Among the most important consequences of the development of geographic information systems is the widespread ability of those in government and business, as well as the public more broadly, to create spatial representations. These increasingly may use three or even four dimensions, and may include a wide range of information about points, lines, or areas. And one consequence of the ability to create these representations, especially in the contrast of increasingly vocal demands for a "right to know," is the development of new threats to the privacy of individuals and groups.
In one sense any computerized information system creates such threats, just to the extent that it makes it easier to obtain information; suddenly, one can with a click retrieve information associated with some parcel or area. Where previously one had to dig through masses of paper, during business hours, in a central location, and under the watchful eye of a suspicious bureaucrat, one can now satisfy one's curiosity about neighbors, employers, and potential friends much more easily. In fact, the increasing availability of such information on the Internet makes it possible to snoop in the comforts of one's own home.
But beyond those problems raised by information systems in general, _ geographic_ information systems raise an additional, and rather different set of problems. This is a set that have received less direct attention than the first. These problems arise not from the ability of a user to acquire information already in the systems, but rather from the ability, using the systems, to construct new sets of visual representations. The problems arise where the systems allow the creation of representations that appear to be accurate, but are in the end misleading, or where they allow the creation of representations that are accurate, but that are accurate representations of information that has previously been and ought to remain privileged.
Geographic information systems in this way raise a set of issues that have both ethical and legal moments. They are issues concerning the ways in which one ought and ought not to use these representations; they are at the same time issues concerning the ways in which the use and misuse of these representations ought to be regulated through the legal system.
These issues arise in a number of arenas. Perhaps most obvious is the area of medical data. There the easy availability of data on stigmatized diseases or, potentially, on genetic propensity to disease creates the possibility for misuses damaging to individual and neighborhood alike. Here the issues raised some twenty years ago, in the case of mortgage redlining, have arisen with a new intensity, as a result of the ease and simplicity of analysis and representation brought on by technological change.
Similarly, the recent development, especially in the private sector, of systems of geodemographics has made it increasingly easy to produce systems of data profiles, ones that represent individuals and neighborhoods as having certain social, cultural, ethnic, and economic characteristics. These geodemographic systems are now widely used in the making of decisions about insurance, lending, and the like, and it seems clear that their use will continue to increase.
My paper today is in the form of a proposal for a collaborative research project. This project is an attempt to make sense of the issues posed by these spatio-visual representations. It will examine current practices in the use of the representations. It will analyze the ethical issues involved in their creation and use. It will then analyze the current regulatory framework in which they operate. Finally, it will suggest ways in which this framework can be improved, so that the right to privacy and the right to know are better balanced.
Current practice with respect to spatio-visual representations can perhaps best be seen as growing out of the intersection of three roots. First are a very general set of common-sense principles and practices associated with what might be termed "cartographic ethics." In this category are included the homilies and exhortations that are taught to those learning to produce or read maps - one needs to use due diligence, to be accurate, to be careful in generalizing, and so on. This set of ideas, once relatively untheorized, has lately come to be seen as itself having a history and a politics; the result has been a set of works on the rhetoric of cartography, on maps and ideology, and more generally on the "situatedness" of the map. At the same time, there seems to have been little interaction between the new ways of thinking about the map and the homiletics of the textbook.
A second root of current practice has developed in the United States Census, and in the systems established for the protection of individual records . Over the last several years the Bureau has developed increasingly sophisticated tools, designed to prevent data users from working backwards from aggregate results to individual cases. From an initial set of practices designed to prevent one from using simple arithmetical tools to replace intentionally omitted low-valued cells in matrices, the Census has moved to far more sophisticated techniques of "data masking," ones that move values into adjacent cells and even create dummy households. If these techniques have been primarily aimed at data represented in tabular form, it remains that these techniques have impacts on such data when they are mapped.
A third root of current practice, one related to the last, is the legal and regulatory framework sur- rounding any release of certain forms of data. For example, by state statute medical data typically may not be released for small geographic areas. Here, too, regulations designed for data released in _ any_ form have implications for those data once they are released in the form of maps.
The first part of the project will address current practices of spatio-visual representation. It will address the following questions:
In one sense, I have already pointed to two limitations of the current practices associated with spatio- visual representation: the homiletics of teaching do not make recognition of the best available understanding of the nature of cartographic practice, and institutionalized practices do not deal explicitly with the question of the ways in which special problems may be raised by cartographic representations. The second part of this project will develop a more thoroughgoing analysis of these and other limitations. It will do so by addressing the following questions:
This is a multidisciplinary project. It will involve work in and the cooperative efforts of experts on:
On the basis of the answers to the above three questions, the project will finally lay out a general picture of the ways in which one can move from generalized data to cartographic representations, while at the same time minimizing the likelihood of violations of privacy of individuals and groups.