Traditionally, geographic databases have been viewed as containers of maps, a perspective that is
reinforced by the familiar layer-cake view of GIS. The map captures knowledge of the world, and
provides input to a database through processes of digitizing and scanning. In some cases, the
process of creation of the map is measurement like, as in photogrammetry and other aspects of
topographic mapping. But in other cases the process uses the knowledge of a field expert, such as
a soil scientist, or forester, and is not likely to be replicable from one observer to another.
In the latter case, the map becomes a kind of communication channel between the expert and the
user, a means of transferring the expert's knowledge and interpretation of a complex real world in
a form useful to the end user. Several trends are making the operation of this communication
process increasingly critical. The widespread adoption of GIS has meant that integrative studies
are now being done that combine input from a variety of disciplines, but without specialized
knowledge of each individual layer of data. For example, global climate models are now
incorporating information on soil properties, but soil science is not likely to become a major
element in the training of atmospheric scientists. Increasing emphasis on sharing of
scientific data, and the widespread popularity of the Internet for scientific communication, have
meant a steady increase in the physical and intellectual separation between data producer and data
user. In this context, the notion that spatial metadata might allow a user to determine the
fitness of a data set for use in a particular project takes on even greater significance-are we
really asking a soil scientist to educate an atmospheric scientist through the medium of a metadata
record?