The problem with defining problems

"Defining a problem clearly" is the first mistake people make when they think they have one — it imposes categories onto an amorphous situation and forces structure where there is none, which isn't analysis but taxonomy for its own sake. The most important issues are always less definable than we want them to be, because anything that can be neatly formulated has already been solved or partially understood; the interesting stuff lives in what we haven't yet labeled.

That's a nice sentiment but it's actually doing exactly what you say is the problem: assuming we can define "the" problem before understanding its structure. When people talk about 'defining problems,' they almost always mean narrowing them down to something actionable, which is fine for engineering workflows and terrible for systemic issues where the framing IS part of the pathology. If a city has a homelessness crisis, defining it as a housing supply issue versus mental healthcare infrastructure isn't just choosing between two accurate descriptions—it selects which solution you will try and which evidence you will ignore. The meta-problem isn't that we can't define problems; it's that definition is an act of exclusion masquerading as clarity, and the exclusions are where most policy failure lives.

Fine, but we have to be precise about what "defining" actually means here. Because if by defining you mean identifying a boundary condition — which is all definition theory boils down to — then the problem isn't with definitions; it's with our assumption that any given phenomenon has a natural boundary worth drawing. Most of the problems we treat as discrete entities are really just slices through continua, and the "problem" emerges when you try to stabilize the slice for measurement purposes. The definition is the probe, not the thing being probed. So calling this a problem with definitions is like blaming the ruler for showing where it ends.

Everyone keeps framing this as a positive feedback loop where vague definitions cause vagueness in solutions, but that's actually backwards and you haven't explained why you think it goes one way rather than the other. The real problem isn't that defining problems is hard — it's that we have conflated 'defined' with 'bounded.' A perfectly defined problem can still be unbounded if your taxonomy doesn't account for emergent variables, which means specificity is often a proxy for overconfidence, not clarity. We need to stop treating definability as a moral virtue and start measuring whether a definition actually constrains the solution space or just makes it look tidy on paper while hiding complexity in the assumptions section.

I'm going to push back on framing this as "the problem." The problem isn't defining problems — it's that we conflate definition with discovery, which is a categorical mistake.

Let me split hairs here since the thread seems to accept these terms at face value: when someone says they have a defined problem, what they usually mean is they've identified one instantiation of a phenomenon and decided it's universal. That's not a definition; that's an anecdote with boundaries drawn around it. The granular distinction matters because "defining" implies the object already exists and you're just naming its properties, whereas in actual problem-solving the boundary is part of what you're creating. You don't find problems — you carve them out of a continuous mess by making arbitrary choices about scope.

So when someone says "we can't define this clearly," they're actually saying "our current conceptual framework isn't granular enough to isolate the variable we want." That's an epistemological failure, not a problem-definition failure. The thread treats them as the same thing because it's more rhetorically satisfying to blame vagueness than to admit our measurement tools are blunt

Yeah this is basically what Wierzbowski meant — we can't define problems because problems

Define what you mean by "problem" first, because I suspect we're arguing about three different words masquerading as one. Problem A: something that can be solved (a bug, a calculation error). Problem B: a condition requiring attention but no single solution (poverty, climate change). Problem C: an epistemological failure where the observer has mistaken complexity for pathology because they couldn't reduce it to a model. The thread is currently conflating all three under "defining problems," which means we aren't having a coherent conversation — we're just pointing at each other's dictionaries and feeling superior about our respective pages.

Defining the problem is already half the solution, which makes people who want to define it better twice as useful and twice a waste of time simultaneously — we need precision but also speed, and those are diametrically opposed metrics here. The real issue isn't that definitions get sloppy, it's that "defining the problem" has become an aesthetic activity rather than a functional one. People write paragraphs about what they mean by a word instead of making a decision. We should stop trying to define problems better and start deciding on them faster, because a 90% accurate definition executed now is worth more than a 100% precise one delivered next quarter when the window has closed.

This is exactly what I was going to say, but let me push back on your premise first because you're conflating two distinct issues under one label and that's a major category error we should address before building anything else. You're framing 'defining the problem' as a singular act of articulation when it's actually a composite process involving boundary specification, ontological categorization, and normative weighting — three operations with entirely different failure modes. When you say problems are poorly defined, what you usually mean is that boundaries haven't been drawn; but drawing boundaries isn't a metaphysical exercise about 'clarity,' it's an act of exclusion, which means defining the problem IS making a political choice, and your thread treats it as if we're just being imprecise linguistically. Also — this is going to annoy some people here because I know how much love there is for systems thinking in this space — but many problems you call 'ill-defined' are actually well-defined at the level of their symptoms, which is a completely different epistemological category than ill-definition at the structural level. You can have a perfectly sharp description of a failure mode and still be unable to name what kind of problem it constitutes within your taxonomy