, my possible commuting zone, or locations of potential employers?

eg. Borders of Beds, Herts, Cambs: Is that the East, East of England, East Anglia, South East, East Midlands, South of England, etc. etc.? This confusion makes on-line searching very cumbersome and unreliable.

It depends on what you’re doing.

Pretty much everyone knows the postcode of their home, so things which relate to the position of your home (ie most government dealings) can/should use that.

If you ask people who their MP is, only a third know, and a third of them get it wrong. You can’t use it as the basis of anything reliable.

If you can’t expect the vast majority of people to know the postcode, then it’s useless as a location mechanism. The BBC seems to work with drill down maps, which (depending on the map you’re looking at) can roughly match onto connurbations. This may or may not make sense depending on what you’re doing.

As for your holiday home, it depends. What happens if you ask your vistors where they think it is? Maybe “the middle of nowhere” is just the location you need, although you might want to go with something that people wont think is in Scotland. Context is everything; and it also depends on where you are standing.

As a WIBBI, http://gaze.mysociety.org/ does the place name to location conversion for you. How that then goes back to ‘the “closest” big place is …” is a second, hard, step. Maybe if there was access to road/rail/bus info (“from this place the reasonable sized town that is fastest to get to is X”). That may suffice (“roughly halfway between Plymouth and Wick”)

Have you any ideas for a more logical, better understood way of describing geographical locations in the UK; especially for on-line search purposes?

I see that you’re having trouble with various look-up tables.

As described in my 2005 post: Where am I?

Different organisations have very different ways of sub-dividing the UK. This greatly complicates life as a customer, supplier, taxpayer, Internet user, etc.

For example our holiday business, officially in “Mid Devon” near the Somerset border, is badly served by agents whose database divides the county arbitrarily into “North Devon” & “South Devon.”

Most services etc. come from towns nearby in Somerset but bus passes etc. stop at the Devon county border.

I built the current (local, v3) directionless (and v2 before it).

> Has Google changed the rules so such systems dont work?

They’ve not changed anything.

Google has an API which you can call and it return results. It works great most of the time, but, unfortunately, sometimes the perl code which calls Google dies and kills the process serving you the web page. That’s what you saw.

I have no idea what is causing the problem (I’ve seen it on our Freebsd server and my OpenBSD workstation). It’s not a problem with the interface or google itself, it’s a problem with the implementation. I’m sure there’s a common WIBBI here.

The biggest threat to directionless is not from Google (we could switch to another API relatively easily if any of them offered the functionality we needed, they didn’t). Search is a big area.

While your questions about Google Earth and sketchup have legs, I just thought that I should point out that google hasn’t killed directionless even if they did, we have a backup plan.

Cheers
Sam
http://www.disruptiveproactivity.com/