in translation
Copyright - 2010, tous droits réservés
© IGD 2010

Decision engineering in a strained industrial
and business environment
Copyright : This article is licensed under the GNU Free Documentation License (GFDL)
Janusz Bucki, SP, Paris, France
Sylvain Roth, Naval support procurement, DGA, Paris, France
In the current highly complex, changing, and exacting environment, we need more efficient approaches to designing and modeling systems and organizations using state-of-the-art information technologies in order to answer the “doing more with less” conundrum. In particular, experience shows that process analysis does not automatically translates in the best information systems and organizations possible. Man Machine Interface is a key issue there and actual user behavior is seldom what the IS or organization designers anticipated - quite often for very valid reasons. Waste of effort or sustaining inefficient activities in no longer endurable. The MOD is engaged in a major BPR effort to increase efficiency and reduce costs. Ships operating and support (overhaul and supply provisioning) is an area of particularly crucial efforts, therefore new ways of ‘getting there’ where searched for. Of particular interest appears the decision analysis approach named after its designer B-ADSc (Bucki - Analyse Décisionelle des Systèmes complexes - decision analysis of complex systems), which is based on new concepts for activity and decision capable ‘autonomous objects’. Although based on a strong theory and using a specific paradigm, the method is readily put to use in a working group environment with people from existing organization supported by a specialist or facilitator of the method. The analysis is documented in a readable, structured way highly suited to provide inputs to simulation techniques and to IS specifications.
The presentation will show how B-ADSc provides a way to overcome the complexity barrier by introducing intelligent actors encompassing the complexity of their own domain of freedom which gives way to both structured and concurrent analysis of processes and organizations. Of particular interest is the reusability of model components and the possibility of an incremental approach. Systems designed by this approach offer good maintainability characteristics as well as safety and fault tolerance features, since human fallibility can be taken into account from the outset. In addition, all aspects of business are considered, from the strategic objectives of top managers to more mundane aspects of decision making or doing the work at the shop level.
1. The essence of the problem
Theoretical concepts used at first to design, assess or maintain automated systems have now extended their influence to cover broad management and organizational areas. In the early 70s, available theories and dynamic models could not describe nor explain correctly the workings of real business operations.
However later advances in Information Processing and Technology, while enabling automation of administration as well as production activities, have later raised new organizational issues and called for deep changes in the very nature of business management and corporate control.

Thus business management and control can be redefined as a means to regulate the purpose driven behavior of any agent or business unit, allowing for their individual strategic objectives.
However the pace of changes in the business world and its volatile environment, as well as in the IT and IDE have hindered the growth of effective, proven tools for optimal design or process reengineering of dynamic enterprises.
The traditional basis of sound management rests on decisions based on predictive techniques and strategic planning, in conjunction with a quest for stability, to lend credence to the available knowledge, tools and their consequences in terms of decisions.
Many believe, like Chandler, that large conglomerates will emerge as a direct consequence of available information technology, as for instance early rail transportation corporations, more or less compelled to expand by absorbing smaller companies, in order to take advantage of long distance telegraph communication services, enabling remote management of subsidiaries.

Can we allow unbounded expansion and scope of IT services to drive alone all the changes occurring in the business world ?
Is organization to be a mere consequence of technology ?
Current Information Systems can really be viewed as complex sociotechnical systems and must therefore drive the organizational change.
One permanent feature in organizational structures is the existence of an operational hierarchy, as a result of work sharing. This operational hierarchy translates in terms of power and authority with higher echelons fixing objectives for subordinates and ensuring the convergence of goals throughout the hierarchical strata. This is critical to safely achieve overall objectives.
Empowerment, or delegation of authority to subordinate levels entails precise links between hierarchical levels and decision categories. Decisions relative to executive corporate management obviously belong to the top echelon, those associated with lower level processes are assigned to lower levels. For the agents within the organization, this implies vast difference in the scope of their area of operation and indeed in the very nature of the information they need to accomplish their missions and objectives.
The effectiveness of the organization is dependent on the coherence of decision chains across all the operational levels and therefore of the comprehensiveness of the sum of all the individual objectives pursued by the agents at any given time.

Well known and supposedly “proven” methods used for BPR are more or less unfit where uncertainty and instability are part of the true paradigm. In the competitive, unstable and fuzzy context, any “efficient” organization must prove reactive and flexible through the sociotechnical system, that is both human and machine interacting components.
Reactiveness is dependent on quick perception of operational context changes, on understanding and assessing its full implications and on the capacity to timely contrive appropriate adjustments suited to both the objectives and available resources in the organization.
Flexibility is in its turn dependent on the learning capabilities of the organization in order to cope with the changing context and hitherto unforeseen situations.
Learning requires the ability to identify elements of knowledge or experimental facts, manage the underlying data and build experience from those memories, transferred from time-limited individuals to the perennial enterprise. This goes far beyond mere technical and data management and may require, to be truly effective, in depth cognitive analysis to establish the appropriate paradigm.
As individual agents learn, their supervisor in turn must acquire new ways to direct their actions while allowing more degrees of freedom. Thus the processes are to be adjusted dynamically to the accumulated experience.

The very incorporation of computer based decision tools into the structure of an enterprise challenges organization experts and their current paradigms as well as their analysis and design tools. Reactiveness, quickness, flexibility, control of thoroughly automated systems can only be achieved through design methods able to integrate and apprehend in toto : “Man and his behavior, Machine and its logic, the Organization and its Information System”. Hence an organization can be viewed either as : ‘decision unit’, ‘storage, retrieval and processing unit’ or as a ‘communication network’.

2. A new "decision centered" analysis method for complex systems.
The design and analysis method named B-ADSc, (Bucki-Analyse Décisionnelle des Systèmes complexes), is a systemic approach, suited to overcome these problems and based on concepts not found in classical ‘functional analysis’ and ‘structured design’ approaches. The cornerstone here is the ‘activity’, a concept upon which any organization may be built. In this line of thought, we relate ‘complexity’ in organizations with the existence of autonomous agents -human or machine- acting together towards the achievement of a common goal. The key point here is the degree of freedom an agent is granted to organize his own tasks, given external objectives to achieve.

An activity will be seen as an elementary center of decision making. We will use general, high level objectives and accumulated knowledge or know-how, to redesign a cost-effective operational hierarchy of activities traditionally associated either with the Management System or with the Information System, now alike in their essence. B-ADSc can supplement Operational Research by allowing to study organizational changes as a result of integrating news processes or ways to do business, which often result from O.R., or knowledge gained from inside or outside experience.

The B-ADSc method allows for representing the organization as a set of tiered operational activities as shown on fig.1. The growth of the organization depends on the design or acquisition of new activities. Their purpose is mainly to allow existing activities to delegate management of sub-processes, hence their associated decisions. As a direct consequence, previously existing activities will stand one level higher in the operational hierarchy. In this approach, delegation of authority or power is the key driver towards more complex organizations, as new activities are created, as shown on fig.2.
An ‘activity’ appears at the intersection of two regulating loops, and is appraised downward, by executives as a means whose behavior is stimulated by commands or objectives, and upward, by the subordinate as a source of new objectives or commands based on decisions made within the activity.
Selected References :
. Chandler The Visible Hand, Harvard University Press, Cambrige, Mass. 1977
. Bucki J., Pecquet P. “l’Analyse Décisionnelle - une Méthode de Conception des Systèmes de Production” - 4th International Congress of
  Industrial Systems Engineering - Marseille / 1993
. Lussato, B. “Modèles cybernétiques - introduction critique aux théories d’organisation” Paris, Dunod, 1992
3. B-ADSc and applications to government activities
The B-ADSc method has already been successfully put to use in various contexts, among which design of flexible process control, improvement of Computer Aided transactions for insurance brokers. Some large companies are currently contemplating using B-ADSc to optimize the organizational changes needed to implement new French labor laws without losing their competitive edge. Wherever people are a critical factor, the method appears a very attractive substitute to others that are proven to fail, to a certain extent, the apothegm “doing right the first time”.

However, in a complex and exacting environment, the choice of any new method cannot be based upon theoretical analysis alone : cultural barriers, the degree of dedication of people involved in conducting the change, the willingness to change their perspective and adopt a cooperative, integrated approach wherever required, are part of various factors to be considered.

Great care must also be taken to ensure the usability of the tools that support the method. The description, textual or graphic, must satisfy both the requirements of software engineers and that of other people : clarity, rigor, unambiguousness, but overall simplicity. The tools must also provide for configuration management and be as far as possible suitable for use through EDI techniques and tools to encompass virtual enterprise and CALS concepts.

The French MOD procurement (DGA) for weapon systems acquisition and support is, like others, put under pressure to improve its effectiveness and provide better support and better solutions to the Forces. Within a global framework, a great degree of freedom is left to components and activities to conduct their required process reengineering. Because Fleet support involves both the Navy, the naval procurement service of DGA (SPN) and industry (mainly DCN), and is particularly strained by the budget restrictions, B-ADSc is used in addition to other, more widespread methods, to broaden the scope of posible changes and promote new concepts supported by inexpensive but fitting simulations. In other areas, B-ADSc is used to explore inexpensively new ways to conduct complex naval operations in a teamed effort, which could result in drastic operational improvements and ship life cycle cost reductions at the same time. This appears as attractive to human factors experts as to designers of naval command and control systems.

To conclude, the systems - technical / human / socio-technical - developed with B-ADSc method should benefit from the same ease of combination and maintenance of components. It is firmly believed that CALS/EDI issues, ILS, man-machine interfaces as well as large organizations and all kinds of information systems could greatly benefit from the new perspective offered by the method.
Decision is no longer viewed as the result of choices or preferences but as the ultimate purpose of a sustained coordination of activities towards the achievement of structured objectives. At the same time reporting to executive levels is achieved in a structured approach that matches the operational hierarchy outlined on the ‘action diagram’. Given a stimulus from ‘above’, a response is expected. In the same way, within the organization, decision making must be checked against the validated objectives of the organization, in a closed loop process that may eventually correct the objectives, to suit the changing context, or to account for the reporting from the activities.

The concept of objective or purpose is pivotal in the decision making process. In the organizational context, objectives tend to be specialized according to the ensuing missions, and structured according to the hierarchy of multifarious activities, to the delegations of responsibilities, to the extension and comprehensiveness of the business executive information system. The most sophisticated approaches to inter-activity communication rest on standardization and require that the objectives are well defined and accepted by all the agents belonging to the organization. The enterprise model must therefore incorporate at least the formal communications mechanisms (actions, decisions, reporting, executive summary). In the language of systemic science, objectives precede and explain decisions. In the language of sociotechnical systems, all the objectives exchange information related to the orders.
Failures in the development projects of executive information systems and data management for the ‘learning enterprise’ may be imputed to improperly accounting for the decision process, although undoubtedly the backbone of the working organization. Obviously, an executive information system cannot work as an add-on to supplement the ‘operating system’ : it must be designed in the organization as an intrinsic component. The B-ADSc method provides a conceptual model able to formalize smart and fashionable production policies, traditionally immune from functional analysis and other modeling techniques. Here, the production policy is no more than a hierarchical set of objectives, hence of decision centered processes that are delegated to an activity based organization. In this context, mechanistic junctions are explicitly controlled by decisional junctions and data flows. As stated before, business management begins at the strategic levels and goes down to the mechanistic junctions with systems or devices dealing with the physical aspects of production.
F6 :
The significant factors here being the number of agents and their autonomy, the B-ADSc method assimilates complexity in organization to the presence of a significant number of agents working together towards common objectives. Our concept of purpose driven organization is indeed similar to that of convergence of objectives in management accounting (cost and profit; performance reporting).

Key parameters of a complex organization are :
- presence of agents (humans or computerized machines), able to pursue their own objectives while keeping in line with shared,
  broader objectives;
- degree of freedom of autonomous agents, dependent on the task distribution and individual knowledge or experience, in the
  management of resources made available to them;
- sharing common resources and capacity, where efficiency results from explicit or implicit procedures, according to the weight
  and significance of the missions or objectives.

A complex organization is capable of :
- transformation, induced by :
  . revised common objectives
  . new individual objectives
  . new knowledge, acquired or derived from experience
  . new resources or tools
  . business process reengineering
- growth, through :
  . mergers and acquisitions of new units
  . alliances, virtual enterprise

Until recently, business organizations have been described through generic functions (procurement, marketing, sales, production, support and logistics,…), whereas analysis of superior objectives allows definition of sub-objectives and their delegation to various activities. Thus so called business functions appear merely as a consequence of the organization and the behavior of the agents. The set of all activities is in itself the management system. This goes as far as mechanistic junctions. In other words, complexity for a business organization lies in the management system, not in the machinery.

The B-ADSc method provides a layout of any organization where decision making is the forefront process. Decision making must not be considered as an incidental or spontaneous by-product of agents but viewed as a part of a bigger operational scope.
F5 :
An activity is physically taken charge of by a pilot, not necessarily human. A pilot can also assume more than one activity; conversely, more than one pilot may team up to deal with a complex activity.

By making an unambiguous distinction between the activity and the pilot, the B-ADSc method allows us to pertinently deal with the three major aspects of an organization :
- operational hierarchy of all the activities, transcribed on the ‘action diagram’, able to express the rationale of the enterprise
  (Why, How)
- organizational structure, outlining the missions and responsibilities or agents and entities,
- executive information system, to provide working and archival data, transmit orders and provide reporting and business
  executive summaries.

The B-ADSc method makes a clear connection between the increased complexity of any business or organization and the growing importance of its ‘autonomous’ agents, allowed to assume their own behavioral changes. Interactions between those autonomous agents must undertake a fundamental mutation with an ever increasing importance vested on the Information System. More precisely, a distinction must be established between two kinds of relationship or intercourse between agents within the organization :

• mechanistic junction, in those instances where the subordinate agent is a mechanism without conscience or the ability to self-assign new objectives. Such an agent is unable to evaluate his own output or work. An image is the piston in an internal-combustion engine whose effort will drive the wheels without any intelligent connection. This is also the premise of Taylor’s organization for the production line where the worker is a mere component of the production system, whose simple tasks must be adjusted to the cadence determined by the line.
The causative sequence in a mechanistic junction is simply : ‘action’
® ‘action’.

• decisional junction, where the subordinate agent is aware of the objectives and missions assigned. He has the authority to determine the proper sequence of actions as well as the ways and means to satisfy the mission requirements. This may be the case of a computer system dedicated to supervise production, given output objectives and rules of manufacturing. It is always the case for workers taken out of assembly line jobs. The causative sequence is here more complex :
® ‘objective’ ® ‘decision’ ® ‘action’ : action gives cause to objective, in turn to give meaning to the agent’s behavior in the knowledgeable conduct of his activity.

In any organizational structure, decisional junctions must be supported by information exchange. A junction with a human being is necessarily decisional. Thus the business organization and the information system must always go together. The B-ADSc method allows us to redefine the Executive Information System as any organization where all intercourses between all kinds of agents (human, machine) are based on decisional junctions.
F4 :
Within any organization, an activity, under supervision by upper levels, is also a ‘decision
unit’, the building block of the decision process as seen from the outside. As apparent of figure 4, the inner workings of this activity, or decision unit is based on two functions :
- decision (Fd) : command of the process, based on directives received from above (to be) and on an internal representation of the process
  under control, (estimated state), and updating expectations (in terms of reporting) for the driven process,
- evaluation (Fe) : based of reporting from the process and inner objectives (or expectations), established during the previous decision. This
  may be also considered as a verification or feed back loop.
F3 :
As shown on figure 3, the activity is defined by three components :
   - purpose : set of objectives or purpose that are assigned to the activity
   - command : management of resources in order to achieve any selected objective
   - means : all kind of resources (human, budget, materiel, information) enabling effective use of knowledge and know-how.
F2 :
prior to new delegation :                                                following delegation :
F1 :