These assertions lead us to argue that the spatial character of decision-making is not so special for CSDM as the type of activities to which CSDM applies. As was recently asserted in (Golay and Nyerges 1994) , we consider that land management, including water resources management, land use planning and management, waste management, transportation planning, and environmental impact studies, is a key domain for which decision-making is simultaneously collaborative and spatial. Land management is collaborative because it involves numerous people and organizations, and it is spatial because of the key role of the territory as an integrating factor. Therefore, although CSDM could also apply to other activities, we will narrow the scope of our reflection to land management activities.
This position paper will firstly show the dangers that could arise from considering decision-making as a purely rational process achieved by a "free individual" (the mechanistic point of view), and from ignoring the reshaping role of the organization of which the decision-maker is part (section 2).
As a consequence of this assertion, we have to inscribe a decision-making process within a mechanism to bring to actors within an organization a pertinent institutional knowledge. This shared knowledge is usually called "distributed cognition" (section 3).
But what is the organizational context of land management ? Do we have to consider the land-planning authority ? Or the transportation planning authority ? At the city or at the state level ? One understands that the organization of land management is defined as flexible, mostly informal links among groups of actors belonging to different organizations sharing land management responsabilities within a common territory (section 4).
Unfortunately, most current system design methodologies do not fulfill the requirements of "distributed cognition support systems". The shortcomings of current methodologies and some new methodological trends pertinent to CSDM support system design are described in section 5.
As a conclusion, some issues deserving further research are suggested (section 6).
Conversely, the filtration of the contextual information through the representation of facts in the problem space could prevent managers from responding adequately and creatively to the actual situation. In other words, the reductionism of the approach implies the loss of criticism on the part of decision-makers.
To improve this situation, Wilson & Wilson (1994) suggest a "reflexive learning" approach to decision-making, where the purposefulness of suggested decisions should always be questioned and verified through a critical review within the organization. (Etzioni 1989) suggests, under the name of "humble decision-making", the conscious use of classic decision-making heuristics to prevent reductionism.
The suggested "organizational view" of decision-making does not imply, however, that an organization would itself be able to make decisions. "Organizations do not think or learn, people do" (Simon 1991) . But the ultimate goal is to create a decision-making environment in which organizational members can regard and understand the organization in new ways (Wilson and Wilson 1994) .
Distributed cognition is based on "rich representations" made by individuals through a synergetic combination of action, dialogue and self-reflexion. This process is called by Boland "hermeneutic inquiry". It allows an individual to acquire a better global understanding of the world through an interplay known as the hermeneutic circle. It relies on social interaction, which is a very effective way of learning (Golay and Nyerges 1994) .
In a system to support distributed cognition, the focus should not be on the individual as a decision-maker, but on the individual as a conversation maker (Boland, Tenkrasi et al. 1994) . It should support the ongoing sense-making dialog among organization members. We could suggest that, at the cognition level, a "Distributed Cognition Support System" is to a Decision Support System what, at the information level, an Information System is to Information Processing:
_________________________________________________ || | | || Information-oriented | Cognition-oriented | |===============================================================| | || | | | Individual ||Information Processing| Decision SS | |______________||______________________|________________________| | || | | |Organizational|| Information System |Distributed Cognition SS| |______________||______________________|________________________|An information system allows the members of an organization to share information that can be processed and aggregated by an individual, whereas a Distributed Cognition Support System allows the members of an organization to share cognition that could be enhanced and appropriated by a Decision Support System.
We have to point out that, in this paradigm, "Small-group decision-making" has, in its classic sense, to be classified as a Decision Support System, and not as a Distributed Cognition Support System. Its aim is actually to support discrete group decisions, and not the continuous sense-making process (Boland, Tenkrasi et al. 1994) . But where should we classify CSDM support systems?
Land management activities are shared by numerous different organizations acting within one unique territory. This territory is source of many causal relations among (spatial) entities (Prlaz-Droux 1995) . As an example, rain falling on a street flows to the next sewer pipe, so that transportation planning has to be coordinated with utility planning ("Coordination is managing dependencies between activities" (Malone and Crowston 1994).
One can thus easily understand that the organization of land management is not a fixed one, but is defined (and constantly redefined) as flexible, mostly informal links among groups of actors belonging to different organizations sharing the responsibility of land management within a common territory. These groups of actors, however, can be seen as some type of cross-institutional organizations, to which the organization theory should fully apply (Pornon 1995) . They can particularly be considered as "a Community of inquirers, or a recognized group [...], for whom the evolving image of contingent truths is significant" (Boland, Tenkrasi et al. 1994) ; in other (and more simple) words: the distributed cognition paradigm applies.
If land management activities have to be coordinated, this coordination is often conflictual, due to the different world-views and values of land managers (Golay and Nyerges 1994) . Therefore, negotiation mechanisms (Jelassi and Foroughi 1989) have to be used to bring them to cooperate. Can a negotiation support system be classified as a Distributed cognition support system? The importance of effective social interaction within the negotiation process hints at a positive answer. And is a negotiation process CSDM?
These gaps are especially critical for the design of systems such as Spatial Decision Support Systems, Distributed Cognition Support Systems or CSDM Support Systems, because of the highly social context in which they are generally used.
Some authors have already suggested new approaches to design computer systems matching those requirements: (Rasmussen, Mark Pejtersen et al. 1990) suggest a highly flexible design process aiming to associate agents of the organization to cognitive tasks to be done. (Turk 1992) proposes a Cognitive Ergonomics Analysis Methodology where cognitive task allocation between users and software plays a central role. (Zachary 1988) proposes a design method for Decision Support Systems which emphasizes the role of the human decision-maker; this method facilitates the application of naturalistic decision processes by the human expert and entrusts the machine with the information processing tasks for which the human brain is limited. Finally, (Boland, Tenkrasi et al. 1994) proposes designing principles for information technology supporting distributed cognition, which rely on a strong epistemology of cognition, sociology and decision-making.
However, we do not have any clue as to which method to use for which type of problem or system. When should we promote an approach based on the paradigm of distributed cognition? And when an approach based on the paradigm of decision-making?
Another problem with land management applications of CSDM is to determine the organizational extent of the system. Should a system be designed for one activity? For one organization? For one part of the territory? And how to cope with the causal relations going across the thematic, organizational or spatial limits of the system?
Finally, the interactive tools necessary to support social interaction for CSDM should be identified and designed. Among other such tools, we might mention interactive sketching to support map design through social interaction.
Etzioni, A. (1989). Humble Decision Making. Harvard Business Review (July-August 1989): 122-126.
Golay, F. and T. L. Nyerges (1994). Understanding Collaborative Use of GIS through Social Cognition: "Do You See what I See ?". NATO ARW on Cognitive Aspects of Human-Computer Interaction , Palma de Mallorca, Spain, In Press.
Jelassi, M. T. and A. Foroughi (1989). Negotiation Support Systems: An Overview of Design Issues and Existing Software. Decision Support Systems 1989(5): 167-181.
Malone, T. W. and K. Crowston (1994). The Interdisciplinary Study of Coordination. ACM Computing Surveys 26(1): 87-119.
Martin, J. (1989). Information Engineering . Englewood Cliffs, New Jersey, USA, Prentice Hall.
Norman, D. (1994). Things that Make Us Smart: Defending Human Attributes in the Age of the Machine . Reading, Massachusetts, Addison-Wesley.
Pornon, H. (1995). Structure des organisations et trajectoires de mise en oeuvre des SIRS. Revue de geomatique (paratre) 5(1):
Prlaz-Droux, R. (1995). Conception d'un systeme d'information reference spatiale pour l'amenagement et la gestion du territoire: approche systemique et procedure de realisation. These publie l'Ecole polytechnique federale de Lausanne.
Rasmussen, J., A. Mark Pejtersen, et al. (1990). Taxonomy for Cognitive Work Analysis. Ris National Laboratory, DK-4000 Roskilde, Denmark.
Simon, H. A. (1991). Bounded Rationality and Organizational Learning. Organization Science 2: 125-139.
Tabourier, Y. (1986). De l'autre cet de Merise. Sysetmes d'information et modeles d'entreprise. Paris, Les Editions d'Organisation .
Turk, A. (1992). Facilitating Task Analysis for Geographic Information Systems Through a Cognitive Ergonomics Reference Model. 11th Interdisciplinary Workshop on "Informatics and Psychology": Task Analysis in Human-Computer Interaction , Schaerding, Austria,
Wilson, F. A. and J. N. Wilson (1994). The Role of Computer Systems in Organizational Decision Making. The Information Society 10(3): 173-180.
Zachary, W. W. (1988). Decision Support Systems: Designing to Extend the Cognitive Limits. Handbook of Human Computer Interaction. Elsevier Science Publishers.
He graduated as a Surveyor, and was during 8 years researcher and lecturer in GIS/LIS at the EPFL. He got his PhD with a doctorate thesis on "Modeling of Spatially Referenced Information Systems and of their specialized domains of use: methodological, organizational and technological aspects".
He was then during 3 years senior-consultant by SIT-Conseil, a GIS consulting company in Geneva, Switzerland. He was in charge of the business analysis group, and realized many studies for state and city administrations in the field of GIS planning and GIS application design.
He was elected as a Professor at the EPFL at the beginning of 1994.
Last year, he spent 8 months as a visiting scholar at the University of Washington, working with Prof. Nyerges on social aspects of GIS design and use.
His research interests are going mostly to: