However there are organisations which do maintain unique identifiers for a great many types of real world objects, e.g. road bridges are numbered within the jurisdiction of a local government's civil works department, telegraph poles are labelled by the local telecommunications organisation, and every computer Ethernet card has a unique number burned in to it during manufacture (we know they are unique even if we don't know where they are). These labelled real world objects are almost invariably man-made or even entirely man-imagined, e.g. land parcel identifications, for the simple reason that natural objects usually permit less precise delineation, e.g. is an estuary part of the coast or the riverbank? This indeterminacy is of several distinct types and has been discussed in detail in a recent conference [1].
The types of feature identity that will be introduced later in this paper must be able to interwork with these pre-existing labelling schemes that are maintained by a variety of organisations with very different construction grammars and quality control standards. A key point is that the labels have to be maintained: mistakes happen and must be corrected, incorrect numbers are applied to real objects and real labels are misrecorded in software systems. Thus the real world label must be related to, but distinct from, any geographic information system's feature identity. For these reasons it is clear that real world labels can never substitute completely for some purposes, even though such labels probably provide more added value than other types of feature identity.