“Tools for Thinking: Modelling in Management Science”, by Michael Pidd is an introductory text on how modelling can support decision making and decision makers. Rather than going into excruciating detail on any particular approach, this book is more of a quick rundown of the kind of techniques that are available, along with their features, strengths and weaknesses.
A key distinction Pidd makes in Tools for Thinking is that of the differences between ‘soft’ and ‘hard’ modelling techniques. Hard techniques typically involve a mathematical approach and may involve forecasting or simulation. Soft techniques on the other hand are less widely recognised, and often involve looking at decision making from a human perspective and
Soft Tools For Thinking
Soft techniques and models are used to explore disagreements between people, and the uncertainties that exist about a system. The goal of soft techniques is usually to arrive at an agreed consensus or commitment to action from the relevant parties. Rather than crunching numbers, soft tools generally involve a more facilitative approach to work through the different perceptions of the reality of a problem.
Examples of soft analysis that Pidd gives include:
- Using decision trees to explore the probabilities of different outcomes
- Problem structuring to get to the bottom of ambiguous problems where there may be no one solution
- Soft systems methodology to explore the human processes behind organisations and decision making
- Cognitive mapping and journey making to provide advice on how conflict may be handled in soft systems
- System dynamics to think about how a system might behave under different conditions
Hard Tools For Thinking
Hard tools consist of what might be considered more conventional models. Hard approaches use mathematics, statistics and logic, and rely on quantification to support decision making.
Examples of hard analysis techniques given by Pidd include:
- Linear programming to find optimum solutions to certain problems
- Computer simulation as a cheaper, safer, quicker alternative to running scenarios on a real system
- Heuristic searching for finding ‘good enough’, if not fully optimised solutions.