It seems to me that coordinates in a spatial system have assumed an importance way beyond their geometric worth. Consider an organization that has large stores of information about their customers. Almost certainly the records will include an address.
Professional "spatialists" get called in when some analysis is required and what's the first thing on the to-do list ? Geocode the records of course.
Why ?
What relevance do the coordinates have for the record ? It already has an address. And the rest of the row of information is related to the address. And how many IT organizations are keen to add another column to a complex enterprise data store with a staging copy, a development copy and a QA copy and all the enterprise processes and DBA attention that requires ? Let alone a column with vaguely sinister implications like X,Y or even worse, an opaque BLOB that has dark secrets within (like WGS84 or EPSG:3162).
What exactly does a customer's insurance number have to do with an x,y coordinate pair ? Or the licensee of a car ? Nothing at all.
Why then should a customer considering the application of location intelligence suddenly need x,y coordinates ? Because those coordinates are needed for spatial analysis and the technology to produce coordinates from addresses for the sideshow that is GIS analytics cannot work from addresses directly.
But what if it could ? What if the addresses could always be recognized reliably and at extreme speed - such speed that the x,y coordinates (and the black magic in the spatial column) would be available when needed but ephemeral otherwise ?
Search would be address based. Attributes would be associated with the object they actually belong to instead of some doubtful point on the ground that might move depending on the geocoder.
There's one downer in this fine scenario and that is the spatial index that would be required for very fast spatial operations. Could this be ephemeral as well ?
