GIS Implementation and the Un-Theory: Some useful concepts from ANT. Eugene Martin Ph.C. University of Washington Department of Geography June 7, 1998   Introduction: Implementation and evaluation of geographic information systems (GIS)  in non-traditional settings (non-western cultures, developing countries or alternative institutional contexts) requires a flexible and context sensitive approach. GIS has been developed in the technologically advanced tradition of western countries and is constrained by its inherent western perspectives (Wynne, 1980; Sheppard 1995). Cultural differences in concepts of time, scale, detail, distance, values, topology and relationships (Mark and Eigenhofer, 1995; Rundstrum, 1995) means that GIS implementation is context sensitive.  Beyond these cultural differences, GIS implementation is also affected by institutional context and organizational interrelationships (Chrisman, 1991; Pinto, 1994).  The application of GIS in other cultures will most certainly require a variety of modifications to suit local needs. Successful investigation and evaluation of GIS installations in non-traditional contexts requires an approach that investigates the interactions between the technology and the specific social or institutional setting. Like GIS itself, the literature on GIS implementation and evaluation  is focused on situations and contexts similar to those in which GIS was developed.   Antenucci (1991) Chapter 10, describes five stages (concept, design, development, operation and audit) that are the foundation for seventeen steps towards successful GIS implementation.  Guptill et. al. (1988) suggests using pre-defined evaluation criteria to serve as the basis for evaluating GIS implementation.   Topical suggestions worth considering in  establishing these evaluation riteria are: standards and guidelines, software functions, hardware components and benchmarking.  Both of these frameworks do not provide consideration for the interactions and negotiations that contribute to the fundamental nature of a GIS implementation in a particular context.  Anderson (1996) proposes a framework that unifies technology transfer with organizational issues in a non-linear and iterative development process.  This dual consideration of the technology and the organization as equal components of GIS implementation suggests that the contentious or prominent issues in development should also be considered during evaluation.  To be effective in treating GIS implementation in non-traditional setting, an evaluation process must consider both the technical and the organizational components and pay particular attention to the interactions between these components.  It is in these interactions that the real story  of how the GIS was implemented and how it functions can be revealed. Development of this necessary component of GIS implementation and evaluation seems unlikely to arise from current directions in GIS theory. The current theoretical framework for GIS is too focused on technical development to offer any critical contribution to social theory of technology society interactions. In exploring the social construction of GIS and the inherent difficulty in multi-participant GIS design, some researchers have turned to Actor Network Theory (ANT), a perspective on societal and technological interactions developed by the Paris group for Studies of  Science and Technology (SST). Several efforts have been made to apply this new approach towards technology in society to GIS.  In a piece characterized as a ‘novel use of spatial information’, Francis Harvey (1997) describes the utility of boundary objects in designing multi purpose and multi participant GIS.  The role of boundary objects is also being considered for their role in the social construction of GIS technology (Harvey and Chrisman, forthcoming). ANT, the larger theoretical body to which boundary objects belongs, is a dynamic and ontologically unusual approach towards technology and the varied roles it plays in influencing society.  In addition to the idea of boundary objects, ANT offers several other perspectives that have utility for understanding the interactive nuances of GIS implementation.  In this paper I examine two main issues concerning ANT and its importance for GIS: 1) the general tenets of ANT and whether ANT is a functional theory; 2) some of the functional concepts of ANT that can be applied to investigate GIS implementation. The Theory of the Actor Network. The elements that have evolved to be called ANT began with the Paris group of science and technology studies in the early 1980s (Callon, Law, Rip, 1986). Parallel efforts in the United States, arose from studies of biology and technology by Adele Clarke, Joan Fujimora, Elihu Gearson and Susan L. Star (Star, 1988).  Their perspective is one that knowledge is local and socially constructed, that society is created by science and science is created by society.  Rather than focusing on the representation produced by science, these theorists have engaged in empirical studies of science aimed at uncovering the processes that produced these representations.  This approach has great potential to generate real understanding on the importance of science and technology for those involved in their creation as well as for researchers in the social sciences (Demeritt, 1996) Some of the hallmarks of studies of science, technology and society (STS), of which ANT is a component, are an ecological perspective, emphasis on ontology at the expense of epistemology and methods that focus on tracing or recording interactions.  Researchers making use of ANT are interested in analyzing not only the product of society/technology interactions, but the process that produces those effects.  The ecological viewpoint expands the scope of analysis to consider the impact and contributions of the non-human actors that participate in interactions between society and techno-science.  This perspective opens the analysis to any element of the ecosphere, biological, technical and inert, as well as all varieties of human participants connected with the phenomenon (Latour, 1987).  To cope with the possibility that every actor should be considered, ANT places an emphasis on ontology rather than epistemology.  All actors are considered equal except in their capacity to do work or for their effect upon, interaction with and ability to impact other actors (Latour, 1997).  By downplaying epistemology and approaching phenomenon without pre-established categories or relationships, ANT seek to focus on the more useful concepts of irreductions, networks, translations and durability (Star, 1988).  The methodological approach to investigating these characteristics of actors is to record their interaction, connections and effects (Latour, 1987).  Star (1988) identifies four ‘maps’ that can be drawn from different perspectives: the scientists, the practices of the organization, communication across domains and a map of responsibility or power. The unusual characteristics of this theory, if it can be called that, have resulted in ANT being widely used (and misused) in applications from politics and gender to animals and nature (See discussion in Law, 1997).  In this process, ANT has developed and departed from its origins and is now widely employed in the creation of meaning that was beyond the original scope of its creators (Law, 1997; Latour 1997b). Despite these apparent abuses, the proponents of ANT adhere to the theory’s foundation of ontological supremacy: a theory should not be judged according to an absolute set of indicators, but according to the work that it does in the world. (Bowker & Star, 1996). If we ignore this line of defense and consider the fundamentals of ANT as if it were a theory like any other, we can explore its strengths and weaknesses.  But what definition of theory should we use as a standard?  Returning to David Harvey’s "Explanation in Geography" (1969, p. 88), a classic, if maligned, work on positivist geography, we find reference to several authors (Hempel, 1965; Ramsey, 1960; Rudner, 1966; and Braithwaite, 1960) who explain theory as a system of statements used to explain a set of facts.  Theories have their own language and logic system that relates primitives and axioms to construct theorems meant to explain the behavior of phenomenon.  In a very real sense, they are constructs disconnected from the observable, but have enormous explanatory power.  These primitives and axioms of a theory establish the 'what' that is in question and the causal relationships that govern or explain behavior.  While this may not be the accepted functional definition of theory for the social science of human geography, it is still a useful series of definitions.  In applying this definition to ANT, we should be looking for primitives (the smallest elements under consideration), axioms (rules governing possible behavior) and theorems (explanations for the behavior of a phenomenon).  Despite this attempt for clear cut categories and explanations, we shall see that even the fundamental aspects of ANT defy the application of these accepted categories. Lets start with actors.  We already know that ANT is willing to accept both human and non-humans as actors and that the primary interest in actors is not their context, but their interactions and effects on other actors.  Latour (1991, 1992) and Law (1992) show us that by suspending our assumptions about the nature of actors, that we are better able to ascertain their true nature and the roles they play in constructing a network.  These authors assert that all interactions between humans are mediated through objects of one type or another (Law, 1992; p. 383).  By removing the analytical divide between humans and objects, we are better able to examine the true nature of interactions that are the building blocks of networks.   This perspective also challenges other traditional divisions between science, rhetoric, society and nature.  The justification in this radical view is that if we want to know the true origins of power and structure, then we must consider all components that collaborate and cooperate in their creation, proliferation and persistence. By asserting that humans and objects are functionally equal, ANT requires us to discard our traditional notions of actors and networks and how they behave.  One character of actors and objects that has some apparent equivalency is the capacity to do work.  Another primary quality of actors that permits them to coalesce into networks is their ability to interact.  The many manifestations of these interactions is examined in ‘translations’, a fundamental element in ANT (Callon, 1985; Latour, 1987; Latour 1997).  Between human actors, the translation of interests is roughly analogous to persuasion and the negotiation of common interests.  Between humans and objects, translation occurs during design when the object is imbued with its purpose, program or script in how it interacts or affects other actors (Akrich, 1992).  Further translation takes place between the object and the actors it encounters as the initial program or script is altered through interaction.  Finally, actors must be in motion: all actors circulate and are fundamentally altered by the other actors that they encounter (Latour, 1991). When actors and their interactions are taken together they form a network.  Latour (1997a) cautions us about the differences between an ANT network and technical or social networks.  Social networks exclude non-human actors as technical networks exclude the human.  ANT networks incorporate both with the linkages consisting of translations and interactions between actors.  Networks assume the heterogeneous nature of these linkages and seeks to understand the reasons for this heterogeneity (Law, 1992).   This uneven distribution in networks requires two types of consideration: temporal and spatial.  The temporal considers the durability of the network and its effects and the spatial considers the nature of circulation within a network.  Both of these characteristics become central focuses in the tracing or delineating networks. Having described the characteristics of actors and networks, it is helpful to now consider some special cases of actors or objects that are particularly active or unusual elements in a network.  The first of these is the ‘quasi-object’ (Serres, 1987; Latour 1997a), an object that both circulates and transforms while circulating.  Following the circulation and transformations of quasi-objects is the process undertaken in the tracing of networks.  Another useful variety of object already mentioned is the boundary object (Star and Griesemer, 1989).  Boundary objects are loosely structured when considered generally and acquire specific structure in a particular context.  In identifying boundary objects, the researcher can trace linkages between networks, even though the boundary object performs differently in each situation.  Inscription devices are objects that do not necessarily operate to define a network (although they can), but they do play an important supporting role.  Inscription devices are any object that records, and thus translates, nature, particularly in a visual representation (Latour, 1987).  Scientists use a multitude of instruments, from seismographs to atom smashers just to name two, that make nature ‘accessible’ through the inscription of measurements as graphs or illustrations.  There are other varieties of objects that populate networks, but these three will be sufficient as an introduction and will serve the purpose of this investigation. Beyond the ‘primitives’ (actors, networks, quasi objects, boundary objects and inscription devices) and some basic ‘axioms’ of their behavior (circulation, translation, heterogeneity, durability and mobility) there are some ‘theorems’ about how networks behave.  Two of these ‘theorems’ are convergence and punctualization.  Convergence is the degree to which the processes of translation and circulation lead to agreement (Callon, 1991).  Convergance is a special case of translation that aligns the elements in a network.  Alignment and durability can lead to punctualization, a point where the network supporting an actor disappears from view.  This takes place when the network components that are responsible for the production of objects or performance of functions are summed up in symbols or artifacts that encapsulate the network (Callon 1991) or, in the case of technology, form a ‘black box’ (Latour, 1987).  An ‘open black box’ refers to a black box that requires significant configuration or reinvention to operate in a particular context (Star, 1992). With the elements described above, we have a fundamental sketch of what ANT is and what it is on its agenda.  From this simple description it is still unclear is whether ANT is a theory or not.  Bruno Latour (1997b) calls into question the naming of ANT with the words ‘actor’, ‘network’ and ‘theory’ because the abstraction of these terms is too extreme.   The actors and the networks in ANT are quite unlike any actors or networks that we encounter elsewhere.  They are boundary objects that have already acquired too much structure to permit their effective use as descriptors of ANT.  There is also the question of ‘theory’. While ANT may appear at first glance to have the necessary foundations to operate as a theory, it has no innate ability to offer explanation of phenomenon or to predict  behavior.  Without this final quality we do not have a theory. So, what is ANT?  The founding theorists and proponents of ANT created this particular perspective on science and technology to permit themselves the leeway and scope to consider all the elements of the phenomenon they study (Star, 1988).  Furthermore, they were quite dissatisfied with approaches and theories of other supporting disciplines because they were unable to account for the variety of actors participating in phenomenon that transcend the individual realms of science, technology and society.  By expanding the scope of analysis to encompass the breadth and variety found in STS, the creators of ANT have forgone explanation.  The power in their accomplishment is in removing the artificial divide between human and non-human, between science, nature and society.  What ANT  lacks in explanatory power is made up for with insight.  Many researchers will encounter ANT and return to their disciplines to re-examine old ground already covered.  The descriptive power offered by ANT will, when combined with other traditions of explanation, generate a more complete understanding of phenomenon and processes. What remains to be seen is if the combination of ANT with other traditions is even possible, considering the iconoclastic assumptions required to adopt the ANT perspective. The above description of ANT and the brief discussion demonstrates that ANT is best utilized for its intended purpose: studies of technology in society.  Lacking explanatory powers, it is at its best when used for descriptive purposes, the ordering of observations and the alignment of interactions.  The following section demonstrates how aspects of GIS, the technology, the implementation context and their interaction, can be identified and observed in a manner that supports evaluation of the implementation. Tracing Networks Through Interaction Intermediaries Michel Callon (1991) has examined the role that intermediaries play in aligning networks.  He discusses four main types of intermediaries: texts, technical artifacts, human beings and money.  The following paragraphs paraphrase Callon’s discussion of these intermediaries with commentary related to assessing the networks that support the implementation and operation of GIS. Texts are the primary vehicle for the transmission of scientific accomplishments.  Scientific texts represent network associations through the citations and references that support the research. In relating the phenomenon of interest and intertwining and arranging support for a particular perspective on that phenomenon, a scientific text defines "the skills, objects and relations of heterogeneous entities" (Callon, 1991, p. 136). The difficulty in assessing scientific texts associated with GIS implementation is that they are relatively uncommon.  GIS implementation is more frequently documented in reports by consultants, memos between departments, or progress reports prepared during the implementation process or the functioning of the system.  The nature of this material is more documentary than scientific, but it still serves as an intermediary that can define and delineate the network of actors involved in implementing a GIS. Callon’s second category of intermediaries are technical objects.  Technical objects align networks through their program of action (Akrich, 1992; Latour, 1992). This program of action dictates the resources required for the object to function, the skills an operator must possess to make the object function and the intended task the object was designed to accomplish. Ascertaining the program of action for a GIS means examining the technical objects that comprise the GIS as well as the task for which the GIS has been implemented. Taken as a whole, a functioning GIS can have a program of action that goes beyond the programs of action of the GIS components. GIS requires a large volume of digital information to support analysis, operators trained in spatial analysis, field technicians who can georeference their observations and end users who can interpret the maps or tabular output from the system.  Interpreting the program of  action of the component elements of a GIS and the GIS as a whole can identify the network elements required to implement a GIS.  Examining these network elements for the degree of agreement between their characteristics and those required by the program of action can serve as a means to identify inefficient or inappropriate linkages that weaken a network. Just as technical objects can order and configure other actors through their program of action, human beings can orchestrate the roles of non-humans.  A computer programmer is unable to work without a computer; a computer requires an operator familiar with its operating system and language to program instructions.  This quality of human skills is not limited to the purely technological, any occupation that is strongly structured by skills will order non-human actors. For GIS, the need to have technicians with specialized skills who can make the system function is obvious.  Beyond the GIS technicians, there are skills, maybe better described as responsibilities, that can invoke the functional role of a GIS.  Decision makers are an example of this skill/responsibility.  These individuals require sufficient information of specific qualities to support their ability to make and defend a decision.  The connection and interaction between the GIS and the decision maker is one of mutual configuration.  The decision maker must learn to make use of the information products generated by a GIS.  Likewise, the GIS must produce results that are suitable to support the decision maker’s needs. Money is the fourth intermediary that Callon describes.  He describes the network potential of money in the context of venture capital that becomes translated into orders, actions and recommendations.  The investment required to initiate a GIS has similar characteristics.  Unlike buying a car or a piece of furniture that has a pre-established role or function, a considerable investment must be made to fit the development of a GIS into a particular situation so it will produce the intended benefits.  Tracing the lineage and translations of the financial investment and the manifestations of that investment resulting in an operating GIS is another network that uncovers the process of GIS implementation. Investigation of these four intermediaries, texts, technical objects, skills and money, can all contribute to understanding the interactions involved in the GIS implementation process.  Tracing these intermediaries can generate an initial functional sketch of the interactions that support the initiation of a GIS.  Once these interactions have been identified, they can be examined for the roles they play and their contribution or hindrance to the effectiveness of the GIS. Social / Technical Interactions. Tracing actor-networks through interaction intermediaries can generate one variety of useful information for evaluating GIS implementation.  Another perspective is to consider the interactions and negotiations that revolve around a particular social or technical element.  Three examples of this are boundary objects, GIS as an ‘open black box’ and the functional capabilities that are provided in commercially available software packages. These examples illustrate interactions between GIS technology and the social context of GIS deployment.  In considering the interaction of the social and the technical, it is tempting to consider the potential effect caused to either the social or the technical aspect of the situation.  Although the following examples are considered from this viewpoint, it is important to remember that ANT interactions reflect alignment of the interacting entities to produce durable networks.  Each of the examples described has equal potential for dual effects across the social / technology elements of the network. Investigating negotiations over boundary objects and understanding the dynamics of the negotiation process is a part of Latour’s (1987) methodology of following scientists and tracing networks.  These negotiations can lead to the adoption of definitions or some stability in procedures for GIS implementation.  Harvey (1997) and Chrisman and Harvey (1998) have already demonstrated the utility of boundary objects in assessing the relationships between GIS technology and people.  More than just the identification of boundary objects, their approach is to examine people’s opinions and relation to the boundary object in question to ascertain differences in perceptions of the boundary object across social groups.  Their examples, wetland classifications and technical standards, illustrate disagreement and localized utilization of entity definitions or procedures that are meant to facilitate consistency across institutions and systems. This is an illustration of interaction between human actors meant to configure GIS applications.  The decision of what is to be treated in a GIS is a prerequisite for determining what treatment will be employed. Definitions that permit measurement and lead to representation in GIS is the result of interaction across the social and the technical to produce a working solution. Bowker and Star’s (1991) notion of an ‘open black box’ is a particularly useful concept for illustrating mutual cause and effect of technology implementation a social context.  An open black box refers to "the intersection of ad hoc practices, globally circulated forms and software, and the indeterminacy of the information so collected." Star, 1992, p.407).  While Bowker and Star (1991) were interested in the classification of diseases in distributed reporting applications, there are parallels for GIS.  Most commercially available GIS software packages come with a fairly common set of tools for conducting spatial analyses and representing space.  The software vendor, in an effort to create the most successful package that meets the needs of the majority of users, includes more capabilities than any one end user can hope to apply.  Once the software arrives in a particular institutional context, the needs and goals of the organization dictate which functional capabilities of the GIS software will actually get used.  This dual aspect of the ‘open black box’, a collection of tools and localized decisions regarding appropriate use, can be an effective foundation for investigating the technology / institution interaction.  Observation of which tools are used for what tasks and the negotiations that led to these decisions can serve as a useful meter for assessing the nature of the GIS implementation. Both of the previous examples illustrate situations where social interactions have an effect on configuring the technology.  The reverse is also true: technology has the potential to configure the social.  Several examples of this situation can be found in the automated delineation of areas using GIS.  One such example is the automated delineation of watersheds.  This process is based on a digital elevation model and a spatial analysis algorithm that delineates the drainage basins in the elevation model.  Another example is the use of buffers, delineation of an area within a fixed distance of an object.  These two functions are capabilities commonly offered in commercially available GIS software packages used by land resource managers .  The availability of these processes in a GIS and the choice of an institution to make use of them in an application requires the institution to accept the process imbedded in the buffer or watershed delineation tool as the default process for delineating management areas.  This is not a difficulty when there is a great deal of agreement between the technology and the real world process being treated by the institution.  Situations where the institution is accepting the software’s tool-kit as the best available approximation for areal delineation is being configured by the available technology rather than develop tools that really meet the users needs. Conclusions Current approaches to assessing GIS implementation are limited by their focus on technology and information products, unable to address the formative interactions between GIS technology and social context.  Actor Network Theory is an alternative perspective on science, technology and society that offers several concepts and strategies useful in assessing these formative interactions.  Tracing the implementing and supporting actor networks behind a GIS has been demonstrated as one way to identify these important interactions.  Once these interactions have been identified, they can be evaluated for their contribution to the stability or durability they contribute to the established GIS.  Network interactions that do not contribute to a network’s durability can be considered interactions that reduce the effectiveness of a GIS.  Boundary objects, GIS as an open black box and GIS areal delineation tools were discussed as examples of focal points of social and technical interaction.  Examination of these focal points in specific situations can yield an understanding of the dynamics contributing to the character of a GIS that might not be encountered through the tracing of the actor-networks.  Negotiation in configuring a boundary object across social divisions can lead to the implementation of definitions or procedures in a GIS.  Considering the elements of the GIS that are employed in a particular situation can reveal the negotiations that underlie a GIS implementation.  The configuring of the GIS open black box during the implementation process can serve as a focal point for revealing the negotiation between technical and social components.  Finally, investigating the effect GIS implementation has upon the social element is another focal point that can reveal the nature of the negotiations and the degree of alignment between the social and the technical actors in the network. Bibliography and Resources Akrich, M. (1992). The de-scription of technical objects. Shaping Technology/Building Society: Studies in Sociotechnological Change. W. E. Bijker and J. Law. Cambridge, Massachusetts, The MIT Press: 205-224. Akrich, M. a. B. L. (1992). A summary of convenient vocabulary for the semiotics of human and nonhuman assemblies. Shaping Technology/Building Society: Studies in Sociotechnological Change. W. E. Bijker and J. Law. Cambridge, Massachusetts, The MIT Press: 259-264. Anderson, C. S. (1996). "GIS Development Process: A framework for considering the initiation, acquisition and incorporation of GIS technology." Journal of the Urban and Regional Information Systems Association. 8(1): 10-26. Anderson, C. S. (1996). "GIS Development Process: A Framework for COnsidering the Initiation, Acquisition and Incorporation of GIS Technology." Journal of the Urban and Regional Information systems Association. 8(1): 10-26. Antenucci., J. C. (1991). Geographic information systems : a guide to the technology.   Chapter 10. New York, Van Nostrand Reinhold. Bijker, W. E. and J. Law, Eds. (1992). Shaping Technology/Building Society: Studies in Sociotechnological Change. Cambridge, Massachusetts, The MIT Press. Bowker and S. L. Star (1996). "How things (actor-net)work: Classification, magic and the ubiquity of standards." : http://alexia.lis.uiuc.edu/~bowker/actnet.html. Braithwaite, R. B. (1960). Scientific Explaination. New York, Harper Tourchbooks. Burrough, P. A. and A. U. Frank (1995). "Concepts and paradigms in spatial information: are current geographical information systems truly generic?"  International-Journal-of-Geographical-Information-Systems 9(2): 101-116. Callon, M. (1985). Some elements of a sociology of translation: Domestication of the scallops and the fisherman of St. Brieuc Bay. Power, Action and Belief. J. Law. London, Routledge & Kegan Paul. Callon, M. (1991). Techno-economic networks and irreversibility. A Sociology of Monsters: Essays on Power, Technology and Domination. J. Law. London, Routledge. Callon, M. (1997). Michel Callon's Keynote Speech: Actor-Network Theory -The Market Test. 'Actor Network and After' Workshop., Centre for Social Theory and Technology (CSTT)., Department of Sociology and Social Anthrpology Keele University. Campari, I., T. C. Waugh, et al. (1994). GIS commands as small scale space terms: cross-cultural conflict of their spatial content. Advances in GIS research. Proc. 6th symposium,Edinburgh, 1994. Vol. 1. London, Taylor & Francis. Chrisman, N. R. (1991). Institutional and societal components of cartographic  research. Advances in cartography. J. C. Muller, Elsevier Applied Science, for International Cartographic                     Association: 231-242. Chrisman, N. R. and F. Harvey (1998 (forthcoming)). "Boundary Objects and the Social Construction of GIS Technology." Environment and Planning A. Duerden, F. and R. Kuhn (1995). Indigenous Land-Use Information Project., http://www.geo.ryerson.ca/~gta/html/native/native.html. Gould, M. D. (1992). "Cultural and linguistic aspects of designing geographic information systems for Spanish speakers." Benchmark 1990, Conference of Latin American Geographers. 17/18: 343-348. Guptill, S. C., D. Cotter, et al. (1988). "A process for evaluating geographic information systems." GIS/LIS '88. Proc. 3rd international conference. 1: 145-151. Harris, T. and D. Weiner (1996). GIS and Society: The Social Implications of How People, Space, and Environment Are Represented in GIS., NCGIA Technical Report 96-7. Harvey, D. (1969). Explanation in Geography. London, Edward Arnold. Harvey, F. (1997). Improving multi-purpose GIS design: Participative design. Spatial Information Theory: A theoretical Basis for GIS. S. C. Hirtle and A. U. Frank. Berlin, Springer: 313-328. Hempel, C. G. (1965). Aspects of scientific explanation. New York. Hirtle, S. C. and A. U. Frank, Eds. (1997). Spatial Information Theory: A theoretical Basis for GIS. Lecture Notes in Computer Science. Berlin, Springer. Latour, B. (1987). Science In Action. Cambridge, MA, Harvard University Press. Latour, B. (1991). Technology is society made durable. A Sociology of Monsters: Essays on Power, Technology and Domination. J. Law. London, Routledge: 103-131. Latour, B. (1992). Where are the missing masses? The sociology of a few mundane artefacts. Shaping Technology/Building Society: Studies in Sociotechnological Change. W. E. Bijker and J. Law. Cambridge, Massachusetts, The MIT Press: 205-224. Latour, B. (1997). On actor-network theory: A few clarifications., Department of Sociology and Social Anthrpology Keele University. Latour, B. (1997). Bruno Latour's Keynote Speech: On Recalling ANT. 'Actor Network and After' Workshop., Centre for Social Theory and Technology (CSTT), Department of Sociology and Social Anthrpology Keele University. Law, J., Ed. (1985). Power, Action and Belief. Sociological Review Monograph No 32. London, Routledge & Kegan Paul. Law, J. (1991). A Sociology of Monsters: Essays on Power, Technology and Domination. London, Routledge. Law, J. (1992). "Notes on the theory of the actor-network: ordering, strategy and heterogeneity." Systems Practice 5(4): 379-393. Law, J. (1997).  'Topology and the Naming of Complexity' (DRAFT). http://www.keele.ac.uk/depts/so/staff/jl/pubs-JL3.htm, Department of Sociology and Social Anthrpology, Keele University. Law, J. (1997).  'Traduction/Trahison: Notes On ANT'.,  Department of Sociology and Social Anthrpology Keele University. Law, J. (1997). Traduction/Trahison - Notes on ANT, Department of Sociology and Social Anthropology, Keele University. Lentnek, B. (1992). "The impact of computer based information systems upon Latin American planning agencies." Benchmark 1990, Conference of Latin American Geographers. 17/18: 325-329. Mark, D. M. and A. U. Frank. (1989.). Concepts of space and spatial language. Auto-carto 9. Proc. symposium, Baltimore, MD, 1989. Falls Church, VA., ASPRS/ACSM: 538-556. Mark, D. M. (1992). "Representation of geographic space in natural language, minds, culture and computers." Benchmark 1990, Conference of Latin American Geographers. 17/18: 331-336. Openshaw, S. (1993). "A concepts rich approach to spatial analysis, theory generation, and scientific discovery in GEI using massively parallel computing." Working Paper 93(27). Ramsey, F. P. (1960). The foundations of mathematics, and other logical essays. New Jersey, Paterson. Rudner, R. S. (1966). Philosophy of social science. New Jersey. Rundstrum, R. A. (1995.). GIS, indigenous peoples and epistemological diversity. Cartography and Geographic Information Systems - Special issue on GIS and Society. E. Sheppard and T. Poiker. 22(1): 45-57. Sheppard, E. (1995). "GIS and society: Towards a research agenda." Cartography and Geographic Information System 22(1): 5-16. Smith, B. (1992). "The use of geographic information systems in development planning in Latin America." Benchmark 1990, Conference of Latin American Geographers. 17/18: 337-342. Star, S. L. (1992). "The Trojan door: Organizations, work, and the 'open Black Box'." Systems Practice  5: 395-410. Wynne, B. (1980). Technology, risk and participation: On the social treatment of uncertainty. Society, technology and risk assessment. J. Conrad. New York, Academic Press: 173 - 208.