Examination Tips Vol.00005 | How to Stop Spending Too Long on One Exam Question

To stop spending too long on one exam question, decide what the next short period of work will accomplish before you give that question more time. An exam question does not deserve another minute merely because it has already received several. It deserves another minute when you have a useful next action, a realistic stopping point and enough time left to protect the rest of the paper. These examination tips turn that distinction into a practical routine: finish the necessary work, extend deliberately when justified, or leave a clear return note and move on where the assessment permits.

This guide to exam time management focuses on a particular performance problem: one question keeps borrowing time from everything that follows. You may be stuck, but you may also be answering correctly, adding unnecessary explanation, searching for a perfect phrase, checking an already adequate solution or repeatedly promising yourself that you are almost finished. The test-taking strategies here address those local overruns through original worked decisions, short exercises, answer comparisons and review methods. They are not a universal instruction to rush, skip difficult work or obey a fixed two-minute rule.

Learning how to stop spending too long on an exam question requires three forms of attention: attention to the actual demand, attention to the work that remains on this question, and attention to the obligations elsewhere in the examination. Keep those three visible and a move-on decision becomes more concrete. Rather than asking, “Am I giving up?”, ask, “What would another minute produce, and what would that minute prevent me from doing elsewhere?” This volume is a hands-on companion to the broader How Exam Time Management Works guide.

The 50-second route

For a question you have already answered, go to the sections on completion tests and unnecessary polishing. For a question that is progressing but taking longer than expected, use the bounded-extension method. For a question that is not progressing, use the short decision check here, then consult Vol.00004 on recovering from a stuck question for the underlying problem-solving work. For a paper that is already behind schedule, use the remaining-work examples rather than trying to win back every lost minute immediately.

The smallest usable routine is: name the remaining job; name the next action; choose a stop point; then continue, finish or park. A remaining job might be “state the unit”, “explain the causal link” or “test the negative root”. A next action is something you can actually do, not “think harder”. A stop point might be completing that action or reaching a practised checkpoint. Parking means recording exactly what is unfinished, not making a vague promise that you will remember it later.

How to use this decision manual

Adrian, Jo, Aisha, Ryan, Ben, Mira, Clara and Ethan are fictional teaching characters used throughout this series. Their scripts, marks, timings and practice histories below are invented illustrations, not research participants, testimonials or predicted outcomes. The examples let us stop an examination at a decision point and examine the alternatives without pretending that a real candidate knows the mark scheme or the future.

Read the actual assessment instructions before applying any technique. An examination may restrict navigation, separate sections into independently timed blocks, prohibit returning to a submitted response or specify what can be written on the question paper. Approved extra time, rest breaks, assistive technology and other access arrangements also change the relevant conditions. None of the invented timings here replaces those rules. A flexible written practice paper is not a model for every examination in the world.

The chapters move from the decision itself to worked quantitative and written examples, then to return systems, practice design and frequently asked questions. You do not need to read the entire manual immediately before an examination. Use one case that resembles your own difficulty, attempt the decision before reading the commentary, and test the resulting routine on fresh practice material. Long explanations are for learning; the routine carried into the examination should be short.

This is also a companion rather than a replacement for the ecosystem’s explanatory articles. For the causes of early-paper overinvestment, see Why Students Spend Too Long on the First Exam Questions. For persistence after progress has stopped, see Why Students Stay on a Stuck Exam Question Too Long. Our job here is narrower: rehearse the actual decision to close, extend or leave one question.

1. Adrian’s answer was improving; his paper was getting worse

Imagine a practice assessment with three extended responses. Adrian has a workable answer to the first. His position is clear, two relevant reasons are explained, and his conclusion addresses the demand. He then thinks of a third example. It is a good example. He adds it. While writing, he notices that the opening could be more elegant. He rewrites the opening. The revised opening now seems to require a smoother transition, so he improves that too.

Nothing in this sequence looks like idleness. Adrian is working, and some of the changes genuinely improve the response. That is why the overrun is hard to recognise. A teacher looking only at the first answer might praise its development. A teacher looking at the whole script would also see the final response ending halfway through its explanation. The relevant comparison is not between the first answer before and after polishing. It is between two possible completed papers.

Jo pauses the practice replay at the moment Adrian considers the third example. She asks, “What job would that example do that the existing evidence does not do?” Adrian answers, “It would show that I know more.” That may be true, but it is not yet a reason to spend the time. The task requires a supported explanation, not an inventory of everything he knows. If the third example establishes a missing exception or answers an overlooked part, it has a specific function. If it merely repeats support already present, its value is less clear.

The better decision is not automatically to stop at two examples. There is no universal two-example rule. The better decision is to compare the next addition with the actual task and with the untouched work elsewhere. Adrian should continue if he discovers that one required dimension is missing and can address it coherently. He should finish if the answer already does the required work and his next changes are mainly cosmetic. He should not use his affection for the topic as a substitute for that comparison.

Now change the case. Adrian has written two examples, but neither explains why the stated cause would produce the observed result. The answer looks long but the central reasoning is absent. A brief sentence making that connection is not optional polishing. It is completion work. Ending immediately because he has reached a preferred number of paragraphs would be a different mistake. The move-on routine must recognise the distinction between making an answer complete and making a complete answer more comfortable to leave.

This is the first important lesson: productive activity is not the same as justified additional activity. An extra paragraph can be well written and still be the wrong use of the remaining time. Conversely, another thirty seconds can be worthwhile when it supplies a missing unit, a required conclusion or the final comparison. The decision concerns the next piece of work in context, not a moral judgement about whether effort is good.

2. What does “too long” actually mean?

“Too long” is not an intrinsic property of a question. Ten minutes can be excessive for a brief item and insufficient for a substantial response. It can be sensible early in a paper with ample remaining capacity and damaging near a fixed section boundary. A candidate with approved access arrangements may have a different legitimate time envelope from a classmate. Before deciding that a duration is excessive, specify the task, the rules and the work that duration is displacing.

For this manual, a single-question overrun means continuing to spend time on one item after you have a reason to reassess that spending, without making a deliberate decision about its effect on the rest of the assessment. Exceeding an initial estimate is a warning, not proof of failure. Estimates are imperfect. The failure is allowing the warning to pass unnoticed, or noticing it and granting an unlimited extension merely because stopping feels uncomfortable.

Suppose Mira planned six minutes for an item. At six minutes she has completed the reasoning and needs to copy a final value with its unit. A short extension has a clear deliverable. In another version, she has six lines of algebra but no idea whether the chosen relationship applies. “I have already written so much” is not a deliverable. The visible quantity of work cannot tell us by itself whether continuing is sensible.

A useful warning therefore has two parts. The clock says that the question has consumed its expected share. The work says whether a useful completion is close, whether a specific repair is available, or whether the search remains open-ended. A timing rule that ignores the work can force premature exits. A work rule that ignores the clock can allow one appealing problem to absorb the paper. The purpose of a checkpoint is to bring both sources of information together.

Do not make the definition depend on the mark you eventually receive. A decision can be reasonable under the information available and still lead to an incorrect answer. It can also be poorly justified and succeed by luck. During review, distinguish decision quality from outcome. Ask whether the candidate had a plausible next action, respected the conditions and protected other obligations. Then separately examine whether the subject knowledge and execution were correct.

This distinction matters for learning. If every successful last-minute guess is praised as excellent timing, the learner may repeat a risky pattern. If every sensible decision to park an uncertain item is criticised because the solution looks obvious afterwards, the learner may become reluctant to move on. Review the information available at the time. Do not replace the candidate’s uncertainty with the examiner’s later knowledge.

3. Keep three views of the question open

The first view is the task view: what must the answer accomplish? In a calculation this includes the requested quantity, conditions and representation. In an explanation it includes the relationship that needs to be made clear. In a comparison it includes both objects and the relevant basis of comparison. A response is not finished merely because the page contains enough writing to look like an answer.

The second view is the remaining-work view: what, specifically, is still missing? “Finish the question” is usually too broad to guide an extension. “Check whether the second root is allowed” is better. “Improve the essay” is too broad. “Add the limitation that changes the final judgement” is better. The more precisely you can name the remaining work, the easier it becomes to distinguish completion from indefinite elaboration.

The third view is the paper view: what must still happen elsewhere before time expires? A candidate can be locally successful while becoming globally incomplete. The untouched final section does not interrupt politely to request its share. It simply remains unanswered when the clock ends. A brief paper-level check is therefore necessary whenever a question asks to exceed its planned place in the sequence.

These views do not require three written checklists at every item. During practice, you may write them to learn the distinction. During an examination, a compact internal sentence can be enough: “This asks for two explained differences; I have only one; I can add the second now, but I must leave time for the final task.” That sentence contains more useful control information than “I should probably hurry.”

Consider Aisha answering a source question. The task view says she must explain one inference and support it from the passage. The remaining-work view shows that the inference is present but the support is not identified. The paper view shows two short questions remaining and a reasonable margin. Adding one relevant textual reference is a defensible completion action. Copying the entire paragraph is not required by the same analysis, even though copying would also count as activity.

Now consider Ethan answering a data question. He has described the main trend and the specified exception. He wants to calculate three additional percentages to make the answer look more rigorous. The task does not request them. The remaining-work view identifies no missing requirement. The paper view shows an untouched compulsory explanation. His additional calculations should not receive automatic permission merely because numerical work feels objective.

4. Separate time already spent from work still possible

Once a minute has passed, continuing cannot recover that minute. It can only produce something with a later minute. This simple observation is the basis of a useful question: “Starting from where I am now, is this the best justified next action?” You do not need to deny that the earlier effort mattered. It may have produced knowledge, intermediate results or a nearly complete answer. You do need to distinguish those useful products from the mere fact that time has elapsed.

The psychology literature describes a tendency to continue an undertaking because resources have already been invested. Arkes and Blumer’s study of sunk cost is a classic investigation of that phenomenon. Its existence does not prove that every learner who persists on a difficult item is displaying the same mechanism, nor does it validate a particular examination stopping rule. Here it supplies a caution: previous investment can be a poor reason for additional investment when the future alternatives have changed.

Ben has spent eight minutes on a proof. He says, “Leaving now would waste all eight minutes.” Jo asks what would make the next two minutes different from the previous two. Ben can sometimes answer: he has discovered a useful identity and can now connect the required statements. That is new information, and it matters. On another occasion he cannot name a new route. In that case the eight minutes already spent do not turn an undefined next attempt into a good one.

Keep the useful products of earlier effort. Preserve a valid diagram, a correct equation, a selected quotation or a partial outline. These may lower the cost of returning later. Do not erase sound work just to make the page feel like a fresh start. However, preserving work is different from granting the question unlimited access to the clock. You can respect previous effort by recording it clearly and still decide that the next minute belongs elsewhere.

There is a related trap in self-description. “I am the kind of student who never gives up” can be an admirable commitment to long-term learning. It is not a complete rule for allocating a fixed examination period. A learner can refuse to abandon a topic over the school year while temporarily leaving one question during a paper. Persistence in education and persistence on one item are different timescales. Confusing them can make a sensible examination decision feel like a betrayal of character.

A better identity is practical rather than absolute: “I follow useful work, and I protect the whole task.” That identity permits sustained effort when there is a reason to sustain it. It also permits a clean exit when another attempt has become open-ended. The aim is not to make students indifferent to difficult questions. It is to stop an individual question from acquiring authority over every other obligation in the paper.

5. What the external guidance supports—and what it does not

University study-support resources commonly advise candidates to inspect instructions, recognise question demands and manage the time available. Northwestern’s test-taking guidance, for example, draws attention to qualifications in wording and to examining answer choices carefully. Caltech’s guidance treats taking a test as an active process and encourages reflection afterwards. Those resources provide useful principles, not a universal number of seconds that every question should receive.

The finish, extend and park framework in this manual is an original teaching aid built around the constraints of a finite task. Its worked timings are examples to reason about, not experimentally established thresholds. A school, teacher or learner should test whether the routine improves completion, accuracy and recovery in the relevant assessment. A routine that increases abandoned work or careless errors needs adjustment, even if it sounds efficient.

In particular, do not infer that every minute transferred from a difficult question to an easier one produces a predictable number of marks. You do not know the future score in advance. A question that looks easy may contain a trap; an apparently difficult question may have a short route you have not yet recognised. The practical aim is a defensible allocation under uncertainty, not a promise of mathematically optimal marks.

For the same reason, the examples distinguish visible facts from estimates. “There are twelve minutes left” is a clock observation. “I can complete this in thirty seconds” is a forecast based on the remaining work. “This will earn two more marks” is usually less certain unless you are reviewing against an explicit teaching rubric. During the examination, use mark values to understand scale, but avoid treating speculative credit as a guaranteed return.

The next chapters make the decision concrete. You will see candidates who should continue, candidates who should stop, and candidates who should repair one small omission before leaving. This range matters. A method that always recommends moving on has not learned to discriminate. A useful method sometimes protects another minute of careful work and sometimes protects the rest of the paper from that same minute.

6. Build a completion test before you build a stopping rule

A stopping rule is only useful when the learner knows what an adequate response needs to contain. Otherwise, “move on” can become permission to leave an essential step unfinished. Start with a completion test: a small set of observable conditions derived from the prompt and the subject’s expectations. Do not confuse this with knowing the examiner’s exact mark allocation. You can often identify whether you have answered both parts or included the requested unit without knowing how every mark will be distributed.

For a numerical task, the completion test might be: the required quantity is identified, the method is valid, the result is expressed in the requested form, and any specified interpretation is supplied. For a comparison, it might be: both objects are discussed on the relevant basis, and their difference or similarity is explicitly connected. For a short explanation, it might be: the claim is stated and the reason that makes it follow is visible. These are examples of task-specific tests, not templates to paste onto every subject.

Suppose a question asks, “Calculate the average speed and state the unit.” Clara has written a division correctly but has not evaluated it or supplied the unit. Her answer is not complete merely because a method is visible. If she can finish the calculation reliably in a short time, the completion test supports continuing. If the same question also asks whether the speed was constant, a numerical result alone still leaves a distinct demand unanswered.

Now suppose a different prompt asks for one reason, and Clara has supplied a precise reason with the necessary explanation. She considers adding a second, partly remembered possibility. The completion test identifies no missing job. Adding the doubtful possibility may consume time and make the response less clear. Moving on is not a refusal to think deeply. It is a decision that the current task has been completed at the level supported by the candidate’s knowledge and the available time.

A completion test should also distinguish required communication from decoration. A labelled conclusion can help a reader find a result; an ornamental border cannot repair missing reasoning. A sentence clarifying a pronoun can make an explanation understandable; replacing every ordinary word with a more elaborate synonym does not automatically improve it. When time is short, preserve the communication that carries meaning and remove the changes whose purpose is mainly to make the answer feel impressive.

During review, take three of your own completed answers and write the completion test you should have used. Then mark the point in each answer at which the test was first satisfied. Everything after that point deserves a question: did it add a missing qualification, improve accuracy, or repeat what was already sufficient? This exercise can reveal time spent after completion, which a simple record of start and finish times would not explain.

7. Use review points, action limits and real deadlines differently

A review point asks you to reconsider. An action limit bounds a particular attempt. A deadline ends the available opportunity. These are not interchangeable. Treating a rough estimate as an absolute deadline can force unnecessary abandonment. Treating an actual submission deadline as a flexible estimate can leave the response unsubmitted. A learner needs to know which kind of boundary is being used.

A review point might occur when you have used the time originally expected for a question. It does not mean, “Drop the pen regardless of what remains.” It means, “Check the remaining job and the paper.” If one final substitution will complete a sound solution, a short extension may be justified. If you are starting an entirely new approach with no clear completion path, the same checkpoint may support parking the item.

An action limit is narrower. You decide to test one specific possibility, inspect one uncertain line or locate one relevant passage segment. At the end of that action, you evaluate what changed. You do not silently convert the action into a second, third and fourth search. A useful action limit produces information even when it does not solve the question: you learn that a candidate route does not fit, that a copied value is wrong, or that the needed evidence is not in the place you expected.

A real deadline comes from the assessment conditions. Examples include the end of an independently timed section or the final instruction to stop writing. You must follow it. Your private preference for finishing a sentence does not alter the official boundary. In practice, build the act of stopping or submitting into the rehearsal so that it is not treated as an administrative surprise after the intellectual work has ended.

Mira writes a review point into a practice plan for the end of her first response. When she reaches it, she discovers that her conclusion is missing. She gives herself a bounded completion action: state the judgement already supported by the answer. She does not open a new paragraph of evidence. Her action has a clear output and a natural end. This is very different from telling herself, “I will keep improving it for a little while.”

In another practice, Mira’s first response is incomplete because she has chosen an argument she cannot support. A deadline does not turn unsupported claims into adequate evidence. She must either use a defensible alternative within the remaining time or close the response honestly at the level she can justify. The timing method cannot manufacture missing knowledge. Its function is to prevent the knowledge gap from consuming every other opportunity.

8. Grant a bounded extension, not an emotional extension

An extension should answer four questions: what will I do, what will count as completion, what is the limit, and what work elsewhere am I protecting? The answers need not be written during an examination. They should, however, be concrete enough that someone reviewing the decision could understand why the extra time was granted. “I feel close” is not as informative as “I have the two candidate values and only need to test the condition.”

Ben reaches a checkpoint while solving an equation. He has found two roots. The prompt restricts the variable to a positive interval, and one root clearly lies outside it. His extension is: inspect the interval, reject the invalid root and state the remaining result. This is a bounded finishing task. The existence of a rejected root does not require him to repeat the entire solution if the algebra has already been checked appropriately.

Compare a less useful extension. Ben has written a long expression and says, “Another minute should make it work.” Asked what he will do, he says, “Keep rearranging.” Asked what would show that he should stop, he says, “When I get an answer.” The proposed action has no informative limit. He is not estimating a near-complete procedure; he is authorising an indefinite search in the language of a short extension.

A better version is to identify a cheap test: substitute a simple permitted value to see whether the expression behaves as expected, check the line where a sign changed, or compare the chosen formula with the givens. The test is not guaranteed to help. Its advantage is that the learner can tell whether it produced information. If it does not, the decision returns to the paper rather than dissolving into another unexamined minute.

Do not turn this into a rigid rule that only one extension is ever allowed. A large response can contain several legitimate stages, and new information can change the situation. The safeguard is that extensions must not renew themselves automatically. Every meaningful overrun should still be accountable to the task and the remaining assessment. An extension that was reasonable before an unexpected delay may no longer be reasonable afterwards.

There is also a difference between extending an item and borrowing from an unspecified reserve. “I will take it from checking time” can become a phrase that hides repeated overspending. Checking time is not an infinite account. If you reduce it, recognise what you are reducing: a completion sweep, a targeted review of uncertain work, or the mechanical task of ensuring answers are recorded correctly. Decide deliberately rather than treating the final margin as money that is never counted.

9. Use remaining capacity to test whether an extension fits

The following arithmetic is a teaching illustration, not a prescribed examination schedule. Suppose a learner has forty-two minutes left. From practice with similar tasks, the remaining compulsory work is estimated to require twenty-eight minutes. The learner also wishes to preserve eight minutes for completion checks and submission requirements. That leaves an estimated six-minute margin: forty-two minus twenty-eight minus eight. The estimates may be wrong, but the calculation makes the proposed borrowing visible.

If the current question has a specific three-minute completion route, that route may fit within the margin. The learner should still check its importance and reliability, but the proposal is not obviously incompatible with the remaining plan. If the proposed route is “try several methods until one works”, a six-minute margin does not make the search automatically sensible. A margin is capacity, not a command to spend it.

Now change only one fact. The remaining compulsory work is likely to require thirty-eight minutes rather than twenty-eight. Preserving the same eight-minute margin would require forty-six minutes, but only forty-two remain. The learner is already four minutes short of that plan. An extra three minutes on the current item would deepen the mismatch unless some other task or reserve is consciously adjusted. The question has not changed, but the decision about it has.

This example shows why “always stay if you are close” is incomplete advice. Being close matters, but so does what remains elsewhere. It also shows why “always leave when the estimate expires” is incomplete. A small, well-defined finishing action may fit easily when the paper still has capacity. The correct comparison is conditional, and the relevant conditions can be observed or estimated more explicitly than a general feeling of urgency.

Do not perform this arithmetic afresh at every short question. That would create a new time-consuming task. Use it during practice to understand the logic, and during the examination at meaningful transitions or when a large overrun threatens the plan. For ordinary items, a practised awareness of the next checkpoint may be sufficient. The calculation is most useful when the learner otherwise keeps granting extensions without seeing their cumulative effect.

When your remaining-work estimates are uncertain, avoid presenting the result as exact spare capacity. “About six minutes, assuming the final task goes as in practice” is more honest than “I have six free minutes.” The word free can encourage careless spending. A margin partly absorbs uncertainty. It is not automatically available for optional polishing, and it does not remove the need to observe whether the next task is unfolding as expected.

10. Recognise progress without measuring it by page length

Progress means that the work is moving closer to the required answer, not merely that the pen is moving. In a calculation, useful progress may be reducing the number of unknowns, establishing a necessary relationship or completing a valid transformation. In an explanation, it may be identifying the causal link that was missing. In a reading task, it may be locating the evidence that distinguishes two interpretations. These changes can be small on the page and large in their effect.

By contrast, copying the givens for a third time may add lines without changing the state of the problem. Rephrasing the same point may add words without answering another part. Searching an already checked paragraph again may add effort without new evidence. These actions are not always useless; sometimes copying exposes a mistake or rephrasing clarifies meaning. The question is whether that is what is actually happening, rather than whether the activity looks studious.

Aisha uses a simple practice question: “What do I know now that I did not know at the last checkpoint?” If the answer is a useful fact, relationship, eliminated possibility or completed requirement, she has evidence of progress. If the answer is only “I have spent more time”, the case for continuing is weaker. This question is a review aid, not a demand to narrate every thought during the examination.

There is a danger in demanding visible progress too quickly. Some reasoning requires a period of quiet organisation before a useful representation appears. The method should not punish every pause. This is why personal practice history matters. If a certain task reliably needs a brief planning interval, that interval belongs in the estimate. The problem is not silence; the problem is continuing indefinitely without any way to distinguish a necessary planning stage from a repeating loop.

To train the distinction, review a worked solution in stages. Stop before each new line and ask what the line changes. Does it introduce a relation, discharge a condition, identify a case or communicate a conclusion? Then review one of your own unsuccessful attempts. Locate the first point at which later lines stopped adding useful information. That point may be where a local repair or a timing checkpoint would have been more useful than further expansion.

For extended writing, progress can be assessed by function. An introduction establishes the answer’s direction; a paragraph supplies a reason and evidence; an analysis explains the connection; a qualification limits the claim appropriately. If three successive paragraphs perform the same function with no added distinction, the answer may be lengthening without developing. The remedy is not necessarily shorter writing everywhere. It is writing that earns its place by performing a needed function.

11. Learn five clean ways to leave a question

The first closure is complete and leave. The response satisfies the task as far as you can reasonably judge, and the appropriate local check has been performed. State or mark the final answer in the required place and continue. Do not create a return obligation merely because leaving feels final. A completed answer does not need to remain mentally open throughout the rest of the paper.

The second closure is finish one essential detail and leave. You have a sound response but one small required component is missing: a unit, a second label, the final comparison or the requested conclusion. Complete that component without using it as an invitation to reopen the whole answer. This closure is particularly useful for learners who either leave too abruptly or turn every final detail into another revision cycle.

The third closure is record the frontier and park. The response is incomplete, but useful work exists. Record the last reliable result and the next unresolved job in a permitted place. The note should help a later version of you resume efficiently. “Need to check the sign when moving the second term” is more useful than a large question mark. Parking is appropriate only when the examination permits returning.

The fourth closure is submit the best defensible response within a one-way process. Some assessments do not allow a return after a response is submitted or a section is closed. In that setting, a parking method is not available. Use the remaining allowed opportunity to produce the best response you can justify, follow the stated rules and move forward. Do not import a two-pass strategy into an environment that does not support it.

The fifth closure is stop because the official time has ended. This is not a strategic choice you can override. Follow the invigilator’s or platform’s instructions. Practising this closure matters because some learners treat the final act of recording or submitting as if it occurs outside the timed assessment. It may not. Learn the permitted process beforehand and rehearse it under the same conditions wherever practical.

These closures make “move on” more precise. They also prevent a false choice between perfect completion and total abandonment. A learner may be able to supply a missing detail, preserve a valid partial route, or submit a limited but honest answer. Which option is appropriate depends on the question and rules. What should disappear is the unexamined middle state in which the learner remains attached to the item without deciding what the next work is for.

12. Rehearse the decision aloud before expecting it to become brief

During ordinary study, practise a short explanation of the decision: “The result is correct, but the unit is missing; I will add it and leave.” Or: “The first part is complete, the next part needs a new method, and my section checkpoint has arrived; I will preserve the result and return later.” These sentences expose the logic. Once the logic is familiar, the internal version can become much shorter.

A teacher can test the explanation by changing one condition. What happens if returning is not permitted? What happens if the next question depends on the missing result? What happens if the current answer contains an obvious contradiction? The learner should update the decision rather than repeat a memorised phrase. This is how a routine becomes conditional knowledge rather than another slogan about being efficient.

Do not ask the learner to justify every movement in a full mock examination. That would alter the task and consume the time the method is intended to protect. Separate coaching from simulation. In coaching, pause and discuss. In simulation, use the routine independently, record only minimal permitted cues, and discuss the consequential decisions afterwards. The eventual test is whether the learner can act sensibly without a tutor supplying the stop signal.

The next workshop puts this framework into concrete subject tasks. Each scenario supplies the information available at the decision point, then asks whether to finish, extend or park. Attempt the decision before reading the explanation. The point is not to agree with a slogan. It is to identify which fact makes a particular use of the next minute defensible.

13. Quantitative workshop: make the decision before reading the answer

The tasks in this workshop are original illustrations. They are intentionally short enough that the timing decision can be inspected separately from a long syllabus lesson. A real assessment may assign different marks, expect different working or prohibit particular tools. Use the examples to practise deciding what additional work is for. Do not transfer their invented time allowances or marking assumptions to another paper.

For each case, answer two questions before reading the commentary. First, what is the smallest remaining job that would make the response more complete or more correct? Second, does the information supplied justify doing that job now? You may discover that the same final answer can arise from a good decision in one situation and a poor decision in another. The timing judgement depends on the state of the work, not merely on the topic.

Case A: the calculation is finished, but the answer is not

Task: A cyclist travels 1.8 kilometres in 6 minutes. Calculate the average speed in metres per second. Clara has written 1,800 ÷ 360 = 5. Her review point arrives. The official end of the examination is not close, and her next question is available. She is about to leave the answer as the number 5 because she has been told not to exceed her planned time.

Decision: finish the required unit and leave. The numerical work already expresses the distance and time in the requested units. The answer is 5 metres per second, or 5 m/s. Adding the unit is a clear, brief completion action. Clara does not need to repeat the conversion merely because the review point has arrived. She needs to close the task she has almost completed.

The tempting error is to treat the timer as if it knows the state of the response. A time estimate is not a replacement for reading the output requirement. In this example, the remaining work is small and identifiable. If Clara moves immediately and never returns, she creates an avoidable omission. The lesson is not that units always deserve extra time regardless of circumstances. The lesson is that a review point should trigger a completion check rather than a reflexive escape.

Change the facts. Suppose Clara has written 1.8 ÷ 6 = 0.3 and labelled it m/s. Now the problem is not a missing label. The calculation gives kilometres per minute, not metres per second. She must either convert 0.3 kilometres per minute to 300 metres per minute and then divide by 60, or redo the calculation consistently. The final answer is still 5 m/s, but the remaining job is larger and should be recognised accurately.

Change the facts again. The invigilator has instructed candidates to stop. That official instruction governs the action; a private completion rule does not authorise further writing. This boundary is why the manual distinguishes a review point from a real deadline. During practice, identify which boundary you are simulating so that the exercise does not teach the wrong response to the final instruction.

Case B: one valid root remains to be selected

Task: A rectangle has width x centimetres and length x + 2 centimetres. Its area is 48 square centimetres. Find its dimensions. Ben has correctly formed x(x + 2) = 48 and factorised x² + 2x − 48 as (x + 8)(x − 6). He has written x = −8 or x = 6. He has used more time than expected and is considering leaving both values without interpretation.

Decision: complete the interpretation if the paper still permits the brief action. A physical width in this problem must be positive, so x = −8 is rejected. The width is 6 centimetres and the length is 8 centimetres. A quick substitution checks that 6 × 8 = 48. Ben has a valid route and a small remaining job. His reason to continue is the unresolved condition, not the number of lines already written.

Notice what the completion action does not include. It does not require solving the quadratic again with a second method merely because two methods are available. It does not require drawing a carefully shaded rectangle unless the task demands such a diagram. It does not require explaining the entire theory of quadratic equations. The extension is bounded by selecting the admissible value and stating the dimensions.

A useful return note, if Ben genuinely must park the task, would be “roots found; reject negative width; state both dimensions”. That note preserves the frontier. “Quadratic unfinished” is less useful because it makes the return look like a larger job than it is. Good timing records distinguish a missing method from a missing final interpretation.

Now suppose the factorisation is wrong. Ben has written (x + 6)(x − 8), which expands to x² − 2x − 48 rather than the required expression. Being one line away from an answer does not make the route correct. A local expansion check exposes the sign problem. The next action is to repair the factorisation, not simply to choose the positive root. “Nearly finished” must be evaluated against valid work, not visual proximity to the bottom of the page.

Case C: a local copying error does not require a new method

Task: Two notebooks and three pens cost 13 units of money. Four notebooks and one pen cost 11 units. Find the price of each item. Adrian intends to use simultaneous equations but copies the first relationship as 3n + 2p = 13. He correctly copies the second as 4n + p = 11. His resulting numbers seem awkward, and he starts searching for a completely different approach.

Decision: perform one local input check before changing methods. The first equation should be 2n + 3p = 13. From the second, p = 11 − 4n. Substituting gives 2n + 3(11 − 4n) = 13, so −10n = −20 and n = 2. Then p = 3. The original method is adequate; the representation of the givens was wrong.

The timing issue is not simply that Adrian is slow at algebra. He is about to spend time solving the wrong diagnosis. Restarting with another method may reproduce the same copied quantities and therefore the same error. A small check of the input relationships has a clear purpose and can prevent a much longer unproductive restart. This is a justified bounded action even though the question has already exceeded its estimate.

After the repair, Adrian should not automatically rewrite every line for neatness. If the correction can be made legibly and the final route remains understandable under the assessment’s rules, a complete fresh copy may not be needed. The communication requirement is that the valid solution can be followed. A desire for an immaculate page should not silently become a second version of the overrun.

For practice review, record two separate events: the original copying error and the decision to restart. Fixing only algebra speed would miss the control problem. Conversely, praising Adrian merely for moving on would miss a cheap repair that was available. The useful lesson is to identify the smallest test capable of distinguishing a broken method from a broken input.

Case D: another decimal place will not repair the wrong average

Task: A vehicle travels at 60 kilometres per hour for 30 minutes and then at 30 kilometres per hour for 60 minutes. Find its average speed for the whole journey. Ethan writes (60 + 30) ÷ 2 = 45 kilometres per hour. He then spends time deciding whether the answer should be written as 45.0 or 45.00.

Decision: stop polishing the representation and inspect the method. The first stage covers 30 kilometres and the second also covers 30 kilometres. The total distance is 60 kilometres over 1.5 hours, giving an average speed of 40 kilometres per hour. The two speeds do not apply for equal durations, so their simple arithmetic mean does not represent the journey’s average speed.

This case illustrates a misleading form of visible near-completion. Ethan has a numerical answer and is attending to precision, which can feel like a final quality check. But a precision decision is downstream of the more important question: does the calculation represent total distance divided by total time? The next useful action is a method check, not further formatting of 45.

Would it be reasonable to continue beyond the review point? That depends on whether Ethan recognises a repair he can execute within the remaining constraints. Once he sees the two journey stages, the route is short and specific. If he does not know what average speed means, repeatedly adjusting decimals will not help. The timing review should reveal whether the missing work is a local correction or a larger conceptual gap.

The counterexample is important. If the vehicle travelled at the two speeds for equal durations, the simple average would give the overall average speed. The lesson is not “never average speeds”. It is “check the condition that makes a shortcut valid before spending time polishing its output”. A good move-on routine must be sensitive to such conditions rather than treating every familiar-looking computation as already secure.

Case E: answer the data question, not an imagined investigation

Task: A teaching table records a cart’s distance from its starting point at four times. At 0, 2, 4 and 6 seconds, the recorded distances are 0, 6, 12 and 18 metres. The question asks for the average speed over the six seconds and one observation about the recorded intervals. Aisha calculates 18 ÷ 6 = 3 m/s and notes that each two-second interval shows an additional six metres.

Aisha then begins writing about possible friction, the motor, the track surface and whether the cart might have accelerated between observations. Some of those are sensible questions for an investigation, but they are not all part of the stated task. Decision: complete the requested response and move on. The answer can state the average speed and the equal recorded distance increments without inventing a detailed physical explanation.

Care is still necessary. The table gives sampled positions; it does not by itself reveal every instant between them. Aisha should not turn the observation into an unsupported claim that the instantaneous speed was exactly constant at all times. The appropriate stopping point is an accurate statement of what the supplied data supports. This is a case where a short qualification can be more useful than a long speculative explanation.

Now alter the question to ask whether the readings are enough to establish constant speed throughout the journey. The limitation becomes central rather than optional. Aisha should explain that equal average speeds over the recorded intervals do not rule out variation within those intervals. The same thought that was a detour in the first version becomes required reasoning in the second.

The timing lesson is therefore functional: relevance is determined by the demand. A learner cannot decide that extra discussion is wasteful merely because it is extra, nor that it is valuable merely because it is scientifically thoughtful. Identify the job the discussion performs. Finish when the actual question is adequately answered, not when every interesting question raised by the situation has been explored.

Case F: a targeted trace check beats five complete retraces

Task: Start with total = 0. Process the values 2, 5 and 3 in order. At each step, replace total with twice its current value plus the next value. State the successive totals and the final total. Jo writes 2, then 9, then 21. She has checked the update rule once but feels uneasy and starts retracing the entire sequence for the fifth time.

Decision: finish and leave unless there is a specific unresolved discrepancy. The updates are 2 × 0 + 2 = 2; 2 × 2 + 5 = 9; and 2 × 9 + 3 = 21. The requested intermediate states and final state are present. A further identical trace is unlikely to be as informative as a check directed at an actual risk, such as accidentally updating with the previous input rather than the current one.

Suppose instead Jo’s sequence is 2, 9, 19. Now there is a local arithmetic error in the final update. Rechecking that transition is necessary. It does not require erasing the correct first two states or writing a second full explanation of the algorithm. The remaining job is “correct twice nine plus three”, not “start the computing question again”.

A useful practice contrast is to change the update rule to twice the sum of the current total and the next value. The results then differ: 4, 18 and 42. This variation teaches why checking the exact rule once can be more valuable than repeating the same misread rule many times. Repetition is only a check when it has a reasonable chance of revealing the relevant error.

The broader lesson applies beyond computing. A repeated calculation can reassure without testing a new failure point. Before allocating another verification pass, name the risk it will inspect. If the only reason is a general feeling that the answer might somehow be wrong, use the established completion check and protect the next task rather than reopening the response indefinitely.

Case G: do not finish the wrong quantity elegantly

Task: A right triangle has perpendicular sides of 9 centimetres and 12 centimetres. Find its perimeter. Ryan calculates one half times 9 times 12 and writes 54 square centimetres. He notices that his working is untidy and begins recopying it. The question’s review point arrives while he is still improving the layout.

Decision: stop the recopying and repair the task interpretation. The perimeter requires all three side lengths. The hypotenuse is the square root of 9² + 12², which is 15 centimetres. The perimeter is 9 + 12 + 15 = 36 centimetres. Ryan’s original answer was an area. A neat presentation of that calculation would not answer the question.

This is not a reason to ignore legibility. It is a reason to order checks sensibly. Meaning precedes cosmetic presentation. A brief reread of the requested quantity can reveal a larger problem than a long effort to make the existing answer look polished. Once the correct route is identified, Ryan can decide whether a bounded repair fits the remaining time.

If Ryan does not remember the relationship needed to find the third side, the task state is different. He has identified the misreading but still lacks a method. That situation belongs partly to the recovery work in Vol.00004. The timing decision here is to avoid spending additional minutes polishing known-irrelevant work. Recognition of the wrong quantity should change the action immediately, even if the full repair is not yet available.

In a review log, write “answered area instead of perimeter; then recopied before checking demand”. That description is more actionable than “careless and slow”. It identifies both the subject-task mismatch and the sequence that amplified the time loss. A later practice can target the output check before recopying, rather than demanding faster handwriting as a cure for a different problem.

Case H: the last necessary line is not the same as another possible proof

Task: Explain why the sum of the first n positive integers is n(n + 1)/2. Mira writes the sum forwards and backwards. She pairs the first term of one copy with the first term of the reversed copy and notices that each pair totals n + 1. There are n such pairs across the two copies. She has not yet stated why division by two is required.

Decision: supply the final connection. The pairing has counted two copies of the original sum, so 2S = n(n + 1), giving S = n(n + 1)/2. This closes the explanation. The required final line has a precise logical function. Leaving it out because the answer already looks mathematical would leave the reader to supply the decisive connection.

After closing that argument, Mira considers adding an induction proof as well. Unless the prompt specifically requests induction, multiple methods or some further justification, a second complete proof may not be the best use of examination time. A valid alternative method is valuable during learning, but its existence does not make it compulsory in every response. This distinction is especially important for learners who enjoy exploring elegant alternatives.

Change the task so that it explicitly asks for proof by induction. The pairing argument alone would no longer satisfy the method condition. Mira would need the relevant base case and inductive step. For example, from S(k) = k(k + 1)/2, adding k + 1 gives (k + 1)(k + 2)/2, which establishes the next case after the base is shown. Here the method requirement changes what completion means.

The lesson across these eight cases is not that short final actions should always be taken or that second methods should always be avoided. It is that the candidate should identify the specific requirement being completed. The next minute belongs to a task, not to a feeling. When the required work is done, the rest of the paper deserves the opportunity to begin.

14. Written-answer workshop: stop adding words when the job is done

Written answers can overrun without ever feeling finished. There is almost always another sentence that could be improved, another example that could be mentioned or another qualification that could be added. The difficulty is not solved by demanding that every answer be short. Some questions genuinely require developed reasoning. The useful distinction is between words that perform a missing function and words that merely prolong a response whose relevant work is already present.

The University of Waterloo’s guidance on long answers emphasises planning, relevant application and making the conclusion clear. It does not equate a strong answer with length alone. In this workshop, the passages and responses are original teaching material. The commentary examines their functions; it does not claim to reproduce an official examination board’s marking scheme.

Case I: an inference needs support, not an invented biography

Passage: “The team had enough wood for one more bench. Mira laid the shorter pieces aside and measured the doorway again. ‘A wide seat is useless if we cannot get it inside,’ she said. Only then did she begin cutting.” Question: What does Mira’s second measurement show about her approach to the task? Support your answer with one detail from the passage.

Aisha answers: “She is practical and checks that the bench will fit before committing the wood. She measures the doorway again and points out that the seat would be useless if it could not be brought inside.” This response makes an inference and supports it with relevant detail. Aisha then considers adding that Mira must be an experienced carpenter, probably learned from a past mistake and is likely the group’s leader.

Decision: finish the supported answer and move on. The extra biography is not supplied by the passage. It also does not perform a missing job in the question. The answer already links the repeated measurement to a practical constraint. Adding unsupported history would consume time while moving away from the evidence boundary.

There is a subtle alternative. Suppose Aisha had written only “She is careful.” That is a plausible inference, but the requested supporting detail is missing. In that version, one additional sentence is completion, not padding. The task’s second sentence determines the difference. A move-on habit that notices only that an answer has been written would miss the support requirement.

For practice, underline the exact words in the candidate response that perform the inference and the exact words that supply support. If a sentence does neither, ask why it is present. It may add a useful clarification, but it should have an identifiable function. This exercise trains a stopping point based on evidence and demand rather than on how full the answer space looks.

Case J: two descriptions need one explicit connection

Task: A fictional study room offers two booking arrangements. Arrangement A allows one two-hour booking per day, made a week in advance. Arrangement B allows two one-hour bookings per day, made on the same day. Compare their usefulness for a student whose schedule changes frequently. Jo writes a clear description of A, followed by a clear description of B, but does not connect either arrangement to the changing schedule.

Jo has reached her review point. She thinks her answer needs a more attractive opening. Decision: use the next brief action to make the missing comparison, not to decorate the introduction. She could add: “B is more adaptable to a schedule that changes on the day, while A offers a longer uninterrupted block but requires earlier certainty.” This directly connects the arrangements to the criterion in the question.

The answer should not claim that B is always better for every learner. A student who needs a continuous two-hour session might value A despite the advance commitment. The actual question supplies a particular user need, and the response should stay within that frame. A short qualification can clarify the comparison without expanding into an essay about every possible study-room policy.

Notice how small the useful remaining job is. Jo does not necessarily need to rewrite both descriptive paragraphs. The missing operation is comparison in relation to the stated condition. Recognising that operation makes the extension bounded. “Make it sound better” would not provide the same stopping point.

In a timed rehearsal, give the learner three candidate additions: a decorative opening, a third description of A, and the explicit comparison above. Ask which addition most directly repairs the response. Then ask how the answer would change if the user needed uninterrupted work rather than a changeable schedule. The exercise teaches that good additional time is spent on the missing relationship, not automatically on the longest available addition.

Case K: listing events is not yet explaining the link

Task: In a fictional control system, a robot is programmed to stop when its distance sensor reports an obstacle less than 0.30 metres ahead. The sensor reports 0.20 metres. Explain why the robot stops. Ben writes: “The sensor reports 0.20 metres. The robot stops.” He then begins a long description of the robot’s wheels, casing and possible uses.

Decision: supply the rule that connects the observation to the action. A concise answer is: “The reported distance, 0.20 metres, is below the programmed stopping threshold of 0.30 metres, so the controller activates the stop instruction.” The explanation makes the conditional relationship visible. The additional description of hardware is not required by the supplied task.

This is another case where a small addition is more important than a large expansion. Ben’s original response contains the beginning and end of the process but omits why one leads to the other. A sentence that supplies the condition completes the reasoning. A paragraph of loosely related facts does not necessarily do so, even if those facts are accurate.

Change the prompt to ask why a faulty sensor could cause an unnecessary stop. Now the response needs an additional distinction between the reported distance and the actual environment. The original threshold explanation is still relevant, but it is no longer the whole task. Ben should not use a memorised short answer when the demand has changed. Efficient completion is exact completion, not minimal wording regardless of the question.

Review this case by asking what each sentence enables the examiner to understand. The useful sentence reveals the decision rule. The irrelevant expansion provides background that the question did not request. The timing problem becomes easier to solve when the learner can identify these functions instead of estimating quality by the amount of technical vocabulary on the page.

Case L: a qualification can finish an argument

Task: Two invented classroom reports are supplied. In Report A, group members solve a problem independently, compare their methods and correct an error together. In Report B, one member completes the work while the others copy the answer without discussion. Using these reports, evaluate the claim that group work always produces better learning. Adrian explains the benefits shown in Report A but ignores Report B.

Adrian’s answer is fluent and already long. He plans to add another paragraph praising discussion. Decision: address the limiting evidence before adding more support for the same side. The word always makes the second report important. A defensible conclusion is that the reports illustrate how collaborative comparison can support learning, but merely placing learners in a group does not establish that all members are engaging with the task.

This conclusion is deliberately limited to the supplied fictional reports. It is not a research finding about the overall effectiveness of group work. The response should distinguish what the examples support from a universal claim about all classrooms. The important completion work is to use the contrasting case and adjust the strength of the judgement.

Would an extra paragraph be justified after the checkpoint? A short paragraph addressing Report B may be essential because the current answer has not evaluated the full claim. Another paragraph restating Report A’s advantages is less clearly useful. Both additions could be well written, but they perform different functions. The timing method should select the function the response lacks.

For a contrast exercise, remove the word always and ask only how Report A illustrates one possible benefit of group work. The required response becomes narrower. Discussing Report B may no longer be necessary. This change demonstrates why candidates should not use a fixed essay structure as the sole stopping test. The strength and scope of the actual claim determine how much qualification is needed.

Case M: the story needs an ending more than a new first sentence

Task: Write a short narrative in which a character makes a difficult choice and accepts a consequence. Clara has developed a scene in which a student must decide whether to admit breaking a borrowed object. Her opening is readable, the conflict is clear and the choice has been made, but the consequence has not yet appeared. She has a limited writing margin and begins replacing the opening sentence with a more dramatic version.

Decision: complete the missing narrative consequence. The task contains more than a request for an engaging beginning. Clara might show the character admitting the damage, agreeing to repair or replace the object, and facing an initially disappointed response. The exact story is hers to choose. What matters is that the narrative resolves the demanded relationship between choice and consequence.

A stronger opening could improve the piece, but it cannot replace the absent ending. The paper-level cost is also significant: revising the first paragraph can create new changes throughout the story while the required conclusion remains unwritten. The best next action is the one that brings the whole response into alignment with the task.

This does not mean that endings should always be rushed or that introductions never deserve revision. If the opening contradicts the intended events, confuses the reader or fails to establish the relevant situation, a repair may be necessary. Here, however, the opening is already functional. The learner’s preference for improving a familiar part is drawing time away from an unfinished required part.

During practice, mark a narrative draft by function: situation, pressure, choice, consequence and resolution, using only the elements appropriate to the actual prompt. Then identify which missing function would most improve completeness. This is more useful for timing than telling the learner to “write faster” while leaving the order of effort unchanged.

Case N: uncertainty alone is not a reason for another full review

Task: A fictional laboratory instruction says that a container must remain closed except when a sample is being added. A multiple-choice question asks which action does not follow the instruction. The options describe keeping it closed during waiting, opening it briefly to add a sample, leaving it open while preparing the next sample, and closing it after the sample is added. Ryan selects the action that leaves it open during preparation.

Ryan has identified the exception correctly, but he feels uncomfortable because he has selected the same option letter several times elsewhere. He rereads the item repeatedly. Decision: compare the selected action once more with the actual instruction if needed, then move on. The pattern of letters on unrelated questions is not evidence that this response is wrong. The content of this item provides the relevant reason.

A meaningful reason to reconsider would be discovering that the question asks for the action that does follow the instruction, or noticing another qualifier that was missed. Those are changes in the task model. A general feeling that “there cannot be another C” is not the same kind of information. The move-on routine should distinguish evidence-based revision from a search for a more comfortable pattern.

Scoring and navigation rules still matter. This case does not establish a universal guessing strategy, and it assumes the learner is allowed to leave the item and continue. If an assessment has penalties for incorrect responses, special selection rules or one-way submission, those conditions need to be understood separately. The local lesson is about spending additional review time on a specific uncertainty rather than on unrelated answer patterns.

To rehearse, ask the learner to name the evidence that would justify changing the response. If no relevant evidence is available and the instruction has been checked, the candidate can close the item without requiring complete emotional certainty. Examination completion is not the same as eliminating every possible feeling of doubt.

15. Triage revisions by function

When a written answer is nearly complete, possible revisions can be grouped by what they repair. A meaning revision corrects the claim, reasoning or relationship. A requirement revision supplies a missing part, evidence restriction or requested form. A clarity revision makes the existing meaning understandable. A cosmetic revision changes presentation without materially altering meaning, completeness or readability. The boundaries are not perfect, but the distinctions make the next action easier to discuss.

Consider the sentence, “The second arrangement is useful because it is useful for students.” Replacing the repeated phrase with a specific reason is a meaning revision. Consider, “The second arrangement is more flexible,” when the question asks for evidence from the supplied description. Adding the same-day booking detail is a requirement revision. Consider an ambiguous pronoun that makes it unclear which arrangement is being discussed. Replacing the pronoun with its noun is a clarity revision.

Now consider changing “useful” to “advantageous” when the original sentence is already precise and readable. That may be stylistically acceptable, but it does not automatically deserve priority over an unanswered question. The timing decision should not be controlled by the fact that cosmetic revisions are easy to notice and pleasant to make. Easy work is not always the most important remaining work.

Do not treat grammar as always cosmetic. A misplaced negative, unclear tense relationship or broken sentence can alter meaning. Likewise, punctuation can determine how a comparison or condition is understood. The relevant question is whether the proposed change makes the response more accurate, complete or intelligible. The labels above are tools for deciding, not a hierarchy that excuses unreadable writing.

In a rehearsal, take one paragraph and list four possible revisions, one from each category. Give the learner a short remaining margin and ask which change should be made first. Then change the paragraph so that the previously cosmetic feature becomes a genuine ambiguity. This forces attention to function. The learner should not memorise that a particular surface feature is always low priority.

16. Aim for minimum complete, not minimum effort

A minimum complete response contains the reasoning and communication needed to answer the actual task without unnecessary expansion. It is not a shortcut around subject knowledge. It is not a promise of full marks. It is a way to identify the point at which the answer has performed its essential jobs so that later time can be allocated consciously.

For a short comparison, minimum complete may be one precise sentence. For a substantial evaluation, it may require several developed paragraphs, evidence and a qualified judgement. For a proof, it includes the logical connection that establishes the claim, even if that connection takes more time than a familiar calculation. The phrase minimum complete should always be followed by “for this task under these conditions”.

One danger is using a short-answer habit in an extended-response assessment. Another is using an essay habit in a question that asks for one specific observation. Both arise when the learner carries a preferred output size into the task instead of deriving the output from the demand. The move-on routine must be tied to the question, not to the candidate’s usual page length.

Practice the distinction with three versions of one response. The first omits an essential link. The second supplies the link clearly. The third adds two paragraphs that restate the same point. Ask the learner to explain why the second is more complete than the first and why the third is not automatically more valuable than the second. This makes the stopping point visible without glorifying brevity for its own sake.

When the essential work is done, close the response deliberately. Read the final sentence against the question, ensure the answer is in the required place and begin the next task. The ability to leave a complete answer is part of examination performance. It allows knowledge to appear across the whole paper rather than being concentrated in the first response that felt comfortable to polish.

17. A return note should save a restart, not create another writing task

Leaving an unfinished question is easier when returning will not mean reconstructing everything. A good return note preserves three things: where the reliable work ends, what remains unresolved, and what action should be tried next. It is a memory aid for the candidate, not an additional answer for the examiner. Use only a place and method permitted by the assessment. In some settings, a platform’s own flag may be the available mechanism; in others, a brief note in working space may be appropriate.

Compare two notes. “Come back” tells you that the question is unfinished but not why. “Equation formed; check conversion from minutes to seconds before substitution” identifies the state and next action. The second note can prevent a complete restart. It can also prevent carrying an uncertain quantity into the next attempt as if it had already been verified. Its value lies in precision, not length.

For a written answer, the equivalent might be “comparison made; second source still needed” or “argument complete; final judgement must answer ‘how far’”. These notes distinguish a missing piece of evidence from a missing argument. A learner who sees only a half-filled answer space may assume the entire response needs rebuilding. A functional note shows that a smaller completion action is possible.

Do not turn note-making into a miniature essay about the failed attempt. A long account of every method tried can consume the very margin you are trying to preserve. Include only information that will change the return action. An uncertain assumption should be labelled because it affects what can safely be reused. A record that you felt frustrated for three minutes usually does not belong in the examination note, though it may belong in the later review.

Try a delayed-return exercise outside examination conditions. Work until a teacher-designated pause, leave a compact note, complete another unrelated task and then resume. Measure how much work has to be reconstructed. If the note merely identifies the question number, improve it. If writing the note takes longer than the reconstruction it prevents, shorten it. The aim is an efficient handover between your earlier and later attention, not a perfect archive of the attempt.

18. A parked question needs a place in the remaining plan

A flag is not a reservation of time. You can mark six questions for return and still have no realistic opportunity to revisit all six. That is why parking should include awareness of the remaining plan. A possible return point might be after finishing the current independent section, after reaching all compulsory questions, or during a deliberate final review. The appropriate point depends on the structure and rules of the assessment.

Suppose Clara has parked three items. One needs a final unit, another needs a local arithmetic check, and the third requires a method she has not found. If the paper allows free return, treating the flags as identical can be inefficient. Their remaining jobs differ. The unit may be a brief completion; the arithmetic check has a specific target; the method search is less predictable. The candidate should inspect what remains rather than automatically revisiting the oldest or most emotionally irritating flag first.

This is not a claim that short repairs must always outrank substantive questions. The larger question may carry a major compulsory demand, and newly available insight may make it tractable. The point is that return order should be based on the current state of the work. A question that was blocked twenty minutes ago may now have a clear route. A question that felt nearly complete may reveal a serious error when reread.

A useful practice record has separate statuses: complete, incomplete with a known next action, incomplete with no known route, and recorded answer needing a specific check. These are teaching labels, not a recommendation to write a code beside every real examination item. During training, they reveal why a list of question numbers alone may be insufficient. During an examination, a simpler permitted notation may carry the same distinction.

At the final sweep, distinguish unanswered work from uncertain work. An answer that exists but feels imperfect is not the same as a required subpart that has no response. Before spending the last available period polishing an uncertain answer, establish whether any required work has been overlooked. This is one place where the paper map from Vol.00001 becomes more useful than an intense focus on the current page.

19. Leaving one part can be sensible without leaving the whole question

A multi-part question is not necessarily one indivisible commitment. Some later parts depend on an earlier result; others ask an independent interpretation, use a supplied value or test a different aspect of the same context. Before abandoning an entire question because one part has overrun, read the dependency carefully. Do not assume that numbering alone proves that every later part is inaccessible.

Imagine a fictional data question. Part (a) asks for a calculated gradient. Part (b) asks what the graph’s vertical axis represents. Part (c) asks candidates to use a stated gradient of 4 for a separate calculation. A learner who is stuck calculating the gradient in part (a) may still answer part (b) and use the explicitly supplied value in part (c). The valid route comes from the actual wording, not from inventing a result or assuming that every paper offers the same arrangement.

Now consider a different question in which part (b) genuinely requires the result from part (a), with no supplied alternative. The learner must understand the assessment’s expectations for showing work and any stated marking arrangements. Do not promise that an examiner will award follow-through credit, and do not manufacture a convenient value merely to make the later work look complete. A transparent partial method may be useful where appropriate, but the credit rules belong to the assessment.

The time decision should recognise the dependency’s cost. A brief repair to an early result may unlock several later tasks, making it more valuable than an isolated one-mark repair. Conversely, an open-ended struggle on an early part may prevent the learner from discovering that later parts are accessible independently. The correct action requires a small structural read, not an automatic instruction to stay or leave.

During practice, draw the dependency relationships after completing a paper. Which later answers actually required earlier results? Which merely shared a passage or diagram? Which supplied their own data? This review can improve future orientation without turning the examination itself into a diagramming exercise. The goal is to stop one local overrun from being mistaken for a reason to surrender every related opportunity.

20. Protect the work already done in a shared source

Questions that share a passage, table or case create a different timing problem. Reading the source may support several answers. If one item becomes difficult, moving far away and returning later can require rebuilding that context. The cost does not make leaving wrong, but it belongs in the decision. A useful route may be to finish other accessible questions attached to the same source before switching to an unrelated section, where the rules permit.

Aisha has read a passage carefully and answered its first three questions. The fourth asks for an inference she cannot support confidently. The fifth asks for a clearly stated reason from the final paragraph. If the sequence is flexible, she may answer the fifth while the source remains familiar, then leave a precise note for the inference. That differs from rereading the entire passage repeatedly for the fourth while the fifth remains untouched.

However, shared context does not authorise careless transfer. Evidence that answers one question may not answer another. Aisha must still inspect each command and scope. “I have already read this passage” is not equivalent to “I have already interpreted every question about it”. The source familiarity saves orientation work, not the need to identify the demand.

For a data set, a compact record of verified values can reduce repeated searching. A learner might note that the relevant comparison is between the first and last rows, while labelling an unresolved unit conversion. But copied values introduce their own risk. Before reusing a note, ensure that it corresponds to the correct column, condition and question. The purpose of preservation is to save valid work, not to carry an unnoticed error into several responses.

In review, identify whether the overrun came from the difficult question itself or from repeatedly reacquiring its source. The remedies differ. A weak inference may need reading practice. Repeated source searching may need better orientation and question-to-source mapping. A general command to reduce time on the question will not reveal which process consumed the time.

21. Change the routine when navigation is restricted

The examples so far often allow a return. Treat that as an explicitly stated condition of the example, not a universal feature of examinations. A digital assessment may have its own navigation rules, and a written assessment may use separately timed sections. The candidate should learn the actual process from the official instructions and permitted familiarisation materials. Do not discover a one-way boundary by casually pressing a control during the examination.

In a one-way task, the three views still help: identify the demand, identify the remaining work and recognise the remaining opportunities. What changes is the closure. You cannot park an answer for a later visit if the system does not permit one. You must make the best defensible decision within the available conditions, record or submit the response as required and proceed. The absence of a return route makes an accurate local completion check more important.

In an independently timed section, time left in a later section may not be available to repair an overrun now. A global total can therefore mislead. A learner may think there is an hour remaining in the examination while only a short period remains in the current section. The relevant boundary for the immediate decision is the one governing the current work.

Do not generalise from one platform to another. A visible review flag does not necessarily mean that every submitted response can be reopened. A save indicator does not necessarily mean that the final submission step is complete. Conversely, a candidate should not assume a restriction that the actual instructions do not impose. The safest preparation is to rehearse the documented workflow, including recording, navigation and submission, rather than to rely on broad advice about online examinations.

If a technical problem occurs, follow the permitted reporting procedure. Do not spend a large portion of the examination inventing unsupported workarounds or assume that extra time will automatically be granted. The timing routine concerns your allocation decisions; official staff or platform procedures govern incidents. Distinguishing those responsibilities can prevent a technical difficulty from becoming a second, self-created time-management problem.

22. An open book can become a time sink too

When resources are permitted, searching them is part of the work. It is not time outside the assessment. A learner can spend too long on one question while appearing to use an allowed advantage sensibly. The useful question is what information the search is intended to obtain and what would count as finding it. “Look through the chapter” is less bounded than “locate the definition that distinguishes these two terms”.

Jo faces a permitted-resource task asking her to apply a stated principle to a case. She already knows the principle but wants the wording to match her notes exactly. She spends several minutes locating the original sentence. If exact quotation is not required, that search may not be necessary. The important work may be applying the principle to the case, which no amount of copying will do for her.

In another question, the exact definition or a specified reference is required. Now a targeted lookup can be legitimate completion work. The difference is not whether books are good or bad for performance. It is whether the search serves the actual demand. A permitted resource should reduce uncertainty relevant to the answer, not become a comforting activity whenever independent reasoning feels difficult.

Prepare resource navigation under the same restrictions that apply to the assessment. Learn what is allowed, how material is organised and how quickly you can find specific information. Do not create annotations, indexes or digital tools that the rules prohibit. If a resource search repeatedly takes too long in practice, review whether the difficulty is organisation, unfamiliar terminology or an unclear question model.

A bounded search has an exit. If the needed item is not found in the expected place, decide whether another targeted route is justified or whether you should use what you can support and continue. Reopening the same pages without a new search idea is the resource equivalent of rearranging the same equation indefinitely. The visible activity changes, but the underlying overrun is the same.

23. A worked late-paper decision: recover the plan, not the lost minutes

Consider an invented practice paper with thirty minutes remaining. Adrian is still polishing an answer that now satisfies its main requirements. Three obligations remain: a short data task estimated at six minutes, an extended explanation estimated at fourteen minutes, and final completion checks estimated at four minutes. These estimates total twenty-four minutes, leaving a six-minute margin for uncertainty or specific repairs.

Adrian proposes spending another five minutes improving the current answer’s opening. That would reduce the remaining margin to one minute, even if every later estimate proves accurate. The current response has no identified missing requirement. The proposal is therefore hard to justify. A more defensible choice is to close it, complete the data task and preserve a substantial uninterrupted period for the explanation.

Now change the current answer. It contains a serious contradiction that can probably be repaired in two minutes by replacing a conclusion that does not follow from the evidence. The repair has a specific function and may fit within the estimated margin. Adrian should still keep it bounded: correct the contradiction and leave, rather than treating the repair as permission to rewrite the whole answer.

After the data task, suppose he has only twenty minutes left because it took eight rather than six. The plan needs another update. Fourteen minutes for the explanation and four for checks leaves two minutes of margin. The earlier six-minute figure no longer governs the situation. A timing plan is useful only while it is compared with what actually happened.

A common reaction is to try to recover the two unexpected minutes by writing the explanation at an unrealistic speed. That can replace one overrun with rushed interpretation or incomplete reasoning. First inspect the response’s required functions. Remove optional elaboration, not essential links. Use a concise plan and direct sentences. The goal is to fit a complete, defensible response into the actual remaining opportunity, not to punish yourself for the earlier delay.

The phrase “recover time” is therefore slightly misleading. Passed minutes cannot be recovered. What can be recovered is control over the remaining work. The candidate can stop unnecessary additions, identify accessible tasks, protect required conclusions and prevent another unbounded extension. That is a practical recovery even when the original schedule can no longer be achieved exactly.

24. A timing routine must fit the learner’s legitimate conditions

Do not equate a slower response with a poor decision. A learner may need more time to read, write, process a complex representation or use an approved support. The relevant question is whether the available time is being used deliberately under that learner’s actual assessment arrangements. Comparing every student with the fastest writer in the room can obscure both genuine access needs and specific habits that could be improved.

Use the authorised conditions in rehearsal. If extra time or rest breaks have been approved, understand how they operate in the particular assessment rather than assuming that every accommodation simply adds an interchangeable block of working time. Obtain procedural clarification through the school or assessment provider where necessary. This manual does not determine eligibility, interpret an individual arrangement or replace official instructions.

For a learner who writes slowly, repeated full recopying may be especially costly. For another learner, frequent clock checks may interrupt concentration more than a small number of section-level review points. A student using assistive technology may need to account for navigation or response-entry steps that are absent from a handwritten example. The timing method should be adapted to those observed processes, not imposed as a single pace requirement.

Keep the review respectful and specific. “You spent four minutes rewriting a sentence that was already clear” identifies an action. “You are too slow” labels a person without explaining what to change. The former can lead to a concrete experiment: preserve the clear sentence and use that period to complete a missing part elsewhere. The latter may increase pressure while leaving the actual allocation problem untouched.

The aim is neither to normalise avoidable overruns nor to dismiss every difficulty as a habit. Distinguish the skill being assessed, the access conditions, the knowledge needed and the allocation decisions within the available time. A useful routine works with that whole picture. It helps the learner make better choices without pretending that all learners, tasks and examinations have identical constraints.

25. Measure the overrun you actually want to change

A start time and finish time reveal duration, but not why the duration occurred. Two learners may each spend twelve minutes on a question. One uses the full period to construct a required argument; the other completes the argument in eight minutes and spends four minutes recopying an already legible introduction. A timing intervention should not treat those records as equivalent. To improve allocation, identify the activity inside the interval.

In a coached practice, record the first point at which the response satisfies a reasonable completion test. Then note any later work and its function. Was it a correction, a required addition, a useful clarity change or optional polishing? This record can reveal post-completion time without assuming that every revision is wasteful. Some late revisions prevent serious errors. The category must be justified by the response, not assigned merely because the clock has advanced.

Also record premature exits: occasions when the learner left a small, identifiable required action unfinished even though a reasonable opportunity remained. A method that reduces average question time while increasing these exits is not necessarily improving performance. The goal is better allocation, not the smallest possible duration. Keep accuracy and completeness visible alongside speed.

A third measure is the quality of parked work. How many items were deliberately left for return? How many were actually revisited? How much reconstruction was needed? How many remained unnoticed at the end? A flagged question that never receives its promised return is different from a deliberate final decision not to invest more time. The record should distinguish forgetting from conscious prioritisation.

These measurements belong mainly to training and review. Do not burden a real examination with an elaborate logging exercise that the rules do not permit or that consumes the candidate’s attention. A teacher can observe a short coached task; a learner can reconstruct major events immediately after a mock; a permitted minimal note can capture a decisive checkpoint. Use the least intrusive method that supplies reliable information.

26. Read an invented timing log without inventing a causal success story

Here is an invented record from a thirty-minute rehearsal. It is not a student result or evidence that the method raises scores. The learner spends nine minutes on Question A, seven on B, eight on C and six on D. The durations sum to thirty minutes. The record alone does not show whether the allocation was sensible; it only shows where the time went.

The annotated script adds useful detail. A was substantively complete after six minutes; its final three minutes were spent recopying a clear paragraph. B needed a difficult but relevant explanation throughout its seven minutes. C was left with one requested label missing after eight minutes because the learner moved abruptly at a checkpoint. D was still incomplete when the thirty-minute boundary arrived.

The most direct practice hypothesis concerns A: could the learner preserve the adequate paragraph and transfer some of those three minutes to required work elsewhere? That is a plausible test, not a guaranteed result. The next paper will contain different demands, and the learner may not identify completion at exactly the same point. A good review proposes a change that can be observed rather than announcing a fixed future mark gain.

C supplies a second lesson. Moving earlier is not always the answer. The missed label suggests that the exit routine needs a small completion check. The training target is therefore two-sided: stop unnecessary work on A-type tasks and avoid leaving easy-to-identify requirements unfinished on C-type tasks. A single instruction to “move on faster” would address only one side and might worsen the other.

Suppose the next comparable rehearsal takes twenty-eight minutes and all four responses are complete. That is encouraging, but it does not isolate the effect of the timing routine. The questions may be easier, the learner may know the topic better, or the response format may be more familiar. Record those differences. Repeated improvement across varied but appropriate tasks is more informative than one favourable comparison presented as proof.

Do not compare a first attempt on an unseen paper with a second attempt on the identical paper and attribute all speed gains to better stopping. Remembering the questions and solutions changes the task. Repeating a paper can be useful for learning, but it is a different test from transferring a routine to fresh material. Keep the purpose of the exercise clear in the record.

27. Design practice so that the decision, not the slogan, is tested

A learner who has just read a chapter on moving on may start leaving every difficult question quickly. That is not yet mastery. The practice set should include cases where staying is justified, cases where a final small action is necessary, and cases where another extension is not supported. Without those contrasts, the exercise rewards obedience to the latest instruction rather than discrimination.

Use tasks whose subject demands are accessible enough that the allocation decision can be discussed. If every question is far beyond the learner’s current knowledge, most pauses will reflect missing content. If every question is effortless, there will be no meaningful pressure to extend. A useful training set includes fluent work, manageable challenge and a small number of genuine blockages, chosen with the learner’s level in mind.

Separate the timing intervention from the content repair. After a mock, a learner may need to relearn a method and also improve the decision to stop a failed attempt. Both matter, but they answer different questions. Content practice asks, “How do I solve this kind of problem?” Allocation practice asks, “What do I do with the next minute when my current route is or is not producing useful work?”

Use a baseline before changing the routine. Observe what the learner already does well. Some students leave efficiently but fail to return; others return reliably but reread too much; others need no elaborate timing aid for short items but lose control on writing tasks. A narrowly chosen intervention is easier to evaluate than replacing the entire examination routine at once.

The following twelve-session sequence is an adaptable teaching programme, not a validated dose of training. Sessions can be shorter or longer, repeated or combined according to need. Do not force a student through increasingly tight timings when the evidence shows that comprehension or knowledge requires repair first. The purpose is to establish a reliable decision routine under appropriate conditions.

28. A twelve-session programme with a different job in every session

Session 1: locate the first point of adequate completion

Choose three previously completed responses: one calculation, one short written answer and one longer response if appropriate to the learner. Read each prompt and agree on its essential jobs. Then inspect the response line by line to identify the first point at which those jobs were satisfied. Do this without timing pressure. The learner is learning what completion looks like, not racing to find it.

Classify later additions by function. A correction should not be counted as needless polishing; a repeated example should not be called essential merely because the learner worked hard on it. End with one personal completion cue, such as “both requested quantities stated” or “claim, evidence and explanation connected”. Use the cue on a fresh short task and check whether it supports an accurate stopping point.

Session 2: distinguish a missing detail from a missing method

Prepare several partially completed answers. Some should lack only a unit, label, final substitution or explicit comparison. Others should have no valid route or should be based on a misread question. Ask the learner to name the remaining job before completing anything. The important observation is whether the learner can distinguish a small closing action from a larger unresolved problem.

After the classification, allow selected repairs. Compare the learner’s estimate with the actual work required. A response that looks nearly finished may contain a conceptual error; a sparse response may need only a short inference. Record one case where visual appearance misled the estimate. This creates a reason to inspect task function rather than page fullness at future checkpoints.

Session 3: practise an extension with a named output

Use a moderately challenging task and pause at a planned review point. The learner must propose a next action and identify what its completion would look like. Reject vague proposals such as “do more” or “try harder” until they become concrete. A suitable proposal might be to test an admissible value, expand a factorisation, locate a cited sentence or write the missing causal link.

Allow the action, then ask what changed. Did it produce a result, eliminate a possibility or reveal a different error? If nothing changed, practise ending the extension without automatically granting another. The success measure is not whether the whole question was solved. It is whether the learner gave the additional time an identifiable purpose and evaluated its result.

Session 4: leave a note that survives a delay

Interrupt an unfinished practice question at a meaningful point. Ask for a permitted-style return note containing the reliable frontier and the unresolved job. Then have the learner complete a different task before returning. Do not allow an immediate return, because the exercise is intended to test whether the note supports reconstruction after attention has moved elsewhere.

On return, observe whether the learner restarts from the beginning, reuses an uncertain assumption or resumes accurately. Compare two versions of the note. The shorter one is not automatically better; the useful one saves more reconstruction than it costs to write. End by reducing the note to the minimum information that still supports a clean return.

Session 5: stop polishing a complete written answer

Give a short writing task and ask the learner to produce a complete, direct answer. Once it meets the agreed requirements, present several possible additions: a missing qualification, a repeated example, a decorative opening and a clarity repair. Some additions should be useful under the prompt; others should not. Ask the learner to choose only changes with a specific function.

Then place a second required task beside the first. The learner must decide whether the next period belongs to revision or to the untouched response. This makes the opportunity elsewhere visible. Discuss the difference between improving one paragraph in isolation and producing a complete set of answers. Do not reward awkward brevity; reward relevant completeness and a deliberate transition.

Session 6: preserve accuracy when a checkpoint arrives

Use a short mixed set in which several items are close to completion at their review points. Include one missing unit, one overlooked second part and one already complete answer. The learner should not respond to every checkpoint by immediately leaving. The task is to inspect the remaining job and distinguish a justified closing action from unnecessary continuation.

Review any premature exits as carefully as overruns. Ask whether the learner could identify the missing requirement at the time and whether a brief completion was realistically available. A lower total duration accompanied by more incomplete answers is not the intended outcome. Adjust the cue until it supports both closure and forward progress.

Session 7: work inside a shared-source cluster

Choose a passage or data set with several questions, including one that is deliberately more demanding. Let the learner identify which items can be completed independently and which depend on prior work. Where the practice rules permit, test whether finishing accessible questions within the same source reduces unnecessary rereading compared with repeatedly leaving and reacquiring the context.

Afterwards, inspect every answer against its own demand. Saving source-reading time must not become permission to use the same evidence indiscriminately. Record whether the difficulty lay in source location, question interpretation or subject reasoning. The next practice should address the actual bottleneck rather than treating the whole source cluster as a single slow task.

Session 8: recognise when a return route is unavailable

Create a clearly labelled practice condition in which each response is final once the learner moves forward. Explain that this is an invented simulation and not a statement about their real examination. The learner must adapt the closure routine: check the local requirements, make the best defensible response within the allotted opportunity and proceed without a promised return.

Compare this experience with a flexible-navigation practice. Ask which parts of the routine remained useful and which changed. The three views still apply, but parking does not. End by checking the actual assessment’s documented navigation rules, rather than letting the simulation itself become a new assumption about what will happen in the real examination.

Session 9: make a remaining-work estimate after an overrun

Begin a short paper with a deliberately imperfect plan. After a task takes longer than expected, ask the learner to identify the remaining required work and revise the plan once. The learner should not spend the whole remaining period calculating a perfect schedule. The objective is a practical decision about what must still be completed and where optional elaboration can be removed.

Review whether the new plan counted checking and response-entry requirements realistically. Also review whether the learner tried to recover the delay by rushing essential reading. A successful revision protects the remaining task functions. It does not merely distribute an unrealistic shortage across all later questions and call the arithmetic a solution.

Session 10: run a full segment without coaching interruptions

Use a fresh segment long enough for several decisions to interact. The teacher should not announce every stopping point or rescue each overrun. The learner uses the practised cues independently under the stated conditions. Collect the script and any minimal permitted flags. Discuss the major decisions afterwards while they are still reasonably easy to recall.

Evaluate the whole segment: required completion, accuracy, justified extensions, unnecessary polishing and return reliability. Avoid judging the routine solely by one difficult question. A locally impressive rescue may still have displaced too much later work; an unsolved item may have been handled sensibly while preserving the rest. The review should retain that paper-level perspective.

Session 11: transfer the routine to a different response format

If previous practice focused on calculations, use a suitable written response. If it focused on essays, use a source-based or technical task. Keep the learner’s knowledge level appropriate. Ask which completion conditions change and which allocation questions remain stable. The goal is not to force identical timings across subjects but to recognise the same decision structure in a different form.

Look for false transfer. A learner may attempt to give an essay the same short closing check used for a numerical answer, or may write an extended explanation where a precise result is requested. Correct the task model first. The transferable skill is identifying the remaining job and its cost, not copying the surface routine unchanged.

Session 12: write a personal rule and test its counterexample

Ask the learner to write one practical rule supported by the training record. For example: “When I have answered the demand and my next change is only a repeated example, I close the answer and check the next required task.” The rule should identify a condition and action. Avoid global promises such as “I will never spend too long again”.

Then construct a counterexample. What happens if the supposed repeated example actually supplies the missing second source? What happens if a final label is absent? What happens if navigation is restricted? The learner should refine the rule to accommodate the relevant condition. End with a short fresh test. A rule that survives a meaningful contrast is more useful than a slogan that sounds decisive but fails when the question changes.

29. Decide whether the routine is helping

The clearest evidence of useful change is a pattern of better-controlled decisions on fresh, appropriate tasks. Look for fewer unexplained extensions, less time spent after adequate completion, fewer unnecessary restarts and more reliable returns. At the same time, check that accuracy, legibility and completion have not deteriorated. These observations support a practical judgement about the learner’s routine; they do not establish a universal causal effect.

Review failures with the same specificity as successes. If the learner leaves too early, strengthen the completion test. If the learner recognises an overrun but cannot end it, rehearse the closure and return note. If every question overruns despite disciplined decisions, revisit content fluency, task estimates and legitimate assessment conditions. A local stopping routine cannot compensate for every other difficulty.

Keep one or two changes active at a time. A learner trying to remember a dozen new rules may spend more attention managing the rules than answering the paper. Once a useful habit becomes dependable, shorten its cue. The final examination routine should feel like a small set of decisions that supports the work, not a second examination running alongside it.

For further practice design, use the ecosystem’s How Timed Practice Works guide and the companion training examination timing at home article. They address the broader relationship between practice, accuracy and pace. This volume’s narrower contribution is a set of observable decisions about when one question should stop receiving additional time.

30. Decision clinic: identify the fact that changes the answer

The following cases are not a test of whether you remember the phrase “move on”. They test whether you can identify the fact that makes moving on, finishing or extending appropriate. Some deliberately withhold an important fact. In those cases, the correct learning response is to name what is missing rather than manufacture certainty. During an actual examination, use the information available in the instructions, the question and your current work.

Decision 1: five minutes on a three-mark question

A learner says, “I have spent five minutes on a three-mark question. Should I leave?” That statement gives a duration and a mark value, but it does not show the remaining job, the paper’s duration, the task type or whether returning is allowed. Five minutes may be a warning, yet the numbers alone do not establish the best immediate action.

A useful response is to inspect the current state. If the answer is valid and only its final label is missing, close it. If the learner is starting a speculative new approach and substantial required work remains, parking may be defensible where permitted. If the question contains a shared source supporting later parts, the surrounding structure may change the choice. The deciding information is not simply whether five exceeds three.

The exercise exposes a common misuse of minutes per mark: treating a planning ratio as if it were a complete diagnosis of every item. Ratios help distribute attention, but they do not tell you whether a specific next action is useful. State the missing fact before issuing a universal timing instruction.

Decision 2: an extra optional answer

An invented paper instructs candidates to answer exactly two of three extended questions. Aisha has completed her two chosen responses and begins outlining the third because she recognises the topic. Her second response still contains an unclear final comparison. There is no instruction saying that an extra answer will substitute automatically for a weaker one.

The legitimate next action is to follow the stated selection rule and improve the required submitted work where useful. Familiarity with an unchosen topic does not create a new obligation. Aisha should not assume that extra work is harmless or that an examiner will choose the most favourable combination. The assessment’s instructions determine which responses are required and how additional work is treated.

Change the instruction to permit answering all three with a clearly stated scoring rule, and the decision environment changes. The point is not that extra answers are always forbidden. It is that a timing choice must be made within the actual answer-selection rules, not a candidate’s hopeful interpretation of them.

Decision 3: the clock has moved, but the answer has not

Ben has checked the same algebraic line three times. Each time he reads it in the same direction and concludes that it might be right. He has not compared it with the previous line, expanded the brackets or tested a permitted value. He proposes another identical check because the result still feels unfamiliar.

The next useful action is not automatically another pass. Name the uncertainty: was a sign changed incorrectly, was a denominator lost, or does the result fail an expected condition? A targeted check can then test that risk. If no specific discrepancy is found and the response satisfies the completion test, leaving may be more defensible than repeating the same reassurance loop.

A new check should have a chance to reveal information that the previous checks did not reveal. This does not mean that every check must use a completely different method. It means that “again” is not, by itself, an explanation of what the extra time is intended to accomplish.

Decision 4: a newly discovered contradiction

Clara has written a conclusion saying that the supplied evidence supports a claim in every case. Her preceding paragraph explicitly describes a counterexample. She notices the contradiction just as she reaches her planned move-on point. The remaining paper still has a reasonable margin, and the conclusion can be revised without rebuilding the whole answer.

This is a strong case for a bounded repair. The problem concerns the meaning of the answer, not its elegance. Clara can align the conclusion with the evidence by limiting the claim appropriately. She should not merely delete the counterexample to preserve the stronger conclusion, nor begin a new argument unrelated to the supplied material.

After repairing the contradiction, close the response. The repair does not grant an unlimited licence to polish every paragraph. A useful extension ends when its named job is done. This case protects careful work from a crude interpretation of timing discipline.

Decision 5: an answer-space mismatch after a skip

Ryan skips a question in a practice paper that uses a separate numbered answer sheet. He then records the next answer in the skipped question’s row. He notices the mismatch two items later. His instinct is to keep going because correcting the sheet will cost time and he is already behind.

The mismatch is a concrete recording problem that can affect several answers. Follow the assessment’s permitted correction process and realign the question numbers before continuing. Saving a minute by leaving responses in the wrong places is not useful time management. The required work includes communicating the answer in the designated location.

For later practice, add a brief numbering check at the point of resuming after a skip. Do not respond by banning all skips regardless of the paper. Repair the specific weakness in the return and recording system. A strategy’s implementation costs should be measured and improved, not ignored.

Decision 6: an attractive idea arrives after parking

Mira parks a difficult interpretation and later realises that two supplied details can be connected. She now has a plausible answer route. However, she is in the middle of a different response whose argument is also developing. The paper permits returning, and no immediate section boundary is approaching.

The new idea matters, but it does not necessarily require an instant jump. Preserve a compact permitted cue if useful, finish the current meaningful unit of work and reassess at a natural transition. An insight is a reason to update the return plan, not an automatic instruction to interrupt every other task.

When Mira returns, she should still test the idea against the original prompt. A remembered connection can feel more convincing than it is. The timing method allows new information to improve the plan while avoiding repeated, impulsive switching between unfinished answers.

Decision 7: the spare margin has already been spent

Ethan began a practice paper with a six-minute contingency margin. He used two minutes on one repair and three on another. At a later question, he says, “I can use my six-minute margin for this.” He has remembered the original plan but not the subsequent spending.

The arithmetic leaves one minute of the original margin, assuming all other estimates still hold. Ethan needs to compare the current remaining time with the current remaining work. A plan is not a permanent entitlement. Every extension changes the available capacity, and later estimates may also have changed.

This case does not call for constant bookkeeping during every item. It calls for a realistic update at meaningful checkpoints. The danger is not imperfect arithmetic alone; it is using an outdated plan to justify a new decision. The next minute must be assessed against the present paper, not the paper as imagined at the start.

Decision 8: a correct answer in an unfinished format

Jo has correctly calculated an exact fraction in a practice question that explicitly requests a decimal to two places. She considers the mathematical work finished and wants to move on. In a different question, the instruction requests an exact form and she has only a rounded decimal. Both situations involve a valid calculation that has not yet been communicated in the required form.

The completion test must include the output condition. Where a brief, reliable conversion remains possible within the rules and time, perform it and leave. Do not argue that presentation is always lower priority than mathematics; here the representation is part of the mathematical task.

Conversely, if no additional precision is requested and the answer is already appropriately expressed, producing more digits may add no useful function. The deciding feature is the specified form, not a general preference for fractions, decimals or longer numerical output.

Decision 9: the easiest remaining action is not the required one

Adrian has an incomplete explanation that lacks evidence from a required source. He finds it easy to improve the title and difficult to locate the evidence. He spends the final part of the question’s allocation refining the title, then reports that he was working continuously.

Continuous work is not the relevant completion measure. The title may be readable already, while the missing evidence remains central. Adrian should use a targeted source search or produce the most defensible response he can support within the available time. He should not substitute easy decorative work for the required difficult function and treat both as equivalent progress.

In review, avoid blaming the learner merely for preferring easier work. Identify the trigger: an unclear source-search plan made a cosmetic action attractive. A future rehearsal can teach how to define a bounded evidence search and how to close the response if the necessary support cannot be found.

Decision 10: no single question explains the delay

A learner says, “I never spend very long on any one question, but I still run out of time.” The script reveals small delays everywhere: repeated rereading of instructions, several full recopies, frequent page searches and repeated clock checks. No isolated question looks dramatically excessive.

The local overrun method is only part of the remedy. Add up the small repeated processes and inspect the paper-level workflow. Perhaps the learner needs clearer answer-location tracking, more fluent retrieval or a less intrusive checkpoint system. Do not force every timing problem into the category of staying too long on one difficult item.

The series separates these problems for a reason. Vol.00005 concerns the decision to stop allocating more time to one question. Paper-level budgeting, question reading, retrieval and recording have their own jobs. Recognising that boundary is itself good diagnosis: use this routine where it fits, and connect to the appropriate broader guide where it does not.

31. A teacher conversation that changes the next attempt

Imagine a review conversation after a mock. Adrian says, “I knew the material, but there was not enough time.” A response such as “You need to work faster” may be true at a very broad level, yet it does not identify a change. The teacher instead asks to see the script at the point where the first major overrun occurred. The purpose is to locate a decision, not to challenge the learner’s sincerity.

The teacher asks, “What was missing from this answer when you started the final paragraph?” Adrian says, “I wanted another example.” The teacher follows with, “Which requirement would that example satisfy?” Adrian rereads the prompt and notices that the two existing examples already support the explanation. The final paragraph repeats the same relationship. Now the discussion has an observable object: an addition whose function was unclear.

Next the teacher asks, “What was still unanswered elsewhere?” Adrian points to the final task. The comparison is no longer between a plain answer and an impressive answer. It is between optional elaboration here and required work there. The teacher can then propose a specific rehearsal: on the next suitable task, use the completion test before adding another example and move to the next required item once the test is satisfied.

Adrian objects, “But what if two examples are not enough?” That is a useful objection. The teacher should not replace one rigid habit with a rigid two-example limit. They examine what the question requests and what the examples actually establish. A third example could be necessary if it supplies a missing case or required source. The rule becomes conditional: add another example when it performs a needed function, not merely because another example is available.

The conversation ends with an observable test rather than a promise. Adrian will attempt a fresh task, mark the point of adequate completion in the coaching version and compare later additions with the remaining paper. The teacher will check whether the new routine preserves quality and improves coverage. This creates a feedback loop grounded in work rather than a repeated appeal to willpower.

The same conversation can be adapted to Jo’s repeated checking or Ben’s full restarts. Ask what uncertainty the next check would resolve, or what evidence shows that a new method is needed. Then test a smaller action. The form of the conversation remains stable: identify the state, inspect the proposed work, consider the alternative use of time and agree on a specific next experiment.

32. Eight personal cues, eight different safeguards

Adrian’s cue is “What job does the next paragraph do?” His risk is continuing with fluent material after the demand is satisfied. His safeguard is not a word limit imposed on every answer. It is a functional check before adding another example or rewriting an adequate opening. He should be able to name the missing requirement, not merely the additional content he could produce.

Jo’s cue is “What uncertainty will this check resolve?” Her risk is repeating verification without a new target. Her safeguard is a specific discrepancy, condition or output requirement. Once the relevant check is complete, she can leave without waiting for a feeling of perfect certainty. If she finds a real error, she repairs it; the cue does not tell her to ignore evidence.

Aisha’s cue is “Which evidence is required here?” Her risk is widening a source search beyond the actual question. Her safeguard is a bounded reference target and a clear distinction between supplied evidence and plausible invention. She spends additional time when it secures necessary support, not when it merely collects more interesting material.

Ryan’s cue is “Review point or real deadline?” His risk is responding to a rough timing estimate with an abrupt exit, or treating an official boundary as flexible. His safeguard is to identify which limit governs the decision. A short completion action can be justified at a review point; official stopping instructions remain binding.

Ben’s cue is “Find the last reliable step.” His risk is restarting a whole solution when one copied value or transition is wrong. His safeguard is a local check before changing methods. He preserves sound work, labels uncertain assumptions and uses a new route only when the diagnosis supports it. This can save reconstruction without pretending that every problem has a quick repair.

Mira’s cue is “What capacity remains now?” Her risk is remembering the original margin after several extensions have used it. Her safeguard is an updated view at meaningful transitions. She does not continually recalculate the paper, but she does not spend an obsolete reserve twice. Her plan changes when the observed work changes.

Clara’s cue is “Leave a useful frontier.” Her risk is parking accurately but returning inefficiently. Her safeguard is a compact note identifying the unfinished job and any uncertainty that must not be carried forward. She tests the note after a delay in practice and shortens it until it saves more time than it consumes.

Ethan’s cue is “Scale is not guaranteed return.” His risk is treating mark values as precise predictions of what another minute will earn. His safeguard is to use marks to understand the task’s importance while inspecting the actual remaining work. He avoids spending excessive time calculating an imaginary optimum instead of answering the paper.

These cues are examples for the fictional residents, not personality categories into which real students must fit. A learner may show several patterns on different days or subjects. Choose a cue from observed work, test it and revise it when the evidence changes. The purpose of the character stories is to make distinct decisions visible, not to label a student permanently.

33. Frequently asked questions about spending too long on exam questions

How long should I spend on one exam question?

There is no single duration that fits every question. Start from the actual paper’s available time, required work, mark distribution and task types, then use practice to test the estimate. The estimate should create a review point. At that point, inspect whether you have a brief necessary completion, a specific repair or an open-ended search. Those states justify different decisions even when the clock reading is the same.

Do not import a two-minute or one-minute-per-mark rule from another assessment without checking its assumptions. Such figures may fit an example and fail elsewhere. A useful timing plan protects the whole paper while allowing sensible local adjustment. The detailed budgeting problem is separate from the local decision this volume teaches.

Should I always skip a question when I am stuck?

No. First determine whether a short, targeted action could resolve the blockage and whether the paper permits returning. A copied value, missing unit or overlooked condition may be repairable without a full restart. An undefined search with substantial required work still untouched is a different situation. Skipping becomes useful when it is a deliberate allocation decision with a valid return or closure process.

For the diagnosis of the blockage itself, use Vol.00004. This volume asks when additional time remains justified. Do not confuse the two questions. Understanding why you are stuck can identify a repair; it does not automatically authorise unlimited time to carry out every possible repair during a finite paper.

What should I do when I am almost finished?

Replace “almost” with a named remaining job. If the job is to state the final dimensions, add a required source detail or correct one identified arithmetic step, you can assess a bounded extension. If the job is to discover a method, settle an unclear argument or make the answer feel more impressive, the remaining work may be less predictable than the phrase “almost finished” suggests.

Check that the preceding work is reliable. A response can look one line from completion while being based on the wrong quantity or an invalid assumption. A short verification of the relevant condition is more useful than trusting the visual closeness of an answer line.

Is moving on the same as giving up?

Not necessarily. In a flexible paper, moving on can preserve work for a later attempt while protecting other required tasks. In a one-way assessment, it may be the necessary closure after the best defensible effort within the conditions. Neither decision determines how much effort you should devote to learning the topic after the examination.

Long-term persistence and single-item persistence operate on different timescales. You can return to a difficult method in study, seek feedback and master it without allowing one examination question to consume every remaining opportunity. The useful standard is deliberate effort with a purpose, not refusal to leave under all circumstances.

How do I know whether I am polishing or completing?

Compare the proposed change with the question’s requirements. A missing causal link, second requested reason, relevant source reference or correct unit is completion work. Replacing an already clear sentence with a more decorative version may be polishing. The distinction is functional: what does the change enable the response to do that it could not do before?

Do not categorise every language change as cosmetic. Grammar, punctuation and word choice can alter meaning. A clarity repair may be necessary even late in the answer. The test is whether the revision improves accuracy, completeness or intelligibility in a way that matters to the task, rather than merely making the response more pleasing to its writer.

Should I leave a difficult high-mark question for later?

The high mark value makes it important to understand the task and protect a realistic opportunity to attempt it. It does not mean that the question deserves unlimited immediate time. Inspect its dependencies, the route available, the remaining compulsory work and the navigation rules. A substantial question with a clear method may deserve protected attention; an unbounded attempt should still trigger review.

Do not confuse leaving temporarily with failing to reserve any return capacity. A high-value task parked at the end of an already full plan may never receive the opportunity you intended. Record its remaining job and consider its place in the paper-level schedule.

Can I use marks to predict exactly what another minute will earn?

Usually you cannot know that precisely during the examination. The printed mark allocation tells you the scale of available credit, but you may not know whether your existing work is correct or how much a proposed addition will improve it. Treat estimates of likely credit as uncertain. Do not let an elaborate imaginary calculation of mark returns consume time that could be spent on the task.

During review, an official or teacher-provided marking guide can help identify which functions mattered. Use that feedback to improve future task recognition, while remembering that a new question may distribute its demands differently. Mark awareness should sharpen judgement, not create false precision.

What if I never feel certain enough to leave an answer?

Use a completion test and a targeted check rather than waiting for a particular feeling. Ask whether the response answers the demand, whether the required form is present and whether a specific unresolved discrepancy remains. If a real discrepancy exists, address it within a bounded plan. If only general unease remains, another identical review may not supply useful information.

Practise this distinction on low-stakes material before relying on it in a major assessment. A teacher can help compare the feeling of uncertainty with the actual quality of the work. Persistent distress that substantially disrupts functioning deserves appropriate support beyond a timing checklist; this manual is an educational decision aid, not a clinical assessment.

How do I avoid forgetting a skipped question?

Use a consistent permitted marker and a deliberate completion sweep. A marker should identify what remains, not merely express frustration. At a suitable return point, inspect the parked items against the actual remaining time. Before the assessment ends, verify that every required question and subpart has been considered, rather than assuming that a flag guarantees a later visit.

Test the system in practice. If skipped items are repeatedly lost, simplify the notation or improve the return checkpoint. Also check answer numbering when using a separate response sheet. A safe skip includes accurate recording and retrieval of the unfinished task, not just leaving its page.

Should I erase incomplete working before moving on?

Do not automatically erase valid work. A correct relationship, labelled diagram or useful outline can make a later return more efficient and may be relevant to the assessment of the response. Follow the paper’s rules for presenting and correcting work. Clearly distinguish a rejected route or uncertain assumption from the reasoning you intend to submit.

Preserving work does not mean presenting contradictory answers without clarification. If a method is known to be invalid, correct or identify it appropriately. The objective is to retain reliable progress while making the final response understandable, not to leave every attempted idea as if all were simultaneously endorsed.

What if the next question depends on this answer?

Read the dependency rather than assuming it. A later part may use the result, supply an alternative value or ask something independent about the same context. If a brief repair unlocks several later tasks, that can strengthen the case for the repair. If later work is independent, do not let one blocked part conceal accessible opportunities.

Do not invent a result or presume a particular follow-through marking rule. Use the actual instructions and legitimate subject reasoning. During preparation, examine appropriate marking guidance with a teacher so that you understand the expected presentation of partial work without relying on rumours about guaranteed marks.

How often should I check the clock?

Use meaningful review points rather than a universal frequency. A section transition, a planned major-task boundary or a genuine overrun can justify a check. Looking after every sentence may disrupt the work; ignoring the clock until the final warning can conceal drift. The appropriate pattern should be tested with the learner’s task types and authorised conditions.

A clock check should lead to a useful comparison: what has been completed, what remains and whether the next proposed action fits. Repeatedly observing that time is passing without changing any decision is not the same as managing it. Keep the check brief and connect it to an action.

What should I change when I am already behind?

Inspect the remaining required work and remove unnecessary elaboration before rushing essential reasoning. Close answers that are already complete, preserve clear partial work where a return is allowed and protect the functions that later responses must perform. Update the plan using the time actually left, not the original margin.

Do not try to repay a delay by reading every later prompt carelessly. That can create new errors and further repairs. A concise, task-aligned response is different from a hurried response to the wrong question. Recover control over the remaining work even when the original schedule is no longer achievable.

Can the same stopping routine work across subjects?

The decision structure can transfer: identify the demand, name the remaining job, assess the next action and protect the rest of the assessment. The completion test must change. A proof, an inference, a calculation, a narrative and a technical trace do not have identical requirements or natural stopping points.

Use the quantitative and written workshops to practise that distinction. Transfer the questions you ask about the work, not a fixed number of minutes or sentences. When a new format causes trouble, inspect its task requirements before concluding that the whole timing method has failed.

Should I start using a complicated new system the night before an examination?

A demanding new routine may introduce additional decisions before you have tested whether it helps. Close to an assessment, favour a small, familiar cue grounded in your actual practice, such as checking for a missing required part before leaving or naming the purpose of another revision. Do not replace every established habit with an elaborate sequence merely because it appears in a guide.

The longer programme here is designed for training. Its purpose is to make the eventual routine simpler. Use an immediate review to confirm the real instructions, answer locations and known personal risks, then reserve substantial experimentation for suitable practice rather than treating the live examination as the first trial.

What should a parent or tutor look for in a marked paper?

Look for the distribution of completion, not just the final score. Are early answers expanded beyond the demand while later tasks are unfinished? Are there repeated full recopies? Do parked questions remain blank? Are brief final requirements missing because the learner moved abruptly? These patterns suggest different interventions.

Ask the learner about the information available at the decision point. Avoid judging every choice with hindsight from the solution. The useful review identifies an observable action to test next time, such as stopping after the evidence requirement is met or checking a specific transition before restarting a method. It should not reduce the learner to a label such as careless or slow.

What is the first useful action after a mock in which I ran out of time?

Choose one consequential overrun and reconstruct it. What did the question require? What had you completed? What did you do next? What remained elsewhere? Identify whether the problem was missing knowledge, a poor estimate, unnecessary expansion, repeated checking or a failure to close and return. One well-understood event can guide a more precise practice than a general promise to be faster.

Then inspect whether the same pattern appears elsewhere in the script. A single example may be unusual. A repeated pattern provides a stronger reason to train that particular decision. Keep the next intervention small enough that you can tell whether it helped.

Does this method guarantee that I will finish the paper?

No. Completion also depends on the assessment’s demands, subject knowledge, reading, execution, legitimate access conditions and events during the examination. This method addresses one controllable part: whether additional time on a question is being granted deliberately. It cannot turn an unknown method into knowledge or create working time beyond the official limit.

The appropriate test is whether the routine improves your decisions on fresh practice without damaging accuracy or completeness. Retain what helps, repair what causes premature exits and seek the right support for difficulties outside the routine’s scope. A bounded, honest tool is more useful than a promise that no examination can surprise you.

34. Find the right companion, not another version of the same advice

Within this manual, use the completion test when you do not know whether the answer is finished; bounded extensions when useful work remains; the quantitative workshop for calculations and technical reasoning; the written-answer workshop for evidence, comparison and response development; return notes for clean parking; and the practice programme for sustained training.

For orientation before answering, read Vol.00001: How to Read an Exam Paper Before You Start Answering. For interpreting the exact demand, read Vol.00002: How to Read Exam Questions Carefully Without Wasting Time. For selecting a starting route, read Vol.00003: How to Decide Which Exam Question to Answer First.

The broader operating framework remains Examination Craft. The paper-level explanation remains How Exam Time Management Works. Where the recurring issue is the meaning of printed marks in time allocation, the explanatory companion is Why Students Ignore Mark Allocation When Managing Exam Time. These resources have different jobs; they should be used together rather than read as competing universal formulas.

The external references used in this volume are Northwestern’s test-taking strategies, Caltech’s test-taking guidance, Waterloo’s long-answer guidance, and Arkes and Blumer’s paper on sunk cost. They support the limited principles attributed to them in the text. They do not endorse this particular framework or establish the invented timings as universal standards.

35. The routine to carry into practice

When a question asks for more time, pause long enough to identify the decision. What is still required? What will I do next? What will count as the end of that action? What remains elsewhere? If the response is complete, close it. If one essential detail is missing and a brief completion is justified, finish it. If a specific repair has a reasonable bounded opportunity, attempt it. If the search remains open-ended, preserve useful work and move according to the paper’s rules.

After leaving, make the transition real. Read the next task rather than continuing to rehearse the previous difficulty while your eyes move over new words. If a return is planned, trust the permitted record enough to engage with the current question. If new information arrives, update the plan at a suitable point. Do not allow the unfinished item to keep taking attention after you have stopped giving it writing time.

After practice, review the decision with the script in front of you. Preserve justified extensions; remove automatic ones. Protect required details; stop optional expansion. Improve return notes that fail; simplify checks that consume more attention than they save. The method becomes useful through these adjustments, not through repeating the words finish, extend and park.

The final principle is straightforward: give the next minute a job, not merely a hope. Examination time management is not a contest to leave every question quickly. It is the discipline of spending enough time to do the required work, noticing when the next addition no longer deserves priority, and allowing the rest of the paper to receive the attention it requires.

Build the paper-wide budget behind this decision: Examination Tips Vol.00006 | How to Budget Time According to Marks works through required and optional marks, minutes-per-mark calculations, reading and checking allowances, section checkpoints and revised budgets after an overrun. Use Vol.00006 to allocate the paper’s available time, then return to this volume to decide whether one question deserves a further bounded action.