Decision Making / 4 / 4.9 / 4.9.5

4.9.5 Tame Vs. Wicked Problems—Traceability

There is an exhaustive list of permissible operations that can be used to solve a tame problem. There is no exhaustive, enumerable list of permissible operations to be used for solving a wicked problem.

Complex, dynamic problems cannot be encapsulated in what Kline (1995) describes as a complete invariant paradigm. More importantly, the notion of a complete invariant paradigm is an impediment to our thinking. It provides a ‘comfort zone’ where fundamental assumptions are most unlikely to be challenged. Whilst we remain unchallenged, we remain comfortable with no desire to change, thereby holding onto deeply ingrained assumptions and ways of thinking.

A complete invariant paradigm exists when there is a set of laws proven sufficient to faithfully, and completely, describe observed behaviour, and exceptions to these laws have not been found.

Newtons Laws, the physical laws describing motion are an example. Whilst it is acknowledged that Einstein’s Theory of Relativity challenges Newtons Laws under certain conditions, for physical problems at the macro-level Newton’s Laws form a complete invariant paradigm. Problems which fit within a complete invariant paradigm can be solved by an exhaustive list of permissible operations; they are tame.

There are no equivalents to Newtons Laws, no complete invariant paradigm, when it comes to describing the behaviour of complex systems.

Another consequence of the concept of the complete invariant paradigm is that throughout our formal education we are taught to think in the context of closed systems and taking a reductionist approach to problem solving. This involves reducing to component parts, analysing the parts then reconstructing.

Caution is needed here because not all problems are amenable to a reductionist problem-solving approach. Some are not readily reduced to component parts. Even when they can be so reduced, the component parts do not make sense individually. They only make complete sense in the context of the whole. So, frequently it does not make sense to attempt to trace the contribution of each component to total system functionality. This is best demonstrated by an example.

‘A bicycle is made up of a frame, two wheels, pedals, a drive chain, saddle, handlebars, etc. Even when assembled we only have a machine. When we combine the bicycle and rider we also have control and motive power. The resultant combination is a highly efficient form of personal transportation.

The bicycle and rider combine to form a system with ‘emergent properties’.

In this instance, it does not make sense to break down and prescribe a complete functional test for each component of the system. Remove any critical component and the system falls apart (Stevens, Brook, Jackson and Arnold, 1998: 94).’

It does not make sense to prescribe a complete functional test for the brain of the rider. But the rider’s brain processes eyesight, balance, and muscle coordination. This involves complex kinestatic feedback mechanisms. So the rider’s brain only makes sense in the context of the rider/bicycle transportation system. Similarly, systemic ‘wicked’ problems cannot be broken down and solved by application of exhaustive sets of permissible operations applied to discrete elements.

Complex, dynamic problems are characterised by interacting feedback loops. In some systems these can be very large in number. For example, the human body contains over a thousand chemical feedback loops (Kline, 1995).

In general, feedback loops may be positive (reinforcing), or negative (balancing) in nature, made up of series of positive or negative links:

‘a positive link means that if the cause increases, the effect increases above what it would otherwise have been, and if the cause decreases, the effect decreases below what it would otherwise have been … a negative link means that if the cause increases, the effect increases below what it would otherwise have been, and if the cause decreases, the effect increases above what it would otherwise have been (Sterman, 2000: 139).’

Interacting feedback loops, with or without embedded delay mechanisms, form the structures that produce complex, dynamic patterns of behaviour. The dominance of one loop may give way to that of another, positive feedback may give way to negative feedback, and vice versa.

Under conditions of shiftinging feedback loop dominance and stochastic delay, the behaviour of a wicked problem may become exceedingly complex. Wicked problems have been described as non-linear, tightly-coupled, self-organising, adaptive and policy-resistant (Sterman, 2000: 22). Forrester (1971) first described the counter-intuitive response of wicked problems to corrective action: policies or strategies intended to correct a problem result in counter-intuitive response.

The structure and behaviour of wicked problems cannot be described in conventional ways. Similarly, wicked problems are not amenable to conventional problem-solving methods such as linear algebra or laying out the relationships in a spreadsheet where it is not possible to build in the feedback loops. Further, there are only a limited number of clearly enumerated operations which might be performed in the process of finding a solution.

In contrast, strategies for correcting wicked problems are developed through:

• activities which result in mapping the structure of the problem;
• identifying influential mechanisms within the problem space; and
• identifying ‘pressure points’ (Coyle, 1996: 222) or ‘leverage points’ (Senge, 1990: 64; Klein, 1998: 111-119) to which intensive management effort might be applied.

Convention would suggest we might identify all paths through the problem space. Even though the structure of the problem may be described, shifting feedback loop dominance produces time-dependent responses which are more likely to be stochastic than deterministic: traceability becomes problematic. Tests for each possible effect would require tracing through each possible path and doing so for the full range of parametric values likely to be encountered. So, there is no exhaustive, enumerable list of permissible operations to be used for solving a wicked problem.