How to Plan Properly | Build an Error-Correction Budget

A marked paper can create more work than the learner can possibly finish.

One Mathematics mock contains twelve wrong answers.

An English essay returns with margin comments on relevance, development, grammar, evidence and timing.

A Science paper contains missing facts, incomplete mechanisms, graph-reading errors, unit mistakes and rushed final answers.

A Humanities response shows weak evidence selection, two inaccurate details, one unfinished paragraph and a conclusion that never quite answers the qualifier.

The common reaction is simple:

Fix everything.

Write every mistake into an error log.

Copy every correct answer.

Redo every question.

Add every weak topic to the revision timetable.

Do another paper to see whether the problems disappear.

Then another paper creates another list.

The correction queue grows faster than the learner can close it.

This is how an examination plan can become overloaded by its own feedback.

The mistake is not that the learner is correcting errors.

The mistake is assuming every visible error deserves the same amount of scarce preparation capacity.

It does not.

Some errors expose a high-value missing prerequisite that affects many later questions.

Some are one-off slips.

Some are repeated global controls such as missing qualifiers or poor time allocation.

Some are low-frequency difficult questions whose repair cost is enormous.

Some are already fixed by the time the paper is returned.

Some correct answers are actually fragile and deserve more attention than a wrong answer.

Some wrong answers do not need to enter the active plan at all.

A strong learner therefore needs more than an exam error log, mistake notebook or folder of corrected papers.

The learner needs an allocation system.

An error-correction budget decides how much scarce time, attention, feedback and retesting each mistake is allowed to consume before the next examination.

This is not a budget for making mistakes.

It is a budget for correcting them intelligently.

The 50-Second Read

Every error should pass through seven questions before it becomes active revision.

1. What exactly failed? Knowledge, retrieval, question reading, selection, execution, reasoning, expression, transfer, timing or control?
2. Did it repeat? One isolated error and a recurring error should not receive the same allocation.
3. How many marks does the mechanism expose? Count downstream risk, not only the mark value of one question.
4. How repairable is it? A two-minute cue and a six-week foundation rebuild are different investments.
5. How much runway remains? Deep correction can be rational at Day 90 and irrational at Day 3.
6. What evidence proves the correction worked? Fresh retest, delayed return, transfer or integrated paper?
7. What work leaves to fund it? Correction is not free. Every active repair needs capacity.

Use five states:

• Log: record enough information to preserve the signal.
• Active correction: the mistake has earned scarce repair capacity.
• Monitor: not enough evidence yet; wait for recurrence or another sample.
• Maintain: repaired enough that only occasional checking remains.
• Drop or contain: full repair has poor return inside the remaining runway.

The core rule is:

Do not budget correction by how annoying an error feels. Budget it by expected examination value.

This page continues the How to Plan Properly series. Plan by Marks, Not by Chapters owns mark-aware allocation. Plan Around the First Weak Link owns the response after a diagnosis. Stop Adding Work and Decide What to Drop owns subtraction from overloaded plans. Use Mock Results to Rewrite the Plan owns the post-mock structural rewrite. How to Use Past Papers Properly owns the attempt–mark–diagnose–repair–reattempt loop. This page owns a different decision: how much correction capacity each discovered error receives once the plan has more mistakes than it can fully process.

What This Page Owns

This page owns the transformation:

error evidence → correction priority → allocated repair capacity → retest → release.

It does not own generic mistake tracking.

An error log is a registry.

A mistake notebook stores examples and patterns.

An exam wrapper prompts reflection.

A marked paper supplies evidence.

An error-correction budget decides what happens next when those sources generate more possible work than the learner can carry.

The distinction is important because recording an error is cheap.

Correcting it properly can be expensive.

Correction may require:

• diagnosis;
• re-teaching;
• targeted practice;
• feedback;
• a fresh retest;
• a delayed return;
• transfer to a changed context;
• timed integration;
• later maintenance.

The plan needs to know which errors deserve that entire chain.

What High-Quality Study Guidance Gets Right

Strong external study guidance already agrees on an important starting point: a returned assessment should not end with the score.

Carnegie Mellon University’s Exam Wrappers ask students to identify strengths and weaknesses, examine how they prepared, characterize recurring errors and adapt future learning.

The University of Waterloo’s exam-wrapper guidance similarly asks students to identify difficult questions, classify error patterns and decide what they will do differently next.

UCL’s guidance on preparing students for exams using feedback recommends identifying common mistakes and linking feedback to resources that help students improve.

Retrieval Practice guidance also emphasises that feedback is more useful when it explains why an answer was correct or incorrect and supports transfer rather than merely announcing a grade.

These ideas establish the feedback loop.

The planning problem begins one step later.

Once the learner can see many weaknesses clearly, which of them gets tomorrow’s best hour?

Which receives ten minutes?

Which is monitored?

Which is contained?

Which is ignored because the repair cost is now too high?

That is the correction-budget problem.

Why “Fix Every Mistake” Fails

The phrase sounds disciplined.

It can produce terrible allocation.

Imagine a learner completes a 100-mark paper and loses thirty marks.

Ten marks are lost through one repeated question-reading problem.

Six marks disappear because one algebraic execution error spreads across several topics.

Four marks come from a deep advanced topic that would require several weeks of rebuilding.

Three marks come from one unusual difficult item.

Two marks come from a unit slip.

Five marks disappear because the final page was rushed after too much time was spent earlier.

A “fix everything equally” plan may schedule six separate correction projects.

A correction-budget plan sees leverage.

Question reading may expose many future marks.

The algebraic execution issue may be a high-fan-out weak link.

The late-paper timing loss may be a global control issue.

The deep advanced topic may move to a long-term or containment state depending on runway.

The unusual item may be logged and left alone.

The unit slip may need one small final-answer cue rather than a topic review.

Same paper.

Completely different plan.

An Error Is Not Automatically a Repair Project

This is the first principle.

Wrong answers are evidence.

Evidence must be interpreted before it becomes work.

An error can be:

• random noise;
• a repeated mechanism;
• a missing prerequisite;
• a fragile correct method executed badly;
• a timing consequence;
• a question-reading failure;
• a low-frequency edge case;
• a symptom of another upstream weakness;
• a one-off lapse caused by unusual conditions;
• a high-value global control problem.

Logging preserves the signal.

Budgeting decides whether the signal deserves intervention.

The Error-Correction Budget Has Four Currencies

Students often think only in hours.

Correction consumes at least four currencies.

1. Time

Diagnosis, practice, feedback and retesting require minutes and hours.

2. Attention

Active correction uses high-quality cognitive capacity. A difficult weak link should not always be scheduled in a low-energy window.

3. Feedback Capacity

Some errors can be self-corrected. Others need a teacher or tutor. Feedback bandwidth is finite.

4. Fresh Evidence

Every repair should eventually face a fresh question, delayed return, transfer task or integrated paper. Fresh validation material is itself a resource.

A correction plan that budgets only the first practice session underestimates the real cost.

There Is Also a Fifth Currency: Switching Cost

Every active correction project adds a context switch.

Algebra.

Essay relevance.

Science graphs.

History evidence.

Timing.

Checking.

If all six are active, the learner spends cognitive energy moving between repair programmes.

This is why work-in-progress limits matter.

A strong correction budget deliberately keeps some known errors waiting.

The Correction Queue

Treat discovered mistakes as a queue.

Not all errors enter active work immediately.

A useful queue contains:

• Active: receiving repair capacity now.
• Next: high priority but waiting for capacity or a dependency.
• Monitor: needs more evidence before activation.
• Maintenance: repaired enough to preserve cheaply.
• Containment: known weakness whose full repair is currently uneconomical.
• Closed: no further active correction unless recurrence reopens it.

This is more useful than a giant undifferentiated mistake notebook.

The Correction Backlog

A correction backlog forms when errors are discovered faster than they are processed.

This is common during heavy past-paper and mock phases.

The learner completes another paper because the schedule says “past paper.”

The paper generates fifteen errors.

Only five are corrected properly.

The next paper arrives.

Another twelve errors appear.

Soon the learner has a large archive of known problems but no completed repair loops.

Measurement has outrun learning.

The Correction-Backlog Limit

Set a limit.

One significant unprocessed paper may be enough to pause another full paper.

Two unreviewed essays may be enough to pause more full writing.

A backlog of unresolved timed sections may be enough to return to targeted work.

The exact number depends on the subject and learner.

The principle is:

Do not discover errors faster than the plan can decide what to do with them.

Error Debt

Error debt is not the same as having mistakes.

It is the accumulation of known error patterns that continue to generate mark loss because their correction loops remain open.

One uncorrected sign-control pattern can create debt across many future Mathematics questions.

One qualifier-reading problem can create debt across several essays.

One poor checking routine can create repeated losses across papers.

High-fan-out errors create expensive debt.

Correction Debt Versus Error Debt

Error debt is the mark-loss risk that remains.

Correction debt is the unfinished work created by trying to fix it.

The learner can have low error debt but high correction debt if they schedule too many unnecessary repair tasks.

The learner can have low correction debt but high error debt if they avoid analysing mistakes.

A good budget controls both.

The Error Registry

An exam error log or mistake notebook remains useful.

But keep it lean.

For each meaningful error, record enough to support a future decision.

Source.

Question or task family.

What happened.

Likely mechanism.

Mark exposure.

Recurrence.

Next action.

Next evidence date.

State.

Do not copy the entire question unless necessary.

Do not write long decorative reflections.

The registry should serve the budget.

The Difference Between an Error Log and a Correction Budget

An error log asks:

What did I get wrong?

Why?

What is the correct approach?

When will I review it?

A correction budget asks:

How much capacity does this error deserve relative to every other known weakness?

What work leaves to fund the repair?

What is the minimum intervention that can recover the marks?

When should the repair stop?

What evidence closes it?

The budget is the allocation layer above the log.

Do Not Log Every Wrong Answer as a Separate Item

Ten questions can represent one mechanism.

Logging them separately creates false complexity.

Suppose Ben makes five negative-sign errors in different Mathematics topics.

The useful registry entry is not five separate topic problems.

It is one recurring sign-control mechanism with five examples and high fan-out.

Cluster before budgeting.

Do Log Correct Answers for the Wrong Reason

Error correction should not be restricted to visibly wrong answers.

A correct guess is fragile.

A correct answer reached through invalid reasoning can fail on the next surface.

A low-confidence correct answer may reveal unstable retrieval.

These items belong in the evidence pool.

The budget can decide whether they deserve active correction or monitoring.

High-Confidence Wrong Answers Deserve Attention

A confidently wrong answer can indicate a misconception.

Misconceptions can have high fan-out because the learner is applying an incorrect model consistently.

Use a discriminating question.

Ask the learner to explain the reasoning before showing the correction.

These errors can deserve more budget than a careless slip worth the same mark.

The “Careless Mistake” Problem

Search language often includes careless mistakes in exams.

The phrase is understandable.

It is usually too broad for a correction budget.

Careless can mean:

• misread qualifier;
• copied number incorrectly;
• negative sign lost;
• unit omitted;
• calculator mode wrong;
• answer not transferred to the correct box;
• unanswered subpart;
• rushed late in the paper;
• changed a correct answer without evidence.

Each needs a different control.

The budget should replace “be more careful” with a named error signature and a small intervention.

The Error Signature

An error signature is a recurring pattern that the learner can recognise.

Examples:

• sign flips after expanding brackets;
• percentage calculated from the wrong base;
• qualifier ignored in essay prompts;
• cause stated without mechanism;
• evidence described without linking to claim;
• unit omitted at final answer;
• difficult question held too long;
• correct multiple-choice answer changed from discomfort.

Error signatures are ideal budget units because they often cross several questions.

The Recurrence Test

How often has the error appeared?

Once.

Twice.

Across several contexts.

Across several papers.

Across several subjects.

Recurrence increases confidence that a real mechanism exists.

But recurrence alone is not enough.

A repeated low-value issue can still have lower priority than a new global timing failure.

The Mark-Exposure Test

How many future marks can this mechanism plausibly affect?

Do not count only the marks lost in the current question.

A one-mark sign error can recur across ten questions.

A prompt misreading can damage an entire essay.

A Science mechanism omission can affect multiple topics.

A timing error can expose the entire final section.

Mark exposure is forward-looking.

The Fan-Out Test

How many downstream performances depend on this correction?

A prerequisite with high fan-out deserves more budget.

Basic algebra.

Question-reading control.

Evidence selection.

Scientific causal reasoning.

These can influence many later tasks.

The Repair-Cost Test

What does full correction actually require?

One self-cue?

Ten fresh questions?

A teacher explanation?

Several retrieval returns?

A week of prerequisite rebuilding?

Three full essays with feedback?

Budget the whole chain.

The Repairability Test

High repair cost does not always mean low priority.

If the error is foundational and the examination is months away, deep repair may be rational.

If the examination is three days away, the same repair may have poor return.

Repairability is cost interpreted through runway.

The Diagnostic-Confidence Test

How certain are we about the cause?

If the diagnosis is weak, do not allocate a large repair budget yet.

Use a small discriminating test.

For example, if a Mathematics error could be either formula retrieval or algebraic execution, ask one question that separates them.

Budget diagnosis before treatment.

The Transfer-Risk Test

Will the correction survive a changed surface?

A learner may fix one exact question and still fail the next version.

High transfer risk increases the need for fresh varied validation.

This adds cost to the budget.

The Timing-Risk Test

Does the correction hold under realistic time?

Some errors disappear untimed and return under pressure.

If the real examination is timed, the correction budget must include a timed gate where relevant.

The Maintenance-Risk Test

Will the repaired skill decay without return?

High-forgetting knowledge may need spaced maintenance.

A stable control routine may need only occasional checking.

Budget the maintenance tail.

The Opportunity-Cost Test

What cannot happen if this correction enters?

Another subject’s active repair?

A full paper?

Sleep?

Retrieval maintenance?

Teacher feedback on another essay?

Correction should compete against the best alternative use of capacity.

The Error-Correction Priority Formula

A useful qualitative model is:

Correction priority ≈ mark exposure × recurrence × fan-out × repairability × confidence ÷ total correction cost.

This is not a numerical formula.

It is a decision lens.

A high-value error tends to expose many marks, recur, affect several downstream tasks, be reasonably repairable and have a clear diagnosis.

A low-value correction may have tiny exposure, low recurrence, low fan-out, high cost or weak diagnostic confidence.

Add Runway to the Formula

Runway changes the result.

At Day 90, deep repairs have time to mature.

At Day 30, medium-depth repairs remain possible.

At Day 14, leverage and repair speed matter more.

At Day 7, short global controls become attractive.

At Day 1, most deep correction should stop.

The budget is dynamic.

Add Probability of Recurrence

Some errors are unlikely to reappear because they depended on an unusual question.

Others are structural and likely to recur.

Probability matters.

A low-frequency advanced edge case may not deserve the same correction budget as a recurring denominator error or qualifier-reading failure.

Add Severity of Recurrence

If the error returns, how expensive is it?

One mark?

An entire paragraph?

The final twenty marks of a paper?

Severity multiplies recurrence risk.

The Correction Budget Matrix

High exposure / low repair cost: activate immediately.

High exposure / high repair cost: major project if runway allows; containment if it does not.

Low exposure / low repair cost: fix if spare capacity exists or if the cue is easy.

Low exposure / high repair cost: monitor, defer or drop.

This matrix prevents emotionally loud but low-value mistakes from consuming the week.

The Recurrence–Cost Matrix

Repeated / cheap: high priority.

Repeated / expensive: investigate fan-out and runway.

One-off / cheap: small correction or cue.

One-off / expensive: usually monitor unless consequence is high.

The Confidence–Consequence Matrix

Clear cause / high consequence: act.

Clear cause / low consequence: proportionate correction.

Unclear cause / high consequence: diagnose first.

Unclear cause / low consequence: monitor.

This matrix protects the learner from expensive repairs built on guesses.

The Correction States

Each meaningful error can move through a state machine.

Observed.

Classified.

Budgeted.

Active repair.

Fresh retest.

Delayed return.

Transfer.

Timed or integrated validation.

Maintenance.

Closed.

Reopened if recurrence returns.

This is far more useful than simply writing “corrected” beside a question.

Observed Does Not Mean Active

This state distinction is important.

The learner can know an error exists without giving it today’s capacity.

Observation preserves information.

Activation spends resources.

Do not confuse them.

Classified Does Not Mean Solved

Naming the mechanism is only diagnosis.

“This was a retrieval error” does not recover the mark.

The learner still needs a suitable intervention and evidence gate.

Corrected Does Not Mean Closed

Reading the solution or understanding the teacher’s explanation is not enough.

The learner must reproduce the performance independently.

For important errors, fresh evidence matters.

Fresh Success Does Not Always Mean Maintenance

Immediate success can be supported by memory of the correction.

Use a delayed return when retention matters.

Use changed surface when transfer matters.

Use timing when time matters.

Maintenance Does Not Mean Forgotten

Maintenance is an intentional low-cost state.

The issue is stable enough that it no longer deserves active repair volume.

It still receives occasional evidence.

Closed Does Not Mean Impossible to Reopen

Errors can recur.

Write reopening criteria.

Two repeats.

One high-consequence recurrence.

Failure under integrated conditions.

State transitions keep the budget responsive.

The Active-Correction Limit

Set a work-in-progress limit.

One or two major active corrections may be enough for a student with several subjects.

Small cheap cues can coexist with them.

But five deep correction projects create fragmentation.

Keep the active frontier narrow.

The Correction Queue Should Be Prioritised Weekly

Once a week, review the queue.

Which active error passed its gate?

Which waiting error has become more urgent?

Which monitored error repeated?

Which containment decision should remain?

Which maintenance item decayed?

The budget moves as evidence changes.

The Correction Budget Should Be Rewritten After Major Mocks

Use Use Mock Results to Rewrite the Plan.

Mocks can reveal that an old correction has succeeded, that a new global error has appeared, or that the total workload no longer fits.

After a mock, the correction budget should often change.

The Budget Should Shrink Near the Examination

As runway shortens, fewer deep errors should remain active.

The plan becomes more selective.

Short high-return repairs.

Known error cues.

Timing controls.

Maintenance.

Containment.

By the last twenty-four hours, most correction projects should be closed.

The 90-Day Error-Correction Budget

At ninety days, deep repair can be rational.

Foundations.

Vocabulary systems.

Algebraic prerequisites.

Reading inference.

Scientific models.

Long-term writing development.

The budget can fund longer projects because there is time for repeated feedback, spacing and transfer.

The 60-Day Error-Correction Budget

At sixty days, deep repair remains possible but should begin facing authentic application.

Do not spend two months fixing isolated exercises without transfer.

The budget should include retesting and mixed conditions.

The 30-Day Error-Correction Budget

At thirty days, prioritise high-fan-out and high-recoverability errors.

Use the 30-Day Runway.

Limit large rebuild projects.

Protect enough time for full-paper integration and final validation.

The 14-Day Error-Correction Budget

At fourteen days, distinguish repair from containment aggressively.

High-value short chains remain active.

Deep low-repairability gaps may move to containment.

Strong areas keep maintenance floors.

The 7-Day Error-Correction Budget

The Final 7 Days can support one or two major active repairs, not a historical archive of every mistake.

Use the most recoverable marks.

Compress the error log into final personal error signatures.

The Last-24-Hours Correction Budget

The Last 24 Hours should not become backlog-clearing day.

Use only small familiar corrections and cues.

Protect performance.

Examination Morning Correction Budget

There is almost no correction budget on Examination Morning.

One forgotten formula can be reactivated.

One familiar cue can be reviewed.

Deep projects are closed.

The system has handed over to execution.

The Mathematics Error-Correction Budget

Mathematics creates many visible mistakes, but the correction budget should be organised around mechanisms rather than chapter names.

A wrong calculus answer can come from calculus knowledge.

It can also come from algebra.

It can come from sign control.

It can come from copying a value incorrectly.

It can come from calculator mode.

It can come from method selection.

It can come from running out of time and rushing the final line.

If every wrong calculus question becomes “revise calculus,” the correction budget is too coarse.

Mathematics Budget Category 1 — Representation

Can the learner translate words into equations, diagrams, tables or graphs?

Representation errors often have high fan-out because many later operations depend on the initial model.

If Alicia repeatedly builds the wrong equation from word problems, this may deserve more budget than a difficult isolated manipulation later in the paper.

Mathematics Budget Category 2 — Selection

Can the learner choose the appropriate method without the topic label?

Selection errors deserve mixed practice, not more blocked worksheets.

The correction budget should pay for discrimination.

Mathematics Budget Category 3 — Execution

Does a known procedure break during manipulation?

Sign changes.

Fractions.

Expansion.

Substitution.

Rounding.

Units.

These errors are often cheap to repair and therefore high-return.

Mathematics Budget Category 4 — Verification

Does the learner have a reliable check?

Substitution back into the equation.

Estimate of magnitude.

Unit consistency.

Graph sense.

Domain or condition.

A small checking cue can recover repeated marks without large revision volume.

Mathematics Budget Category 5 — Control

Does one difficult problem consume disproportionate time?

Does the learner leave routine questions unfinished?

Does checking happen too early or not at all?

These are whole-paper budget issues.

Use timed sections or full papers after local correction.

Mathematics Scenario: Ben’s Five Errors Are Really One

Ben’s mock contains mistakes in algebra, trigonometry, functions, coordinate geometry and calculus.

The first version of his mistake notebook lists five topics.

Closer analysis shows the same mechanism in four of them: a negative sign is lost when a bracketed expression is rearranged.

The correction budget does not fund five topic reviews.

It funds one sign-control repair.

Local fresh items.

Then transfer across several topics.

Then a mixed timed section.

One correction project replaces five apparent weaknesses.

Mathematics Scenario: Clara’s Error Log Is Full of “Careless”

Clara has twenty entries labelled careless.

The label is useless.

Reclassification shows:

• six qualifier or condition misses;
• five copied-number errors;
• four unit omissions;
• three premature rounding errors;
• two arithmetic slips.

The budget prioritises qualifier reading and copied-number control because they recur most and affect many topics.

Units receive one final-answer cue.

Rounding receives a small rule.

The isolated arithmetic slips remain monitored.

The giant “careless” project disappears.

Mathematics Scenario: Ethan’s Deep Topic Gap

Ethan loses eight marks on one advanced topic and six marks through timing across the rest of the paper.

Ten days remain.

The advanced topic would require several prerequisite sessions.

The timing issue can be improved through exit rules and paper allocation.

The correction budget may put the advanced topic into containment and fund the timing repair.

This is not a statement that the topic is unimportant.

It is a runway decision.

The English Error-Correction Budget

English creates a different budgeting problem because one upstream error can contaminate a long response.

If a learner misreads the exact qualifier, every later paragraph may be linguistically competent and still lose relevance.

If the learner’s evidence selection is wrong, beautiful expression cannot repair the argument.

The correction budget should therefore trace the first divergence.

English Budget Category 1 — Task Interpretation

Command.

Qualifier.

Audience.

Purpose.

Required content.

These errors can have enormous fan-out.

They often deserve early budget priority.

English Budget Category 2 — Evidence and Inference

Does the answer match what the text actually supports?

Does the learner distinguish literal detail from inference?

Does answer scope fit the marks?

These can be corrected with short high-resolution question sets.

English Budget Category 3 — Organisation

Does each paragraph have a job?

Does the sequence build the response?

Do ideas drift?

A correction budget may fund planning drills before full essays.

English Budget Category 4 — Development

Is the learner making assertions without evidence or reasoning?

Does a story jump from event to event without causal development?

Does an argument repeat rather than deepen?

Use targeted paragraph work.

English Budget Category 5 — Language

Grammar, vocabulary and sentence control matter.

But the budget should distinguish recurrent language mechanisms from isolated stylistic imperfections.

Do not allocate hours to every awkward sentence.

English Budget Category 6 — Timing

Does planning take too long?

Does later writing deteriorate?

Does editing disappear?

Timing errors deserve section-level or full-response correction after local skills are stable.

English Scenario: Aisha’s Essay Has Forty Corrections

Aisha receives a heavily marked essay.

If every teacher annotation becomes a separate task, the week is destroyed.

The correction budget clusters the comments.

Many vocabulary comments are local.

Several grammar errors share one tense-control pattern.

But the largest mark loss comes from partial misreading of the qualifier.

The active budget funds qualifier parsing first.

Tense control gets a smaller maintenance drill.

Local word-choice suggestions do not all become active projects.

English Scenario: Ryan Writes Too Much

Ryan’s answers are accurate and sophisticated.

He loses the final section because his earlier answers are too long.

The correction budget should not add vocabulary or comprehension practice.

It funds answer-scope training.

One-mark, two-mark and extended answers are practised to appropriate length.

Then a timed section validates the change.

English Scenario: Maya’s Grammar Errors Are Real but Low-Leverage

Maya has several grammar errors, but her main score loss comes from weak evidence selection.

The temptation is to fix visible grammar because it is easy to circle.

The budget ranks by mark exposure.

Evidence selection receives active correction.

One recurring grammar pattern receives a small cue.

Minor one-off grammar slips remain logged but inactive.

The Science Error-Correction Budget

Science error correction should distinguish factual knowledge from model use.

A wrong explanation can come from a missing fact.

It can also come from an incomplete causal chain.

It can come from a variable being misunderstood.

It can come from reading a graph incorrectly.

It can come from writing a memorised model answer that does not fit the changed context.

Science Budget Category 1 — Knowledge

Definitions.

Relationships.

Processes.

Facts.

If genuinely missing, targeted learning is needed.

Science Budget Category 2 — Conceptual Model

Can the learner explain why the phenomenon happens?

Model weaknesses often have high transfer cost.

Science Budget Category 3 — Variables and Experimental Logic

Independent variable.

Dependent variable.

Controls.

Fair test.

Measurement.

Repeated errors here can spread across many experiments.

Science Budget Category 4 — Data and Representation

Axes.

Scale.

Units.

Trend.

Comparison.

Evidence.

Conclusion.

These are often highly repairable.

Science Budget Category 5 — Causal Explanation

Cause.

Mechanism.

Effect.

If the middle disappears repeatedly, this deserves a global explanation cue.

Science Budget Category 6 — Transfer

Can the learner use the model in an unfamiliar context?

Authentic varied questions should enter the budget.

Science Scenario: Mira Knows the Facts but Loses Explanation Marks

Mira’s mistake notebook is full of topic labels.

Respiration.

Heat transfer.

Forces.

Photosynthesis.

Review shows the same error across all four: she states the cause and outcome but omits the connecting mechanism.

The correction budget creates one cross-topic mechanism project.

Three fresh explanations.

One delayed return.

One timed section.

Four topic problems collapse into one reasoning correction.

Science Scenario: Graph Errors Versus Content Errors

A learner loses six marks across three graphs.

The content is known.

The budget should not allocate six new chapter reviews.

It funds graph orientation, scale, trend and evidence statements.

Then fresh graphs validate the change.

The Humanities Error-Correction Budget

Humanities learners can accumulate enormous correction lists because every essay contains factual, structural, evidential and evaluative dimensions.

Budget by argument mechanism.

Humanities Budget Category 1 — Retrieval

Can the learner access organised knowledge without notes?

Humanities Budget Category 2 — Question Interpretation

Cause?

Comparison?

Extent?

Most important factor?

Evaluation?

Misreading the task creates global loss.

Humanities Budget Category 3 — Evidence Selection

Does the evidence actually answer the question?

Strong memory can make selection harder if everything feels relevant.

Humanities Budget Category 4 — Reasoning

Does the learner explain why the evidence supports the claim?

Humanities Budget Category 5 — Evaluation

Can the learner weigh factors or evidence using a criterion?

Humanities Budget Category 6 — Timing

Does the learner overplan, overwrite or leave evaluation to the final minutes?

Humanities Scenario: Ryan’s Knowledge Is Not the Problem

Ryan gets many factual details right but scores modestly because his essays describe evidence instead of using it.

The correction budget removes one broad reading session.

It funds evidence-ranking and claim–evidence–reasoning drills.

Then one timed paragraph validates the change.

More knowledge would have been the wrong investment.

The Multiple-Choice Error-Correction Budget

Multiple-choice mistakes require careful classification because the final answer hides the reasoning.

A wrong answer can come from missing knowledge, partial truth, misreading, distractor attraction, calculation, changed answer or guessing.

A correct answer can also be fragile.

Record confidence where useful.

Multiple-Choice Budget Priority

High-confidence wrong answers deserve diagnosis.

Repeated distractor patterns deserve control rules.

Low-confidence correct answers deserve fresh validation.

Random isolated guesses may remain monitored.

The Essay Error-Correction Budget

Full essays are expensive correction objects.

Do not rewrite the whole essay every time.

Identify the first failure.

Prompt interpretation?

Thesis?

Evidence selection?

Development?

Organisation?

Language?

Timing?

Then practise the smallest unit that contains the mechanism.

A plan can fix paragraph relevance with three paragraphs rather than three full essays.

The Oral Error-Correction Budget

Oral practice generates feedback on structure, fluency, pronunciation, responsiveness, idea development, timing and recovery.

Do not activate all of them simultaneously.

Choose the bottleneck.

One session may budget structure.

Another may budget follow-up responsiveness.

Another may budget recovery after losing the thread.

Progress should be observable.

The Listening Error-Correction Budget

Listening errors can come from vocabulary, attention, preview, distractors, changed information, final-state tracking or note-taking overload.

Generic “listen more” is not a budget.

Name the mechanism and allocate accordingly.

The Open-Book Error-Correction Budget

Open-book assessments can produce navigation errors rather than knowledge errors.

The learner spends too long searching.

The budget may fund resource indexing, conceptual maps and retrieval of core structure.

Do not respond to every slow lookup by adding more notes.

The Computer-Based Error-Correction Budget

Separate academic from technical friction.

Typing.

Scrolling.

Navigation.

Submission.

Digital calculator.

Screen reading.

The correction budget should not misclassify interface problems as subject weakness.

The Practical Error-Correction Budget

Practical errors can involve procedure, setup, safety, observation, measurement, recording and timing.

Use appropriate supervised practice.

Do not try to correct practical execution only through written notes when real procedural practice is needed.

The Cross-Subject Error Budget

Some error signatures cross subjects.

Question-reading qualifiers.

Unit checking.

Evidence scope.

Overwriting.

Getting stuck.

Failure to check.

These can have enormous fan-out.

A cross-subject correction can deserve more budget than a narrow topic error.

The Cross-Subject Qualifier Problem

A learner repeatedly misses words such as except, most, best, compare, to what extent, under these conditions and justify.

This appears in Mathematics, Science, English and Humanities.

The budget should create one global question-reading control.

Underline or restate the exact demand before execution.

Then validate across subjects.

The Cross-Subject Timing Problem

A learner stays too long on difficult tasks across several papers.

The root is not subject content.

It is a control rule.

Train exit-and-return globally, then calibrate each subject’s thresholds.

The Cross-Subject Checking Problem

Some learners do not check anywhere.

Others overcheck everywhere.

The correction budget can fund a general checking architecture plus subject-specific error cues.

The Cross-Subject Retrieval Problem

If several subjects show knowledge decay, the problem may be the maintenance system rather than each topic individually.

Strengthen spaced retrieval across the portfolio.

Do not create dozens of separate chapter emergencies.

The Cross-Subject Overload Problem

If correction backlogs appear in every subject, the total plan is over capacity.

This is not an individual-error problem.

Reduce paper volume, active work-in-progress and duplicate homework.

Use capacity planning.

Correction Budget by Evidence Source

Homework Errors

Useful for local mechanisms.

Do not assume they predict examination performance perfectly.

Quiz Errors

Useful for retrieval and recent learning.

Can reveal early patterns.

Class-Test Errors

Useful for authentic time and task conditions.

Past-Paper Errors

Useful for transfer, selection and paper structure.

Mock Errors

Useful for integrated system behaviour.

Teacher Feedback

Useful for answer form, reasoning and misconceptions.

Tutor Observation

Useful for locating where the process first diverges.

Self-Observation

Useful for confidence, hesitation and strategy—if calibrated honestly.

Budget confidence should reflect evidence quality.

Not All Evidence Has Equal Weight

A repeated error under independent exam-like conditions deserves more confidence than an error made during first exposure with heavy support removed suddenly.

Use context.

The correction budget should weight evidence rather than count marks mechanically.

The One-Off Error

One-off errors should often be logged and monitored rather than activated.

Exceptions exist.

A one-off high-consequence error—such as misunderstanding a compulsory paper instruction—can still justify a strong correction.

Use consequence and recurrence together.

The Repeated Error

Repeated errors deserve increasingly strong attention because they indicate the prior correction was insufficient or absent.

But do not repeat the same repair automatically.

If an error returns, ask why the first intervention failed.

Wrong diagnosis?

Insufficient practice?

No transfer?

No delayed return?

Timing pressure?

The budget may need a different operation, not more volume.

The Resurrected Error

An error can return after months of stability.

Do not necessarily reopen the full project.

Use one small discriminating retest.

If the mechanism is still mostly stable, increase maintenance.

If broad failure has returned, reopen active correction.

The Hidden Error

Correct answers can hide fragile reasoning.

Low confidence, lucky guesses and invalid methods deserve monitoring.

Use fresh evidence before declaring the capability secure.

The Expensive Error

An error can be important and still be too expensive to repair fully now.

The budget should permit strategic containment.

Recognise standard cases.

Secure first steps.

Protect partial credit where relevant.

Set a time cap.

Move on.

The Cheap Error

Some errors need only a cue.

Check units.

Read the qualifier.

Write the mechanism.

Use new evidence before changing an answer.

Cheap high-value errors should be corrected quickly.

The High-Fan-Out Error

Prioritise.

Basic algebra.

Question interpretation.

Causal reasoning.

Evidence selection.

These can unlock multiple downstream marks.

The Low-Fan-Out Error

A rare specialised edge case may be logged but deferred.

Do not let one dramatic difficult question hijack the correction budget.

The Error That Is Actually a Scheduling Problem

Sometimes the learner knows the answer but attempts the task when exhausted, rushed or after too much switching.

If errors concentrate in low-quality study windows, the correction may be scheduling rather than content.

Move the task.

The Error That Is Actually a Capacity Problem

Repeated unfinished corrections can indicate that the plan contains more work than the learner can carry.

Do not moralise.

Reduce load.

Budget correction inside real capacity.

The Error That Is Actually a Feedback Problem

A learner may repeatedly practise a wrong method because no reliable feedback arrives.

The correction budget should fund expert feedback if self-correction is not accurate enough.

The Error That Is Actually a Resource Problem

A resource may explain poorly, provide too little transfer or use outdated formats.

Do not assume repeated failure is always the learner.

Change the resource when evidence justifies it.

The Error That Is Actually a Motivation Problem

Sometimes work is not attempted.

But “motivation” can hide overload, ambiguity or lack of next action.

Before allocating a motivational intervention, make the task specific and small.

What exactly must be corrected?

The Error That Is Actually a Confidence Problem

A learner may know the method but repeatedly abandon it from doubt.

Use confidence prediction and representative evidence.

The correction budget may fund assurance rather than content.

The Error That Is Actually a Perfectionism Problem

Overchecking, overwriting and reluctance to move can cause timing loss.

The budget should fund completion criteria and stop rules.

The Error That Is Actually an Independence Problem

The learner performs correctly only with prompts.

Remove support progressively.

The correction is not more explanation.

It is independent retrieval and execution.

The Error That Is Actually a Transfer Problem

Local worksheets are strong.

Past papers are weak.

Do not repeat local practice forever.

Fund changed-surface transfer.

The Error That Is Actually a Timing Problem

Untimed performance is stable.

Timed performance collapses.

Use the timing-entry architecture.

The budget should pay for the exact time-loss mechanism.

The Error That Is Actually a Maintenance Problem

Previously secure knowledge disappears after neglect.

Do not rebuild from zero.

Restore a small spaced maintenance floor.

The Error That Is Actually a Paper-Control Problem

The learner knows everything needed but loses marks through question order, getting stuck or no final checking.

Use full-paper controls.

Do not add content.

The Weekly Error-Correction Budget Meeting

Keep it short.

1. What new error signatures appeared?
2. Which repeated?
3. Which active correction passed its gate?
4. Which maintenance item failed?
5. Which waiting error now deserves activation?
6. Which error no longer deserves capacity?
7. How much correction backlog exists?
8. What work is being removed?

Then set the next active queue.

The Daily Error-Correction Budget

A daily correction budget can be tiny.

One active weak-link block.

One maintenance cue.

One retest.

The learner does not need to review the entire mistake notebook every day.

The Five-Minute Mistake Review

If time is short:

Open the error registry.

Select the highest-priority active signature.

Cover the solution.

State the correction rule.

Attempt one fresh item.

Record the state.

Stop.

The Fifteen-Minute Mistake Review

Use one recurring pattern.

Reconstruct the original reasoning.

Explain the correction.

Attempt two or three fresh variations.

Identify whether the error has moved from active to stabilising.

The Thirty-Minute Correction Block

Use for a higher-value mechanism.

Five minutes diagnosis.

Ten minutes repair.

Ten minutes fresh transfer.

Five minutes update and next gate.

The block has an exit.

The One-Hour Correction Block

Reserve for genuine high-leverage work.

Do not fill an hour with copying corrections.

Teach or reconstruct the model.

Practise the mechanism.

Remove support.

Test fresh.

Schedule delayed return.

The Correction Budget After a Past Paper

Do not immediately log every wrong question separately.

Classify.

Cluster.

Rank.

Activate only the high-value mechanisms.

Use How to Use Past Papers Properly for the learning loop.

The Correction Budget After a Mock

Use the full portfolio.

Which old repairs held?

Which new bottleneck appeared?

Which strong subject can release capacity?

Which errors deserve only containment now?

Mocks should rewrite the budget, not merely add to it.

The Correction Budget After Teacher Feedback

Teacher annotations can exceed correction capacity.

Ask the teacher or tutor which two comments are most consequential if priority is unclear.

Find repeated mechanisms.

Do not attempt to perfect every stylistic detail simultaneously.

The Correction Budget After Tuition

A tuition session may discover several weaknesses.

The tutor should identify which one becomes home practice.

Everything else can be logged or waiting.

A session that creates ten active projects may reduce follow-through.

The Correction Budget After School Homework

School homework can double as fresh retest or transfer.

Merge functions.

If the school set already contains the needed fresh questions, do not add duplicate correction worksheets.

The Correction Budget Across Several Subjects

Every subject can produce a mistake notebook large enough to consume the week.

Use a portfolio correction budget.

One major active correction per high-risk subject may be enough.

Stable subjects keep maintenance.

Low-repairability issues wait or contain.

The Correction Portfolio Table

Subject.

Error signature.

Mark exposure.

Recurrence.

Repair cost.

Runway.

Current state.

Next gate.

Weekly budget.

This makes trade-offs visible.

Do Not Budget by Emotion

The most embarrassing mistake can feel urgent.

The most difficult question can feel important.

The lowest subject score can feel dominant.

Use evidence.

Emotion is context, not allocation.

Do Not Budget by Teacher Ink

A heavily annotated answer may contain many low-value local comments.

A lightly annotated answer may contain one major conceptual failure.

Correction priority should not be proportional to the amount of red ink.

Do Not Budget by Topic Size

A large chapter is not automatically high priority.

A small global control can recover more marks.

Use exposure and repairability.

Do Not Budget by Fear

Students often overallocate to the subject or topic they fear most.

Fear can be poorly calibrated.

Use representative evidence.

Do Not Budget by Familiarity

Students may prefer correcting familiar easy errors because success comes quickly.

This can crowd out the actual bottleneck.

Use high-value active correction even when it is less comfortable.

Do Not Budget by Difficulty Alone

A difficult error is not necessarily a valuable error.

Hard questions can have low recurrence and high repair cost.

Budget by expected return.

Do Not Budget by Frequency Alone

A frequent one-mark slip may deserve a simple cue, not hours of practice.

Use recurrence plus consequence plus repair cost.

Do Not Budget by Marks Alone

A ten-mark difficult topic may require enormous repair.

A six-mark global timing issue may be cheap and repeat across papers.

Marks matter, but so do cost and probability.

The Correction Budget and Prioritisation

General prioritisation asks where scarce attention should go across all tasks.

The error-correction budget asks a narrower question inside the correction queue.

Which mistakes deserve how much repair?

This page is therefore a child mechanism of wider prioritisation, not a duplicate.

The Correction Budget and Capacity Planning

Capacity planning defines how much work the learner can carry.

The correction budget spends only a portion of that capacity.

Do not allocate 100% of available study time to fixing mistakes.

New learning, maintenance, retrieval, integration and recovery still need space.

The Correction Budget and Work in Progress

Too many active corrections slow progress.

Close loops.

Do not keep every error half-repaired.

The Correction Budget and Buffers

Mocks and papers can reveal new high-value errors.

Keep some uncommitted response capacity after major assessments.

Otherwise new evidence cannot change the week.

The Correction Budget and Recovery

Do not finance correction through chronic late nights.

If the queue does not fit, triage it.

The budget must respect the learner’s usable capacity.

The Correction Lifecycle

A mistake should not remain an undifferentiated item in a notebook forever.

It should move through a lifecycle.

Signal.

Classification.

Priority.

Repair.

Retest.

Transfer.

Maintenance.

Retirement.

The budget exists to control movement through this lifecycle.

Stage 1 — Preserve the Signal

Before reading the solution, reconstruct what the learner actually did.

This matters because once the correct answer is visible, the original reasoning can become difficult to remember honestly.

Write one sentence.

“I thought the percentage should be calculated from the final value.”

“I read ‘compare’ as ‘describe’.”

“I knew the formula but could not retrieve it.”

“I guessed between B and D.”

“I ran out of time before reaching the question.”

This preserves the mechanism evidence.

Stage 2 — Classify the First Divergence

Where did the answer first leave the correct path?

Before knowledge?

At retrieval?

At question interpretation?

At selection?

At execution?

At expression?

At timing or control?

Correction should usually begin at the first useful divergence rather than the final visible error.

Stage 3 — Decide Whether It Deserves Budget

Run the priority tests.

Recurrence.

Mark exposure.

Fan-out.

Repair cost.

Runway.

Diagnostic confidence.

Opportunity cost.

The result is a state: active, next, monitor, maintenance, containment or closed.

Stage 4 — Choose the Smallest Corrective Operation

Do not use a full chapter review when one cue will work.

Do not use one cue when the learner lacks the entire conceptual model.

The repair must match the mechanism.

Knowledge gap → learn or reconstruct.

Retrieval gap → closed-book retrieval and spacing.

Selection gap → mixed discrimination.

Execution gap → focused procedural practice.

Reasoning gap → explicit relationship building.

Expression gap → answer-form practice.

Transfer gap → varied fresh contexts.

Timing gap → timed sections and control.

Stage 5 — Remove Support

A repair is not ready for evidence if the learner still needs the original hint.

Remove worked examples.

Remove teacher prompts.

Remove topic labels.

Remove answer frames gradually where appropriate.

The learner should produce the performance independently.

Stage 6 — Fresh Retest

Use a new question.

Fresh evidence prevents surface memory from pretending to be learning.

The retest should target the same mechanism without copying the original problem exactly.

Stage 7 — Delayed Return

Some corrections fail after time passes.

Schedule a delayed return when retention matters.

The delay can be short or longer depending on the runway and skill.

The key is that the original correction context has faded.

Stage 8 — Transfer

Change the surface.

New numbers.

New diagram.

New essay theme.

New source.

New data representation.

New wording.

The learner should recognise the invariant structure.

Stage 9 — Timed or Integrated Validation

If the real examination adds time pressure or whole-paper interaction, the correction should eventually face those conditions.

Local success is not enough for a global control error.

Use timed sections or integrated papers where justified.

Stage 10 — Maintenance or Retirement

If the mechanism holds, reduce the budget.

Move to occasional maintenance or close it entirely.

Success should make the plan cheaper.

The Correction Loop Must Have an Exit

A correction without an exit can consume unlimited time.

Write the gate before the repair.

For example:

Three fresh independent successes.

No recurrence across two mixed contexts.

One delayed return.

One timed section.

Then maintenance.

The exact gate depends on the mechanism.

The Correction Loop Must Have a Reopening Rule

Maintenance is not permanent immunity.

Write what reactivates the issue.

Two recurrences.

One high-consequence failure.

Failure under full-paper pressure.

This creates a controlled return rather than panic.

The Error-Log Review Rhythm

A mistake notebook should be reviewed according to state, not reread from page one every week.

Active entries: frequent review and fresh practice.

Stabilising entries: delayed retest.

Maintenance entries: occasional sampling.

Closed entries: no routine review unless recurrence returns.

This prevents the log from growing into an impossible archive.

The Error-Log Retirement Rule

Retire an entry when:

• the learner can explain the corrected model;
• fresh independent tasks are successful;
• the error does not recur across changed contexts;
• timing holds if timing matters;
• the issue survives a delayed return.

Retirement is important because a mistake notebook that never shrinks eventually becomes unusable.

The Mistake Notebook Should Contain Representative Errors, Not Every Error

If ten questions share one mechanism, keep one or two representative examples.

Record the recurrence count.

Do not paste ten near-identical questions.

The notebook should preserve patterns, not paper volume.

The Mistake Notebook Should Record Corrections That Future-You Can Use

A copied model answer may be difficult to transfer.

A short rule can be more useful.

“Only cancel common factors, not terms in a sum.”

“Read the qualifier before retrieving examples.”

“State cause → mechanism → effect.”

“Change an answer only with new evidence.”

“Leave after the time threshold and return.”

These are portable corrections.

The Mistake Notebook Should Record the Next Test

Every active entry should know how it will be challenged again.

Fresh question tomorrow.

Delayed retrieval Friday.

Mixed section Sunday.

Next mock.

Without a next test, the notebook stores insight but does not prove change.

The Mistake Notebook Should Record State

Active.

Stabilising.

Maintenance.

Closed.

Containment.

This is more useful than endless highlighting.

The Error-Log Search Intent: “How Do I Stop Repeating the Same Mistakes?”

The answer is not merely “write them down.”

Repeated errors stop when the learner:

1. identifies the mechanism;
2. uses the correct repair;
3. removes support;
4. retests fresh;
5. returns after delay;
6. tests transfer;
7. uses the right cue under exam conditions.

Writing is only the first preservation step.

The Search Intent: “How Do I Review Exam Mistakes?”

Review should not mean reread.

Reconstruct the original decision.

Classify the first divergence.

Explain the correction.

Try again without the solution.

Then decide whether the error deserves more budget.

The Search Intent: “Should I Make a Mistake Notebook?”

Yes if it stays small enough to use and feeds actual correction decisions.

No notebook format is valuable by itself.

The notebook should reduce repeated mark loss, not become another study craft project.

The Search Intent: “How Often Should I Review My Error Log?”

Review frequency should depend on state and runway.

Active high-value errors may return frequently.

Maintenance errors may need occasional sampling.

Closed errors do not need constant review.

Do not use one fixed review frequency for every entry.

The Search Intent: “What Should I Write in an Error Log?”

Write only what changes future performance.

Source.

Mechanism.

Portable correction.

Priority or state.

Next test.

Do not copy long solutions unless the solution itself is the learning object.

The Search Intent: “How Do I Fix Careless Mistakes?”

First stop using “careless” as the diagnosis.

Name the exact error signature.

Then install the smallest control at the point of failure.

Qualifier check.

Sign check.

Unit check.

Final-subpart scan.

Answer-change rule.

Carefulness becomes a behaviour, not a personality request.

The Search Intent: “How Do I Learn From Mistakes in Exams?”

Treat the mistake as data about a process.

What did I think?

Where did the process diverge?

What repair matches the divergence?

How will I prove the repair holds?

Then change the plan.

The Search Intent: “What Do I Do After a Mock Exam?”

Use the mock rewrite owner to change the plan, then use the correction budget to allocate capacity among the errors the mock exposed.

The two systems work together.

Scenario 1 — Twenty Errors, Three Mechanisms

A learner gets twenty questions wrong.

The error log looks disastrous.

Clustering shows:

nine errors from weak retrieval;

seven from one question-reading problem;

four one-off difficult items.

The correction budget activates retrieval and question reading.

The four difficult items remain monitor or containment.

Twenty errors become two active projects.

Scenario 2 — One Error, Huge Consequence

A learner misreads a compulsory instruction and answers the wrong question.

It happened once.

Recurrence is low.

Consequence is enormous.

The budget still activates an instruction-reading routine because one future recurrence could be catastrophic.

Scenario 3 — Repeated One-Mark Unit Errors

Six unit errors appear across two papers.

Repair cost is tiny.

Add a final-answer unit cue.

Use several fresh questions.

Then maintain.

This is a cheap high-return correction.

Scenario 4 — One Twelve-Mark Advanced Question

The learner cannot solve one advanced problem.

The question is rare.

Repair cost is high.

Four days remain.

The correction budget may log and contain it.

Protect easier marks elsewhere.

Scenario 5 — The Same Error Returns After “Correction”

The learner reread the solution and redid the original question successfully.

A fresh question fails.

The first correction was surface-specific.

The budget now funds model explanation and varied transfer, not another repetition of the original question.

Scenario 6 — The Student Has a Beautiful Error Log but No Improvement

The notebook is colour-coded and complete.

Nothing is retested.

The budget identifies the failure: the log is a storage system, not a correction system.

Reduce logging time.

Increase fresh retesting.

Scenario 7 — The Student Keeps Recopying Correct Solutions

Copying creates familiarity.

The budget moves effort toward independent reconstruction and fresh questions.

Use the solution only to repair the model.

Scenario 8 — The Student Corrects Every Wrong Answer Immediately

Immediate correction can be useful.

But if every error receives full attention, the session may never progress.

Classify quickly.

Activate high-value errors.

Log low-value ones for later.

Scenario 9 — The Student Never Reviews Errors Until the Final Week

The correction backlog becomes enormous.

High-value mistakes repeat for months.

The budget should process errors continuously enough that the final week contains a small set of personal signatures, not an archive.

Scenario 10 — The Student Reviews the Entire Notebook Every Day

Maintenance volume explodes.

Use state-based review.

Active entries only.

Sample maintenance.

Closed entries stay closed.

Scenario 11 — The Parent Wants Every Mistake Fixed

The intention is understandable.

The plan cannot afford it.

Show the budget matrix.

Which errors expose the most marks?

Which are repeated?

Which are cheap to repair?

Which are too deep for the runway?

Use transparent trade-offs.

Scenario 12 — The Tutor Finds Five Weak Links in One Lesson

Choose the first useful weak link with the highest leverage.

Log the others.

Do not send five separate repair programmes home.

Scenario 13 — The Teacher Returns a Paper Covered in Comments

Cluster the comments.

Which are examples of the same mechanism?

Which affect marks most?

Which can be corrected independently?

Which need teacher clarification?

The budget turns annotation volume into priorities.

Scenario 14 — The Student’s Weakest Subject Has the Most Errors

Do not allocate proportionally to error count.

Some errors may be deep and expensive.

Another subject may contain cheaper recoverable marks.

Use portfolio expected return.

Scenario 15 — The Student’s Strongest Subject Has a New Global Error

A strong subject can deserve correction budget if a high-fan-out control fails.

Strength is not immunity.

Use one bounded repair and preserve the rest on maintenance.

Scenario 16 — The Student Is Already Behind

Do not turn every missed task and every error into backlog.

Rebase.

Choose the highest-value current corrections.

Drop obsolete historical errors.

Scenario 17 — The Student Is Ahead

Do not invent correction work simply because capacity exists.

Use light monitoring and ceiling work after the main error budget is stable.

Scenario 18 — The Student Has One Week Left

Compress the error registry.

Which three to five signatures are most likely to recur?

Which can still be repaired?

Which require containment?

Which should be ignored now?

The final-week budget is intentionally small.

Scenario 19 — The Student Has One Day Left

No historical clean-up.

Review the short final cue list.

One fresh item only if a narrow risk remains.

Protect sleep and performance.

Scenario 20 — The Student Has Months Left

Use the opportunity to correct deep foundations properly.

Do not postpone high-fan-out weaknesses because they are expensive.

Early runway is exactly when deep correction can pay back.

Failure Mode 1: Logging Without Budgeting

The notebook grows but priorities remain vague.

Repair: assign state and next action.

Failure Mode 2: Budgeting Every Error Equally

One-off slips receive the same attention as global weak links.

Repair: use mark exposure, recurrence, fan-out and repair cost.

Failure Mode 3: Calling Everything Careless

No mechanism exists.

Repair: name the exact signature.

Failure Mode 4: Calling Everything a Topic Gap

Selection, timing and expression are misdiagnosed as content.

Repair: classify the first divergence.

Failure Mode 5: Correcting the Final Answer Rather Than the First Weak Link

Symptoms return.

Repair: trace upstream to the earliest useful cause.

Failure Mode 6: Copying Solutions

Familiarity replaces independent performance.

Repair: reconstruct and retest fresh.

Failure Mode 7: No Fresh Retest

The learner never proves the correction generalised.

Repair: use fresh equivalent material.

Failure Mode 8: No Delayed Return

Immediate success is mistaken for retention.

Repair: schedule a later retrieval.

Failure Mode 9: No Transfer Test

The correction works only in familiar form.

Repair: change the surface.

Failure Mode 10: No Timed Validation

The correction disappears under time.

Repair: add timing if the real performance requires it.

Failure Mode 11: No Exit Gate

Active correction continues forever.

Repair: define maintenance criteria.

Failure Mode 12: No Reopening Rule

A recurrence triggers panic or is ignored.

Repair: define the evidence that reactivates the issue.

Failure Mode 13: Too Many Active Corrections

Attention fragments.

Repair: use a work-in-progress limit.

Failure Mode 14: Correction Backlog Keeps Growing

Measurement outruns repair.

Repair: pause new papers or reduce task scale.

Failure Mode 15: The Student Tries to Clear the Entire Backlog

Old low-value mistakes consume current runway.

Repair: re-rank by present value.

Failure Mode 16: The Student Never Deletes Entries

The error log becomes unusable.

Repair: retire stable signatures.

Failure Mode 17: The Student Deletes Too Early

One immediate correct attempt is treated as mastery.

Repair: use delayed and transfer gates for important errors.

Failure Mode 18: High-Confidence Wrong Answers Are Ignored

Misconceptions persist.

Repair: prioritise diagnosis.

Failure Mode 19: Low-Confidence Correct Answers Are Ignored

Fragile performance looks secure.

Repair: validate fresh.

Failure Mode 20: Every Correction Uses the Same Method

Rereading becomes the universal answer.

Repair: match intervention to mechanism.

Failure Mode 21: Every Error Gets Another Worksheet

Practice volume increases without diagnosis.

Repair: choose the smallest corrective operation.

Failure Mode 22: Every Error Gets a Full Paper

Large measurement is used for local problems.

Repair: use targeted questions and sections.

Failure Mode 23: Deep Errors Are Avoided Because They Are Expensive

Early runway is wasted.

Repair: fund high-fan-out foundations when time permits.

Failure Mode 24: Deep Errors Are Pursued Too Late

The final week becomes a foundation rebuild.

Repair: switch to containment and recoverable marks.

Failure Mode 25: Teacher Feedback Is Treated as a To-Do List

Every annotation becomes equal priority.

Repair: cluster and rank.

Failure Mode 26: Parent Anxiety Expands the Budget

More hours are added without subtraction.

Repair: keep the capacity ceiling.

Failure Mode 27: Tutor Homework Expands the Budget

Each lesson adds tasks to an already-full week.

Repair: replace lower-value work.

Failure Mode 28: School Homework and Correction Work Duplicate Each Other

Two tasks perform the same function.

Repair: merge functions.

Failure Mode 29: The Budget Ignores Strong Areas

Maintenance disappears.

Repair: preserve cheap stability.

Failure Mode 30: The Budget Ignores Recovery

Correction is financed by fatigue.

Repair: reduce active queue.

Failure Mode 31: The Budget Uses Score Instead of Mechanism

Low score gets more time regardless of why.

Repair: allocate to recoverable causes.

Failure Mode 32: The Budget Uses Topic Frequency Without Current Evidence

Generic high-frequency topics crowd out the learner’s actual bottleneck.

Repair: combine examination value with current learner state.

Failure Mode 33: The Budget Ignores Official Scope

A past paper under-samples an important area and the learner drops it incorrectly.

Repair: use the syllabus or specification as the coverage boundary.

Failure Mode 34: The Budget Treats All Past Papers as Equally Representative

Outdated or unusual material drives corrections.

Repair: weight evidence quality.

Failure Mode 35: The Budget Does Not Change After Success

Old weak links stay expensive.

Repair: release capacity.

Failure Mode 36: The Budget Does Not Change After Failure

The same ineffective intervention repeats.

Repair: revisit diagnosis or repair method.

Failure Mode 37: The Student Uses the Error Log Only Before Exams

Feedback arrives too late.

Repair: process meaningful errors continuously.

Failure Mode 38: The Student Turns the Error Log Into Notes

Entries become long summaries.

Repair: keep portable correction rules and retests.

Failure Mode 39: The Student Reviews Errors Passively

Recognition replaces retrieval.

Repair: cover the answer and reproduce.

Failure Mode 40: The Student Measures the Notebook, Not Performance

Number of entries cleared becomes the goal.

Repair: measure whether the error stops costing marks in fresh representative work.

The Parent Error-Correction Guide

Parents can help most by protecting prioritisation.

Ask:

Which mistake is repeating?

How many marks can it affect?

What is the proposed fix?

What evidence will prove the fix works?

What are we not doing to make room?

Avoid asking for every wrong question to be redone immediately.

The Tutor Error-Correction Guide

A tutor can provide diagnostic compression.

Five visible errors become one upstream mechanism.

The tutor should:

1. claim the first useful weak link;
2. repair the mechanism;
3. remove support;
4. set a fresh retest;
5. define the home budget;
6. reduce duplicate homework.

The student should leave with a smaller, clearer queue.

The Teacher Error-Correction Guide

Teacher feedback gains planning value when it identifies patterns.

“You repeatedly omit the mechanism.”

“Your method is correct but sign control fails after rearrangement.”

“Your examples are relevant to the topic but not the qualifier.”

“Your conclusion appears only after time has nearly expired.”

Pattern feedback helps students budget correction.

The Student Error-Correction Guide

A mature learner should be able to say:

I have twelve logged mistakes, but only two active correction projects. One is a high-fan-out sign-control error worth several recoverable marks. The other is a timing problem that affects the final section. Four one-off errors are monitored, three old errors are maintenance, and the advanced geometry gap is in containment until after the nearer examination.

That is correction ownership.

The One-Page Error-Correction Budget

Error signature: what keeps happening?

Source: where did it appear?

Mechanism: what failed first?

Recurrence: how often?

Mark exposure: how many marks can it affect?

Fan-out: what else depends on it?

Repair cost: what will full correction require?

Runway: how much time remains?

State: active, next, monitor, maintenance, containment or closed?

Budget: how much time or feedback this week?

Next gate: what proves progress?

Stop rule: when does the active budget end?

The Five-Minute Error-Budget Audit

1. Which error costs the most repeated marks?
2. Which error has the highest fan-out?
3. Which error is cheapest to repair?
4. Which diagnosis is uncertain?
5. Which active correction can now move to maintenance?
6. Which old error should be dropped or contained?
7. What work leaves to fund the active queue?

The Fifteen-Minute Error-Budget Audit

Group the week’s new errors.

Update recurrence.

Estimate mark exposure.

Review active-state gates.

Reallocate capacity.

Set one fresh validation task.

Stop.

The Monthly Error-Budget Audit

For longer preparation:

Which error signatures disappeared?

Which keep reopening?

Which interventions are failing?

Which maintenance floors are too low?

Which resources produce useful correction?

Which subjects consume disproportionate correction capacity?

Does the portfolio need structural change?

The Final-Week Error Card

By the final week, compress the entire error registry into a small card.

Only the most recurrent high-value signatures remain.

For example:

• read the qualifier;
• check signs after brackets;
• state the mechanism;
• evidence must answer the claim;
• leave after the threshold;
• check units;
• do not change without new evidence.

This is the final return of months of correction work.

The Independence Test

A learner owns the error-correction budget when they can answer:

• Which mistakes are only signals?
• Which are active correction projects?
• Which errors repeat?
• Which have the highest mark exposure?
• Which have high fan-out?
• Which are cheap to repair?
• Which are too deep for the current runway?
• What work leaves when a new correction enters?
• What fresh task proves the correction worked?
• When does the correction move to maintenance?
• What evidence reopens it?
• How large is my current correction backlog?

This is different from having a notebook full of red crosses.

Frequently Asked Questions

What is an exam error log?

An exam error log is a record of meaningful mistakes, their causes, corrections and future review actions. It becomes more useful when entries are clustered by recurring mechanisms and linked to fresh retesting rather than stored as isolated wrong questions.

What is a mistake notebook?

A mistake notebook is a paper or digital record of questions, error patterns and corrections that a learner wants to revisit. A good notebook stays concise, records why the mistake happened and supports reattempts after a gap.

What is an error-correction budget?

An error-correction budget is the planning layer that decides how much scarce time, attention, feedback and retesting each discovered error deserves relative to every other weakness and study demand.

Do I need both an error log and an error-correction budget?

They serve different jobs. The log stores signals and patterns. The budget allocates repair capacity. A simple system can combine both in one table by adding priority, state and next-gate fields.

Should I write down every mistake?

No. Record representative errors and recurring mechanisms. Ten nearly identical mistakes can often become one entry with a recurrence count. Keep the system small enough to use.

Should I record correct guesses?

Yes when the reasoning is fragile or confidence is low. Correctness alone does not prove stable knowledge or method selection.

Should I record careless mistakes?

Yes, but replace “careless” with the exact mechanism: qualifier missed, sign lost, unit omitted, copied value wrong, answer changed without evidence, or another identifiable pattern.

How do I stop making the same mistakes?

Identify the recurring mechanism, use a repair matched to the cause, retest on fresh material, return after a delay, test transfer and use an exam-time cue where relevant. Simply rereading the old correction is often insufficient.

How often should I review my mistake notebook?

Review according to state. Active high-value errors need frequent work. Stabilising errors need delayed retests. Maintenance errors need occasional sampling. Closed errors do not need routine review unless they recur.

How many errors should be active at once?

There is no universal number, but a small work-in-progress limit is usually more effective than many simultaneous deep repairs. One or two major active corrections plus small cues may be enough for many learners.

How do I know which mistake to fix first?

Compare mark exposure, recurrence, fan-out, repair cost, remaining runway, diagnostic confidence and opportunity cost. High-exposure recurring errors with clear causes and affordable repairs are strong candidates.

Should I fix the highest-mark question first?

Not automatically. A high-mark error may be rare and expensive to repair, while a smaller global error may recur across many future questions. Use expected recovery, not question size alone.

Should I fix easy mistakes first?

Cheap high-value corrections are attractive, but do not avoid a major bottleneck merely because it is harder. Balance leverage and cost.

What if I do not know why I made the mistake?

Budget diagnosis first. Use a small discriminating question, reconstruct your reasoning, ask a teacher or tutor, or compare similar examples. Do not commit large repair time to an uncertain cause.

How soon should I correct mistakes after a test?

Soon enough that your reasoning is still reconstructable and enough runway remains to act. A short break after an emotional or exhausting assessment can be useful, but leaving errors unprocessed for weeks reduces their planning value.

Should I redo the same question?

Yes as one step, but do not stop there for important errors. Use fresh equivalent questions to prove the mechanism transferred rather than relying on memory of the original solution.

How many times should I redo a mistake?

Use evidence rather than a fixed count. Continue until the learner can perform independently on fresh material, after a delay and under relevant conditions. Then reduce to maintenance.

Should I use spaced repetition for mistakes?

Spacing can be useful, especially for retrieval-related errors and important recurring knowledge. But not every correction needs flashcards or the same review interval. Match review to mechanism.

Should I review mistakes before an exam?

Yes, but the final review should be compressed to the most recurrent, high-value signatures. Do not reread an enormous historical archive immediately before the paper.

What if my error log is too long?

Cluster duplicate mechanisms, retire stable entries, archive old low-value errors and keep only active, waiting and maintenance items visible. The log should shrink as learning succeeds.

What if I keep making new mistakes faster than I fix old ones?

You have correction backlog. Reduce measurement volume, limit active work-in-progress and prioritise the highest-leverage mechanisms before doing more full papers.

Should I do another past paper if I have uncorrected errors?

Not automatically. If the existing paper already revealed a clear active weak link, repair it first. Another paper is useful when it answers a new important question or validates change.

Should I spend all revision time fixing mistakes?

No. New learning, retrieval maintenance, transfer, full-paper integration and recovery still need capacity. The correction budget is one part of the whole preparation portfolio.

How should parents use an error log?

Parents can help ask which patterns repeat, what the proposed correction is, and what evidence will prove it worked. Avoid using the log as a list of everything the child did wrong.

How should tutors use a mistake notebook?

Tutors can use it as a diagnostic registry, but should cluster errors, find the first useful weak link, choose a small repair and define the fresh retest. The notebook should guide teaching rather than become homework volume.

How should teachers use error patterns?

Specific pattern feedback helps students act. Naming repeated mechanisms such as missing causal links, misread qualifiers or execution errors gives students a better basis for correction than broad comments alone.

What if an error comes back after I thought it was fixed?

Reopen it according to your criteria. Determine whether the failure was retention, transfer, timing or the original diagnosis. Use a different intervention if the first repair only worked superficially.

When should I stop correcting an error?

When the mechanism is stable enough under fresh, delayed and relevant transfer conditions to move to maintenance, and the next unit of correction has lower value than other work.

When should I abandon an error?

Full repair may be strategically abandoned or contained when mark exposure is low, repair cost is high, runway is short or higher-value work competes for the same capacity. Record a containment rule if the weakness can still affect the paper.

Does this work for university and professional exams?

Yes. The exact error categories differ, but the allocation logic transfers: preserve the signal, diagnose the mechanism, rank by expected value, repair selectively and validate under representative conditions.

Does this apply outside Singapore?

Yes. Use the official syllabus, specification, paper format and marking rules for the relevant examination. The correction-budget architecture is location-neutral.

The Rule to Keep

Mistakes are valuable.

But value is not created by collecting them.

Value appears when the right mistakes change future performance.

Record enough to preserve the signal.

Classify the first divergence.

Cluster repeated mechanisms.

Measure mark exposure.

Estimate repair cost.

Respect the remaining runway.

Choose a small active queue.

Fund the repair by removing lower-value work.

Retest fresh.

Return after delay.

Test transfer.

Test timing where timing matters.

Move stable work to maintenance.

Retire what stays fixed.

Contain what cannot be rebuilt economically.

Do not discover errors faster than you can use them.

Do not let a mistake notebook become an archive of guilt.

Do not let every red mark become an active project.

Do not finance correction through sleep or overload.

And do not waste the mistakes that reveal the first weak link.

A strong examination plan does not promise to fix every mistake. It builds a correction budget that spends scarce preparation capacity where an error is most likely to become a recovered mark.

Continue Through the Planning and Examination System

• How to Plan Properly | Use Mock Results to Rewrite the Plan — how integrated evidence changes Plan Version Next.
• How to Use Past Papers Properly | Attempt, Mark, Diagnose, Repair, Reattempt — the learning loop after a paper.
• How Past Papers Fail | Why Doing More Papers Can Leave the Same Weaknesses Untouched — why measurement can outrun correction.
• How to Plan Properly | Plan Around the First Weak Link — how diagnosis changes the operating week.
• How to Plan Properly | Plan by Marks, Not by Chapters — mark-aware allocation.
• How to Plan Properly | Stop Adding Work and Decide What to Drop — subtraction from overloaded plans.
• How Capacity Planning Works | You Cannot Schedule 120% of a Student — the hard ceiling on active work.
• How to Plan Properly | Decide When Past Papers Enter the Plan — when authentic examination evidence should enter.
• How to Plan Properly | Decide When Timed Practice Begins — when the clock becomes a useful training condition.
• How to Plan Properly | The 30-Day Runway to Examination Day
• How to Plan Properly | The Final 14 Days
• How to Plan Properly | The Final 7 Days
• Examinations & Assessment Hub
• Study & Learning Methods Hub

---

eduKateSG — a mistake becomes useful when the plan knows whether to fix it now, monitor it, maintain it, contain it or let it go.

The Correction Budget Is a Queueing Problem

Once a learner begins doing serious practice, errors arrive continuously.

Homework creates some.

Past papers create more.

Mocks create many at once.

Teacher feedback adds another layer.

Tuition can expose hidden weaknesses that were not visible on paper.

The correction system therefore behaves like a queue.

If arrival rate exceeds processing rate for too long, backlog grows.

The answer is not simply to process faster.

The learner can also reduce arrivals, reduce task size, improve classification and stop admitting low-value errors into active work.

Four Ways to Control the Queue

1. Reduce Error Arrival

Pause unnecessary full papers.

Use smaller targeted checks.

Do not produce new evidence faster than the plan can use it.

2. Increase Processing Efficiency

Cluster duplicate mechanisms.

Use portable correction rules.

Merge school work with retesting.

Use expert feedback where diagnosis would otherwise take too long.

3. Tighten Admission

Not every error enters active correction.

Use priority gates.

4. Retire Aggressively After Evidence

Once a correction holds, release it.

Do not keep successful items active because they used to be weak.

The Correction Queue Has Service Levels

Not every item needs the same speed.

Immediate: high-consequence global errors, missing required instructions, high-fan-out misconceptions.

This week: important recurring errors with manageable repair cost.

Monitor: uncertain or one-off errors.

Long-term: deep foundational issues with enough runway.

Containment: important but uneconomical before the examination.

Service levels prevent every red mark from becoming an emergency.

The Correction Budget Is Also an Investment Portfolio

Each repair has an expected return and a risk.

A cheap unit-check cue has small cost and reliable return.

A deep conceptual rebuild has high cost but potentially high fan-out.

A speculative correction based on one ambiguous error has uncertain return.

A full-paper timing repair can protect many marks but needs realistic validation.

The learner should hold a portfolio of corrections rather than betting everything on one visible problem.

Correction Return

Return is not merely “I got the original question right.”

Return means future mark loss is reduced.

A correction earns return when it:

• reduces recurrence;
• improves transfer;
• protects downstream tasks;
• reduces time loss;
• releases other capacity;
• stabilises confidence and control.

The best corrections often do several at once.

Correction Risk

Every correction project has failure risk.

The diagnosis may be wrong.

The repair may be too narrow.

The learner may perform only with support.

The correction may not survive time pressure.

The skill may decay.

High-risk corrections need stronger validation.

The Correction Payback Period

How long before the repair produces useful examination performance?

One cue can pay back today.

A retrieval routine may pay back over several returns.

A reading habit may take weeks.

A deep algebra foundation may take longer but unlock many topics.

Payback period matters because the examination date is fixed.

The Correction Half-Life

Some repairs decay quickly if not revisited.

New vocabulary.

Formula retrieval.

Fine distinctions.

Other repairs become durable habits.

Question-reading routines.

Checking rules.

Representation habits.

The maintenance budget should reflect the likely half-life.

The Correction Portfolio Should Contain Different Horizons

Short horizon: cheap execution cues.

Medium horizon: retrieval, selection and answer-form repair.

Long horizon: foundations, language depth, reading maturity.

A student months from the examination can fund all three.

A student seven days away should have mostly short-horizon corrections active.

The Budget Should Distinguish Prevention From Repair

Repair fixes a known error.

Prevention reduces the probability of future error.

For example, correcting one unit mistake is repair.

Installing a final-answer unit check is prevention.

Strong correction budgets look for preventive controls with high fan-out.

Preventive Controls

Read the exact qualifier.

Check sign after a bracket transformation.

State the mechanism.

Use evidence before inference.

Leave difficult questions at the threshold.

Scan for unanswered subparts.

Change answers only with new evidence.

These are low-cost preventive investments.

The Budget Should Distinguish Correction From Compensation

Sometimes the learner cannot fully repair a weakness before the examination.

Compensation reduces damage.

A deep geometry weakness may be compensated by recognising standard cases, writing known relationships and leaving advanced variants after a time cap.

A vocabulary weakness may be compensated by choosing precise familiar language rather than risky decorative words.

Compensation belongs to containment.

The Budget Should Distinguish Symptom Control From Cause Repair

Symptom control can be rational near the examination.

A cue can prevent an error even when the deeper habit is not fully rebuilt.

Long runway favours cause repair.

Short runway may favour symptom control if it reliably protects marks.

The plan should know which it is doing.

The Budget Should Distinguish Local From Systemic Corrections

Local correction affects one question family.

Systemic correction affects many parts of preparation or performance.

Question reading.

Retrieval maintenance.

Exit rules.

Checking.

Systemic corrections often deserve priority because of fan-out.

The Budget Should Distinguish Stable From Context-Sensitive Errors

Some errors happen everywhere.

Others happen only under time, fatigue, unfamiliar representation or specific question formats.

Correction conditions should reproduce the context that triggers the error.

Do not declare a timing-sensitive error fixed using only untimed practice.

The Budget Should Distinguish Individual From Environmental Errors

A noisy environment, wrong calculator, poor digital interface or unclear printing can affect performance.

Do not turn every environmental issue into a learner weakness.

Correct the environment where possible.

The Budget Should Distinguish One Bad Day From a Pattern

Illness, poor sleep or unusual stress can create noisy errors.

Record context.

Seek repetition before expensive structural change unless consequence is high.

The Budget Should Distinguish Practice Failure From Exam Failure

An error in low-stakes learning can be desirable information.

An error in the real examination costs marks.

Practice should allow enough difficulty to expose weakness while leaving time to repair it.

Do not make practice so safe that important errors first appear on exam day.

Correction Budget and Productive Errors

Not every practice error is undesirable.

Retrieval attempts that generate errors can reveal what is not yet accessible.

Challenging transfer can expose hidden assumptions.

The correction budget should not aim for errorless practice.

It aims for useful errors followed by effective feedback and correction.

Correction Budget and Feedback Quality

Feedback can be:

right/wrong;

correct answer;

worked solution;

explanation;

diagnosis;

next action.

Higher-value errors may deserve richer feedback.

A one-off unit slip may need a cue.

A misconception may need explanation and contrastive examples.

Budget feedback depth.

Contrastive Correction

Some errors improve when the learner compares the wrong and correct models directly.

Why is this method tempting?

What condition makes it wrong?

What cue distinguishes the cases?

Contrastive work is especially useful for selection errors and misconceptions.

Minimal-Pair Correction

Use two nearly identical questions where one requires Method A and the other Method B.

Ask what changed.

This can repair discrimination efficiently.

The correction budget funds information-rich examples rather than raw volume.

Counterexample Correction

A misconception can be challenged with a case where the learner’s rule fails.

Then reconstruct the more accurate rule.

Use carefully chosen counterexamples for conceptual errors.

Self-Explanation Correction

Ask the learner to explain why the correct answer works and why the original route failed.

This can reveal whether the correction is genuinely understood.

Teach-Back Correction

Ask the learner to teach the corrected mechanism in simple language.

If the explanation collapses, understanding may still be fragile.

Teach-back is useful for high-value conceptual repairs.

Worked-Example Fading

For difficult procedures, begin with a worked example, then remove steps gradually.

Do not jump directly from explanation to full independent performance if the cognitive load is too high.

The budget can include scaffold fading.

Retrieval-Based Correction

For memory errors, do not solve them with more passive reading.

Retrieve.

Check.

Space.

Return.

Memory correction needs memory operations.

Interleaved Correction

For selection errors, mix related problem types.

The learner must choose.

Blocked repetition can hide the decision.

Transfer-Based Correction

For surface dependence, vary the context.

Keep the underlying relationship constant.

Ask the learner to name the invariant.

Timed Correction

For errors that return under pressure, add time gradually.

Use the timing progression rather than jumping immediately to maximal simulation.

Integrated Correction

For whole-paper control, only integrated tasks can provide final evidence.

Exit rules, question order and late-paper accuracy need sections or full papers.

The Error-Correction Budget Decision Tree

Start with an error.

Did the learner have the required knowledge?

If no, teach or reconstruct.

If yes, could they retrieve it?

If no, retrieval repair.

If yes, did they interpret the task correctly?

If no, question-reading repair.

If yes, did they choose the right method or evidence?

If no, selection repair.

If yes, did execution succeed?

If no, procedural repair.

If yes, did the answer express the required reasoning?

If no, expression or reasoning repair.

If all of these work untimed, did time or control break performance?

If yes, timing or paper-control repair.

Then rank whether the repair deserves budget.

The Correction Budget Red–Amber–Green System

Red: repeated, consequential, clear mechanism, repairable. Active.

Amber: uncertain, moderate consequence, needs another sample. Monitor or next.

Green: stable after fresh and relevant evidence. Maintenance or closed.

Containment can sit outside the colour system: known red weakness whose full repair is not economical now.

The Correction Budget Stoplight Is Mechanism-Specific

Mathematics can be green on knowledge and red on timing.

English can be green on language and amber on relevance.

Science can be green on content and red on data interpretation.

Do not label an entire subject one colour.

The Correction Budget Dashboard

Keep a small dashboard.

Active corrections.

Waiting corrections.

Correction backlog.

Maintenance failures.

High-fan-out signatures.

Next gates.

The dashboard should help choose, not become another administrative project.

The Correction Budget Burn-Down

As examination day approaches, active correction count should generally shrink.

Deep projects close or contain.

Maintenance becomes cheaper.

Final error cues compress.

If the active correction count is still expanding in the final days, the plan may be discovering faster than it can repair.

The Correction Budget Burn-Up Warning

Warning signs:

every new paper adds several active projects;

nothing moves to maintenance;

the error log grows every day;

sleep is shrinking;

correction blocks are replacing all retrieval and integration;

the learner cannot name the top two priorities.

Stop and rebase.

The Error-Backlog Reset

If the backlog is already huge:

1. archive everything temporarily;
2. pull only errors from the most recent representative work;
3. cluster recurring mechanisms;
4. identify high-fan-out historical signatures that still recur;
5. activate a small queue;
6. leave the rest archived unless future evidence reopens it.

You do not need to clear every historical mistake to recover control.

The Error-Log Reset

If the notebook has become too complicated, rebuild it with five columns.

Error signature.

Mechanism.

State.

Correction rule.

Next test.

Everything else is optional.

The Parent Reset

If parents have been using mistakes as a pressure list, change the conversation.

Which two errors matter most?

What evidence supports that?

What is the repair?

What can be ignored for now?

This protects the child from carrying the whole archive emotionally.

The Tutor Reset

If tuition sessions are overloaded with corrections, choose one mechanism per major lesson block.

Repair deeply.

Retest.

Assign a precise home return.

Do not race through every red cross.

The Teacher Reset

If feedback volume is high, identify priority codes.

For example:

P1 = high-impact recurring.

P2 = important but local.

P3 = polish or optional refinement.

The exact system can vary.

The principle is to help students see consequence.

The Student Reset

Say:

I do not need to fix my history. I need to fix the highest-value mechanisms that are still costing marks now.

This sentence can rescue an overloaded revision plan.

Additional Scenarios

Scenario 21 — The Student Has 100 Error-Log Entries

Do not start at entry 1.

Filter by current recurrence and mark exposure.

Archive the rest.

Scenario 22 — The Student Has No Error Log at All

Start with the most recent representative paper.

Build only the first few meaningful signatures.

Do not reconstruct months of history.

Scenario 23 — The Student Logs Only Wrong Questions

Add high-confidence wrong, low-confidence correct and correct-for-wrong-reason cases when they reveal fragile mechanisms.

Scenario 24 — The Student Logs Everything and Feels Worse

The system is storing evidence without prioritisation.

Hide closed and low-value entries.

Keep only the active queue visible.

Scenario 25 — The Student Repeats the Same Error in Different Subjects

Look for a global control.

Qualifier reading.

Checking.

Time allocation.

One systemic repair may serve several subjects.

Scenario 26 — The Student Corrects Perfectly at Home but Fails in Mocks

The missing condition may be transfer, timing, independence or fatigue.

Escalate validation rather than repeating home correction.

Scenario 27 — The Student Improves in Mocks but the Notebook Still Looks Terrible

The notebook is stale.

Retire entries.

Let performance update the registry.

Scenario 28 — The Student Has One Recurring Error Worth Only One Mark

If repair cost is seconds and recurrence is high, fix it.

Cheap recurring marks still have value.

Scenario 29 — The Student Has One Error Worth Many Marks but It Is Very Rare

Use probability and repair cost.

One difficult edge case should not automatically outrank global routine losses.

Scenario 30 — The Student Keeps Getting the Same Topic Wrong for Different Reasons

Do not use the topic label as the budget unit.

Separate knowledge, selection, execution and timing mechanisms.

Scenario 31 — The Student Gets Different Topics Wrong for the Same Reason

This is exactly where clustering creates leverage.

One mechanism.

One active correction project.

Scenario 32 — The Student Corrects Errors but Cannot Explain Them

Use teach-back.

If the learner cannot explain why the original method failed, the correction may be superficial.

Scenario 33 — The Student Can Explain but Cannot Execute

Move from conceptual explanation to procedural fluency.

The budget changes operation.

Scenario 34 — The Student Can Execute but Cannot Select

Use mixed discrimination.

Stop giving topic labels.

Scenario 35 — The Student Can Select but Cannot Finish in Time

Use target pace, sections and control rules.

The correction has moved downstream.

Scenario 36 — The Student Can Finish but Cannot Check

Create checking reserve through earlier efficiency and exit rules.

Scenario 37 — The Student Checks but Does Not Know What to Check

Build a personal error-signature checklist.

Checking becomes targeted.

Scenario 38 — The Student’s Error Rate Rises When Tired

Use capacity and recovery evidence.

Do not treat every late-night mistake as a new content weakness.

Scenario 39 — The Student Makes Errors Only in One Teacher’s Papers

Investigate format, wording, difficulty and alignment.

The evidence may reveal transfer or representativeness issues.

Scenario 40 — The Student Makes Errors Only in Past Papers

Local classroom support may be masking selection and transfer weakness.

Budget authentic exam-form practice.

The Correction Budget Quality Checklist

• Every active error has a named mechanism.
• Duplicate errors are clustered.
• Priority reflects mark exposure and recurrence.
• Repair cost is visible.
• Runway affects the decision.
• New work has a funding source.
• Every repair has a fresh test.
• Important repairs face delay and transfer.
• Timing is tested where relevant.
• Stable errors move to maintenance.
• Old entries can retire.
• The active queue stays small enough to execute.

The Correction Budget Smell Test

Something is wrong if:

the notebook only grows;

every error is active;

nothing is dropped;

every correction uses rereading;

full papers continue despite correction backlog;

parents and tutors keep adding work;

sleep funds the correction queue;

the learner cannot say what proves a correction is complete.

The Correction Budget Green Flags

Good signs:

several wrong answers collapse into one mechanism;

active work becomes more specific;

the notebook shrinks through retirement;

fresh retests replace passive review;

stable subjects require less capacity;

mock results change the queue;

the final week contains only a short list of high-value cues.

The Correction Budget for High Performers

High performers may have fewer errors but higher perfectionism.

Do not spend excessive time polishing low-consequence imperfections.

Prioritise global controls, difficult transfer, precision and ceiling opportunities only after the floor is stable.

The Correction Budget for Borderline Pass Students

Protect standard accessible marks.

Prioritise high-frequency knowledge, core procedures, question reading and time control.

Contain difficult edge cases.

The Correction Budget for Students Chasing a Top Grade

Routine mark leakage becomes expensive because the target margin is small.

Fix global execution and checking errors before spending most capacity on exotic difficult questions.

The Correction Budget for Younger Learners

Keep the system simple.

One mistake.

Why?

Better rule.

Try again.

Adults can handle the deeper allocation logic while gradually transferring ownership.

The Correction Budget for Secondary Learners

Students can begin managing states and recurrence themselves.

Teach them to distinguish topic from mechanism and active from logged.

The Correction Budget for Senior Learners

Older students can manage portfolio trade-offs, correction backlogs, feedback capacity and simulation timing more independently.

The Correction Budget for University Learners

Use assessment objectives, marking criteria and course-specific expectations.

The same allocation logic applies even when errors involve complex proofs, essays, cases or laboratory work.

The Correction Budget for Professional Examinations

Question banks can generate huge error volumes.

Cluster mechanisms, use confidence, prioritise recurring high-yield weaknesses and preserve fresh validation.

Follow the current official examination format and rules.

The Correction Budget for Open-Response Subjects

Feedback capacity becomes especially important.

One well-marked paragraph can sometimes create more learning than three unreviewed essays.

Budget tasks according to available feedback quality.

The Correction Budget for Objective Tests

Confidence annotation and distractor classification can increase diagnostic resolution.

Do not rely on right/wrong alone.

The Correction Budget and Digital Tools

Digital systems can help tag, sort and schedule errors.

Automation can reduce clerical load.

But the core decisions remain conceptual:

What failed?

What matters?

What is worth fixing now?

What proves it is fixed?

A better app cannot compensate for a poor diagnosis.

The Correction Budget and AI-Generated Practice

Fresh practice can be useful when aligned carefully.

Generated questions should match the mechanism, level and examination demand.

Official materials remain important for format and marking validity.

Do not create endless questions simply because generation is cheap.

Correction capacity is still scarce.

The Correction Budget and Privacy

Student error logs can contain personal performance data.

Use appropriate privacy practices when storing or sharing them digitally.

The correction architecture does not require public disclosure of marks or weaknesses.

The Correction Budget and Motivation

A shrinking active queue can be motivating because the learner sees closure.

Show movement.

Active → stabilising → maintenance → closed.

Do not present the error log as an ever-growing record of failure.

The Correction Budget and Identity

Use mechanism language.

“I lose signs after brackets.”

Not “I am careless.”

“I misread qualifiers under time.”

Not “I am bad at English.”

Mechanisms can be changed.

The Correction Budget and Independence

The long-term goal is for the learner to own the allocation.

Which error matters?

Why?

What is the smallest repair?

What will prove it?

When do I stop?

This is planning maturity.

The Final Error-Correction Contract

I will not treat every wrong answer as an equal obligation. I will preserve useful error evidence, identify the first meaningful mechanism, rank it by recurrence, mark exposure, fan-out, repairability and runway, spend correction capacity only where the expected return justifies it, and require fresh evidence before I call the error fixed.

This contract turns mistake review into resource allocation.

The Final Canonical Boundary

Use an error log to remember mistakes.

Use a mistake notebook to store representative patterns and portable corrections.

Use exam wrappers to reflect on preparation and error patterns.

Use past-paper correction loops to repair specific attempts.

Use mock-result replanning to rewrite the broader preparation system.

Use the error-correction budget to decide which errors deserve how much scarce repair capacity inside that system.

That is this page’s job.

The Final Rule

A strong learner will still make mistakes.

A strong plan will still discover new ones.

The objective is not to eliminate error from practice.

The objective is to stop wasting error.

Capture the signal.

Find the mechanism.

Cluster recurrence.

Measure exposure.

Respect cost.

Respect runway.

Fund a small active queue.

Remove lower-value work.

Repair.

Retest.

Transfer.

Time it where necessary.

Release the capacity when it holds.

Then let the next error compete for the budget rather than entering automatically.

The best mistake notebook is not the one with the most entries. It is the one whose important entries disappear from future papers because the plan spent correction capacity well.

Continue Through the Planning System

• How to Plan Properly | Use Mock Results to Rewrite the Plan
• How to Use Past Papers Properly | Attempt, Mark, Diagnose, Repair, Reattempt
• How Past Papers Fail | Why Doing More Papers Can Leave the Same Weaknesses Untouched
• How to Plan Properly | Plan Around the First Weak Link
• How to Plan Properly | Plan by Marks, Not by Chapters
• How to Plan Properly | Stop Adding Work and Decide What to Drop
• How Capacity Planning Works | You Cannot Schedule 120% of a Student
• How to Plan Properly | Decide When Past Papers Enter the Plan
• How to Plan Properly | Decide When Timed Practice Begins
• Examinations & Assessment Hub
• Study & Learning Methods Hub

---

eduKateSG — mistakes are evidence. Correction capacity is scarce. Planning decides which evidence earns the repair.