Bo Strangert (RD14)

On general sources of uncertainty in project planning 

Some recent notes on the planning of development projects have touched upon difficulties of handling task uncertainty (RD12, RD13). The conclusion was that management of uncertainty and diversity is a question of devising appropriate R&D approaches. These should focus on the challenges of complexity, that is, complex adaptive systems characterized by diversity, dynamics, uncertainty, and non-transparency (R13).

Reducing uncertainty

It is reasonable to focus on how to reduce task uncertainty. Handling complex tasks and real adaptive systems require means to infer and control their structures and functions. Characteristics like diversity, dynamics, and non-transparency contribute to the uncertainty or entropy (in terms of information theory) of systems as well as the tasks to handle them. Thus, the line of reasoning here is those complex phenomena are characterized by uncertainty and, therefore, handling them should be based on that insight and not trapped into old or conventional lines of thought.

It can be said that development projects aim at reducing uncertainty about needs or problems by generating means-end solutions. One complication is that the development process itself, although supposed to generate constructive solutions, may introduce latent uncertainty and bias, too.

In particular, it was suggested in RD13 that the use of qualitative modeling could improve the planning of complex social systems and tasks when statistical reasoning is inappropriate because of unique case qualities and contexts.

Qualitative analysis means first that it will be necessary to identify possible sources of uncertainty. An important next step is to try to survey their causal structure and enquire into means of influencing relevant sources. This is of course a very complicated undertaking that hinges on the unique qualities and conditions of the particular case. Probably, no general and straightforward procedure exists. Yet, a preliminary view of miscellaneous analytical problems is possible.

Preliminary locating sources of uncertainty

One prime task in a development project is to explore what is known, believed to be true but not proven, or not known about a given phenomenon or system. Another obvious task is to scrutinize what shall be accomplished and why. The consequential question is how the uncertainty about these two tasks could be reduced.

For example, reflecting on the following simple questions about a case can shed some light on general analytical difficulties in the development process:

What are the sources of uncertainty about

• I - a given existing complex adaptive system or phenomenon in reality?
• II - the expected outcome of a project application to that complex phenomenon?
• III - the entire associated activities of this development process?
• IV - components of the development process, e.g.,
• the set of knowledge domains and expertise in CD&E,
• psychological and social attributes of the project work,
• project management.

Note that the questions I-IV are about categories of representation, not about the order of reflection which should jump freely among the categories. Perhaps it is natural to start with the prescriptions of the project task (II) and reflect on some necessary conditions for task achievement (III), regarding the states of reality (I) and goal fulfillment (II).

Questions of category l are about basic conditions of reality that are relevant for a given project. It may help to think of uncertainty as a lack of knowledge in understanding, predicting, or controlling a complex phenomenon. Identifying a source of uncertainty indicates a need for structure and possibly gives a clue about how to conceive and develop it. Answers to type I questions entail a set of hypotheses about sources of uncertainty, H₁, H₂,…H_k. Preferably, hypotheses should be formulated as cause-effect propositions. It is a matter of accumulating knowledge continuously during the whole development process. The suggested R&D approach is to make this an ongoing and dynamic CD&E effort to collect and test empirical evidence.

However, the sources I-IV are entangled in quite complicated ways. At the project start, sources of category I represent an uncontaminated reality baseline for investigation, but soon they will change dynamically depending on both external influences and project interventions (II-IV). Sources of category II are to a limited degree manipulated through the instruments of category III, although these interact in ways that are difficult to disentangle. As a consequence, task uncertainty is supposed to change dynamically in quality and quantity, which calls for a multi-categorical and interactional analysis.

Thus, the formulation of the category I question is dependent on the object of development (i.e., a product or service according to the project’s goal or task) and its associated category II questions. Previous RD reports commented upon the difficulties of avoiding bias in the initial perspective of the project; this is crucial for establishing a true and project-specific baseline of reality. 

Uncertainty reduction in development tasks can be regarded as decision making: searching, designing, testing, and choice of alternative solutions. The answer to questions of type II entails a set of expected outcomes of applications: E₁, E_2,…E_n. It is advantageous to formulate the expectancies as means-end propositions about how uncertainty can be reduced.  A few are stipulated already in the project directives, but most are constructed during the design process. The conditions of `the formal architecture of a project plan belong to category II (see RD5).

Thus, questions of type I-II are about the object and the goal of development, while questions of type III-IV concern sources of the design process in itself (i.e., the human activities to conduct the development project). The latter questions are nested and can dynamically influence each other, directly or indirectly. For example, conditions of the lower level IVb can influence the higher level III (i.e., the set of all process components except IVb), but III can also influence IVb. The conditions of ’process coordination’ belong to the categories III-IV (see RD5).

Previous notes on this subject emphasized how important it is to form a suitable initial perspective of planning and the need to reconsider and develop this perspective recurrently.

Why conjoint modeling of object development and the design activities?

The suggested R&D approach is difficult to conceive and apply. However, there seem to be some strong reasons to justify it. A few arguments will be summarized here; an extended discussion will follow in forthcoming notes.

First, the development of an object (product or service) that is complex or is planned to exist in a complex context depends intricately on the design methodology used. Sometimes this is not a great problem when proven technology is used. However, it is a major complication when task uncertainty is considerable and related to sources of both complex object qualities and the lack of an established and efficacious methodology for design and evaluation.

Given cases of complex object development, the conventional procedures of theory and methods may not be sufficient to meet acceptable standards of verification and validation of results. This is especially the case when psychological qualities and social systems are central. The outcome, therefore, seems to run a risk of being accidental and inappropriate according to traditional scientific criteria.

In development projects about complex tasks, verification of requirement specifications or goal fulfillment should not be regarded sufficient when

• the original requirement specifications or goal statements are insufficient,
• it is difficult to produce explicit and testable means-end propositions continually,
• the design operations include psychological and social factors,
• external factors interfere during development or when the application context is unpredictable.

The conclusion is that a ”technological approach” with well-known and regularly practiced means is unsuitable because task uncertainty is so large in complex cases. The sources of uncertainty exist in the development object and its context of the application as well as in the methodology of design and investigation. Because these conditions prevail for most complex adaptive phenomena or systems, verification judgments should be subordinate to validation requirements. In this case, validity means that the developed product or service will function in its real context according to the criteria of internal and external validity as well as construct validity.

Therefore, a crucial point will be to develop the methodology of conjoint quality modeling. The efforts should concentrate on means to model and investigate the interaction between all significant design operations and the dynamic conditions of the object of development in realistic contexts.

Some general aspects of model specification

Focus on the object of development means modeling the characteristics of the phenomenon or task to be investigated and designed. Case-based modeling would be preferred because cases include those characteristics that are representative of complex objects and tasks: systems of uniquely interacting components, dynamics, and uncertainty (M1).

Sometimes it can be a kind of ”natural relation” between a case and its context; in other cases, the boundary between them is blurred or artificially imposed (as in most research). If the case is conceived as a complex adaptive system, the boundaries between system and contexts will be specified gradually as a consequence of modeling their structure and interactions.

For example, in the reported case of intelligence operations (see e.g., RD10, RD1, A), there were at least three different perspectives that could be used for system delimitation (focus on functions of tools, tactics, or strategical operations). The solution was a balanced network of the three system perspectives on intelligence operations. Both micro and macro levels of analysis were possible.

Conceptual development involves the gradual construction of multi-categorical structures. One example is the case of intelligence preparation of the battlefield (IPB), which is an instrument for categorizing crucial factors as the first step before a commander makes a joint decision about action (see A2, Section 9).

The relational structures of qualitative factors can formally be suitably and easily represented by Boolean algebra.  Modeling can be further nuanced by extending formalization to include modal logic. For example, the concepts of ’uncertainty’ and ’dynamics’ express modes of possibility and time; such modalities may be formally represented in cases where probability measures are not viable. A strong argument for qualitative modeling is that decisions frequently concern differences between qualitatively structured alternatives. An example is a commander judging the best and worst outcomes of possible actions.

Scenario construction, simulation, and gaming should be used to satisfy the need for a variety of possible courses of events and contexts, including cause-effect structures. Field research is a cornerstone in experimentation; the use of systematical interventions in real contexts and organizations is an incomparable source of knowledge about complex phenomena.

Focus on management and instruments of development means modeling for advancing an innovative and quality assured development process. The reason for joint modeling of the object (product or service) and the means of development is to counteract bias but promote diversity of innovation. Arguments for formal modeling of ’process coordination’ are presented in RD12.

Modeling of process coordination is particularly crucial for selecting and joining sources connected with knowledge domains, agent activities (participation of decision-makers, expertise, end-users), and management procedures.

Exploring qualitative modeling of complex practices. The issue of preconditions for qualitative modeling will be further discussed and reexamined based on experiences from some previous projects about organization development. The outcome will be reported in forthcoming notes.

References on this website

RD1. A case approach to study applied research methods

RD5. Remarks on planning design and dynamics of goal analysis.

RD10. Choice of perspectives on counteracting collateral damage:

A fictitious case of planning command and control systems.

RD12. Perspectives on coordination of a planning process.

RD13. Perspectives on projects for R&D: some issues to be contemplated.

M1. Om fallstudiemetodik och vetenskaplig förankring i professionella verksamheter.

A1. Applied research methods for development projects.