![[Navigation imagemap]](../../images/bookshelf.jpg)
|
Summary of position paper
HA presented some first observation from a central for rescue services (911 central).
The general problem is that all people in the central control room should have some idea of a current event to be handled, at the same time as they should each take an independent responsibility, in order to work efficiently.
During daytime each operator has special responsibility for the fire-brigade, ambulance and breakdown lorries/police/extended services respectively. The specific operator is responsible for knowing where each car in the "his/hers" service is geographically and what each unit is doing. In practice all the operators have got a quite clear image the presence or absence of cars at their ordinary sites, as they seldom ask the responsible person about this.The formal task allocation prescribes a distribution of responsibilities: those who are charged with keeping track of ambulances should only be doing this, others, responsible for fire trucks should only be doing this, etc. This formal distribution is counteracted by a computer system, which is introduced to keep track of the events. The computerised collective memory helps the operators to coordinate their action via the computer, while the prescriptive organisation requires them to either distributing the task immediately or telling each other when any new action is undertaken.
It seems that the computer which is designed to centralise the information in fact also decentralises the information as everybody cannot be updated on everything that is written into the computer.
As a consequence the team runs the risk that, when there is very high work load, all information get very fragmented, that is everyone knows a little about every event, but no one has a full picture of any specific detail.
Presentation and discussion
In his presentation, Henrik attempted to relate the various papers to each other, and questioned the current trend of moving from generalities to particularities. Are we ready for this transition?
Henrik proceeded by claiming that we must have models, that are temporal fixations, of work to proceed and build new system. He proposed that we must both study the particulars and the whole as they go together. This might sound just as one of this clichees. But take as an example the question of responsibility. The organisation often has to know exactly who is responsible, what procedures they have used, (procedures are often designed in order to be able to tell who is responsible) etc. That is why there exist general requirements. If we neglect such and only study particular activities, we might miss quite a lot of important stuff which belong to the cognitive system., e.g. that people forget, do not notice, have a limited attention span. It should also be noted and that responsibility in teams is always distributed over artifacts and people and that the actual decision, or working process is a collective accomplishment. This means that the ideal case of prescriptions and regulative procedures are seldom met.
The emphasis on particulars might have problem for us in designing better system. If we can accept a model then it follows that we can discuss this model and compare with other models in different respects. The conflict between the enthnography informed and the traditional science perspectives might do more harm that good. We should be able not only to say what people do, but also to predict what they can do and what will happen because of the intended changes. For this we need to do models, not only descriptions.
When we say that we should design systems which promote competence, we do not acknowledge that "human-artifcts" are joint cognitive systems. This means that introducing new tools, procedures or technology, forces us to work in new ways. This is not promoting competence. it is simply changing the conditions. After the new artifacts, the actors need new form of knowledge and do not need their old forms. Reification of our views of cognition and computer systems as separate pieces is still restraining our imagination, even if we try to change our concepts by introducing a new vocabulary.
Henrik proposed that when we talk about collective memory and collective learning it is exactly all those things that traditional cognitive psychology has ignored, as they are not inside of peoples heads: symbols, tools, journals, cheat sheets, negotations, gestures, collaborative action etc. Does this mean that we throw all cognitive psychology babies out with the bathwater? Second, we have so many different concepts that only are partly different; common information space, articulation of work, coordination mechanisms, joint cognitive system, distributed cognition, situated action, etc. How about our competence at articulating our own work in the research community?
|to top|
Much of the CSCW effort regarding collective or organizational memory has been directed to develop automation or automated aids that will fulfill the role of a collective memory base usable by a particular work force. There are several problems with this notion:
Distributed cognition facilitates applying a cognitive framework to a system. This directs observation and analysis to focus on the system's representational states and processes relevant to a particular task. The use of an object in a particular situation can be seen as the processes of coordinating with a particular representational state and interpreting it. These processes result in a task relevant representation; that is, a representational state that has been given meaning relative to its use and the situation.
What does this say for designing automation for a complex dynamic system? One thing it points out is that no matter how well designed much of the usefulness of the "collective memory" must come from the users themselves. This also means that the context of use will change over time, whether that change is anticipated or not."
In her presentation, Christine pointed out how we as researchers often look at the objects at a workplace out of their context. The practitioners' use of the objects are contextualized. With the introduction of IT new use develops, but we shall take note of the resources used in the "old" situation so that the new tools do allow these to work. Explaining use is similar to explaining to your mom (somebody who is totally naive).
An important aspect of artifacts is that they perform precomputations for us, so that we do not have to do so much complex reasoning. Memory is always recontextualized, that is adapted to the current situation. The conclusion of this for design is that the appearance of an artifact should support immediate perception and remind persons of what has to be done. Designers of IT-systems often assume that IT is the whole solution and forget the support that already exists. Examples of such support are the slide rule, "cheat sheets", customizing notes.
Another aspect of artifacts are learnability, that is design should be conceived in terms of access (visibility for instance), appropriateness, use.
|to top|
Medical decision is not always the result of an individual process of analysis. In complex, unusual and risky situations, decisions are taken by groups of highly trained specialists who meet in order to reach a common understanding of the case and to decide on a therapeutic procedure. We are studying these situations of collective decision making with two objectives. The first one is to reach a better comprehension of the way in which these groups work in order to optimize their functioning. The second objective deals with knowledge elaboration. Beyond the case-specific therapeutic decision, do practitioners build new general knowledge while processing successive cases ? How could this knowledge elaboration process be assisted ? Our long-term goal is to help the group in practicing a meta-functional activity (Falzon, 1994), that is an activity that, although triggered by the occurrence of specific operational problems, is aimed at the production of new knowledge and new cognitive or organizational tools that can be re-used in later situations.
The text goes as follows. The first section describes the general context of the situation under study. The second section analyzes the functioning of a specific group, and more specifically the elements of knowledge that form the shared knowledge space, the modes of elaboration of that knowledge space, and finally the trace that is presently kept of the work of the group (as compared to the effective functioning of this group).
In his presentation, Pierre Falzon stressed the importance of meta-functional activities. He pointed out that expertise lies in adapting old knowledge to new cases and that knowledge is goal-oriented and context-sensitive. By metafunctionality is meant the elaboration of new knowledge and new tools through a reflexive activity. The following examples were given:
The important lesson is that space has to be provided for these activities.
Catherine Sauvignac presented some interesting details from observations of specialists discussing the treatment in oncology. The therapy is a result of a collective activity, where the disease is evolving, the consequences of a treatment are difficult to predict, and several disciplines must be involved. It was found inter alia that in the writing of a report, to be stored in a decision committree folder, some elements of the discussion were eliminated. Such omitted elements included justifications of decisions, rejected solutions, as well as pieces of general knowledge that were provided in the course of the discussion of a case. On the other hand some information was reconstructed such as elements of the patient's file that were not mentioned in the committee's discussion.
The discussion concerned the changes in collective memory made from meeting to report and the functions of these changes. It was suggested that the changes served the purpose of efficiency but not the purpose of learning.
|to top|
The general question posed was: Do environmental resources have an impact on acquisition of knowledge and on the construction of a collective memory?
The context is a satellite control room,where three controllers work on shift. They have similar competencies and there is no division of labour.
In this particular environment, it seems to be the case that visual and hearing access to everybody's activities make coordination possible. This access is due to the open tools and the open environment.
Two interesting concepts were introduced (both taken from Hutchins, 1990):
|to top|
All organisations are complex and the corresponding control mechanisms, for coordination and cooperation, reflect the relative degree of complexity. Increased complexity gives way for still more complex control mechanisms. A couple of strategies exist in order to handle the control functions. A military command and control (C2) organisation is structured, very information dense and is designed to make possible the coordination in a complex organisation. Division of labor usually is very well developed with clear responsibilities and interfaces between people and organisation. Problems in the usually very dynamic environment are handled along several approaches and defined in terms of actions and counteractions according to real or imaginary plans (the latter when an opponent is studied). Through the use of doctrines, standard operating procedures, rules and training, a basic ability to respond in some way is constructed. This leads to the concept of control of the environment and the devlopment of tactics and means to do it. One theoretical base is systems theory and thinking, which unfortunately seems to be all-encompassing and used as an axiom instead of critically analysing the meaning of it.
In this working paper the military organisation serves as an example of a complex organisation, where a lot of common knowledge and information systematically is developed and maintained for the unity of effort. To a certain extent the occupational culture is visible in the form of heroes, symbols and rituals, themselves different forms of visualising the values of the leaders. The suggestion is that some has lost any active role, while others, but perhaps less evident expressions, still play critical roles in forming the structure which facilitates coordination and communication. They are therefore elements in the control information, and contributes to the organisational learning. This information is part of the organisational memory (OM), and is also a visible common information space (CIS), open to see for members of a unit. The hypothesis is that these expressions and the messages that they carry, are worthwile to analyse if information technology (IT) is considered as a general supporting technology. If not so, some aspect of the information flows or parts of the CIS/OM may be lost and hard to reshape. Possibly the forma-tion of an occupational culture and making it an important part of the CIS and OM is a phenomenon in other organisations than the military, while these reflections have a more general interest. Some observations during a staff exercise are described towards a background of theories about organisations, organisational culture and communication theory. The implications of the observations for development of supporting technologies are discussed briefly.
Per-Arne Persson presented the thoughts behind his position paper and also posed some questions related to collective learning and Information Technology: It is said that "when formed prior to hostilities, coalitions have time to concur on doctrines, tactics, techniques and procedures"
He asked: what time is neccessary? What common techniques, procedures, knowledge is developed?
What about the time needed for review of plans? For wargaming? What about communication, interoperability, liaison?
Coordination is performed by implicit and explicit means.
| Implicit | Explicit |
| Shared mental models |
Requests for transfer
of resources, information |
| Training | Coordination messages |
| Leader's periodic update | Directing-control |
| Anticipation of needs | Standards |
These different means are related to the type of control information in the following way:
|
Control information
Operative (immediate) |
Directive
(long-term) |
|
|---|---|---|
| Explicit | Order, command | Training, goals |
| Implicit |
Procedures
Standards |
Culture, practice
ethos |
The final questions and conclusions were formulated as follows:
The organizational memory should be supported by structure, ethos, organisation, rules and procedures. These form the Common Information Space.
Is the organizational memory mainly concerned with non-dynamic knowledge?
Is collective memory something which takes place simultaneously in several individuals? Is it selective and adapted to the individual's task and situations? Does its value decrease if it is not simultaneous?
These questions were so encompassing that we had to postpone answering them in order to see if we could simplify them by further perspectives on the problem.
|to top|
Collective learning and collective memory are quite complex topics not only because they involve several interdisciplinary aspects that make it difficult even to define them in an effective way but also because it is not clear which are the purposes they serve. In other words, why an organization and its members should pay a price for them? We are not claiming that this question has no answer, rather that the answers are usually very generic and cover a broad set of motivations: among the others, improving the capability of handling exceptions (up to emergency and time-critical situations), reducing the cost of the turn-over, enhancing the consciousness of the organizational members about their dynamic context to improve creativity or positive attitudes to cooperation. The point is that it is hard to imagine a common way to deal with all these requirements in a single approach and strategy for 'realizing' collective learning and collective memory. And, of course, to identify any computational support to these processes.
As a first evidence, each organization motivation for constructing a collective memory carries specific learning techniques and their combination, among the several that have been considered: by teaching, by doing, by examples, etc. As a second evidence, each motivations identifies different pieces of information that is worthwhile to learn and/or to record, and different ways to relate them for future retrival.
Bearing in mind this premise, our focus is on collective learning and collective memory motivated by the encreasing of people's capability to deal with exceptions and critical situations in contexts showing dynamic demands of by the organization and its environment. Moreover, our focus is on articulation work more that on the work content specific to each field of work.
From the articulation work perspective, the successful handling of exceptions and critical situations is the result of many concomitant factors, among which we would like to underline the following:
While not exhaustive the above points show that (part of) the management of emergencies is deeply rooted in the everyday work, either as memory of experiences or as a capability of interpreting of what is going on or is expected in the near future. Consequently, any (technological) support to the emergency management should be deeply coupled with the (technological) support to the everyday work on the one hand, and on the other hand this latter has to be conceived, from the very beginning, as a provider of information potentially exploitable when the critical situation arises.
In her presentation, Carla suggested the following uses of collective memory:
for innovation
for increasing cooperation
for managing turn-over
for handling of exceptions,
for dealing with critical situations
If we imagine that we have to extract knowledge to a collective memory for future use, which knowledge has to be extracted? The knowledge may concern the field of work, and thus be domain specific, and it may concern articulation work, which makes it situation specific. The articulation work is concerned with creating conditions for coordination.
Carla is working with a concept called "coordination mechanisms". By this is meant a set of protocols and artifacts for coordinating behaviours. These structure the information and reflect the behaviour. In a computerized form, the coordination mechanisms should be malleable and linkable. Two types of coordination mechanisms were suggested:
classifications - which are related to the field of work
"workflows" - which are related to articulation work.
|to top|
Peter reported on a study performed in a "microworld" made for studying how people manage their coordination problem in a dynamic environment. Four people with limited views of the world try to coordinated their actions to fight a simulated forest fire. In previous studies it had been found that local coordination was effected to a great extent by communicating intentions, since prediction of the others activity is the best way to handle the delays associated with effort. This particular study was designed to investigate the effect of communicating intentions by visual means rather than by explicit communication. In social perception results indicate that we perceive intentions directly from body movement. A similar perception of intentions was attempted in the context of fire fighting by visualising information about movement of the fire fighting units on a blackboard shown to all subjects in the group. It was found that the efficiency of fire fighting was not increased in the experimental group getting visualized information about intentions. However, the explicit communication of intentions decreased.
|to top|
In this paper we discuss a situation where we apply various computer-based techniques, in particular knowledge-based systems, for supporting knowledge management in organisations. These techniques may be instrumental for collective learning and organisational memory, understood as teh ability for individuals to jointly develop their knowledge and understanding of subjects crucial for a given organisation, as well as the ability for the organisation as such to access and manage the same knowledge.
Three specific issues of salience in this context were raised as topics for discussion, namely
These matters are not well understood in general and viewing them in the context of collaborative work presents additional interesting aspects.
After Sture Hägglund's presentation, some discussion about the utility of a knowledge based approach arose. Some people were critical about the knowledge extraction procedure and the resulting decontextualisation of knowledge. Others were considering the use of knowledge based systems and wondered if there were any detailed studies of such use. Actually, it seems that these issues have not yet been addressed to any great extent.
|to top|
One of my research tasks is to work on the development of so called knowledge graphs. Knowledge graphs can be viewed as a particular kind of semantic networks. Such a graph can be used as a partner in the decision process. I have no empirical data regarding how the cooperative aspects here look like. The knowledge necessary for the decision process has an own nature, and probably needs an own representation. I would be happy to get more insight in all this. The graphs fit in the domain the group is working in.
One of the essential differences between knowledge graphs and semantic networks is the explicit choice of only a few types of relations. In semantic networks most attention is given to the representation of the concepts in the network, while in the knowledge graphs most attention is directed to performing inferences on the relations. The construction of knowledge graphs starts with the extraction of information from texts. The result is a list of concepts, labelled points, and a list of typed links between the points. These form the author graph. The most important type of link is the causal relation. The next step is called concept identification, where various author graphs are combined into one graph by identifying points with each other. Here synonyms and homonymns are taken into account. The result is a compiled graph. This graph is further investigated in a procedure concept integration where one tries to find interesting substructures, and in a procedure link integration where new links are inferred from the given ones. The result is called the integrated graph.
In case cooperation for process management is time critical and expertise is distributed (in terms of competence as well as time and place) it should be possible to integrate specific knowledge one has (to be investigated is which knowledge) with the available graph. Complementing knowledge is integrated, knowledge on which disagreement exists is indicated. This representation is returned to the experts (to be investigated is how this can be done best). The experts can use it in their process of decision making. Because there might be several sources of knowledge it might be a good idea to use this all in a CSCW environment (to be investigated). Alternatives for the knowledge graphs might be cause maps and case based reasoning. To be investigated is why the graphs are preferred.
The work is in process and it cannot yet be judged to what extent aknowledge graph may be useful in a dynamic situation.
|to top|
Kari Kuutti
Dept. of Computer Science & Information Systems
University of Limerick, Limerick, Ireland
and
University of Oulu, Dept. Information Processing Science
Linnanmaa, FIN-90570 Oulu, Finland
E-mail: kuutti@rieska.oulu.fi
Liam Bannon, who is the coordinator of a COST 14 group concerned specifically with "Organizational memory and Common Information Spaces" , presented some existing ideas about organizational memory. In particular, he pointed out two difficulties with the concept: one concerned the "organization" as a collection of individuals versus an entity by itself, the other concerned with the static view of organizational memory versus the dynamic view of organizational remembering. He reminded of the classic work by Sir Frederick Bartlett, which concerned "remembering", not memory, and which, together with work from contemporaries in the Soviet Union such as Vygotsky and Zinchenko pointed to the constructive nature of remembering.
The consequences of this view is that cooperative work cannot be facilitated simply by the provision of a shared database, but requires the active construction of the participants of a common information space, where the meanings of are debated and resolved, at least locally and temporarily.
Thus, a collective memory must to some extent rely upon developing a common meaning. However, this implication conflicts with the idea that "meaning" is developed in use, and that the same data may get different meanings in different contexts. The solution might lie in both developing a common information space and allowing people using it access to the actors and contexts in which the information was originally produced.
Most participants in the group were willing to accept the idea that "remembering" is an important concept to use besides "memory". However, the consequence for collective management of dynamic processes were not easy to see, and we turned to a general discussion.
|to top|
All these reasons are important in the context of dynamic processes.
Accident analyses for instance are supported by the legal demand of storing information in the "black box" of aeroplanes. Analyses of medical cases are supported by the patient journals, which may be used for reflection and learning as well as for justification of actions taken. In a time-stressed situation, the precomputations inherent in routines and checklists serve to alleviate cognitive load from the actors. In situations where the same information is used repeatedly, the reuse of information serves the same purpose.
These situations can then be regarded as "distributed remembering", where the demand on memory is spread over people and artifacts.
The problem of using information storages arises when context of use changes. Then we should learn from the fact that people forget. A lasting collective memory may hinder the natural forgetting caused by changing context. How can we counteract the conservative force lying in a collective information storage? If only minor changes are needed, people must be prepared to update the collective memory. If major changes are required, or even restructuring, the memory repository may be a major obstacle which has to be thrown over board. Who are willing to take that responsibility?
Although the workshop thus ended in new questions, some concepts were better understood after the workshop and the topic of collective handling of dynamic processes was to further illuminated by the discussions held. In particular, the usefulness of collective memory for precomputations in and learning from a dynamic situation should be stressed.
|to top|