search and folk geographies

ideas of location are many layered. Official boundaries are constantly rewritten to account for changing demographics which affect electoral boundaries and allocation of resources. Long lost official boundaries live on memory, they become folk geographies blurring with their replacements and no longer reinforced by official documentation.

the historic county of middlesex

middlesex in england moved constantly through geographies over a hundred years, it's last hold on official remit faded in the early 21st century. Though a governmental boundary for Middlesex might no longer exist, a folk idea of Middlesex is persistent. Search engines will return results for 'Eating out in Middlesex' so it cannot be ignored as a continuing economic geography.

grounding perception

laces in Middlesex also have a location in current Ordinance Survey geographies: Place a pin on a map and it exists both in Middlesex and… say Surrey. If 15 places claim to be in Middlesex and define a position then we may begin to approximate Middlesex as it currently exists in shared perception. We could claim other places in Surrey, that fall within the convex hull of our 15 places, as being *in Middlesex*.

addressing issues

middlesex continues to be an allowed part of Royal Mail addresses: these form another geography overlaid upon governmental division. A UK address is a routing address, passing, not down through containing geographies but, across sorting offices to a postman's delivery route. An address may contain a different county, a different town, than the actual location. This routing information is continued into the postal code: as a resource locator it holds limited value unless one holds the key to its encoding. With search both the routing geographies of the postal system (postal town/county and postal code) assume a value that potentially exceeds that of positional geography.

which is correct?

resource information is contextual. We must always ask questions about the reasons for accessing a resource. If we wish to deliver information we follow the routing requirements of the system, a phone number, a mail address. If we want location information then large amounts of routing information may be irrelevant, even misleading: Ignore the postal town of a location when looking at statistical information if the postal town falls into a different statistical container. For search all the geographies may be relevant - we may need to *reason* which geography is most relevant to any given search but otherwise *weak ontologies* have been given enough additional meaning to allow reasoning to be machine driven.