Wicked problems and the framing of the problem

Year 2 / Module 9 Interaction design methods and critique / Unit 3 / Wicked problems and the framing of the problem

HCI2101 · Unit 3 of 10

Design problems are not given but constructed, and because problem and solution co-evolve the engineering model of defining then solving misdescribes design; the wicked problems concept is a useful warning against technocratic confidence and a weak diagnosis, because it neither grades wickedness nor says what to do.

Wicked problems and the framing of the problem

In 1973 Rittel and Webber gave planning a phrase for a suspicion designers already held. The hardest problems are not solved by defining them and then applying a method. Ritchey (2013) reproduces the ten criteria by which Rittel and Webber (1973) characterised such problems. A wicked problem has no definite formulation, no stopping rule and no test of a correct solution.

This unit, the third of the ten, makes three claims. Design problems are constructed in the act of addressing them, and they are not given in advance. Problem and solution co-evolve, so the engineering model of define then solve misdescribes what designers do. The wicked problems concept is therefore a well founded warning against technocratic confidence, and it is also a weak diagnosis. It neither grades how wicked a problem is nor says what to do about one. The useful successor is not the label. It is the move the label points at, the framing of a problem. The two units before it examined a method and the school that sells one; this one turns to the problem they take as given.

The problem is not given, and expert designers frame rather than define

Head (2022) reconstructs the original argument. Scientific and technocratic approaches to social problems are inadequate, because they overlook different stakeholder perspectives in the framing of those problems. Reliable scientific knowledge fits only issues whose variables are measurable. Head (2008) makes the difference operational as complexity, uncertainty and divergence between stakeholders. Buchanan (1992) carried the term into design, arguing that design has become a new liberal art of technological culture.

Ritchey (2013) draws out the consequence. Rittel and Webber objected that the systems approach assumed a planning project could be organised into distinct phases. For wicked problems, they said, this scheme does not work:

One cannot understand the problem without knowing about its context; one cannot meaningfully search for information without the orientation of a solution concept; one cannot first understand, then solve.

The point is not that planning is hard. The point is that the first phase of the canonical process, understanding the problem, presupposes the answer it is supposed to produce.

The claim is also what the studies of expertise find. Cross (2004) reviewed research on expert and novice designers. Novices work depth-first, while experts work top-down and breadth-first. He notes that the protocol studies of design bear the contrast out only in part. More telling, experts move rapidly to early solution conjectures. They use those conjectures to explore and define problem and solution together. Cross concludes that problem framing captures best what designers do.

The phase model misdescribes design because it fixes the problem first

Ritchey (2013) notes that problems are wicked and tame only a potiori, with a gliding scale between them. Coyne (2005) presses further. Revisiting the 1973 essay, he argues that wickedness is not aberrant. He argues that the formulations of professionalism are the deviations, because they treat formal rules and calculation as the norm of rationality. In his pragmatic reading all problems are wicked, and the distinction dissolves.

That costs the label its information. But the dissolution redirects attention. It moves from a class of problems to the relation between a problem and those who must act on it. Ceschin and Gaziulusoy (2016) give the design-side version. Reviewing design for sustainability, they categorise its approaches under four innovation levels. They show the field moving from a product-centric focus towards system-level change. The further a problem sits from the product, the less the literature says about what to do.

The label is a warning, not a diagnosis

Two questions must be separated. Does the label warn against something real? Does it diagnose a case well enough to guide action? The literature supports the first and withholds the second. Lonngren and van Poeck (2021) analysed a sample of 55 papers. They found that the term is not always applied consistently, that many meanings attach to it, and that no consensus exists on its utility. Peters (2017) states the sharper version. The concept has become a fad, and many problems labelled wicked fail a strict definition of it. Rastegar (2024) lists insufficient definition, wrong application and neglect of reflexivity among the criticisms. He reports that a spectrum of wickedness is now accepted.

The term does one thing well: it refuses the assumption that the problem statement is given. It does one thing badly: it does not tell a team what to do on Tuesday.

A workable diagnosis grades complexity, uncertainty and stakeholder divergence

Three proposals make the diagnosis do work, and they disagree about where it should sit. Farrell and Hooker (2013) reject the claim that design is demarcated from science by having wicked problems. Each of the ten features derives from agent finitude, system complexity or problem normativity, which science and design share. No binary demarcation follows, and wickedness becomes a matter of degree.

van Bueren, Klijn and Koppenjan (2003) locate the diagnosis elsewhere. They analyse an environmental decision from a policy network perspective. They explain impasses and breakthroughs through cognitive, social and institutional factors. Head (2008) offers the version closest to an instrument, since complexity, uncertainty and divergence can each be graded low, medium or high. His own illustration shows the grading in use, and a stakeholder who cannot read the consultation document, or answer it on a slow connection, is not in the count. The two cases he works through are Indigenous disadvantage and climate change policy. He reads Australia's refusal to ratify the Kyoto Protocol between 1996 and 2007 as a political symbol of value divergence rather than a failure of evidence. Climate policy is several linked problems, and they cannot be settled one at a time. Turnbull and Hoppe (2019) push furthest, reframing wickedness as problematicity, the political distance between those who question it.

The constructive answer is framing

If the diagnosis is weak, the affirmative claim must stand alone, and it does. Dorst (2015) describes frame creation as a nine-step approach to open, complex, dynamic and networked problems. He argues that it focuses not on generating solutions, but on creating new approaches to the problem situation itself. Instead of labelling a problem, the designer changes the frame in which it is posed. Cross (2004) supplies the independent evidence that the move is characteristic of expertise.

The limit is documented too. Ritchey (2013) presents general morphological analysis as a method for structuring social messes. It decomposes a problem into parameters and possible values. A frame can be built and explored, and no structuring method decides whose aims prevail.

Contested question: is wickedness in the problem or in the relation to it?

The first position holds that wickedness is a property of certain problems. Its canonical form is the ten criteria reproduced by Ritchey (2013). Its modern version is the graded characteristics in Head (2008). A problem can therefore be assessed before anyone discusses it. The second position holds that wickedness is relational. Coyne (2005) argues that the rationalist professional model, not the problem, is the outlier. van Bueren and colleagues (2003) find impasse and progress explained by network interaction. Turnbull and Hoppe (2019) define problematicity as political distance. A third position cuts across both. Farrell and Hooker (2013) argue that the distinguishing features are not special to design at all.

What would settle it is an instrument and a comparison the literature lacks. The first position makes a prediction. Independent assessors, given a problem statement and no access to stakeholders, will agree on its complexity, uncertainty and divergence. Their score will predict how the project unfolds. The second position predicts that one statement given to two stakeholder groups will score differently. It predicts that a change in the network, rather than the statement, will change the outcome. Neither prediction has been tested at scale. Head (2008) and Farrell and Hooker (2013) supply the dimensions to be scored.

The two positions rest on different evidence. The property position rests on a set of conceptual criteria and a graded definition, so it describes problems without having been used to score one and then watch the project unfold. The relational position rests on one network case analysis and two philosophical reframings, which are interpretive rather than comparative. The only counting in the section is a mapping review of 55 papers, and it reports inconsistency rather than a test. Until a study scores a statement with independent assessors and then follows the project, the designer should grade the problem on complexity, uncertainty and divergence and record the stakeholder divergence separately, so that if the two disagree the record shows which one moved the outcome. The strongest objection to the unit's thesis is that a graded diagnosis by an outsider may predict outcomes better than the framing move the unit recommends, since the dimensions are at least open to scoring while frame creation has never been compared against a problem statement left alone.

Where the argument is heading

Named work points in one direction. Head (2022) records that scholars have worried the term has become a catchword. They have called for the stark contrast to be refined. Rastegar (2024) reports that a spectrum of degrees of wickedness is now accepted. Ceschin and Gaziulusoy (2016) show the problem levels design now addresses.

The sources this unit rests on

Buchanan, R. (1992). Wicked problems in design thinking. Design Issues. doi:10.2307/1511637 Buchanan carries the wicked-problems vocabulary into design, arguing design thinking places problems within four orders from signs to systems, and that designers have no privileged access to the wicked problems they address. Conceptual essay.
Rittel, H. W. J. and Webber, M. M. (1973). Dilemmas in a general theory of planning. Policy Sciences. doi:10.1007/BF01405730 Rittel and Webber argue that policy problems are wicked rather than tame: they cannot be definitively described and have no solutions in the sense of definitive and objective answers. It is a conceptual essay that names ten properties. It is the primary source for the criteria the unit discusses. It measures nothing and was not tested.
Coyne, R. (2005). Wicked problems revisited. Design Studies. doi:10.1016/j.destud.2004.06.005 Coyne revisits Rittel and Webber and argues the wicked-versus-tame contrast rests on a residual rationalism, so the professional formulations are the deviation and all problems are wicked. It is a conceptual critique.
Farrell, R. and Hooker, C. (2013). Design, science and wicked problems. Design Studies. doi:10.1016/j.destud.2013.05.001 Farrell and Hooker reject the claim that design is demarcated from science by having wicked problems: each of the ten features derives from agent finitude, system complexity or problem normativity, which both share. Conceptual analysis.
Head, B. W. (2008). Wicked problems in public policy. Public Policy. https://www.semanticscholar.org/paper/Wicked-Problems-in-Public-Policy-Head/566b52b2bfe2219cf0388c2af5f76bd6104dbe13 Head makes the difference operational as complexity, uncertainty and divergence between stakeholders, each gradable, and illustrates the grading with Australian policy cases. It is a commentary, not an empirical study.
Peters, B. G. (2017). What is so wicked about wicked problems? A conceptual analysis and a research program. Policy and Society. doi:10.1080/14494035.2017.1361633 Peters argues the wicked problems concept has become a fad and that many problems labelled wicked fail a strict definition of it, then sets out a research programme for testing which genuinely are. No test is offered.
Turnbull, N. and Hoppe, R. (2019). Problematizing wickedness: a critique of the wicked problems concept. Policy and Society. doi:10.1080/14494035.2018.1488796 Turnbull and Hoppe reject the basis of wickedness on philosophical and practical grounds, proposing instead that problems be characterised by the uncertainty and disagreement they carry. It is a critique, not a study.
van Bueren, E. M., Klijn, E.-H. and Koppenjan, J. F. M. (2003). Dealing with wicked problems in networks: analysing an environmental debate from a network perspective. Journal of Public Administration Research and Theory. doi:10.1093/jpart/mug017 Van Bueren, Klijn and Koppenjan analyse an environmental decision from a policy network perspective, explaining impasses and breakthroughs through cognitive, social and institutional factors. It is a single case analysis.
Dorst, K. (2015). Frame Innovation: Create New Thinking by Design. MIT Press. https://mitpress.mit.edu/9780262550963/frame-innovation Dorst sets out frame creation as a nine-step approach to open, complex, dynamic and networked problems, developed from expert designers' practice. The decisive move is creating a new approach to the problem situation. It is a practitioner method, not a comparison.
Ceschin, F. and Gaziulusoy, I. (2016). Evolution of design for sustainability: from product design to design for system innovations and transitions. Design Studies. doi:10.1016/j.destud.2016.09.002 Ceschin and Gaziulusoy review design for sustainability and map its approaches across four innovation levels, from products to socio-technical systems. It shows the further a problem sits from the product, the less the literature specifies. The literature is thin there.
Ritchey, T. (2013). Wicked problems: modelling social messes with morphological analysis. Acta Morphologica Generalis. https://www.swemorph.com/pdf/wp.pdf Ritchey reproduces Rittel and Webber's ten criteria and presents general morphological analysis as a way to structure social messes, decomposing a problem into parameters and values. It builds a frame without deciding whose aims prevail.
Lonngren, J. and van Poeck, K. (2021). Wicked problems: a mapping review of the literature. International Journal of Sustainable Development and World Ecology. doi:10.1080/13504509.2020.1859415 Lonngren and van Poeck map 55 papers on wicked problems and find the term applied inconsistently, many meanings attached to it, and no consensus on its utility. It is the unit's only counting here. Mapping review.
Head, B. W. (2022). The rise of wicked problems: uncertainty, complexity and divergence. In Wicked Problems in Public Policy. Palgrave Macmillan. doi:10.1007/978-3-030-94580-0_2 Head reconstructs Rittel and Webber's argument that scientific and technocratic approaches are inadequate for social policy because they overlook stakeholder perspectives in framing problems. It is a book chapter and reports no test.
Rastegar, S. H. (2024). Wicked problems in public policy: origins, concepts and critiques. Iranian Journal of Public Policy. https://jppolicy.ut.ac.ir/article_98289_en.html Rastegar traces the origins of the wicked problems concept and lists the criticisms it has attracted: insufficient definition, wrong application, neglected reflexivity, failure to reach mainstream policy. It records a spectrum of wickedness now accepted. It reports no test.
Cross, N. (2004). Expertise in design: an overview. Design Studies. doi:10.1016/j.destud.2004.06.002 Cross reviews research on expertise and reports that expert designers, unlike depth-first novices, move rapidly to early solution conjectures and use them to explore problem and solution together. It is the unit's independent evidence for framing. It reports no measurement.

Check your understanding & apply it · 4 questions, 1 essay

1A city authority calls a housing-affordability programme a wicked problem in a public statement, and an advisor concludes that no method applies and the team can only wait for the politics to settle. Which diagnosis does the unit support?

2Two consultants score a housing programme on complexity, uncertainty and divergence. One treats the score as a property of the problem; the other says the score predicts nothing because wickedness is relational. Which judgement does the unit's evidence support?

3An engineer argues that the team's design problem is wicked while the scientists' problem is tame, so no process model from engineering can be imported. Which diagnosis does the unit support?

4A team's brief states that it will understand the problem fully and then solve it. A member quotes Rittel and Webber's objection. What does the unit's evidence imply for the brief's first phase?

5Essay A city authority asks you to advise on a housing-affordability programme that its chief executive has publicly called a wicked problem. Write the briefing you would send. State what the label adds and what it leaves undetermined, say what you would measure before recommending a course of action, and name the observation that would change your advice. Ground the argument in Ritchey (2013), Head (2008), Peters (2017), Turnbull and Hoppe (2019), Farrell and Hooker (2013), van Bueren and colleagues (2003), Cross (2004) and Dorst (2015).

Begin by separating what the label adds from what it hides. The label adds a warning that the unit endorses: as Ritchey (2013) reproduces them, Rittel and Webber's criteria include that a wicked problem has no definite formulation and no stopping rule, so the authority should not expect to close the problem statement before work begins, and it should not treat an early phase of understanding as a precondition for proposing anything. Then state what the label leaves undetermined. Head (2008) reduces wickedness to high levels of complexity, uncertainty and divergence, which is a definition rather than a diagnosis, and it does not tell the authority what to do on any particular Tuesday. Peters (2017) presses the point that many problems labelled wicked fail a strict definition of it, so the word may be doing rhetorical work; a briefing should ask which of the three characteristics is actually high, and for whom. Turnbull and Hoppe (2019) supply the alternative reading, that wickedness is better understood as the political distance between those who question the problem, which is high only when wide distances are maintained, and that has a direct consequence for the programme: the useful work is to find sub-questions that can be closed while the wider disagreement stays open. Farrell and Hooker (2013) then discipline the diagnosis, since the ten features derive from agent finitude, system complexity and problem normativity, which science and design share, so the authority should grade the challenge rather than decide whether the problem is wicked or tame. van Bueren and colleagues (2003) show why the grading must include the network: in their environmental case, impasses and breakthroughs were explained by cognitive, social and institutional factors, and deadlock and progress turned on the interactions of the network rather than on properties of the problem. What I would measure follows. I would ask for the divergence: which stakeholders disagree, about what, and whether the disagreement is about facts or about aims. I would ask for the network: who can block, who can deliver, and what has changed in that set. And I would ask the frames: Cross (2004) reports that expert designers move rapidly to early solution conjectures and use them to define the problem, so I would ask the team to produce two or three candidate frames, following the frame creation practice Dorst (2015) sets out as a nine-step approach to open, complex, dynamic and networked problems, and to show what each frame makes visible and what it hides. The observation that would change my advice is a low divergence score combined with a stable network, since that would mean the problem is being called wicked when its difficulty lies in delivery, and the authority would be better served by a plain programme with a fixed outcome measure.