In the course of analysing the international geography of innovation I keep an eye on the literature focussing on macro-regions. Macro regions at their most simplistic can basically be defined as economic or social regions which ignore conventional jurisdical borders.
A classic in this vein is Joel Garreau's Nine Nations of North America (1981) http://www.garreau.com/main.cfm?action=book&id=3 a very easy read.
An interesting comparison of geographies of similarity is the beyond borders project http://www.beyondbordersbook.com/index.html
Another in a similar vein is Richard Florida 'mega-regions' http://cjres.oxfordjournals.org/cgi/content/abstract/1/3/459.
These take a somewhat similar perspective by looking for regions that can be grouped either by mainly geographic and social characterisitcs (Garreau) or by agglomerations of activity, measured for example by Florida with light emissions data.
But what of interdependencies? Typically ignored because it is data intensive or isn't within the established construct. However, there are obvious examples. The North American auto complex is a 'corridor' that streches from southern Ontario down to Mexico following important highways and railway lines. But that is an easy eaxample because it is largely contiguous. For other sectors California and New York may be more closely linked than regions surrounding New York.
Identifying regions of similarity and continuous geography is something we are familar with, define a piece of land and draw a border around it (which can have significant value in itself) but we are less familiar with network geography.
