Michael Rowe

Wicked problems

note7 min read

A wicked problem is one that resists resolution

Wicked here means stubborn: a wicked problem can’t be definitively formulated, has no point at which it’s finished, and admits solutions that are good or bad rather than true or false. Rittel and Webber (1973) named them to describe what planners kept running into when working in complex social systems. The category matters because the big problems worth working on are almost all of this kind, and because the standard response (define, solve, and close) is what turns a difficult problem into a wicked one.

Wicked problems

One-sentence definition: A wicked problem is one that resists resolution because its requirements are incomplete, contradictory, and changing, and because any attempt to address one aspect reveals or creates others.

Rittel and Webber (1973) set out ten characteristics. All ten, in their order:

  1. There is no definitive formulation. How you state the problem already contains a theory of its solution. Different stakeholders state it differently, and each statement is legitimate.
  2. There is no stopping rule. Nothing tells you the work is complete; you stop because you’ve run out of time, money, or patience.
  3. Solutions are not true or false, but good or bad. There is no test that settles it, which is why disagreement persists among reasonable people who are all reasoning well. (Rittel and Webber’s own phrase is “good-or-bad”; the common gloss “better or worse” softens it.)
  4. There is no immediate and no ultimate test of a solution. Every intervention sets off consequences that keep unfolding, and some of the later ones will outweigh the intended ones.
  5. Every solution is a one-shot operation. You can’t trial an intervention on a health system without changing the system you were trying to learn about, so there is no learning by trial and error.
  6. There is no enumerable set of potential solutions. Nobody can list the options and pick the best, because the list is never complete and new options keep being invented.
  7. Every wicked problem is essentially unique. There is enough shared structure to recognise the type of problem but never enough to transfer the solution from one to another.
  8. Every wicked problem can be seen as a symptom of another problem. Ask why a problem exists and the answer is another problem: students disengage because the curriculum is overloaded, which is because accreditation demands it, which is because the profession fears risk, and so on. Each “why” moves you up a level, and there is no level at which the chain stops. Where you choose to intervene is a judgement, and no amount of analysis will turn it into a finding.
  9. The discrepancy can be explained in many ways, and the explanation chosen determines the resolution. Poor NSS1 scores can be read as a teaching problem, a communication problem, an estates problem, or a cohort problem, and each reading points to a different intervention. The choice of explanation is where most of the politics happens.
  10. The planner has no right to be wrong. Consequences land on people, so the scientist’s licence to be provisionally mistaken, to publish a hypothesis and let others refute it, isn’t available. This is the cruellest of the ten.

The relationship to complexity

Complexity science describes systems with open boundaries, non-linear cause and effect, feedback loops, and dependence on their own history. The wicked-problems literature describes what happens when someone tries to solve a problem inside such a system. They are the same phenomenon seen from two positions. A complexity theorist looking at a health system sees feedback and non-linearity; a planner trying to change that health system experiences those same properties as a problem that won’t hold still, can’t be tested, and turns out to be a symptom of something else. Rittel and Webber’s ten characteristics are the view from the planner’s chair.

That’s why the two literatures, complexity science and the wicked-problems tradition in planning and public policy, keep arriving at the same recommendation from different starting points: iterative, contextualised, incremental action that shifts a pattern over time, in place of a better specification of the problem (Head & Alford, 2015; Fraser & Greenhalgh, 2001).

Wickedness can be manufactured

The usual reading of wicked problems treats wickedness as a property of the terrain: the problem is hard because the world is complicated, and better management would get us closer to a solution. Eoyang and Mennin (2019) argue the reverse for health professions education. Every institution, clinical setting, and student differs, and the field responds to that variability by intensifying standardisation, rigour, and accountability to competencies in every facet. Their conclusion is that the search for consistency and control in such a highly variable system leads inevitably to intractable, wicked problems. On this reading some of the wickedness is produced by the management itself, by insisting on uniformity in a system whose variability was never a defect. Cleland, Patterson, and Hanson (2018) make a parallel argument about selection and widening access, problems that have absorbed decades of effort while remaining unresolved, as wicked problems always are.

Coping strategies for wicked problems

Roberts (2000) identifies three coping strategies, each with a real cost:

Authoritative strategies tame the problem by putting it in the hands of a few. Reducing the number of stakeholders genuinely reduces complexity, because the competing perspectives are eliminated at the start. But those perspectives were part of the problem, so this is also a failure mode.

Competitive strategies set opposing positions against each other and pick the best. Solutions get compared. But the adversarial framing discourages the knowledge-sharing that would have produced better ones.

Collaborative strategies engage all stakeholders to find something that works for all of them. They’re slower, and depend on a level of trust that often isn’t there, but this is the only one of the three that doesn’t achieve tractability by discarding part of the problem.

Networks matter here for a structural reason: any problem sufficiently complex to be worth caring about is one you can’t address alone (Conklin, 2005).

Wicked problems in health professions education

Wicked problems are typically where the consequential work of health professions education sits: health inequalities, multimorbidity across fragmented services, the health consequences of climate change, and the integration of care across institutional boundaries. None of these has a solution, all reward sustained attention, and none can be addressed by someone waiting for a tractable version to appear.

They also carry a career implication. The work that moves a wicked problem forward is rarely the work that produces clean outputs on the schedule an institution counts. It’s a real cost of choosing them, and one worth knowing about before you do.

Sources

Footnotes

  1. NSS (National Student Survey) — the annual survey of final-year undergraduates in the UK, whose results feed into league tables and the regulator’s view of teaching quality. A programme with poor NSS scores comes under pressure to act, often before anyone agrees on what the scores mean. Wikipedia ↩

Continue reading

  • Problem-based learning and the structural conditions for productive AI integration

    Higher education's response to AI has focused on the artefact: detecting it, restricting it, and restoring confidence in what students produce. This essay argues that the structural features of problem-based learning — problem-driven inquiry, collaborative knowledge construction, facilitation over instruction, and metacognitive reflection — are the same conditions under which AI integration becomes educationally productive rather than substitutive. The alignment is structural, not retrospective: PBL was designed around these conditions before AI existed. The argument extends further: AI shifts what category of problem PBL can engage with, expanding access to wicked problems previously beyond students' reach. Investing in PBL's structural conditions is simultaneously investing in AI readiness.

  • From teaching to learning: Rethinking education for a world of information abundance

    Traditional education systems are structured around teaching, assuming it inevitably produces learning. This essay argues that this unidirectional model — knowledge flowing from expert to novice — no longer suits a world of information abundance and AI disruption. A networked, learner-centred approach offers an alternative, reconceptualising learning as a complex process where educators shift from knowledge authority to learning facilitator, enabling diverse participants to contribute to collective understanding. This transformation is urgent as AI develops capabilities once exclusive to human experts, and requires reimagining institutional structures around learning networks rather than teaching hierarchies to prepare graduates for complexity and uncertainty.

  • The wicked problem of AI and assessment