What is the essential character of a map? Is it a faithful "image of reality" or is it a "manifesto for a set of beliefs about the world"? This question is central to the effort to understand cartographic representation in the context of digital geospatial databases, which (in some venues at least) are rapidly replacing traditional manually- produced "paper" maps. While paper maps and geospatial databases are both products of human agency working through technological means, new mechanical and digital technologies have led to new modes of cartographic production and new paradigms for the representation of geographical phenomena. According to some critics, the most important change wrought by these technologies is the ascendance of a new geospatial "science" focused on the goal of producing ultimately truthful and objective representations of the external environment. This goal is seen as a byproduct of the new technological means, with its appeals to neo-positivism, reductionism, instrumentalist thinking and naive empiricism in which "reality" is taken as a given (Pickles, 1995; Sheppard & Poiker, 1995).
The assumptions of this new science have been challenged in a number of recent critiques (Harley, 1989, 1991; Wood, 1992). Much of the criticism has focused on the geospatial data "community", a diverse group of individuals whose common interest is the development and application of geospatial technologies. These individuals are depicted primarily as technicians trained to push the right buttons in the right sequence in order to fulfill some programmatic mission. It is asserted that they rarely have to confront the ways in which different forms of representation affect communication (Miller, 1995). They are characterized as harboring the naive belief that reality is uncontested and objectively measurable (Harley, 1989). Most significantly, perhaps, members of this community are seen as unwilling or unable to accept the role that social and institutional values play in the representation of features in geospatial databases. It is asserted that these individuals, by failing to realize that geospatial databases embed the values of society at large and the specific institutions in which they are constructed, have promulgated the epistemological myth of the "cumulative progress of an objective science always producing better delineations of reality." (Harley, 1989, 15).
Users of geospatial technologies must of course be wary of aggrandizing the potential capabilities of these technologies, especially as this pertains to the inherent "truthfulness" of encoded data and derived cartographic products. At the same time there is an equally grave risk of assuming that because these individuals work in a technologically-infused environment they are concerned only with technique and not the larger social issues that inform their work. Users of geospatial databases cannot help but be aware of the fact that there is no single "objective" representation of the external environment that is universally accepted. Like their manually- produced map counterparts, geospatial databases are not intended to be miniature replicas of the external environment. Rather they emphasize some aspects of the environment and suppress others in an effort to convey a particular message and argue for a particular viewpoint. What is contained in a database is a function not only of the nature of the external environment but also the values of the society and institution within which the database was constructed (Turnbull, 1989). Different values correspond to different attitudes about what aspects of the environment are important, and are responsible for the creation of different representations of the same geographic space. Values are embedded at various stages of database production through processes such as abstraction, classification and generalization. At the selection stage values impact database content (i.e., What feature classes and feature instances from the external environment are depicted?), while at the representation stage values impact database form (i.e., What geometric properties are employed in the depiction of a given feature class or feature instance?).
Content and form are manipulated by database producers to convey information in a specific way. The language of a geospatial database, as revealed through content and form, is usually tailored to a certain audience (e.g., geologists, soil scientists, etc.) who understand the language of the database just as they understand the jargon of their discipline. Different databases designed for different purposes show features in different ways, even if they cover the same geographic space and include many of the same features. Of course, the embedding of values is not always deliberate. The values particular to a given society at a given time are often taken for granted and are thus difficult to detect except in retrospect. The hierarchical social structure of eighteenth century Paris, for example, is made apparent cartographically when mansions are represented with symbols that maintain a rapport with what they represent while common homes are represented with a generic point identically reproduced in bulk (Harley, 1989). Because underlying values are often not consciously recognized, maps tend to reflect the social order and to reify and legitimate it. The rules of social order insert themselves into maps in a way that makes the map a commentary on the social structure of the place and time it was created (Harley, 1989).
These broad values form the backdrop for more specific values that reflect institutional characteristics. Institutional mandate is perhaps the most critical institutional characteristic that can affect database content and form. Broadly speaking, mandate defines institutional mission in terms of data collection and dissemination. For specific databases, mandate is formalized as a set of design guidelines that define the rules for data collection and encoding. These rules are especially important for large national mapping projects, where standardization is important, and for agencies that need to document their data collection and encoding processes for potentially contentious issues related to policy enforcement and litigation.
The broad social values inherent in databases may be inescapable and, to the extent that they are taken for granted, not easily documented. However, the values embedded in databases as a function of institutional characteristics can be articulated, documented and communicated to the database consumer. Indeed, producers of geospatial databases now rather routinely document these databases with metadata that can be used by consumers to deduce some of the institutional values impacting the processes of data collection and encoding. This communication process is important since it affects the consumer's understanding of the limitations of a database and facilitates its appropriate use. The primary tools that have evolved to serve this communication process are derived from work on geospatial database quality. Geospatial database quality tends to be viewed rather broadly. It includes the concept of accuracy as a measure of conformance with an external standard, as in fields such as statistics and surveying, but it also includes factors such as resolution (e.g., What is the smallest area that can be mapped?), taxonomic fidelity (e.g., What is the degree of homogeneity in mapped classes?), consistency (e.g., Are any internal contradictions present?) and completeness (e.g., Does the database depict the features it purports to depict?). Many of these dimensions of data quality are included in the Spatial Data Transfer Standard (SDTS), recently adopted as a Federal Information Processing Standard (FIPS) by the National Institute of Standards and Technology.
Completeness is a useful concept in the context of the debate over values. Completeness refers to the relationship between the objects in a database and the abstract universe of all objects. The abstract universe is a reference standard against which the database is compared. Evaluation of completeness thus requires a formal description of both the database and the reference standard. Typically, the reference standard is related to a specific context and defined in accordance with a desired level of abstraction, such as a specific project objective (Brassel et al, 1995). From this standpoint, completeness is defined in terms of the intended contents of a database rather than some abstract and idealized "real world". Even small-scale, generalized databases can be complete (Veregin & Hargitai, 1995). Embedded in this definition of completeness is the notion of a reference standard that is relative rather than absolute. This definition recognizes that each database has a particular set of objectives and that these objectives are the formal expression of the values associated with mandate and other institutional factors. From this perspective, completeness measures the degree to which the objects encoded in the database are consistent with institution- defined design guidelines. The same relative reference standard is used, albeit less explicitly, in other data quality components as well. Indeed even the notion of accuracy is only meaningful with reference to the standard against which accuracy is assessed.
What are the implications for the current debate over values? First, it is an oversimplification to assert that geospatial databases are intended as mirrors of some objective and value-free "real world". Indeed from the standpoint of assessing data quality it is not even necessary to insist that such a world exists. Data quality and its various components, including accuracy, are relative measures in the sense that they are always measured against some imperfect reference standard. Second, it is an oversimplification to state that the geospatial data community is unaware of the significance of values. The reference standard itself is a formal statement of the values that are embedded in the database as a function of institutional mandate. Admittedly, broader social values are embedded in databases unconsciously and uncritically, but this is not necessarily a liability. After all, the ability of a database to communicate is dependent in some measure on its reliance on shared cultural values. Third, values can be articulated, documented and communicated to geospatial database consumers using fairly simple tools. This suggests that consumers can assess the validity of a particular database for a given application as a function of database content and form. Knowledgeable map users have of course always been aware of the limitations of their data. In the digital era, increased reliance on secondary data sources means that the database consumer is more dependent on the particular biases present in the data source, which demands a more formal explication of these biases.
The ideas sketched above suggest that it is possible to
account (to some degree at least) for the social construction of
geographic space within a given institutional setting. This is an
important conclusion, since the alternatives are not particularly
useful for those who wish to continue to produce and use
geospatial data. Some critics have claimed, for example, that
given the dependence on social values it is really not possible
to distinguish between competing representations of the same
geographic space. Thus it has been argued that the artificial
distinction between propaganda and truth must be dismantled, as
must the arbitrary dualism between "art" and
"science" (Harley, 1989). In short, all representations
become equally valid since they are all simply expressions of
one's personal values, or the values of one's culture, or the
values of one's institution, any one of which has no more claim
to legitimacy than any other. This anarchistic epistemology
implies that we have no agreed-upon standard of reference, no way
to argue for one representation over another, and no basis for
communicating biases and assumptions. On the other hand, if
databases are to be more than just personal artistic diversions,
if they are to convey information rather than simply express the
values and viewpoint of their creator, then they must rely on
modes of representation that are meaningful across a spectrum of
users.
Brassel, K., Bucher, F., Stephan, E.-M., & Vckovski, A. (1995). Completeness. In S. C. Guptill & J. L. Morrison (Eds.), Elements of Spatial Data Quality, pp. 81-108. Oxford: Elsevier.
Harley, J.B. (1989). Deconstructing the map. Cartographica, 26(2), 1-20.
Harley, J.B. (1991). Can there be a cartographic ethics? Cartographic Perspectives, 10, 9-16.
Miller, R. (1995). Beyond method, beyond ethics: Integrating social theory into GIS and GIS into social theory. In E. Sheppard & T. Poiker (Eds.), Cartography and Geographic Information Systems (Special Content: GIS and Society), 22(1), 98-103.
Pickles, J. (Ed.). (1995). Ground Truth: The Social Implication of Geographic Information Systems. New York: Guilford.
Sheppard, E. & Poiker, T. (Eds.). (1995). Cartography and Geographic Information Systems (Special Content: GIS and Society), 22(1).
Turnbull, D. (1989). Maps are Territories. Chicago: University of Chicago Press.
Veregin, H. & Hargitai, P. (1995). An evaluation matrix for geographical data quality. In S. C. Guptill & J. L. Morrison (Eds.), Elements of Spatial Data Quality, pp. 167-188. Oxford: Elsevier.
Wood, D. (1992). The Power of Maps. New York: Guilford.