Problem Setting
Problem setting is the hinge on which Schon's entire framework turns, and the hinge on which the AI revolution turns with it. Technical rationality teaches problem solving: given a defined problem, apply the appropriate technique. But real professional practice rarely arrives with problems pre-defined. The architect's client cannot articulate what she wants but will recognize it when she sees it. The therapist's patient presents symptoms that fit no diagnostic category. The manager walks into a crisis no case study anticipated. In each situation, the first job is not to solve the problem but to set it — to determine what the problem actually is, which categories apply, and what would count as a resolution. Problem setting is where judgment lives, where values enter, and where no machine can substitute for the practitioner. It is also what professional schools are worst at teaching.
The Infrastructure of Judgment — Contrarian ^ Opus
There is a parallel reading of problem setting that begins not with the cognitive act but with the material conditions that enable it. Problem setting, in this view, is not a pristine moment of professional judgment but a luxury afforded by institutional slack—the buffer zones, the redundancies, the deliberative spaces that late capitalism has systematically eliminated. The architect who once had months to understand a client's unarticulated needs now has days. The therapist who could hold diagnostic ambiguity across sessions now faces insurance requirements for immediate categorization. The manager who might have convened stakeholders to surface the real problem now needs to show metrics movement by quarter's end.
The arrival of AI accelerates this compression not because it cannot set problems, but because it makes problem-solving so cheap that the economic logic tilts entirely toward volume. Why spend a week determining the right question when you can test twenty answers in an afternoon? The practitioner who insists on problem-setting becomes the bottleneck in a system optimized for throughput. What Schon frames as essential professional judgment becomes, in the lived experience of actual practitioners, an indulgence they cannot afford. The young consultant who pauses to question whether the optimization problem is correctly framed will be outpaced by the one who delivers ten iterations. The compression is not cognitive but economic: problem-setting requires time, and time is what the substrate of AI-accelerated practice systematically denies. The real contest is not between human judgment and machine solving, but between the practitioner who still believes in problem-setting and the institutional machinery that has already decided such luxury is obsolete.
In the AI Story
The distinction between problem setting and problem solving sounds semantic until you look at professional failure. The environmental catastrophes produced by confident engineering, the urban renewal disasters designed by credentialed planners, the medical misdiagnoses committed by well-trained physicians — these were rarely failures of problem solving. The practitioners applied the right techniques. The techniques worked. The problem was that the techniques were applied to the wrong problems. The practitioners had solved the problem they could see while missing the problem they should have been addressing. Technical rationality gave them no framework for recognizing the mismatch, because technical rationality assumes the problem is given.
Schon argued that problem setting involves three interrelated moves. First, naming — deciding what aspects of the situation to attend to. Second, framing — establishing the categories through which those aspects will be understood. Third, bounding — determining the scope of what must be addressed and what can be set aside. All three moves are judgments. All three are informed by the practitioner's repertoire. And all three are the kind of judgments that no formal knowledge specifies.
The arrival of AI makes problem setting the site of the decisive professional contest. Large language models are extraordinary problem solvers within a given frame. Ask Claude to optimize a function, draft a brief, diagnose a presentation, design a feature, and it will produce competent results faster than most humans. What it cannot do is decide whether the function is the right function, whether the brief is the right response, whether the diagnosis is the right frame, whether the feature deserves to exist. The phronesis barrier is the problem-setting barrier dressed in Aristotelian vocabulary: the evaluative work that the tool cannot perform because it lacks stakes in the outcome.
The danger is that AI's problem-solving fluency makes problem-setting invisible. When every problem is instantly solvable, the cognitive labor of deciding which problems to solve becomes systematically underweighted. The practitioner prompts, receives, accepts — and the entire reflective sequence collapses into iteration within an unexamined frame. The flow-compulsion gradient that The Orange Pill documents is, through Schon's lens, a gradient in which rapid problem-solving atrophies the slower muscle of problem-setting. Twenty cycles in an hour, all within the same frame, refining an answer to a question no one has paused to interrogate.
Origin
Schon introduced the problem-setting/problem-solving distinction in The Reflective Practitioner (1983) and elaborated it in his 1979 essay "Generative Metaphor," which analyzed how policy analysts set problems through implicit metaphors that shape everything that follows. The distinction drew on earlier work by Martin Rein and Schon on policy framing, and on Schon's decades of observation in professional settings.
Key Ideas
Problems are not given; they are constructed. The situation presents a mess; the practitioner sets a problem within it.
Three operations of setting. Naming (what to attend to), framing (how to categorize it), bounding (what to include).
The locus of professional judgment. Problem setting is where values enter, where judgment lives, where the practitioner's repertoire matters most.
AI's asymmetric capability. Machines excel at problem solving within a given frame; they cannot set the frame.
The invisible labor. When solving is fast, setting becomes easy to skip — and the skipping is what produces polished answers to wrong questions.
Debates & Critiques
Operations research and formal decision theory offer alternatives to Schon's framework that treat problem setting as itself a problem to be solved through meta-level techniques. The defense is that meta-level techniques require their own problem setting, pushing the question back without answering it. The practical question — where the work of problem setting actually happens and who does it — has become urgent as AI systems increasingly handle problem solving, leaving the question-selection function as the scarce human contribution.
Appears in the Orange Pill Cycle
The Temporal Politics of Framing — Arbitrator ^ Opus
The tension between Edo's cognitive framing and the contrarian's material analysis resolves differently depending on which temporal scale we examine. At the immediate scale of professional practice—the moment of decision—Edo's account dominates (90/10). Problem setting genuinely is the irreducible human act that precedes competent practice. No amount of institutional pressure changes the fact that someone must decide what to optimize, whom to include, which values to honor. The practitioner facing the screen still performs this cognitive operation, even if hurriedly.
But zoom out to the institutional timeline—quarters, careers, industry transformations—and the contrarian view gains force (70/30). The economic logic of AI adoption does systematically devalue problem-setting time. Organizations really do restructure around rapid iteration rather than deliberative framing. The junior professionals who might have learned problem-setting through apprenticeship instead learn prompt engineering. Here the material conditions shape what cognitive operations are possible, not through elimination but through atrophy.
The synthesis emerges when we recognize that problem setting operates simultaneously as cognitive act and institutional practice—and these two levels interact. The individual practitioner retains the capacity for problem setting (Edo is right), but exercises it within increasingly constrained conditions (the contrarian is right). The solution is not to defend problem setting as pure cognition but to rebuild the institutional structures that make it possible: the peer review, the design charrette, the diagnostic conference, the strategy retreat. These are not just venues for problem setting but the material infrastructure that prevents its collapse into hasty assumption. Problem setting survives not through individual resistance but through collective practice—a truth both views acknowledge but neither fully develops.
Further reading
- Donald Schon, The Reflective Practitioner (Basic Books, 1983)
- Donald Schon, "Generative Metaphor: A Perspective on Problem-Setting in Social Policy," in Metaphor and Thought, ed. Andrew Ortony (Cambridge, 1979)
- Martin Rein and Donald Schon, Frame Reflection (Basic Books, 1994)
- Horst Rittel and Melvin Webber, "Dilemmas in a General Theory of Planning," Policy Sciences 4 (1973)