Shuichiro Ogawa
日本語

Notes · updated 2026-07-19

What Shape Does the Problem Design Tries to Solve Have?

Engineering problems usually come with a signal that says when to stop. Once the equation is solved, or the specification is met, you may put down your tools. The problems design faces, however, have no such signal written in from the start.

From the context of urban planning, Rittel and Webber argued that social planning problems have no stopping rule (1973)1. You do not stop because you have fully solved the problem. You stop because time or budget has run out, or because you judge the result good enough. They called this kind of problem a wicked problem, distinguishing it from the tame problems that natural science handles.

The absence of a stopping rule may look, on its own, like a procedural inconvenience. But what Rittel and Webber showed lies deeper. In wicked problems the formulation of the problem is inseparable from the choice of solution, and which formulation to adopt is not settled by evidence alone. Where to draw the line carries the designer’s value judgment.

That value judgment is not something one can avoid by trying to. In Logic: The Theory of Inquiry (1938), Dewey formulated inquiry as the controlled transformation of an indeterminate situation into a determinate one2. Inquiry does not begin from value-neutral observation; it already contains an evaluation of what counts as the problem and what counts as a satisfactory resolution. To say that the act of designing is value-laden does not mean that the designer’s subjectivity creeps in. It means that the very structure of framing a problem and choosing a solution demands evaluation. That this evaluation carries social judgment is borne out by the genealogy in which design meets social science (design-social-science-nexus).

If there is no stopping rule and value judgment cannot be avoided, on what basis does the designer decide “this will do”? Simon placed the concept of satisficing here (1956)3. An agent with limited cognitive resources and information cannot compare every option to select an optimum. Instead it sets a threshold and halts the search once an option that meets it is found. Satisficing rather than optimizing is the realistic form of rationality under bounded rationality.

That completes the three points of the introduction. The absence of a stopping rule, value-ladenness, and bounded rationality are not separate defects. Each is a manifestation of the fact that the problems design handles have a different shape from those of engineering and natural science. From this premise, the sections below trace the foundations of method in turn.

How Far Does Reductionism Carry?

Decompose a complex problem into elements, solve the parts, then reassemble. This procedure of reductionism sits at the core of the successes that modern design and engineering have accumulated.

In The Sciences of the Artificial (1969), Simon argued why complex systems tend to take on a nearly decomposable hierarchical structure4. If interactions within a subsystem are denser than those between subsystems, the system can be treated as a collection of nearly independent parts. Because of this property, a designer can proceed part by part without grasping the whole at once. Reductionism works when this hierarchy actually exists on the side of the world.

The trouble arises when that hierarchy does not hold. In Rittel and Webber’s wicked problems, every problem is a symptom of another, and carving out elements changes the very nature of the problem (1973)1. Buchanan connected this recognition to a general theory of design thinking, arguing that design problems carry an intrinsic indeterminacy (1992)5. In domains of high interdependence, where the solution to a part rewrites the framing of the whole, the very strategy of decomposing and solving in isolation spins its wheels.

If decomposition fails, a standpoint that treats the whole together is required. In a management-science editorial, Churchman warned of the danger in analytic methods that try to tame wicked problems too easily (1967)6. In Redesigning the Future (1974), Ackoff grasped social problems as interlocking messes and argued for redesigning the whole system rather than excising and solving individual problems7. Reductionism and holism do not stand in a relation of superior and inferior. Which is the fitting tool changes according to whether the hierarchy of the problem actually exists.

When a Single Method Is Not Enough

If the shape of the problem is not fixed, the method cannot be fixed to one either.

In Against Method (1975), Feyerabend argued that no single universal method guarantees the progress of science, and raised the banner of epistemological anarchism8. Looking back at the history of science, breaking established methodological rules is repeatedly found to have produced decisive advances. His claim is not that “method does not matter.” It is that which method is effective depends on context, and that the choice of method is itself part of inquiry.

Cross took up this plurality as a mode of knowing proper to design (2001)9. Contrasting the project of design science, which seeks to ground design as an imitation of natural science, with the position of designerly ways of knowing, which holds that design has its own way of knowing, he placed his weight on the latter. Design’s knowledge gropes for problem and solution at the same time, combining several methods as the situation demands. Methodological pluralism is not an absence of methodology but an active stance for responding to wicked problems.

Buchanan and Dorst deepened this stance into an account of the nature of the design problem. Buchanan called design “the art of placement” and argued that its indeterminacy, its having no fixed subject matter, is the very source of design’s possibility (1992)5. In “The core of ‘design thinking’,” Dorst analyzed the design problem as an endeavor that reframes the problem itself rather than solving a given problem (2011)10. Both reject the picture of reaching a solution by a single method.

Thinking in Action, Leaping to a Hypothesis

How does an agent who selects a method as the situation demands actually think?

In The Reflective Practitioner (1983), Schön severed the knowledge of professionals from technical rationality11. Technical rationality applies established theory to a particular situation to solve a problem. But real professionals form judgment in dialogue with uncertain, one-off situations. Schön called this reflection-in-action. The designer reads the unexpected responses that material, drawing, and situation return, and replays the next move. Knowledge is not completed prior to action; it takes shape in action.

The logical form of this thinking is abduction. In addition to deduction and induction, Peirce posited a form of inference that leaps to a hypothesis explaining a surprising fact, calling it abduction12. Deduction draws a conclusion necessarily from premises; induction infers a general law from cases. Abduction is neither, proposing for an observed fact the explanation that “if this hypothesis were true, this fact would follow as a matter of course.” It guarantees no truth, yet it is the only form of inference that generates a new hypothesis.

Dorst argued that this abduction sits at the core of design thinking (2011, 2015)1013. In ordinary problem-solving, the working principle (how) and the desired value (value) are given, and one derives the object (what) that realizes them. This is one form of abduction, but the difficulty proper to design does not lie here. In what Dorst calls frame innovation, the core of design, only the desired value is given, and the working principle and the object must be found at once (2015)13. Under what frame should one view the problem so that a solution rises up? That frame itself is leapt to as a hypothesis. The process that reflective practice described as “thinking in action” can be grasped, at the level of logic, as this double abduction (the validity of methods in design research is treated in design-research-methods-validity).

Not Deciding, as a Design Strategy

In a design that has no stopping rule and leaps to frames by abduction, when and what to fix becomes the question. Decide early and you bear the risk of rework; decide late and you keep your options.

Analyzing Toyota’s product development, Ward, Liker, Cristiano, and Sobek reported the paradox that deliberately delaying decisions can instead make better cars faster (1995)14. They called this the “Second Toyota Paradox.” Rather than selecting one alternative early and building it out, one proceeds while holding multiple alternatives as a set.

Sobek, Ward, and Liker formulated this principle as set-based concurrent engineering (1999)15. Each function keeps a broad set of feasible solutions, probes the overlap of the sets across functions, and gradually narrows the options through cheap testing. In contrast to point-based design, which converges to a single point early, set-based design keeps the options open in phases of high uncertainty.

This strategy meshes squarely with the three premises seen in the introduction. Since there is no stopping rule, premature fixing tends to become a groundless cutoff. Under bounded rationality, holding a set until information accumulates can be more rational than betting on a single point with little information. And in a design that leaps to frames by abduction, keeping the breadth of solutions has a point, anticipating that a promising frame may be found later (the relation between exploration and environmental states is organized in creativity-situated-cognition-environment). Not deciding is not indecision but a design of exploration matched to the shape of the problem.

Drawing the Map Together

The genealogy traced here branches from a single epistemological premise. The problems design handles have no stopping rule, cannot avoid value judgment, and must proceed under bounded rationality.

Accept this premise and reductionism ceases to be an all-purpose tool (Simon, Rittel & Webber, Buchanan), a holistic standpoint is called for (Churchman, Ackoff), and method can no longer be fixed to one (Feyerabend, Cross). As modes of thinking, reflection-in-action (Schön) and the leap of hypothesis (Peirce, Dorst) come to the fore rather than the application of theory. And in time, a strategy that avoids premature fixing and holds solutions as a set (Ward, Sobek) takes on meaning.

These are not mutually independent claims. From the single point that the shape of the problem differs from that of engineering, the same consequence appears in each of the treatments of method, thinking, and time. Read design methodology as a collection of isolated techniques and this coherence is lost. The use of a literature map lies not in the location of individual concepts but in making visible the single thread that runs through them.

References

Footnotes

  1. Rittel, H. W. J., & Webber, M. M. (1973). Dilemmas in a General Theory of Planning. Policy Sciences, 4(2), 155–169. ↩ ↩2

  2. Dewey, J. (1938). Logic: The Theory of Inquiry. Henry Holt and Company. ↩

  3. Simon, H. A. (1956). Rational Choice and the Structure of the Environment. Psychological Review, 63(2), 129–138. ↩

  4. Simon, H. A. (1969/1996). The Sciences of the Artificial (3rd ed.). MIT Press. ↩

  5. Buchanan, R. (1992). Wicked Problems in Design Thinking. Design Issues, 8(2), 5–21. ↩ ↩2

  6. Churchman, C. W. (1967). Guest Editorial: Wicked Problems. Management Science, 14(4), B141–B146. ↩

  7. Ackoff, R. L. (1974). Redesigning the Future: A Systems Approach to Societal Problems. Wiley. ↩

  8. Feyerabend, P. (1975). Against Method. New Left Books. ↩

  9. Cross, N. (2001). Designerly Ways of Knowing: Design Discipline Versus Design Science. Design Issues, 17(3), 49–55. ↩

  10. Dorst, K. (2011). The Core of ‘Design Thinking’ and Its Application. Design Studies, 32(6), 521–532. ↩ ↩2

  11. Schön, D. A. (1983). The Reflective Practitioner: How Professionals Think in Action. Basic Books. ↩

  12. Peirce, C. S. (1931–1958). Collected Papers of Charles Sanders Peirce (Vols. 1–8). Harvard University Press. ↩

  13. Dorst, K. (2015). Frame Innovation: Create New Thinking by Design. MIT Press. ↩ ↩2

  14. Ward, A., Liker, J. K., Cristiano, J. J., & Sobek, D. K. (1995). The Second Toyota Paradox. Sloan Management Review, 36(3), 43–61. ↩

  15. Sobek, D. K., Ward, A. C., & Liker, J. K. (1999). Toyota’s Principles of Set-Based Concurrent Engineering. Sloan Management Review, 40(2), 67–83. ↩

← All Notes · Home