Today I read a paper from 1971 by Horst Rittel and Melvin Webber about “wicked problems” – problems that are intrinsically difficult or impossible to solve in the sense that one can solve a crossword or mathematical proof, or win a game of chess. Wicked problems abound in policy questions and design, and it’s interesting to think about what differentiates them from these other “tame problems”.
The paper defines a wicked problem as one with most of the characteristics in the list below. Bear with me, because being able to spot a wicked problem and thus infer the consequences of that fact is quite a powerful tool for thinking about decision making in pretty much any context. Once you’ve got a clear idea of the concept, you can start seeing them everywhere – in policy such as city planning, in international conflict, project management, personal time management, and even in family Christmases. They’re everywhere, and, unlike tame problems, they’re impossible to solve absolutely, though sometimes they can be resolved partially with relative ease.
- Wicked problems can’t be definitively formulated. To formulate the problem, we must make decisions about how to conceptualize it, limiting our consideration in some way. Different people will conceptualize the problem differently, and these differences might not be readily apparent. Furthermore, a wicked problem may be partially obscured such that we cannot see its full extent without attempting a solution.
- We can’t tell if we’ve completely solved a wicked problem or not. Wicked problems are usually qualitative rather than quantitative and often have fuzzy boundaries. Though we can arbitrarily declare stopping conditions, these conditions are, well, arbitrary.
- We can’t easily rank possible solutions to a wicked problem. Different stakeholders want different things for value-oriented and self-specific reasons. Points 1 and 4 also suggest that the problem will only be known incompletely and there is significant uncertainty about the effects of each solution.
- We can never assess the full effects of a solution. We don’t know for sure in advance whether our solutions will have the desired effect and, after implementing them, its impossible to predict or observe their full consequences over time due to ripple effects.
- Solutions are single-shot. We only get one chance to solve the problem because our solution always changes it, affecting any future solution attempts. This can often make it hard to learn from mistakes.
- The solution set is unbounded. We can’t list or exhaustively describe the possible solutions, so we never know whether we’ve got the best solution (even assuming we can rank the solutions them).
- Every wicked problem is essentially unique. Though two problems may appear fundamentally similar, they are of sufficient complexity such that it is impossible to determine whether apparently minor differences will have any effect. Two cities may be laid out almost identically, but this is not enough for us to conclude that public transport should be laid out exactly the same way.
- Wicked problems are linked to other wicked problems. Almost every wicked problem can be thought of as part of another wicked problem and as having several wicked problems as part of it. Inevitably, this leads to cycles of problems that cannot be solved independently. Crime leads to jail time, which leads to abandoned children, which leads to crime.
- The effects of a wicked problem and its solutions can be explained in many ways. Not only is it hard to formulate the problem and evaluate its possible solutions, it’s very easy, after the fact, to explain its effects in a variety of ways. For example, FDR’s New Deal and WWII spending is hailed by Keynesian economists and left leaning politicians as having ended the Great Depression, while these same policies are damned by market economists as having prolonged its effects. The problem and its solution can be suborned by multiple conflicting narratives, again making it hard to learn from our actions.
- The costs of being wrong are often high. Losing a game of chess doesn’t hurt a lot, but arranging a family get together that goes wrong can be painful. Bad solutions may create problems whose size exceeds the original problem. Given that wicked problems are often policy related, they can adversely affect people’s lives.
This list constitutes a polythetic or cluster definition; that is, problems must have some, but not all of the criteria to be considered wicked. Furthermore, problems possess them to a greater or lesser extent than others, implying the idea of a continuum of problem wickedness. Polythetic definitions are normally used to define complex concepts in philosophy, and the fact that such a definition is required to define wicked problems suggests that they are not a clear or natural category as the paper suggests.
That said, however, the category of tame problems is much clearer. It consists of problems with stopping criteria, clear correctness of outcomes, limited solution action sets with clear results. It seems, then, that wicked problems are perhaps best understood as the set of all problems that are not tame.
One implication of the wickedness continuum is that wicked problems could be made less wicked if we understood the factors that make them wicked. Unfortunately, however, the list of criteria above is primarily descriptive, not explanatory, and so only of use as a starting point. On Tuesday, I’ll be participating in a further discussion on this topic in which I’d like to explore explanations of what makes a problem wicked. This would, I think, give a better definition as well as some ideas for how wickedness might be reduced. Below are some candidate explanations:
- Problem complexity and chaos; in particular the existence of emergent properties and behaviour
- Situation of the problem within an unbounded or open system
- Tight coupling and / or dependence of the components of the problem
- Visibility or availability of information within the problem context
- Conflicting definition or identification of the problem
There’s one last point I want to make. Being written in 1973, the paper gives the impression that the difference between wicked and tame problems maps fairly clearly to the difference between abstract, mathematical or game problems and real political and social problems. It’s interesting to note that in recent years, the term wicked problem has been used to describe problems in software engineering and design that exhibit many of the same properties of the social problems outlined in the paper, demonstrating that it is not abstraction itself that makes a problem tame. It would also be interesting, I think, to look at some of the strategies employed by engineering teams to deal with wicked software problems, and work out if they could be applied to wicked problems in a social or political context. Food for thought, anyway.