45.1An answerable factual question
An answerable factual question fixes two things together: an object to be measured, and the conditions under which an observation would count as evidence for the answer. Take the fictional inquiry already posed: how many step-free entrances are open after 18:00? No count is supplied, and none should be invented. What the question does supply is a shape that any answer must fit.
Naming the object is necessary but not sufficient. "Step-free entrances" marks a candidate set, yet the question becomes fully answerable only once the site, the relevant evening, and the sense of "open" are settled. Conditional supposition: if "open" means unlocked and passable throughout the stated window, then being step-free is necessary for an entrance to enter the count but not sufficient, since it must also satisfy the window. The named quantity compares the number of entrances meeting both conditions with the total set of entrances at that site. It is not a general measure of accessibility, and reading it as one would silently change the task.
Fixing the object and the counting conditions in this way also localizes disagreement. Two respondents may divide over whether a particular ramp counts as step-free, over which hours the window covers, or over whether a door was passable at a given moment. Each dispute is a dispute about classification, scope, or observation, resolvable against conditions the question has already set. The question's answerability therefore lies in this discipline of comparison, not in any expectation that a number will be produced. Whether evidence will be available to settle these disputes is a further matter, which the next step examines. Until then, the inquiry stands as what it is: a factual question with definite answer conditions, awaiting evidence rather than a change of subject.
Consider the fictional question, “How many step-free entrances are open after 18:00?” Its answer would be a count of entrances satisfying three jointly required conditions: being step-free, being open, and their status after the stated hour. Each condition is necessary for an entrance to enter the tally; together they are sufficient for inclusion. The named quantity is therefore the number of qualifying entrances, not a measure of how accessible a building feels, nor a judgment about whether late opening is desirable.
The absence of a count or observation changes the practical situation, not the kind of question. Evidence may be unavailable—no survey exists, records are incomplete—or noisy—a door propped open, an unattended entrance, conflicting reports. In either case the question can remain an empirical one whose answer is currently out of reach. It becomes a normative demand only if what is being asked is how many entrances ought to be open, or whether the building should provide access after 18:00. Those are different questions, and data cannot answer them by themselves.
A qualification matters. If “step-free” or “open” is disputed in application—a ramp with a steep section, a locked door that staff will unlock on request—then the factual question inherits a boundary problem. The difficulty lies in classifying cases, not in converting the inquiry into a demand. Two people may agree on the intended conditions and still disagree about whether a particular entrance meets them. That disagreement can be settled by further specification or by evidence about the entrance; it does not by itself make the question normative.
What this question cannot deliver is an account of overall accessibility or of fairness in opening hours. A count records how many entrances match the conditions at the relevant time. Without an observation, no number follows; the question stands, answerable in principle but unresolved here.
45.2A contested concept
When the key term of a question is itself the site of reasoned disagreement, the question employs a contested concept. In "What should count as equitable access?", the dispute concerns standards rather than predictions: which goods are distributed, to whom, and by what threshold of need an arrangement earns the label equitable. The term thus organizes competing accounts of what would count as getting the matter right.
That organizing role is the point at issue here, but it needs a boundary. Contestedness is neither necessary nor sufficient for open inquiry. Two parties may share a definition of access and still disagree about facts, priorities, or the acceptable cost of a policy; such a dispute need not be conceptual and will not be dissolved by clarifying terms. Conversely, a question built on a stipulated concept can be answered cleanly while the concept stays disputed in other forums. Disagreement over what counts is possible and may persist; it is not inevitable in every use of the term. The classification must therefore attach to the particular claim rather than to the vocabulary as a whole.
Where the classification does hold, it changes what closure would require. Settling a concept-contested question means arguing for one standard against its rivals, not merely gathering more evidence about how the world behaves. A system that treats evidence-gathering as the only respectable route to closure will misdescribe such a question as a defective factual one.
Rittel and Webber's challenge to unqualified optimal-solution language for pluralistic policy problems sits within this theoretical caution about planning (Rittel and Webber, 1973). It does not establish that every local task resists resolution, and the caution should not be extended past what the position claims.
The practical consequence is procedural. Keeping a contested concept in play means keeping its rival standards visible as live positions, so that whichever standard is later adopted can be recognized as a choice among those rivals.
An operational definition fixes what a measurement will count. Consider the illustrative question, "What should count as equitable access?" Suppose, as a piece of prospective design, that access is stipulated to be the share of households lying within a stated distance of a service point. The stipulation permits a particular comparison: locations are sorted against the chosen boundary, and the resulting figure answers a conditional question — which places count as served, given this threshold and this unit of population. What the named quantity compares is proximity of households to service points. It does not compare need, cost, service quality, or the burdens that reaching the service actually imposes.
Two limits follow. First, the definition is sufficient for that measurement but not necessary for understanding access at all. A different but defensible definition might fix on affordability or adequacy, and one definition's success at classifying locations does not show that the others are misguided. Second, the threshold is a stipulation the measurement cannot certify. If the boundary were widened, some places now classified as unserved might be classified as served. That is a possible consequence of changing the stipulation, not an inevitable instability in the figure; under a fixed threshold the classification is straightforward. But the possibility shows that the choice of threshold does work that the resulting number cannot undo.
A critic can therefore accept every figure generated under the definition and still deny that proximity is the right basis for equitable access. The disagreement is not a defect in the measurement, nor evidence that the concept is unusable. It marks a difference between two tasks: classifying locations under a stated rule, which the definition handles, and defending that rule as an account of equity, which the definition does not attempt. The contested concept stays alive at exactly the point where the operational boundary is asked to justify itself.
45.3A demand for justification
A demand for justification asks why an action is acceptable under stated commitments. It puts the action and the commitments on the same argumentative plane: the demand is met only when the stated commitments, together with whatever bridging claim is needed, make the action acceptable. Naming a commitment is therefore not enough. The commitments are necessary as the standard of assessment; they are not sufficient, because the action must still be shown to meet that standard. Even if the commitment is genuine and relevant, showing that it permits the action is a further step; showing that it requires the action is further still. Permissibility and requirement are different verdicts, and an action may be permissible without being inevitable or obligatory.
Applied to the invented restriction question, "Why is this restriction justified?" the demand is directed at the reasons connecting the restriction to whatever commitments are said to govern it. A reply that merely restates the commitment answers only part of the demand. A reply that says the restriction serves some purpose must then show that the purpose is licensed by the commitment, and must not assume that serving the purpose settles whether the restriction is acceptable.
This structure also clarifies what can vary without changing the question. The justification may be strong under one stated commitment and fail under another; the same action can therefore be justified and unjustified relative to different commitments. That possibility is not a defect to be eliminated by a single verdict. It follows from the demand's form: justification is indexed to the commitments stated, and the demand remains open where the commitments themselves are contested or where the bridge from commitment to action is disputed. Uncertainty here belongs to that bridge, not to the fact that a demand for justification has been made.
A demand for justification has its scope fixed by the commitments it invokes. In the fictional case, the question "Why is this restriction justified?" asks for reasons bearing on acceptability under those commitments. The forecast that the restriction will persist answers a different question, even when the forecast is accurate. Accuracy here compares predicted occurrence with actual occurrence; it does not compare reasons with the commitments placed in issue. Persistence is therefore neither sufficient nor necessary for justification. It is not sufficient, since a restriction can persist while remaining unacceptable under the stated commitments. It is not necessary, since the reasons can be adequate even if the restriction is expected to be withdrawn.
Certainty does not repair the gap. Even an unavoidable restriction is not thereby warranted; inevitability describes the space of possible outcomes and its closure, whereas justification concerns whether the action should be taken at all. Treating the two as equivalent would silently convert the normative demand into a question about what must happen, and would thereby lose the possibility of criticizing an outcome that cannot be prevented.
The forecast can still enter the argument, but only as a premise. Conditional on commitments that make expected consequences relevant—for instance, if the stated commitments require weighing a predicted harm—the forecast supplies content for the weighing. Yet the verdict still turns on how the commitments apply, not on the forecast's reliability. A reliable prediction may contribute to a justification without being one, just as a relevant premise is not the conclusion it supports.
This distinction between justification and prediction is an application of the question's type rather than a general scepticism about prediction. What is refused is a substitution: the accurate description of a likely or certain action standing in for the demand that the action be defended.
45.4A question changed by discovery
Discovery can do more than supply an answer: it can show that the question posed assumes something false, or that it collapses a distinction on which the answer depends. In the illustrative case already sketched, the stipulated investigation finds no second entrance. The finding does not return a plain negative to the question as framed; it removes the object that question was about. A question asking how to use a second entrance cannot be answered as written once no such entrance exists.
Rejecting a false presupposition can be an appropriate answer or terminate an inquiry; it need not force a newly formulated question. Two qualifications follow. It is possible that the revision captures everything at issue; it is not inevitable, because the original phrasing may also have obscured a distinction that surfaces only when the premise fails. Suppose, for instance, that the question had already moved between the existence of an opening and its practical availability; only the failed premise makes that slippage visible.
Nor do comparisons survive the change intact. Two formulations may carry a similar label while comparing different things, since the options being weighed were defined by the assumption now withdrawn. A fair comparison would hold the task and inputs fixed, but revision changes exactly those inputs: the earlier comparison cannot be carried over as though only the wording had altered. What changes is the comparison being made, not merely the sentence in which it is stated.
Meanwhile a temporary hours decision can still be made on the narrower grounds that bear on it. Such a decision neither predicts nor declares the wider inquiry resolved; it answers only what its own premises support. Opening the question again, then, is compatible with acting now — the unresolved inquiry remains unresolved even though the false premise has been settled.
Suppose the stipulated investigation finds no second entrance. The right response is to state the revised question and record why it changed, rather than discarding the earlier inquiry as simply worthless. A question that could not survive contact with the evidence is not thereby meaningless: it marked a distinction someone thought worth testing, and the reason for revision preserves what the test established. The record now carries two things—the question currently in force and the grounds on which its predecessor was replaced.
That distinction matters when the record is put to use. Compare two ways of storing the same inputs, holding the task fixed at determining the scope of a later decision. A record containing only the current question leaves no visible link between the investigation and the change; a record pairing the revised question with its reason shows which premise was withdrawn and what remains open. The encodings differ in exactly this respect, and the difference is what makes scope recoverable.
The revised question is also narrower than the inquiry that prompted it. Asking whether access justifies a change in hours is not the same as asking whether all questions about access have been settled. Revising the question is necessary for the record to be accurate, but it is not sufficient to establish that the new question is answerable or that any particular answer follows from the investigation alone.
Accordingly, a temporary hours decision can be justified within the scope the revision supports without being treated as a resolution of the wider inquiry. The decision applies to a bounded situation; the inquiry remains open in exactly those respects the recorded reason did not close. Scope is what the reason preserves: not a verdict on the whole investigation, but an account of which part of it a decision may claim.
45.5A decision despite open inquiry
The stipulated investigation has already done one narrow job: it shows that the supposed second entrance does not exist. That result defeats the particular route hypothesis that depended on it. It does not by itself establish how access works overall, why the question arose, or what arrangements would be justified. The question should therefore be revised and the reason for the revision kept visible, so the earlier inquiry remains part of the record rather than being treated as wasted.
A temporary-hours decision can now be made at a different level. Its immediate question is what hours to adopt for the stipulated period under the facts presently accepted. That decision is not a verdict on the wider access inquiry: adopting hours is neither necessary nor sufficient for concluding that the inquiry has been answered. Nor is the decision inevitable merely because a choice must be made; if the timing constraint is absent, postponement remains a possible response. What the decision does establish is only that, given the accepted facts and the need for a schedule, one workable arrangement is available.
The inference is therefore conditional and limited. If the revised question later yields further findings, the hours decision should be open to revision on stated grounds; if it does not, the decision can stand without pretending the wider question is closed. Only if the decision were accompanied by an independent demonstration about access could it count toward closing the inquiry; the hours choice alone cannot do that. This separates two acts: settling a practical measure for a bounded period, and preserving a research question whose answer is still underdetermined. The temporary measure records what was assumed and what was chosen, while the revision record explains why the original question changed. Keeping those records distinct allows action to proceed without converting an unresolved inquiry into a forecast.
The decision remit should be stated before action: it covers a temporary hours setting, not the wider question of access. Its provisional assumptions are that the stipulated investigation is bounded, that its negative finding about a supposed second entrance is accepted for this decision, and that the arrangement can be reconsidered when the inquiry's remit changes. These assumptions are working conditions, not settled conclusions. If the arrangement is temporary and revisable, that limitation belongs to the decision itself.
Once the entrance claim fails under the stipulated investigation, the question should be revised rather than silently preserved in its old form. The record ought to say which premise was withdrawn and why the inquiry now takes a different shape. This is a narrower inference: a route or expectation is no longer available on the stated terms. It does not establish that the broader inquiry has been answered. Indeed, the negative finding is compatible with several remaining questions, and those questions may still resist resolution.
Accordingly, the temporary hours decision can proceed while the wider inquiry remains open. The fact of action shows only that a bounded choice was made under stated assumptions; it is neither sufficient to demonstrate that inquiry has concluded nor a claim that continued inquiry is impossible or pointless. Proceeding may narrow later options; that is a possible consequence, not an inevitable one, provided the record preserves the changed terms. What the decision settles is the immediate setting within its remit. What it leaves unsettled is the larger question whose terms have changed. Keeping those two judgments distinct allows revision without pretending that every provisional step must be final.
45.6Premature closure and endless deferral
Two failures oppose one another, but they are not mirror images. Premature closure occurs when inquiry stops while a live objection has gone unassessed. That is not the same as acting early: a provisional decision can leave inquiry open, whereas closure withdraws the objection's standing. The damage is specific. An erased objection is not refuted; it is merely unrecorded, so a later revisiting lacks the grounds on which doubt was raised.
Endless deferral is not the mere absence of closure. Inquiry may continue indefinitely at little cost while no choice is required. Deferral becomes a failure only when a needed reversible choice is postponed long enough that the choice is no longer available, or is available only on worse terms. Reversibility is therefore a candidate for earlier action, not a sufficient warrant: an apparently reversible step may carry irreversible components, and the cost of undoing may be poorly known.
Neither failure is inevitable, and the two risks turn on different conditions. Closure is premature when a material objection—one that would change the decision or its remit if sound—remains untested. It is not enough that some objection has been voiced, nor is the mere existence of an objection sufficient to block a decision. Conversely, the presence of an unsettled question is not sufficient reason to defer, since a question can remain genuinely open after a choice is made within a stated remit.
As a conditional design supposition, imagine that each recorded objection names the decision it threatens and the conditions under which it would be reopened. Then a settled decision can preserve a question rather than erase it, and later reconsideration has a definite starting point. Recording is necessary for preservation, because an unrecorded objection cannot be recovered; it is not sufficient, because a record may be ignored. The supposition therefore separates two tasks: testing the objection before closure, and keeping its terms available afterwards.
The Thursday 12:00 deadline sharpens the comparison by fixing the task and its inputs: next week’s temporary-hours schedule must be published using the objections and proposals available by that time. Against this task, premature closure and endless deferral are not symmetrical defects. Closure becomes costly when it suppresses an objection that would change the schedule; deferral becomes costly when it withholds a decision whose value lies in making it before publication. Neither cost follows automatically. An objection may prove immaterial, and a remaining question may not obstruct the decision.
The relevant comparison is therefore not simply how many questions remain open. It is whether an unresolved issue bears on the schedule and whether publication can still be revised. Closure is not sufficient to eliminate an objection if the reservation is preserved in the record; deferral is not sufficient to prevent action if the decision is made at the deadline despite the open question. Conversely, the mere presence of a deadline does not make earlier closure necessary. What matters is whether the available material supplies enough of a basis for a provisional choice.
Two conditional suppositions clarify the consequences. If the published schedule is effectively irreversible, treating every unresolved issue as a reason to postpone may leave the period uncovered, so endless deferral has a clear cost. If the schedule can readily be corrected after publication, the same postponement buys less protection and closure is less damaging, because the objection can re-enter when evidence changes. The same deadline and inputs thus support different balances depending on the consequence at stake.
For the fixed task, an unresolved objection can also be encoded in different ways: as a reason not to publish, or as a recorded reservation attached to a provisional schedule. The first encoding treats non-resolution as decisive; the second permits choice while keeping the question active. Neither encoding is universally appropriate. The reservation works only where the objection is material enough to warrant later reconsideration, while withholding publication works only where the schedule is not needed on time. That dependence is why no general midpoint between closure and deferral answers every inquiry.
45.7Maintaining an active research agenda
An agenda that keeps questions alive must record what has happened to each one, not merely that it exists. Four states do this work: answered, revised, deferred, and newly raised. In a proposed record these are prospective categories; nothing here asserts that the corresponding investigations have been carried out. Their value lies in what they let the record distinguish.
First, "answered" and "deferred" are not parallel judgments. An answer addresses evidential status under a stated framing; a deferral addresses scheduling or priority and says nothing about whether evidence would suffice. Treating deferral as a weak form of answer would make endless postponement look like accumulating knowledge. Conversely, "answered" is sufficient only relative to the question as posed; if the framing changes, the correct state is "revised," which preserves the earlier work as a partial answer to an earlier question rather than retroactively denying it had been answered.
Second, "newly raised" prevents the agenda from being read as a closed inventory. An answer can generate a sharper question that was not visible before, so growth in the list is compatible with progress rather than evidence of drift. The state marks provenance: the question entered because of something the agenda learned, not because the record was padded.
What these states do not supply is a verdict on any particular inquiry's outcome. A question marked deferred may later be answered or revised; a revised question may be deferred again. The categories describe the record's current handling of each entry, and a transition must say why that handling changed, so that a revision is not recorded as a fresh deferral. That is the contribution of the scheme: not a promise of resolution, but a way to say, for each question, what kind of work remains and why.
A long list of open questions demonstrates only that somebody can generate them. It does not show that the agenda is productive, because the count of entries measures the inventory's size, not the chances that any entry will yield knowledge. Compare two encodings of the same questions, for the same task and the same starting material: a bare enumeration, and an enumeration in which each entry is followed by a designated next investigation — the material to examine, the argument to reconstruct, or the comparison to perform. The questions are identical; only their encoding differs. The second version is more informative not because it contains more items, but because it licenses a decision about what to do next.
Even so, guiding one concrete investigation is necessary for a productive entry, not sufficient. A question may point clearly toward a next step that turns out to be idle: the proposed comparison may rest on a category that does not survive scrutiny, or the designated material may not bear on the dispute. Productive research remains possible where each entry guides an investigation; it is not inevitable. Correspondingly, an entry that supports no concrete investigation may still be worth keeping as a deliberately open inquiry, provided it is not being counted as evidence of forward momentum.
The qualification matters most at scale. A prospective design might attach to every question a specified investigation and a record of what has been attempted. Such a design would make the distinction between busy and idle entries inspectable. But unless investigations are actually carried out, the entries are plans rather than results: the encoding improves what the agenda can communicate about scope and direction, and it does not by itself advance any question. What the list can claim, at most, is a readiness to guide the work ahead.