Work Oriented Objects

Work Oriented Objects Object Oriented Modelling of Computer Mediated Cooperative Activities: An Activity Theoretical Perspective by Ole Smørdal Submitted to the Department of Informatics, University of Oslo for partial fulfilment of the degree of Dr. Scient. December 1998

ii

Summary The topic of this thesis is object oriented modelling as an approach to understanding and improving the fit between computer systems, work and organisations. Object oriented models have been applied to capture aspects of the usage world, such as human work, organisational structure, work procedures, and communication in work. This thesis identifies a problem associated with object oriented approaches in mastering the work and organisational context of the computer systems. This in turn makes understanding and improvement of information systems difficult. On a general level this is due to lack of bridges between social and technical concerns in systems development. On a specific level this is due to lack of concepts regarding the role of the computer within work practices. This thesis contributes to object oriented modelling approaches by developing a theoretical framework for understanding the role of computers in the various aspects of work. Activity Theory is used as a theoretical basis for understanding how artefacts mediate various aspects of human activity. Aspects that are identified are production, distribution, exchange and consumption. The thesis also indicates that integration issues are important since the computer may have a role in several aspects at the same time. An object oriented technique for identifying problems related to poor integration is developed, based on the theoretical framework and drawing of rich pictures. A conceptual framework for object oriented modelling, based on the theoretical framework, is developed in the thesis. Three interacting metaphors for modelling are proposed: Physical modelling, theatrical performances and flow. The conceptual framework integrates concepts from two schools of object orientation. The research is part of an ongoing project extending object oriented languages by utilising this conceptual framework. 14. December 1998 preface.fm iii

14. December 1998 preface.fm iv

Thesis Structure The thesis is based on five papers (Contributions 1-5), and a summary. Summary: Work Oriented Objects, Object Oriented Modelling of Computer Mediated Cooperative Activities: An Activity Theoretical Perspective. The summary reviews each contribution below and explains the overall research approach, research questions and main results. Contribution 1: Human Work as Context for Development of OO-Modelling Techniques. This paper is coauthored with Jens J. Kaasbøll and is published in Brinkkemper S (ed.) IFIP WG 8.1/8.2 working conference on principles of method construction and tool support (Method Engineering' 96), (Atlanta, USA). Contribution 2: The Computer s Incorporated Role in Work. This paper is coauthored with Annita Fjuk and is published in Buch NJ et al. (eds.) Proceedings of The 21st Information Systems Research Seminar in Scandinavia (IRIS'21), Sæby Søbad, Denmark, Department of Computer Science. Aalborg University. Contribution 3: Soft Objects Analysis, a Modelling Approach for Analysis of Interdependent Work Practices. This paper is published in Patel D and Sun Y (eds.) 3rd International Conference on Object-Oriented Information Systems (OOIS'96) (London, UK), Springer-Verlag. Contribution 4: Classifying Object Oriented Analysis Approaches with Activity Theory. This paper is published in Orlowska ME and Zicari R (eds.) 4th International Conference on Object-Oriented Information Systems (OOIS'97) (Brisbane, Australia), Springer-Verlag. Contribution 5: Objects@Work - An Activity Theoretical Framework for OO Modelling of Computer Mediated Cooperative Activities. This paper is submitted for publication. v

vi Thesis Structure

Acknowledgements I wish to express a sincere gratitude to colleagues and friends that have helped me in various ways to write this thesis. First to Annita Fjuk and Jens Kaasbøll, who have coauthored two of the contributions in this thesis. I have learnt a lot from encouraging and visionary discussions with them. Kristen Nygaard has supervised this work. It has been a privilege to work in close contact with one of the inventors of object oriented programming and one of the pioneers of participatory design. Pål Sørgaard and Birger Møller-Pedersen have joined Kristen Nygaard in supervising this work in a support group. Joan Greenbaum and Markku I. Nurminen have given me important pieces of advice throughout the process. Annita Fjuk, Tone I. Sandahl and Markku I. Nurminen have coauthored papers that are not included in this thesis. This was important for the initial development of the contributions. I have learnt research design and research methods from cooperation with Eevi E. Beck, Joan Greenbaum, Jens Kaasbøll and Leikny Øgrim, who also are coauthors of empirical reports that was used in developing the contributions. Tone Bratteteig, Kristin Braa, Anders Mørch, Leikny Øgrim and Jens Kaasbøll have discussed functional integration with me in the FIRE-project, which was the starting point of this research. I would also like to thank members of the GOODS team, in which my current research is done: Anders Torvill Bjorvand, Haakon Bryhni, Dag Sjøberg and Kristen Nygaard. Tone Bratteteig, Joan Greenbaum, Jens Kaasbøll, Birger Møller- Pedersen and Dag Sjøberg made useful comments on the summary of this thesis. I would like to thank the workers and management at the two sites for empirical studies for participating and offering me challenging, real world problems to address in this work. My family have supported me beyond any thankgiving. vii

viii Thesis Structure

Table of Contents Summary... iii Thesis Structure...v Acknowledgements... vii Summary: Work Oriented Objects...1 1 Introduction...1 1.1 Problem area...2 1.2 Goal...3 1.3 The Structure of this Summary...3 2 Modelling in Information Systems Development...3 2.1 Cooperative Work and Computers...4 2.2 Models as «Bridges» in IS Development...6 3 Object Oriented Modelling Related to Work...8 3.1 Shift of Perspectives...8 3.2 Two main schools of OO modelling...9 4 Research Approach...10 4.1 Goal for the Theoretical Trajectory...10 4.2 Goal for the Modelling Trajectory...12 4.3 Research Process...12 4.4 Research Questions...14 5 The Computer s Incorporated Role in Work...15 5.1 Leontiev s view...15 5.2 Engeströms model...16 5.3 Interpretations in this research...17 6 Consequences for OO Modelling...18 6.1 Basic concepts...18 6.2 Models as a means for reflection...19 6.3 Models as a structure for systems...19 7 Conclusions...21 7.1 Methodological contribution...21 7.2 Theoretical Contribution...22 7.3 Contribution to OO modelling...22 7.4 Contribution to OO techniques...23 7.5 Implications for Further Research...23 References...24 9

First Contribution: Human work as context for development of object oriented modelling techniques...27 1 Introduction...27 1.1 Suggestions in the literature...28 1.2 Seamlessness in modelling...29 1.3 Overview of the paper...30 2 The development of techniques...31 2.1 Explanations...34 2.2 Theoretical contributions...34 2.3 Suggestions for change of techniques...36 2.4 Current trend: Method engineering...37 3 The domain of techniques...37 3.1 Human work and modelling domains...38 4 A learning cycle bringing in a semiotic relation...39 5 A learning cycle bringing in activity theory...40 6 Conclusion...42 Acknowledgements...43 Biography...43 References...43 Second Contribution: The Computer s Incorporated Role in Work...47 1 Introduction...47 2 An Activity Theoretical Account of Incorporation...49 2.1 The hierarchical structure of activity...50 2.2 Instrument-mediated action...50 2.3 Operationalising collective aspects...51 2.4 The Problem of computer-mediated collective action...53 3 An Interactionist Theory on Action...55 3.1 Action and interaction...55 3.2 Social worlds and effective communication...56 4 Application of the Interactionist Theory Within an Activity-Oriented Approach...56 5 A Conceptual Framework for The Computer s Incorporated Role in Work...57 5.1 Incorporation...58 5.2 The Main Categories...59 5.3 What to Look for?...60 6 A real-world illustration...62 7 Summary...63 Acknowledgements...64 References...64 10

Third Contribution: Soft Objects Analysis...67 1 Introduction...67 1.1 Research approach...69 2 Shift toward including the work setting in object oriented models.69 2.1 Seamlessness...69 3 Interdependency in work...71 4 Handling differences in world views and creation of visions...74 5 The activities of Soft systems analysis...76 5.1 Draw semi-structured rich pictures of the current work...76 5.2 Create root definitions...77 5.3 Make a chart of current work practices...77 5.4 Identify interdependencies and problems hereof by arranging workshops in and across work practices...77 6 Illustration of use...77 6.1 Using Soft Objects Analysis...79 7 Conclusion...81 Acknowledgements...82 References...82 Fourth Contribution: Classifying Approaches to Object Oriented Analysis of Work with Activity Theory...85 1 Introduction...85 1.1 Research Approach...87 2 Representations of Work...87 3 Computers Incorporated into Work Arrangements...89 4 Object Oriented Analysis of Work in an Activity Theoretical Perspective...91 4.1 Two Schools of OOA...91 4.2 Mediating Action and the Aspects of Work...92 5 Explaining Selected OO Approaches...94 5.1 The Tools and Materials Approach...94 5.2 OOA (Aalborg)...94 5.3 OO Task Analysis...94 5.4 Use-Cases...95 5.5 Object Behaviour Analysis (OBA)...95 5.6 Object Oriented Role Analysis and Modelling (OORAM)...95 6 Conclusions and Future Research...95 Acknowledgements...96 References...96 11

12 Fifth Contribution: Objects@Work...99 1 Introduction...99 2 Understanding Computer Mediated Cooperative Work...100 2.1 Empirical background...100 2.2 Basic Unit of Discourse: the Activity...101 2.3 The Driving Forces within Activities...101 2.4 Multiple Activity Systems...102 2.5 The Tool/Sign Duality...102 2.6 Aspects of an Activity...103 3 OO Modelling Revisited...105 3.1 Shift from the Problem Domain to the Application Domain...105 3.2 Models for Reflection and Models in Action...106 3.3 Three Basic Metaphors for Modelling Activities...107 3.4 Intersecting the perspectives...110 4 Conclusions and Future Work...113 Acknowledgements...114 References...114

Summary: Work Oriented Objects Object Oriented Modelling of Computer Mediated Cooperative Activities: An Activity Theoretical Perspective 1 Introduction The motivation for this research is twofold. Firstly, the thesis relates to fundamental changes within the field of information systems research. Approaches within information systems research are becoming increasingly focused on the use and organisational contexts for the technical systems. Some authors refer to this change in terms of a new paradigm for information systems research (Kuutti 1994; Dahlbom 1996). Still, the field is regarded as design oriented, but it is recognised that the impacts of introducing computer systems are fundamentally organisational and social, and it is generally accepted that designing a technical system alone is insufficient. Hence, there is a need in the field for creating «bridges» between technical, work oriented and organisational concerns in design of computer systems. In this effort, many approaches in the human or social sciences represent a potential for contributions to information systems research. Secondly, in the field of object oriented modelling, there has been a shift of perspectives regarding the role of the computer in work settings: from a focus on the computer as a means of control and administration of a problem domain, to a focus that also includes work and organisational issues. I regard this change of perspectives in object oriented modelling as a response to the general shift of perspectives in information systems research. 14. December 1998 final.fm 1

Summary: Work Oriented Objects 1.1 Problem area The problem area for this research is the fit between computer systems and the work and organisational context in which they are used. In order to improve this fit technical, work oriented and organisational concerns must be balanced in IS development (Braa et al. 1993; Braa 1995; De Michelis et al. 1998). Hence, it seems worthwhile to develop frameworks for IS development that bridge these concerns. This research explores one traditionally important tool in IS development; the use of models as a means in a reflection process. In particular, the object oriented (OO) perspective is considered. Object orientation is a universal modelling approach based on concepts developed in object oriented programming languages. Hence, there is no particular support in the object oriented perspective to address work or organisational issues. Approaches that address such issues have therefore extended object oriented modelling concepts by introducing additional interpretations of the models, such as metaphors (e.g. Høydalsvik and Sindre 1993; Bürkle et al. 1995) or social theories (e.g. Carstensen et al. 1995). However, there are still problems in object oriented modelling in handling the interplay between various aspects in work, the organisational context, and the technical system (Contribution 4). In particular I address two problematic areas of object oriented modelling: Firstly, there is weak support for addressing the role of the computer in work practices, other than to regard the computer as a tool or as a computational or storage resource in work. It is argued in the thesis that the new and emerging situations of computer applications cannot be based on a toolmaking perspective 1 alone. Rather, the challenge is to view the computer as an incorporated part of human activity, merging the dichotomy of nontechnological and technological aspects in a unified whole (Dahlbom 1996). The computer is occupying the role as an instrument mediating both the operational and the communicative aspect of human actions. Secondly, there is a tendency in approaches to object oriented modelling to regard human action as a predescribed (prescribed) sequence of tasks to be performed by a user, thus failing to regard the important dynamics between the planned actions and the unforeseen contingencies that are present in any work (Suchman 1987), and thus require reflection and unpremeditated actions by the workers to get the work done. 1. See e.g. Ehn (1988) on the tool metaphor. 2

2 Modelling in Information Systems Development 1.2 Goal The practical goal of my research is to contribute to a better fit between computer systems and the work practices in which they are used. The research goal is twofold: Firstly, to contribute to an understanding of the role of the computer in various aspects of work, such as individual use, the division of labour and communication. Secondly, to further develop the object oriented perspective to take this understanding more into account in models. 1.3 The Structure of this Summary This summary is structured as follows: In Section 2 the role of models in systems development is discussed, and related to technical, work oriented and organisational concerns. Two schools of object oriented modelling approaches, six selected approaches to OO modelling of work, and a description of what I have called the context problem of OO modelling are presented in Section 3. The research approach and research problems are presented in Section 4. The activity theoretical perspective on computers, work and organisation is developed in Section 5. The consequences for object oriented modelling are discussed in Section 6 by presenting a conceptual framework for modelling, including a structured set of metaphors. Section 7 concludes this summary. Contribution 4, Classifying Object Oriented Analysis Approaches with Activity Theory, is mainly summarised in Section 3. Contribution 1, Human Work as Context for Development of OO-Modelling Techniques, is summarised mainly in Section 4. Contribution 2, The Computer s Incorporated Role in Work, is mainly summarised in Section 5. Contribution 3, Soft Objects Analysis, a Modelling Approach for Analysis of Interdependent Work Practices, and contribution 5, Objects@Work -An Activity Theoretical Framework for OO Modelling of Computer Mediated Cooperative Activities, are mainly summarised in Section 6. 2 Modelling in Information Systems Development This section discusses the needs in information systems research for handling new interpretations of computer use, and the consequences these needs may imply for understanding the role of models. The discussion is continued in Section 3 with a focus on object oriented modelling. 3

Summary: Work Oriented Objects 2.1 Cooperative Work and Computers Information systems typically concentrate on a more or less narrow subfield of the work in an organisation. The author s earlier empirical work comprises many examples, such as applications for document processing, case handling, journaling, map-production, traffic-planning, project management, etc. As a consequence, workers are forced to use several different information systems for different subfields of work, even if many subfields are present simultaneously or alternatively in a work situation. Further, the information systems are often based on incompatible or inconsistent perspectives and heuristics. This put extra demands on the workers because they must shift from system to system, re-interpret data in various systems, and cut and paste data from application to application (Braa et al. 1993). Shift towards Collective Work Support Teege (1996) reports similar problems with CSCW-systems, because different CSCW applications are used in different forms of collaborative work, such as work-flow specification, electronic meetings, and collaborative editing. The resulting problems of different metaphors and user interfaces and of a weak support for data exchange between different applications is one reason for the bad acceptance of CSCW systems in practice (ibid., p. 93). During the last 10 years, computerised tools, such as text processing, drawing and spreadsheet applications have been used to articulate thought and knowledge into texts, drawings and schemes. In the late 90ies these tools are increasingly used in collaboratively based work arrangements free from time and place constraints. The increasing dissemination of networks (including the Internet) encourages organisations to divide their labour force geographically (Greenbaum 1995). These new working environments cause a situation in which texts, drawings and schemes are exchanged and articulated by the personal tools of the individuals. A computer-based instrument (together with other mediating instruments like e.g., paper and pencils) should ideally mediate the actions of the workers, without hampering the alternation between the actions. Schmidt (1994) illustrates this in the following way: «(...) the user should not be required to shift to a special editor and leave the word processor normally used for composing letter, writing report, etc. The same applies to CSCW facilities supporting cooperative authoring, conferencing, etc.» (Schmidt 1994, p. 68) 4

2 Modelling in Information Systems Development The operational conditions of the personal tools may hamper the mediation of individual and collective actions taking place in collaborative arrangements. Individual Work and Articulation of Work 1 In any cooperative work community, the individual worker shifts between individual and cooperative activities, depending on the situation at hand. The boundary between the individual and cooperative work is dynamic, and individual and cooperative work are mutually constituting and delimiting each other (Schmidt 1994). However, individual work has to be articulated in some way to get the whole work done (Schmidt and Bannon 1992; Schmidt 1994). The term articulation work refers to putting together tasks and task sequences in order to accomplish the work (Strauss 1988). According to Schmidt (1994) and Schmidt and Bannon (1992), articulation work has to be taken seriously in CSCW because of its coordination aspect: make individual work available for the cooperative work community, and because there are almost inevitably unexpected contingencies that alter the tasks and operations (Strauss 1985). Articulation work is thus an important aspect of understanding cooperative work as interdependence in work (Schmidt 1994). The term interdependence in work does not necessarily indicate that the individuals are working on the same tasks at the same time, or sharing the same material and resources. The term indicate that the individual workers need to coordinate or articulate their activities and worktasks in some way to get the whole work done. Articulation work cannot be considered as the only aspect of cooperative work, but may be one crucial aspect. Inter-human relations and conflicts in work communities «A community of practice is a set of relations among persons, activity, and world, over time and in relation with other tangential and overlapping communities of practice» (Lave and Wenger 1991, p. 98). Communities, which are considered analogous to community of practice, do not necessarily imply copresence, a well-defined, identifiable group, or socially visible boundaries (ibid.). Work communities can be understood in terms of more informal boundaries of work, but do imply participation in any system of actions about the members shared understanding concerning what they are doing and what that means in relation to the objective of their cooperation. However, the members of a work community may have different interests, make diverse contributions to their actions, hold different goals 1. Ideas and perspectives in this and the following sections are a result of cooperation with Tone I. Sandahl and Annita Fjuk (1995). 5

Summary: Work Oriented Objects and viewpoints and have various functional roles. Because of these differences directed towards the individual member, cooperative work may be a frame in which latent conflicts can come into light (see e.g., Borum and Enderud 1981 in their study of system development projects). Cooperative work is thus a complex whole action involving an intricate pattern of interactions, governed by social roles and inter-human relationships (cf. Easterbrook 1993). Inter-human conflicts are then not understood as the opposite of collaboration, but as a inevitable part of the whole process of cooperative work, sometimes reducing stagnation and promoting change, but at other times disrupting successful cooperation. Our basic view on cooperative work may conflict the widely used term group, much applied in the CSCW-literature. The term is used to designate almost any kind of social interaction or is used to designate a relatively closed and fixed aggregation of people sharing the same goal (Schmidt and Bannon 1992). The group perspective considers cooperative work as a small, stable, homogeneous and harmonious «collection» of people. The shared-goal perspective and group perspective have a close relationship to each other in the way that they both consider a cooperative work arrangement as something outside the individual. «something that is definitionally, independent of individuals but which has pertaining affects on their behaviour» (Fulk et al. 1992 p. 9). The individual of the cooperative work process may have different motives and goals that may be very personal, or represent their organisational unit. These different goals and motives of the cooperative work process may be in conflict with the goals and motives of the official cooperative work settings. In practice cooperation, coordination and communication within a cooperative work arrangement «may reflect pressure, constraints or acquiescence to power as much as shared goals» (Thompson and McHugh 1990, p. 18). 2.2 Models as «Bridges» in IS Development This thesis explores object oriented modelling as an approach for improving the fit between computer systems and the context in which they are used. In particular two «bridges» in IS development are needed: 6

2 Modelling in Information Systems Development Bridge between Concerns related to Computers, Work and Organisation In order to address the quality of the fit between computer systems and the use context, a multi perspective approach 1 is needed, because there are many factors that affects this relationship. More specifically the relevant factors could be classified into three main perspectives in information systems development. Technical perspectives, such as design of databases, user interfaces, applications and networks, and the interoperability between these elements. Work oriented perspectives, such as practices of work, the dynamics between individual and cooperative work, and communication. Organisational perspectives, such as the division of labour, power structures, resources, procedures, and organisational goals. This multi-perspective approach is based on similar proposals in Braa (1995) and De Michelis et al (1998). It is generally accepted that information system developers need a framework to understand and observe the complex nature of work practices, in order to suggest improvements of the information systems in use, or determine the demands for new systems. The framework should be abstract in the way that it let information system developers apply various theories covering the differing aspects of cooperative work they find crucial, and the interdependence between these theories. Such aspects are: Individual work, articulation of work, cooperation and communication, and inter-human relations and conflicts. The framework should also allow freedom so that it does not restrict observations and empirical studies too much. On the other hand, the framework should be sufficiently specific so that it lets developers structure the complexity of cooperative work; the objective being to design computer-based artifacts which mediate crucial aspects of cooperative work. Bridge between Development Concerns and Use Concerns In a traditional view of information systems development, the development process consists of distinct activities with predefined goals. In analysis, the functional requirements of the future system are determined. Models are used both in the process of identifying requirements, e.g. in discussions and negotiations, and may be the product of analysis, because models may document parts of the future system (Boehm 1988). In design, the implementation considerations of the future system are determined. Models 1. See Nygaard (1987) on early multi perspective approaches in Information Systems Research. 7

Summary: Work Oriented Objects are used in the process of identifying requirements for the implementation, e.g. as a basis for prototypes, and as a documentation of implementation considerations. In short, analysis models mediate a vision of what a future information system can do, and design models describe how the system is going to do it. This distinction may be problematic when trying to understand and improve the way computer systems fit with the work practice, because IS development should take the actual use of computers into consideration (Contribution 2). 3 Object Oriented Modelling Related to Work There has been a shift of perspectives in OO modelling, in order to address the organisational context for the technical systems. This shift has lead to what I have called the context problem of OO modelling. 3.1 Shift of Perspectives Various object oriented approaches have been used to analyse the realworld domain that the computer system is intended to maintain information about. Lately, object oriented approaches have also been used to capture aspects beyond this domain, and address the usage world, e.g., aspects relating to actors, communication, articulation of work, collective work, task flow, and work procedures (Contribution 4). This is due to a shift of perspectives regarding the role of the computer in work settings; from a focus on the computer as means of control and administration of a realworld domain, to a focus that also includes the computer as a mediator in the usage world, e.g., as in groupware or workflow applications. This shift of perspectives has not been accompanied with extensions of the theoretical foundation for object oriented modelling (Contribution 1). This is a problem for pure object oriented approaches, because OO is a universal approach, and not particularly targeted toward analysis of the work or organisational context of the computer systems. Therefore, many approaches combine OO with some other perspective, such as: A metaphor, such as regarding workers as skilled craftsmen in a workshop. Models are then made of their tools and materials (Bürkle et al. 1995), or regarding software as a bureaucratic organisation (Reenskaug et al. 1996). An orthogonal model (which is not object oriented) that explains an object oriented model (e.g. Jacobson et al. 1994). 8

3 Object Oriented Modelling Related to Work An existing non OO framework, seen in an object oriented perspective (e.g. Wang 1995). Implications from a social theory, (such as articulation work (Strauss 1993)) (e.g. Carstensen et al. 1995). Six OO approaches that address work and/or organisation are reviewed in (Contribution 4). They all have a too simplistic view of human work, due to formal and/or implementation oriented contributions (e.g. prescriptions of work), and a too simplistic view of the role of the computer in work (e.g. as tool for individual work). It is therefore a context problem both in OO research and in OO methods (Contribution 4). 3.2 Two main schools of OO modelling The OO approaches mentioned above can be regarded as belonging to one of two main schools of object oriented modelling, the physical modelling and the role modelling school (Contribution 4). A similar distinction is made for object oriented programming by Madsen (1995), termed the modelling and the reuse school. There is a potential benefit for combining these schools in order to create more powerful modelling concepts for addressing the work and organisational context (Contribution 4). The Physical Modelling School The modelling school 1 of object oriented programming is developed with real-world modelling in mind, were objects and classes in the models represent phenomena and concepts in the real world (Dahl and Nygaard 1965; Dahl et al. 1967; Nygaard and Dahl 1981; Madsen et al. 1993). Thus, the executing program is regarded a physical model, simulating the behaviour of the real world (Madsen and Møller-Pedersen 1988). This idea, originated in a programming context, was later applied in many object oriented analysis approaches (e.g., Coad and Yourdon 1991; Mathiassen et al. 1993). Due to the focus on modelling the real world, the approaches have poor support for representing how s, i.e. the collaboration among objects to do some job, or the organisation of objects in a large system. The Role Modelling School The reuse school of object oriented programming is developed with organisation of software systems in mind, were collaboration between objects and their responsibility in respect to the operations of the system are focused (see e.g. Goldberg and Robson 1983). Some object oriented analysis 1. Also referred to as the Scandinavian school of OO. 9

Summary: Work Oriented Objects approaches are based on this view, and hence focus on collaboration and responsibilities of objects (Rubin and Goldberg 1992; Høydalsvik and Sindre 1993; Reenskaug et al. 1996). Due to this focus, these approaches are suited for modelling aspects of the application domain, like the work organisation, but they are weak on conceptual modelling of the problem domain. Section 6 suggests a framework for OO modelling that integrates many concepts in these schools. 4 Research Approach Based on the general introduction on information systems research and object oriented modelling in the previous sections, I will now go into more details about the research approach for this work. There have been two interrelated trajectories of research: Development of modelling concepts that address work and organisational issues, and development of a theoretical basis for suggesting such concepts. The latter is important because well-defined concepts for modelling need a theoretical basis (Kaasbøll 1996). According to (Contribution 1), the dominant basis for OO modelling concepts are formal theories. In this thesis, I will suggest an alternative based on a social theory. The motivation for this is the need for developing bridges between issues of work, organisation, and the technical issues. Three case studies have been conducted in order to 1) define problems for theoretical investigations, and in order to 2) apply the modelling concepts I suggest in this thesis in a practical setting. 4.1 Goal for the Theoretical Trajectory In Section 2 it was argued that there were practical problems due to poor functional integration, i.e. the fit between computer systems, work and organisation. I also gave a presentation of why this problematic situation occurs, e.g. due to complex dynamics of individual and collective work arrangements, articulation of work, inter-personal relations and conflicts. Many contributions in OO modelling are currently limited to software engineering concepts, and may thus be limited of implementation concerns (Contribution 1). This thesis suggest that OO development should be open for any theoretical contribution in order to improve the situation, because the complexity of the fit between computers, work and organisation needs to be handled in a more fundamental way. 10

4 Research Approach Traditionally, design based on a perspective emphasising the user s confidence in work is often based on a tool-metaphor. In the UTOPIAproject (Bødker et al. 1987; Ehn 1988) the basic idea was that tools for craftsmen should be the basis for development of computer systems. The basic idea was that computer based applications should be designed on the basis of the traditional practical understanding of tools and materials used within a given craft or profession. The computer applications are considered as tools like hammers, trowels, knives, etc. The attention for the craftsman is related to the work and not the tool applied. Similarly, the computer application as a tool, should allow the user to concentrate on the goal of her actions and not on the tool used to carry out her actions. Another design metaphor applied, relevant for CSCW, is the mediummetaphor. This metaphor considers the computer applications as medium for communication among individuals or organisations that wish to be interdependent in some way. The computer applications based on such a perspective are based on various theories understanding verbal communication. Examples are traditional e-mail systems and the Coordinator (Winograd and Flores 1986). The discussion above indicates that a separate tool-perspective or medium-perspective is not sufficient in design of computer systems for cooperative work. Computer applications have different roles in different situations, and these roles are not necessarily identical (Reder and Schwab 1990). The computer applications have then to be incorporated in various cooperative work tasks to offer future users a confident, flexible and effective cooperative work process. From a review of the literature it appears that relatively many researchers and designers within CSCW assume that cooperative settings are closed rationale systems. Rationale systems describe collectives which are oriented toward the pursuit of relatively specific goals and exhibiting a highly formalised social structure (Kling and Jewett 1994). Closed systems mean that all key influences on behaviour come from within the identified organisation. In this sense a closed rationale system is a task system where upper management guide subordinates toward clear goals through explicit strategies (Ibid.). Within this perspective on cooperative work it makes sense to talk about computer support. Designing a system for support requires that system analysts and designers understand the work practices, schedules, resource constraints, and other contingencies experienced by the people who will use the new computerised system. We believe that this is only possible within a perspective of rationale systems. 11

Summary: Work Oriented Objects In order to move away from the closed rationalistic approach Kling introduced what he called Web-analysis (Kling and Jewett 1994). Webanalysis locates computer systems and people in relationship with other people, organisational settings, and technologies on which they depend. Web-analysis makes it possible to examine how computers and people «work» and how they can effectively be sustained and changed. The main point is to regard computer-systems, people, resources, etc. as part of an infrastructure. When looking at computer-systems as a part of an infrastructure, it makes sense to use the term incorporate instead of support (Contribution 2). When talking about a computer-system incorporated in a social setting, the importance of the computer-system impact on social behaviour is emphasised. The goal of the theoretical trajectory is to develop a framework usable for modelling that from a technical, work oriented and organisational perspective addresses how computers are incorporated in the social setting in which they are used. 4.2 Goal for the Modelling Trajectory This thesis has identified a need of extending oo modelling to include the application domain, and a problem with existing OO approaches that attempt to do so, because they are too restrained by the technical concerns in IS development (Contribution 1). Hence, there is a need to develop theory for OO modelling that take work and organisation more into account (cf. Section 3). The approach taken in this paper is to derive consequences for modelling from a wider theoretical framework that address the fit between computer systems, work and organisation. Generally, new approaches to modelling work are balancing between two pitfalls. On the one hand they contribute to more powerful representational forms in respect to work arrangements; on the other hand they should recognise that abstraction and formalisation of work practices inevitably leave something out of the representation (Bannon 1995). The goal of the modelling trajectory is to develop a conceptual framework for OO modelling and OO techniques that bridges technical, work oriented and organisational perspectives. 4.3 Research Process The approach of my research was developed as a part of the research process itself. The approach is described in (Contribution 1), in which Jens Kassbøll and I discuss how OO modelling concepts could be developed with a basis in knowledge about human work. The paper suggests a learning cycle for development of OO techniques, which consists of four phases and transfor- 12

4 Research Approach mations from one phase to the next, see Figure 1 (Contribution 1). My theoretical and modelling trajectories described above are connected in this learning cycle, by 1) letting theory development be a basis for OO modelling development, and 2) putting OO modelling techniques into practice for evaluation and problem detection. Figure 1. A model of technique development. Construction of refined techniques Modeling in practice test derive hypotesis find breakdowns and anomalies search for explainations Theoretical contributions Evaluation and problem detection Modelling in practice Modelling in practice is the area researchers learn from, and also the area they want to improve. Three case studies have been conducted in order to define problems in modelling. A pilot study of Lotus Notes use in a governmental control agency was conducted in cooperation with Eevi E. Beck. The study focused on the duality within Lotus Notes to be used as a tool and as a medium, and how these roles of the computer were integrated in work practices. Two studies were conducted at a municipal agency in Norway dealing with town planning, building permits, and geodata. The agency has three main responsibilities: Keep new and changed buildings complying with current building standards and plans; make overall and detailed plans for the city areas; and keep geodata for the city up to date. The agency faced 13

Summary: Work Oriented Objects problems in using a central archive, journal and case application because it in several respects did not fit the current work practice, and the management of the agency wanted to improve or replace this computer system. Evaluation and problem detection Evaluation and problem detection are triggered by experiences when modelling is not straightforward. The main concern in this phase is to identify problems that stem from the use of some technique in a practical system development context. The problems may be identified due to 1) breakdown in the use of a technique, e.g., some property of the application domain could not be captured in the model, or the appearance of inconsistencies in the model, or 2) anomalies in the model or in the use of a technique, e.g. the resulting model seems strange compared to the application domain. Problems with OO approaches have been identified (Contribution 4). Theoretical contributions When explaining problems in a scientific way and considering ways to avoid them, one has to consult other scientific results and theories. One may try to explain the problems within the frame of the research or possibly creating new theories. A framework introducing activity theory as a basis for modelling was developed (Contribution 2). Construction of refined techniques. When the appropriate theoretical considerations have been made, one may have to adjust the technique and possibly include new mechanisms, metaphors and notation. Hypotheses concerning the techniques and the approach to evaluate the hypotheses are worked out. An OO technique for needs assessment, based on activity theory and informal pictures rich pictures from SSM (Checkland and Scholes 1990) was developed and applied in practice (Contribution 3). Modelling in practice The cycle restarts with using the technique in modelling. Any kind of practice which contributes to learning about the technique and its place in system development is feasible. 4.4 Research Questions The research questions have developed during the research process. The initial research questions were based on the pilot study and literature studies and indicated that it would be useful to replace the concept of support with a concept of incorporation, thus focusing on how computers are interwoven 14

5 The Computer s Incorporated Role in Work with practice. Hence, this research started early to explore what theoretical accounts could be useful to explain the concept of incorporation. What are the various roles of the computer in a work practice, and how do they relate to each other? How can an understanding of these roles be used in modelling? Further exploration suggested that activity theory could be used for explaining the use of the computer as a tool, as a medium, and as a mediator of the division of labour. Further research focused on applying and interpreting activity theory in modelling. How could activity theory bridge technical concerns and concerns related to work and organisation? Literature studies of strengths and weaknesses of current OO approaches in relation to descriptions of work practices and organisation suggested that the underlying assumptions about work in general are inappropriate in OO analysis approaches, and that further research should be done in order to improve this situation. Hence, my research focused on how OO modelling could benefit from a theory about computers, work and organisation. There are several schools in OO modelling, having different modelling perspectives, and this inspired me to try to combine the physical modelling school and the role modelling school in order to address work and organisation. 5 The Computer s Incorporated Role in Work The new learning and work environments of the 90ies, with increasing use of networked computers, call for an understanding of the computers role in work (Contribution 2). 5.1 Leontiev s view A potential candidate is activity theory. Activity theory, which originated as a psychological theory providing a notion of context to human actions in their environment, in the sense that an activity orients a subject in an objective world (Leontjev 1983). The useful concept for IS development and IS research is the notion of artefact mediation, as people do not relate to the objective world directly, but through artefacts like concepts, heuristics, and tools. One of the claims of activity theory is that the nature of any artefact can be understood only within the context of human activity by identifying the ways people use this artefact, the needs it serves, and the history of its development (Kaptelinin 1996). 15

Summary: Work Oriented Objects An activity constitutes a hierarchical structure with inner dynamics, transformations and its own development. (Leontjev 1983). An activity is realised through goal-oriented processes, termed actions. An action can realise different activities as the given action may fulfil different motives. 5.2 Engeströms model Engeström (1987) presents an extension to Leontjev s model of activity, with three interacting entities (the individual, the object and the community) in order to analyse the social phenomenon of human activity. The objective of the model is to take the social context of human activity into account, by including rules of communication and division of labour. The model is illustrated in Figure 2. Figure 2. The four aspects of collective activity Instrument Subject Production Consumption Object Outcome Exchange Distribution Rules Community Division of labour The upper triangle of the model illustrates Leontjev s basic interpretation of human activity (termed production by Engeström), and denotes the relationship between subject and object. This relation is mediated through instruments. The computer may be regarded as an instrument, as it may be used as a tool to change the object, or as a sign orienting the subject in the objective world. The two other aspects represent the collective aspects of human activity: Distribution denotes the relationship between community (e.g. the workgroup or the employees in the organisation) and an object. This relation is mediated through the division of labour. The computer may be regarded a mediator of this division of labour, e.g. that coordination of work, or distribution of tasks, may be done by means of the computer. Exchange denotes the relationship between the a subject and the community. This relation is mediated through rules of social behaviour and communication. The computer may be regarded as a communication 16

5 The Computer s Incorporated Role in Work channel in this relation. E-mail and conferencing software are examples of this role in the work context. Consumption denotes a meta level aspect, the consumption of the outcome of an activity (such as using a service or a product). In the model, human activity is interpreted as a dynamic interrelation between the four aspects. The extended model of human activity shows that an individual is not isolated but is a part of a community, and the activity is affected by the individual s participation within this community. 5.3 Interpretations in this research (Contribution 2) Engeström s model is not without problems in an IS context. In the paper we point to two problems: Firstly, there is a weak connection between the model and the levels of Leontiev (1983). To overcome this problem, we introduce the interactionist theory of Strauss (1988; 1993), and suggest a partial integration of these approaches. Secondly, the duality between sign and tool (Vygotsky 1978) is not captured for collective work arrangements in the model in Figure 2, because this duality is only present in production, while we argue it is useful to regard instruments as mediators also for distribution and exchange. Therefore we suggest a revised model, in which the role of the computer is integrated, see Figure 3. We operationalise our theoretical framework by suggesting a taxonomy for analysis of a current work practice. Figure 3. Computer applications as instruments in cooperative work tools/signs (e.g. computers) outcome subject exchange production distribution object community 17

Summary: Work Oriented Objects 6 Consequences for OO Modelling This thesis aims to bridge technical, work oriented and organisational concerns in modelling. This section explains how the theoretical framework presented in Section 5 is applied as a basis for OO modelling. This thesis distinguishes between conceptual frameworks and notation/languages, and puts the main emphasis on basic concepts, which is presented first. Then, an OO technique for needs assessment that is developed in this thesis is presented. Finally, an integrated set of metaphors for OO modelling is presented. 6.1 Basic concepts The concept of a system is much used in OO approaches in order to model a part of the world that, for some purpose, is regarded as a whole (Dahl et al. 1967; Holbæk-Hanssen et al. 1975; Nygaard and Sørgaard 1987; Mathiassen et al. 1993). From the discussions in Section 5, it follows that a useful basic unit of discourse in information systems development is the activity. While the system concept is applicable to any part of the world that is regarded as a whole, the activity concept is addressing human actions in a social context. In this thesis an activity is regarded as a specialisation of the system concept, as the system concept is defined in Nygaard (1987). By doing so, alignment with current OO approaches is possible, while using activity theory to explain relationships and dynamics within activities. This implies that models should be delimited by the activity in question. The process of delineation is based on a common object of work shared by a community, because an activity is given by its object (Leontjev 1983). The concept of artefact mediation is particular interesting due to the focus on technology in IS development. The computer is an artefact that may mediate several aspects of an activity, and this is the background for suggesting the concept of computers incorporated into work (Contribution 2). An analysis of the mediating artefacts should identify artefacts, such as tools, rules and the division of labour (Kuutti 1991). When it comes to understanding the computers role in the activity, it is useful to keep the duality of artefacts as tools and signs (Vygotsky 1978) in mind (Contribution 2). The computer may be regarded both as tool and as a sign. When regarded as a tool the computer contribute to the realisation of an activity as executing operations only, because computers do not qualify as subjects. Thus, a human subject should always be identified for all actions. 18