Modesty in modelling : on the applicability of interactive planning systems : with a case study in pot plant cultivation

This book is a statement about the applicability of automated support for planning. ]be planning situations considered are those that bear upon the next production cycle of an organization and in which a planner has to allocate 'production goods' to each other. Such situations are for instance tactical planning, production planning, scheduling or timetabling. Drawing from various bodies of literature, a marked position is taken. It is argued that the practical impact of planning systems developed by academics is lower than it could be because formal models play too prominent a part during the development process.

Based on views taken from organization theory, psychology and systems theory, a SCanning Aid for Planning SltuationS (SCAPSIS) is developed dig places thew situations on a continuum from 'messy' to 'well-structured'. Not only the problem situation in a strict sense but also the organizational context are taken into account. It is argued that in practice the majority of planning situations are closer to the 'messy' end of the continuum than to the other. The planners deal with these situations as 'open problems', i.e., problems of which they can change the definition at any moment.

A planning system for such an open problem will have to grant the user a prominent role. Usability precedes normative support both m importance and in chronology: a user interface familiar to the planner and close to his train of thought is the first and most important element of the system to develop whereas a formal model of the planning problem is definitely not. Such a model, if created at an early stage, can act as a baffler to communication for both parties. As a result, chances are that the resulting system ends up solving the developer's formal problem but not the planner's open one.

To come up with a usable system, developer and user must closely cooperate right from the start of a project and aim to make a first usable system as rapidly as possible. Only then, on the basis of experience with this system. can it be considered whether to proceed, for instance by incorporating model-based components in the system.

A modelling technique with some promise is the category of interactive heuristics, i.e. heuristics that allow the user to intervene at each iteration. Desirable properties of such heuristics are discussed. It is concluded that this category merits more attention as an element of interactive planning systems.

The stance taken towards planning situations in this study is that they am instances of decision making in general, embedded in the functioning of the organization. With this in mind, a tool is created to help evaluate decision support systems. This 'QUality ESTimator' (QUEST) takes into consideration both the process of system development and the resulting system as it functions in the organization.

A case study in which a planning system was developed to support cultivation planning at pot plant nurseries serves as an illustration throughout the book. The development of this system is an example of how design ideas that are attractive from an academic developer's point of view can fail to succeed in practice, and how cooperative development and modesty in modelling can lead to a simple but usable system.

The ideas expressed in the book can be readily generalized to non-agricultural domains.