OOPSLA'96Business Object Workshop III


EMPOWER: A Object-Oriented Business Information Systems Framework for Learning Organisations

Nigel Phillips and Dr. Dilip Patel
Centre for Information and Office Systems
South Bank University
http://www.scism.sbu.ac.uk/cios

Email: phillinp@vax.sbu.ac.uk, dilip@vax.sbu.ac.uk

Consultants resources - directory of Consultants related websites.


1.0 Introduction

Object-oriented frameworks offer businesses the opportunity to greatly improve the flexibility and responsiveness of their information systems development effort. Application level frameworks provide generic solutions to well understood tactical problems, interface design, peer-to-peer communications, client-server data manipulation etc. Although successful exploitation of these frameworks is of considerable strategic importance, application frameworks are essentially tactical tools, they address the question 'how' not 'what'. Domain level frameworks fall into two categories, those that support the implementation of well defined strategic options e.g. SEMATECH's Computer Integrated Manufacturing (CIM) 'Works', Taligent (1994) and those that support information architecture development, e.g. Hertha et al (1997). This paper addresses some fundamental problems with architectural frameworks.

Architectural frameworks must enable the modelling of soft, unstructured information, and support planning and design in conditions of high uncertainty. Further they need to simultaneously provide a clearly defined coherent model of the organisation. To be usable by business organisations architectural frameworks must address these issues from within the prevailing business paradigm. The convergence of communication and computing technologies and a managerial focus on process has lead to the conception of organisations as learning organisations where competitive advantage is gained through effective knowledge processing. Nevis et al. (1997) Architectural frameworks must support organisational modelling in these terms, Tricker (1990).

Systems thinking approaches have a good track record of addressing soft, unstructured, 'messy', problem situations. Checkland's (1981) Soft Systems Methodology (SSM) is a popular 'systems thinking' approach to problem solving. Attempts to integrate SSM approaches with information systems design methodologies have proved difficult. These difficulties can be explained in terms of the fundamentally different ontological and epistemological orientation of the SSM and ISD, Jayaratna (1994). ISD adopts a 'realist' attitude while 'systems thinking' approaches are more 'constructivist'.

The root of these problems has been very competently addressed by Brian Cantwell Smith with his philosophy of presence (Smith, 1996). A systematic method for developing a constructive realist approach to the development of organisational engagement is seen as attainable.

This paper demonstrates how Complex Adaptive Systems Theory Gell-Mann (1996), and Smith's 'philosophy of presence' can be used to construct a meta-framework of organisational information systems. Taking Job Design as the fundamental component of Business Organisations. We then demonstrate how the Architectural Framework proposed by Hertha et al. (1997) can be adopted to support this framework. We conclude by identify the tools needed to use such a framework effectively to construct a learning organisation information architecture.

The rest of this paper is structured as follows:- Section 2 reviews object-oriented frameworks. Section 3 examines the main elements of Soft Systems Methodology the difficulties with combining it with ISD modelling approaches. Section 4 briefly introduces our interpretation of Smith's ontology and provides an overview of the complex adaptive systems theory. Section 5 illustrates how adoption of this framework can support the development of an information architecture framework that supports knowledge organisation and organisational learning.

2.0 Object-Oriented Frameworks

A framework provides the structural skeleton within and about which we organise. Frameworks supply structure through the connectivity (of relations) that they provide between concepts (objects) of the area framed. Such structure is amenable to rigorous mathematical analysis Atkin (1974). The concept of an 'Object Oriented Framework' has become very overloaded. In the literature a framework is anything from software that offers flexible solutions to well understood requirements, Taligent(1994), to a strategic policy for understanding information needs (Hertha et al.).

By far the largest effort (particularly from industry) has been in creating engineering solutions to well defined problems, Taligent (1994). From this perspective the main purpose of a framework is to promote code reuse by providing a high level of abstraction for searching object repositories. This approach seeks to document frameworks according to the patterns of object interactions they contain Gamma et al. (1993), Johnson (1992). Such frameworks emerge from developers recognising common structure across implementations and developing class libraries that capture this structure. At the application level a framework typically consists of this class library, a rationale for using it and CASE tools to support its use. Application frameworks do not distinguish between the conceptual model (group of ideas) of the framework and the classes. It is recognised that the success of application frameworks depends on an appropriate domain, business or enterprise modelling framework Taligent (1994).

There is a broad spectrum of approaches to domain level frameworks, distinguished by their relationship to information architecture, and by the business paradigm they employee. Systems integration frameworks aim to implement a particular architectural model, while modelling frameworks aim at safe and effective architectural evolution. The former; build on manufacturing experience, are project based and address tactical concerns, while the latter are associated with commerce, are organisational based and address strategic concerns. The scope of the different approaches reflects differing view of business about the future of organisational structure and the role of information technology. The extremes of the organisation structure debate are well represented in the MIT Scenarios of 'Shifting Networks of Small Firms and All-encompassing "Virtual Countries" Laubacher (1997). Distributed Artificial Intelligence (DAI) approaches believe that competitive advantage lies in replacing fallible people with smart machines, Krin (1997). Human competence approaches believe that competitive advantage lies in a work force with the skills to leverage the machines, Docherty (1997).

Hertha et al. (1997) present an information architecture framework which distinguishes very clearly between the models and the implementation of those models. Their main concern is to ensure that individual application development efforts are selected and deployed so as to address the needs of the business and maximise value. There is a need to avoid code oriented frameworks from becoming 'solutions looking for a problem'. There is little point in improving software productivity if the software produced does not help the business meet its objectives. A central objective of the approach is to improve communication about ends and means throughout the organisation.

In contrast the SEMATECH manufacturing framework is designed to support the implementation of a Computer Integrated manufacturing system for producing semiconductors. It aims to provide an interface structure that allows third parties to design software that can be added to the system on a 'plug and play' basis Taligent (1994). Stewart et al.(1997) Examined an instance of application of a beta version of the SEMATECH 'Works' framework and developed various heuristics as a guide to good practice. The problems these heuristics address fall into two broad categories. Problems cause by the blurred distinction between the framework as implementation and as conceptual model and soft issues of the perceptions, understanding and motivational management of stakeholders.

Stewart (1997) identifies management of soft issues of the 'Works' framework as the most problematic. The evolution of the specification is difficult to manage and they see the need to create a definitive reference model as early in the project as possible. Conformance certification is expensive and requires consensus among stakeholders about suitable tests and fair distribution of the costs of compliance. The "process of awareness, involvement and training is absolutely essential ... we recommend that it be continued and expanded to the limits of the resources available". Hertha et al.(1997) and Ramackers and Clegg (1997) also emphasis the importance of modelling soft aspects. The former through the evolution of models from ideal to pragmatic and the latter through supporting 'what if' business modelling.

3.0 A Hard look at 'Soft' Systems

Checkland (1981, 1990) introduced the idea of 'soft' systems to model aspects of organisations that are not well understood. That is, where there is disagreement or confusion as to what needs to be done and why, where data is unstructured, and working environments are exception rich. In large part Checkland was motivated to develop his Soft Systems Methodology (SSM) because the 'hard' engineering approaches, developed to directly support manufacturing, proved unsatisfactory when applied to less structured business environments.

In SSM the analyst creates Human Activity System (HAS) models of a problem situation from various view points. A HAS models a situation using directed graphs where the nodes are activities and the edges are events (information flows). An activity being considered a system of actions, individual nodes can be expanded and modelled as an HAS in turn.

The Activities are performed by Actors, and result in a Transformation from one state of affairs to another. The performance of each Activity is to the benefit or detriment of some individual or group; the Clients. Individuals and groups within the situation will view the activity according to a particular World view (or Weltanschauung). Some individual or group, the Owners, will be responsible for the activity and it will take place within an Environment which will impose constraints on the Actors.

The analyst first develops a gestalt of the situation and captures the relevant elements in a Rich Picture. A Rich Picture is an idiomatic sketch, that identifies the boundary of the activity, external elements, individuals and groups, activities and events and the relationships between them and highlights areas of concern. The picturing process iterates until the individuals in the situation agree that the Rich Picture is a true representation of their concerns. The analyst then develops a number of Relevant HASs that describe the situation in the picture. Graphs are drawn showing the actions and information flows, and root definitions are developed that identify the Clients, Actors, Transformation, Weltanschauung, Owners and Environmental factors. Monitoring and Control is modelled externally to the activity boundary and considered at three levels, Efficacy, Efficiency and Effectiveness controls. Individual activities within the HAS models are similarly developed and the process iterates until analyst and individuals concerned agree that the models are accurate. The purpose of this process is as much provocative as analytic. It serves to make explicit the differences in Weltanschauung and leads the individuals involved to challenge the understanding of others and question their own. The result should be to identify the root causes of problems. HASs can them be used to explore possible resolutions of the problem.

 

Checkland's work is valuable for two reasons, it provides a conceptual definition of 'system' and makes a clear distinction between reality and our models of it. Checkland's work provides a framework for 'soft' intervention and enquiry. There have been many attempts to integrate Checkland's Soft Systems Methodology (SSM) notably Avison, Wood-harper, 1990, Wilson (1986).

To information systems developers a system is:

"a number of components connected together for a purpose", Pressman (19).

In Checkland's model a system is:

"an abstract idea of a whole having emergent properties, a layered structure and processes of communication and control which in principle enable it to survive in a changing environment" Checkland, (1990).

The differences between these two views are quite profound, Table 3.1.

ISD view of Systems Checkland view of systems
Systems exist in the real world A system is an abstract idea
A component is part of the system if it contributes to achieving the purpose of the system A component is part of the system if it participates in the processes of communication and control
the system is the one true representation of reality the system is one of many interpretations of reality, all equally valid
the analyst seeks the true representation of reality the analyst seeks the most useful interpretation of reality

Table 3.1, comparison of ISD and SSM systems models.

Both approaches lead to a final model which then becomes a working hypothesis of reality and is used to design and plan change. These underlying differences are largely ignored in attempts to integrate SSM into ISD methodologies. Avison, Wood-Harper, (1990) limit their area of concern to computerised information systems, which is defined as a sub set of information systems, which are in turn defined as systems that perform operations on information. Although they adopt the tools of SSM they use the ISD systems definition. It is our view that this ontological difference is the root of difficulties in developing effective soft ISD methods. If we are to develop frameworks that effectively deal with soft aspects then we need a definition of system that includes emergent behaviour and self management and control.

Although we find the SSM systems definition superior to the ISD definition it nevertheless has a number of shortcomings. In particular emergence is not well defined, it is not clear whether the processes of communication and control are part of the emergent behaviour, or in some way superordinate to it. The modelling of these processes leads to problems when shifting from analytical modelling to application design modelling.

 

3.1 Emergent properties

The principle of emergent behaviour is popularly stated as "the whole is greater than the sum of the parts." What this means is open to wide interpretation. Some people would accept the principle but argue that it is only our lack on knowledge about lower level components that make it appear to be true. While others would argue, rather mystically, that it is literally true and no amount of knowledge about components will ever enable full explanation of the emergent behaviour. These difficulties can be resolved if we restate the principle in computational terms.

The principle of emergent behaviour:- there is in general no way that the long term behaviour of a system can be calculated from knowledge of the short term behaviour of its parts.

Computer scientists will recognise this as a restatement of Turing's famous halting problem result. We can therefore treat it as literally true without resorting to mysticism.

SSM follows the cybernetic approach in modelling monitoring and control as separate functional components in feedback and feed forward loops, separate from the transformational components of the system. This is a useful abstraction for analytic purposes. When for instance, we wish to concentrate on the emergent behaviour of an ecosystem, where we wish to sum over the low level processes to understand higher level functioning. Used in this way there is no presumption that there are actual individual components of the system performing this behaviour. It is simply useful to model the system 'as if' they obey rules at the higher level. This is a problem for the ISD analyst leading to designs that run counter to the holistic principles of SSM. If the analysis is to be true to its holistic principles it should aim to identify lower level behaviour that cause the desired global behaviour to emerge. However this would require that short term low level behaviour can be calculated from knowledge of long term behaviour of the system. A moments refection on the number of different ways it is possible to program a computer to produce the same result shows that such identification is not possible.

If our framework is to support the handling of soft issues and ISD development it must enable the development of lower level local components that achieve the desired higher level global behaviour.

3.3 The Objective Observer

A second problem with SSM, particularly when used in ISD, is that the analyst is modelled as separate to, and with an Objective view of, the situation. In SSM the analysts problem is how to help the individuals involved in a situation understand it better. In ISD the analysts problem is to build a computer application on time and within budget that does what people said they wanted it to do. When the analyst analyses they are in the system, and the system they are analysing has an analyst in it. In our early days as computer consultants we were frequently irritated by the additional functionality that clients would 'creep in' whenever it came to finalising a specification. With more experience we realised that we were the cause. By virtue of the questions we had asked and the possibilities we had raised, the clients perception of their world changed, both in terms of what was possible, and what would be most useful. The client couldn't know that they would want X until they understood what it would mean to have Y.

This problem can be resolved if we make systems thinking the users responsibility.

A Root Definition captures an individuals perspective on the work situation. In transferring a description to the model the individual is able to distance themselves somewhat from the situation. By explicitly capturing the individuals Weltanschauung the model remains connected to the reality as experienced by the individual. Modelling the Owners connects the model to the organisation. Modelling the Environment connects to the world it operates in. This serves to connect the less immediate more conceptual aspects of the activity to the reality the individual experiences. The exercise then produces a local disengagement and a global engagement for the individuals involved. The sharing of the models leads to a discourse in what Smith (1996) calls the middle distance. The issues are close enough that they connect to peoples actual experience but far enough away that they can be viewed dispassionately. It is this double facility of SSM: to allow individuals to disengage from their work and consider it objectively, and simultaneously, to pull down and engage with the wider less immediate aspects: that is its main strength.

Therefore we argue that for a framework to support dealing with soft issues it must support disengagement from the local experience and engagement with the global experience, or, more technically, it must support user modelling in the middle distance.

4.0 Complex Adaptive Systems

A Complex Adaptive System:

"acquires information about its environment and its own interaction with that environment, identifying regularities in that information, and condensing those regularities into a kind of "schema" or model, and acting in the real world on the basis of that schema. In each case, there are various competing schemata, and the results of the action in the real world feed back to influence the competition among those schemata" Gell-Mann (1994).

We can form abstract models of such systems characterised by simple sub-processes that interact, in easily understood ways, to cause some form of emergent behaviour. The individual components interact and a global behaviour emerges that could not have been predicted from knowledge of the component parts, and this emergent behaviour then feeds back to the lower level by creating constraining structures.

The emergent behaviour in any number of separate sub-systems can itself become a sub-system that produces a higher level of emergent behaviour. A typical complex system will have several layers discernible by the level of behaviour exhibited by their constituent parts. An important analytical feature of hierarchies is that the behaviour of any particular layer can be examined without the need to understand the behaviour of all the other levels. Although typically the behaviour at the level above and below the level studied will need to be considered.

Fig 4.1 Hierarchy in Complex Adaptive Systems.

Brian Cantwell Smith's (1996) arguments on the nature of metaphysics and reality are both novel and radical. Although we find ourselves in sympathy with his constructions (they "resonate" with our own) we do not claim to have assimilated them to any great degree. Our very sympathy is almost certainly in part due to the sort of inscription errors Smith warns against. Therefore in this analysis we simply identify a number of general evaluative principles that we believe are cognisant of his Philosophy of Presence PoP.

PoP argues that the world is 'One' connected whole, it is not possible for anything to be separate from the world and it therefore cannot be viewed Objectively, not even by a systems analyst. In order to achieve a measure of objectivity this 'Oneness' must be held in mind at all times. Understanding is as much concerned with construction in the world as it is with construction in the minds of individuals. In learning we actively change the world to fit our understanding as much as we change our understanding to fit the world. In order to understand an aspect of the world we must be engaged with it. Engagement necessarily results in an irreducible connection between engagee and engager (subject and object). Such connectedness is essential to an understanding of separateness, and it is the tension between the two that enables the engager to isolate the engagee in the 'middle ground' and register it. Registration is much more than recognition or tokenising. To register an object is not just to know that it exists but to have expectations of it, of one's relationship to it and of relationships to the world and each other. And crucially to see it extended spatially and temporally in both the past and the future. That is to register an apple is to see both an individual apple and a potential source of food, stomach-ache, and a sign of autumn, wind and maggots, the seasons, natures fruitfulness, mans fall from grace with God etc. etc.

A CAS is very much like Checkland's system applying SSM. It is also a PoP registering subject. Schema building is isomorphic to registration. CAS theory provides a number of concepts for explaining the behaviour of CASs.

Attractors in dynamic systems are states into which the system will eventually settle and then remain until some major perturbation throws the system into disarray. Also known as state cycles, this occurs when the system is in a period phase. Each state is connected through the attractor but separate in time. In an office environment Attractors might manifest as routines. In PoP terms, registered as a repeating sequence of activities. When new work is introduced unexpectedly an ad hoc or haphazard set of actions are instigated to deal with it. After a time handling the new work settles into a pattern. As we understand the new work we change the world to accommodate it. The routine established may not be the most effective or efficient but is considerably more so than the ad hoc phase. Once a routine is established it can be difficult to change. PoP explains this difficulty by the connectedness of our registration of the world and our experience of it. Effective change in the routine requires change in the registrations of the individuals involved.

The state space of a system is defined as all the states that a system can be in, arranged so that each possible state is next to all those states that differ from it in only one respect.

Fitness is a comparative measure of how well different systems perform. A fitness landscape is a topographical representation of a systems space where fitter systems configurations form higher peaks (Kauffman, 1993). Systems change can then be considered as movement within state space. Intuitively we can imagine taking any organisation and changing it one small step at a time into any other possible organisation, such a change is termed an adaptive walk. Information systems development can be conceived of as a search in systems space for fitter system configurations and methods as strategies for performing this search. As organisations change, they move through organisation space. Management is then a search through organisation space to create a fitter organisation.

Of central importance to systems engineers is the idea of deforming fitness landscapes, that is as systems change so does the landscape. So that a configuration that is very fit today is likely to become less fit in the future as other systems change and deform the landscape, so there can never be an optimal way of doing things. If the rate of change of the fitness landscape exceeds the development process rate then the systems developed will either be subject to continuous revision, abandoned or if implemented never used.

The organisation seeks to register the world and its place in it, to do this it must engage with the world. Organisations seek to change themselves and the world in order to improve their fitness. Learning is the process of developing fitter registrations. Just by engaging in the world organisations learn. To be a Learning Organisation is to be an organisation that knows that it learns and seeks to improve its fitness through improving its ability to learn. To achieve this the organisation needs to separate its self-knowledge from that of the individuals within it.

 

5.0 An Architectural Framework for Supporting Learning Organisations

In this section we outline our ideas of how the Complex Adaptive Systems model outlined above, combined with our modified version of the Hertha model, provide an organisation modelling framework for learning organisation development. We adopt the ideas of narrative, and discourse to describe monitoring and control and the emergence of knowledge processing behaviour in the organisation. Job design is then identified as the appropriate component level to study. A set of empowering functions is proposed as essential elements in the job design of the knowledge worker. It is argued that effective organisational knowledge processing requires what is otherwise simply good practice to be made explicit through the construction and evolution of a globally (within the organisation) available computer supported model set similar to Hertha's framework. This model set is intended to enable users to view their work in the midddle distance in a similar way to SSM, but, as part of general management practice rather than analytical intervention.

5.1 The Organisational Process

The organisational process is the glue that binds the business processes into a coherent whole. From the point of view of analysis the organisational process has the virtue of being temporally constrained through regulation and common practice. Planning, budgeting and resource allocation are tied to the accounting requirements of the financial year. The structure of the organisational process is therefore quite uniform across a variety of businesses with very different business process structures.

 

5.2 Narratives and Discourse

We use narratives to explain how things are as they are. A narrative tells the story of the organisation, highlighting the important events and their consequences from a particular view point. A narrative represents the political, economic and social environment as understood by individuals and groups in the organisation. Official narratives are enshrined in company reports and strategic plans. In simple terms a job narrative is the story that a member of the organisation would tell a new colleague in order to give them an understanding of the way things are. Alternatively the new colleague constructs narratives to test their understanding, and similarly, all members of an organisation continually refine and revue their narrative in conversation with one another and in the light of experience. This in turn feeds into future assessment of the business and emerges in new company reports and strategic plans.

A narrative can be deconstructed into four threads (or plot lines) each with three temporal aspects the past, present and future. The ideal thread defines the 'good' organisation; how things would be if they were ideal. The state thread describes the external constraints on, and the internal constraints of, the organisation. The pragmatic thread describes what is considered obtainable given the constraints. The operational thread describes current practice.

  Past Present Future
Ideal What the organisation was striving for in the past What the organisation is striving for now The possibilities the organisation might strive for in the future
State The past state of the organisation and its environment The present state of the organisation and the environment. The possible future states of the organisation and its environment
Pragmatic What was believed possible (the strategy followed) What is believed possible (current strategy) What might become possible (strategic options)
Operational How the strategy was implemented. How strategy is implement How strategies might be implemented.

Each individual has several narratives that they use to understand the world. Some of which will be private and not shared with anyone, some will be public within the organisation and some public outside the organisation. For a learning oriented individual no aspect of a narrative is fixed, past narratives are frequently re-assessed as an individuals knowledge and understanding changes. The present is similarly under constant review, as are the options for the future.

The behaviour of the organisation is determined by the interaction of individuals operating according to their personal narratives - the discourse of affairs. Individuals observe this behaviour, they use their narratives to understand it and revise them as they see fit. Strategic management understands the organisation as the sum of these interactions and provides organisational level narratives that explain the organisational view and projects a common vision of the organisation. Management actions are determined in accordance with this common view and the individual views of the implementing managers. This in turn feeds back to individuals and leads to changes in their narratives. Consequent changes in the behaviour of the organisation are compared with this vision and the process continues.

'Visioning' is a management approach to controlling the discourse of affairs. Managers provide the official narratives, which should serve to inform and shape the narratives of individuals. The development of 'Learning Histories' is one such approach, but more generally annual reports and staff newsletters serve to present management's interpretation of past events, indicate what they see as implications for the future and outline the approach to be taken in meeting this future. The extent to which such management vision is effective in directing individuals conceptions will in large part depend on how closely the management interpretation of events and consequences aligns with workers perceptions.

Fig 5.1 Levels of Organisational Behaviour

Figure 5.1 gives a schematic overview of emergent behaviour of an organisation. It is important to appreciate that the behaviour levels relate to behaviour of the organisation as a whole, not to instances of behaviour of the individual members of the organisation. A senior manager thinking or acting strategically is in the lower, operational level. It is from the interaction of the manager with other elements, their understanding and response to her actions from which the strategic organisational behaviour emerges. Similarly an organisation can behave strategically, without individuals doing so. Individuals making reactive responses to the environment can cause a change in the strategic behaviour of the organisation without individuals deciding that it ought to do so.

 

5.3 Job Design

In a review of the impact of IT on job design for managerial, accountancy and blue-collar workers Winfield (1991) detected that middle managers have found their informal peer communications replaced by more formalised structures as strategic management used computers to replace the traditional means of monitoring and control enabling them to be more hands-on in the operational planning. As some compensation for their loss of power and control over workers middle managers have found they have more influence with strategic management in terms of how goals should be achieved. In contrast both accountancy and blue-collar workers have seen a growth in the importance of informal social interactions as a result of IT introductions. In general he concludes that while IT has had a significant impact on job design with much greater autonomy for workers in areas such as task co-ordination and operational decision making, management practice has failed to accommodate these changes for the benefit of the organisation as a whole. He concludes "People, then, are doing different tasks often requiring a different emphasis on skills, but the power contexts in which they are doing these remain the same. Despite opportunities things remain unchanged. Traditionally technology comes first. Work design is seen as a later, non-essential development; to a certain extent it has even been perceived as something of a luxury. Implementation is seen as an unproblematic activity, as simply a mechanical process governed by strict Taylorist principles" Winfield (1991).

A systemic approach to Job Design makes feedback part of the job. Workers should see the results of their efforts, evaluate them for themselves, and have the responsibility to instigate change for the future. Management reviews aim to evaluate outcomes both planned and unplanned, disseminate policy, assess future needs and identify opportunities for development, plan future activities and set targets. In a command oriented organisation staff development is concerned with ensuring that staff have the minimum skills required to deliver the desired outcomes as determined by the command structure. In a learning oriented organisation staff development aims to maximise the range within which current skills can be exercised and identify opportunities for developing new one. The former concentrates on ensuring that the job that must be done gets done, it ensures standards are maintained. The later that everything that can be done is done, it ensures standards are continually raised. These are matters of organisational culture. A learning organisation needs a conceptual framework that empowers workers to continually improve their jobs and organisational performance. Evaluation of management practice in a number of organisations leads us to propose seven functions that should be part of every job design in a learning organisation.

5.4 EMPOWERing Job Design

Every job design should require of the worker that they perform the following seven functions, and actively improve their ability to do so to the maximum of their ability.

In a learning organisation workers must be given as much responsibility for managing their own work as they have the competence to deliver, and should aim to raise the workers level of competence to their full potential. This requires that they have available to them the appropriate knowledge of the organisation and their role in it. That is the organisation, through the management structure, must make available to workers, not just knowledge of what has been decided, but the possibilities being considered. This inevitably increases the level of uncertainty a worker will experience. To compensate for this the worker must be able to influence the choice that will be made.

 

5.5 Adapting the Information Architecture Framework to Fit the Learning Organisation Meta-Framework

The Information Architecture Framework (IAF), Hertha et al. (1997) identifies three models, Reference, EndState and Deployable each with four tiers, Strategic, Process, Application and Technology Fig.5.1. The reference model is ideal and abstract it defines the basic elements relevant to the information needs of the enterprise. It identifies and relates the kinds of information we are concerned with and provides "common concepts and language which multiple groups of people can work with". The EndState model is Ideal and concrete, it identifies the information system we would like to implement. It presents the reference model in terms of component descriptions. A component exhibits certain behaviour under certain conditions it might be software, human or fields of barley. The deployable model is pragmatic and concrete, it represents the actual production systems which 'satisfy the end state to a greater or lesser degree'. The models aim to provide a central point of reference and a commonality of language that ensures good communication and continuity between disparate areas. IAF separates component definitions (which it models in OO fashion) in to functional requirements (what it must do) and operational objectives (how it must do it). The models are evolved through revisions in the light of experience and changes in the environment.

Fig. 5.1Architecture Framework Hertha et al. (1997)

Fig. 5.2 The Reference Model. Hertha et al. (1997)

There are some problems with the Hertha framework.

Communication between the Strategy tier and the technology tier is mediated through the application and process tier. Higher tier request "what within budget" lower tiers respond with "How at a price" Fig. 5.2. This is contrary to the experience of strategic management: "It is becoming clear as firm's consider their future in an information era of superhighways, multimedia and information richness that IT executives should contribute more positively to management thinking by identifying the business threats and opportunities posed by IT. It is evident now that technology influences strategy as well as the other way around." Rockart (1996)

Although it allows purely human activity to be modelled at all levels it does not address job design or human resource management. We feel that this approach is greatly enhanced if, instead of tiers, a network structure is employed (Fig. 5.3). There is little point in deploying technology and applications if you don't have people with the right skills in the right place to use it effectively. Ensuring the right skills is a major objective of reformulated IT steering committees in many leading edge companies Rockart (1996).

Models are not routinely evaluated, revisions are only made when resources cannot deliver the EndState requirements or new strategic requirements emerge. As such it provides a poor support for the organisational process. The management process takes a strategic view of processes, applications, technology and skills 'side by side'. "In an effective relationship, IT and line managers need to work together to understand business opportunities and to choose among technology options, to determine needed functionality, and to decide when urgent business needs demand sacrificing technical excellence for immediate, albeit incomplete, solutions." Rockart (1996). Strategic management requires how at a price information to be developed across functional boundries. The what/how dialogue then wrests the deployable model from the reference model through a consideration of the organisation from the various viewpoints: Strategic, Process, Application, Technology, Physical and Human resources. The tiers would be better modelled as viewpoints that the management process refines from reference views to deployable view (i.e. plans) Fig. 2.3.

Fig. 5.3 A network structure to replace Hertha's Tiers

The overall structure of the model is laid out in the mission statement goals and objectives of the organisation and the Business Unit, Section and individual job designs. This forms the root narrative or initial hypothesis. The management process, evaluation, review planning and resource allocation refines this model through the EMPOWER framework. The three models, Reference, EndState and Deployable are constructed and maintained. The Deployable model is the working hypothesis, what the organisation is actually trying to achieve. It is this model that systems development should address. Individuals maintain a number of alternative hypothesise in addition to the deployable model which are contingent on change.

In an IT rich environment where the desktop computer is used to perform, or record most of the activities in the job design their is a strong isomorphism between the job design and the state of the desktop interface. The EMPOWER models can drill down to the messy actions. Actual activity can be connected to modelled behaviour.

It is the aim of our research to create the tools and data structures that will enable users to Flip to the nice crisp conceptual model and Flop to the messy spongy reality. That is to support engagement and disengagement so that an individual can hold their personal work experience and the organisation experience in the middle distance and consider them both objectively.

6.0 Conclusions and Future Work

Effective deployment of application frameworks requires a domain level architectural framework. Such frameworks must be able to deal with 'soft' information and support planning and design in conditions of uncertainty. 'Systems Thinking' approaches offer the potential for dealing with soft issues but integrate poorly with Information Systems Design approaches. In part this is due to the different goals of an SSM and ISD analyst. The framework needs to handle emergence in a constructive way that avoids translation from a control to a command system. Effective handling of soft issues requires that workers can disengagement from the local experience and engagement with the global experience, or, more technically, it must support user modelling in the middle distance. In addition to the multiple view point of workers the framework must also support a single deployable model that ISD practitioners can use for application development planning. Adopting a Complex Adaptive Systems view of the organisational process enables soft issues to be handled by users as part of the management process. This requires that the whole management process including physical and human resource management as well as IT/IS resources be part of the framework. This can be achieved by concentrating on Job Design as the main component of organisations. Such an approach allows the development of a Learning Organisation by clearly distinguishing between individual learning and organisational learning and making organisational knowledge explicit. The Architectural Framework of Hertha et al. can be adapted to accommodate these changes. For such an approach to be effective integrated tools support is required. From a system developers point of view it is important to appreciate that development is a continuous process. Identification of objects and actions is not enough it is the dialogue about these elements that defines them.

"It is a conceit to assume that the meaning of discourse is built up out of the atomic and stable meaning of words, as we are classically taught. There is every reason to believe that the meaning of the words is just as much affected - adjusted, focused, stretched, made more precise - by the meaning of the embedding discourse."

We have developed a meta-framework that equally supports the viewing the system from a realist ontological perspective of ISD and the constructivist epistemic perspective of systems thinking and demonstrated how this can be incorporated into a domain level framework.

The type of thing that worker wants to register about their work are the types of thing that systems developers want to model. The nature of OO languages is such that it is a relatively straightforward matter to create base object types that implement interfaces for a variety of applications, so that the objects a worker creates when using diary and planning software, can also be queried as an object model of the domain.

Fig 6.1 Main components of freeBASE

Our approach is to develop intranet browsers that support the planning and organisation of work through the desktop with management review and strategic planning software built in. The browser can be re-arranged so that it fits the job of the user. Its state then becomes a model of the job design. All work on the desktop is save under the different activities supported by the job design and form a record of the objects, events and actions that compose the job. The narratives are then passed to a Chronicle Temporal Database Server written in JAVA and stored on an NT Server via Back Office.

The browser must also support:

This paper has been concerned with developing the conceptual framework needed to underpin the browser and interface with Chronicle in a way that supports both operational and design viewpoints. Our next task is to develop some prototype browser components and develop the narrative data structures. At the same time we aim to ensure that it supports the performance measures of Earl's (1996) Information Equity Model.

To date the Chronicle TDMS has only been implemented in SmallTalkDB (Schleifer 1997) and has to be ported to JAVA and interfaced to Back Office.

Hanley's (1996) work on 3D conceptual Modelling of large unstructured data has already started to deliver serviceable tools for data mining and there are possibilities for creating synthetic narratives.

Usability is obviously a major concern particularly with regard to the determination visibility and granularity. We are currently seeking business partners so that we can conduct field trials and add an aspect of action research to the development.

References

Atkin (1974) Atkin, R.H. Mathematical Structure in Human affairs. Heinemann Educational Books.

Avison, Wood-Harper (1990) Avison, D.E. and Wood-Harper, A.T. Multiview, An Exploration in Information Systems Development. Alfred Waller Limited.

Checkland (1981) Checkland, P. Systems Thinking, Systems Practice. John Wiley.

Checkland (1990) Checkland P. Scholes, P. Soft Systems Methodology in Practice. John Wiley.

Earl (1996) M.J. Earl. Information equity A Model for Measuring performance in the Information Age. London Business School. CRIM WP96/2. URL http/www.lbs.ac.uk/crim/.

Docherty (1997) Docherty, Peter and Nyhan Barry. Understanding Industry in Transition. In Human Competence and Business Development: Engineering Patterns in European Companies. Eds. Peter Docherty and Barry Nyhan. Springer.

Enterprise (1994) Enterprise Partners. Enterprise State of the Art Survey. The Enterprise Survey, University of Edinburg.

Gamma et al. (1993) Gamma, Erich. Helm, Richard, Johnson, Ralph and Vlissides, John. Design Patterns: Abstraction and Reuse of Object-Oriented Design. ECOOP '93 Springer.

Gell-Mann (1994) Gell-Mann, M. The Quark and the Jaguar. Little, Brown and Company (UK).

Hanley (1996) Hanley, D. Patel, S. A 3d-Visualisation System for Large WWW Based Document Collections. In Proceedings of 1996 International Conference on Object Oriented Information Systems. Eds. Patel, D. Sun, Y. Patel, S. Springer.

Hertha et al (1997) Hertha , William F., Bennett, Jim E., Post, Frank J, Page Ian M. An Architecture Framework: From Business Strategy to Implementation. In Business Object Design and Implementation. OOPSLA '95 Workshop Proceedings. Springer.

Johnson (1992) Johnson, Ralph. Documenting Frameworks using Patterns. Proceedings OOPSLA '92. Springer.

Kauffman, (1993) Kauffman, S. The origins of Order. Oxford University press.

Krin (1997) Kirn, Stephan. Cooperative Knowledge processing - Research Frameworks and applications. In Cooperative Knowledge processing:- The Key Technology for Intelligent Organisations. Eds Stephan Kirn and Gregory O'Hare. Springer.

Laubacher (1997) Laubacher, R.J. and Malone, T.W. Two Scenarios for 21st Century Organisations: Shifting Networks of Small Firms and All-encompassing "Virtual Countries. MIT Initiative on Inventing the Organizations of the 21st Century, Working Paper. URL http://ccs.mit.edu/21c/21CWP001.html.

Pressman (1994) Pressman, R.S. Software Engineering: A Practioner's Approach. European Edition. McGraw Hill.

Ramackers Clegg (1997) Ramackers, Guus. And Clegg, Dai. Object Business Modelling, requirements and approach. In Business Object Design and Implementation. OOPSLA '95 Workshop Proceedings. Springer.

Rockart (1996) Rockart, J.F., Earl, M.J., Ross, J.W. The New IT Organisation: Eight Imperitives. London Business School. CRIM WP96/3. URL http://www.lbs.ac.uk/crim/

Schleifer (1997). Schleifer, P. A Chronicle Approach to Modelling Temporal Database Objects. PhD Thesis South Bank University.

Smith (1996) Brian Cantwell Smith. On the Origin of Objects. MIT Press.

Steward St.Pierre (1997) Stewart, S.L. and St. Pierre, James A. Experiences with a Manufacturing Framework. In Business Object Design and Implementation. OOPSLA '95 Workshop Proceedings. Springer.

Taligent (1994) Talegent Inc. While Paper:- Building Object Oriented Frameworks.

Wilson (1990) Wilson, B. Systems: Concepts, Methodologies and Applications. Second edition. John Wiley.

Winfield (1991) Organisations and Information Technology - Systems, Power and Job Design. Blackwell.


OOPSLA'96Business Object Workshop III