The Tutor Handbook Vol No.0057 | The Recovery Check

The Tutor Handbook · Volume 0057 · Series ID THB-0057

The rollback is complete.

The learner is back on the earlier route.

Now what?

Alicia has returned from a seven-step checking routine to the simpler three-check routine that once protected signs, units and copied values without destroying paper completion.

Beatrice has stepped back from a rigid paragraph frame to the underlying planning questions that previously gave her structure without making every paragraph look identical.

Denise has reduced full-paper timing pressure to the moderate timed sections under which her Additional Mathematics accuracy was last reliable.

Emily has abandoned the study dashboard that consumed too much attention and returned to the rolling three-day plan she could actually operate.

Faith has regained one light prompt after an over-aggressive fade made unfamiliar tasks collapse into silence.

The route looks familiar again.

That is not yet proof that recovery happened.

A Recovery Check is the tutor’s deliberate verification that a rollback has actually restored the learner’s critical stable functions—not merely restored the old routine—before the tutor begins rebuilding forward again.

The distinction matters.

A routine can be restored while the capability remains degraded.

A timetable can look normal while backlog continues growing.

A learner can receive the old prompt level while still depending on more tutor control than before.

A paper score can rebound because the next paper is easier.

A familiar worksheet can look clean while the changed-condition failure remains untouched.

The previous volume, The Tutor Handbook Vol No.0056 | The Rollback, asked how a tutor returns to the most recent verified stable route without erasing what the failed change taught.

This volume owns the next question:

How do we know the learner has actually recovered enough to make the next forward change meaningful?

Quick Read

Rollback restores a route. The Recovery Check verifies whether the learner’s important functions returned with it.
Do not confuse familiar routine with verified recovery.
Check the exact function that failed: accuracy, initiation, method selection, workload control, transfer, timing, retrieval or learner ownership.
Use the pre-change stable state as the main reference when it is still valid.
Recovery needs at least one representative task, not only easy or familiar work.
Immediate rebound is useful but can be fragile; delayed return strengthens the claim.
Support conditions must stay attached to the evidence.
Recovery is stronger when the learner can perform without the hidden rescue that made the failed route look better than it was.
Check both target restoration and collateral restoration: did the original capability return without leaving a new cost behind?
A score can recover while the mechanism remains weak; inspect the process beneath the total.
One good lesson rarely proves recovery from a multi-week instability.
The Recovery Check should be proportionate to the consequence of moving forward again.
Do not retest every stable capability. Verify where the failed change could plausibly have caused damage.
A rollback can reveal that the old “known-good” state is no longer good enough for the learner’s current demand.
If the old state no longer verifies, reopen diagnosis rather than walking backward indefinitely.
Learning, studying, training, education and improvement require different recovery evidence.
Class 0 to Class 6 tutor functions should verify different control points.
Repair, Alignment and Frontier tuition have different definitions of recovered operation.
Three-student tutorials provide useful recovery evidence because tutor attention naturally moves and unsupported behaviour becomes visible.
The final question is not “Are we back to the old routine?” It is “Are the functions that made the old route trustworthy actually working again under conditions that matter?”

1. What This Volume Owns

This volume owns tutor-side verification after rollback.

The broader systems owner is How Verification Works | Did the Repair Actually Work? on eduKateSG. That article owns the general question of whether a claimed state has actually been reached.

The Tutor Handbook narrows that logic to one practical moment:

A tutor changed a learning route, the newer route became unstable, the learner was rolled back, and now the tutor must decide whether the restored state is real enough to support the next decision.

That is the Recovery Check.

2. What This Volume Does Not Own

The Rollback owns the return to the last verified stable route.

How Recovery Planning Works | Know the Return Path Before You Need It owns the broader route from disruption back to normal operation.

How Validation Works | Are We Solving the Right Problem? owns a different question: whether the restored or repaired state is sufficient for the downstream purpose.

The Stability Window owns how long a new route should be held before judgement.

The Recovery Check is narrower still:

Did the rollback restore the specific capabilities, controls and operating conditions that justified trusting the earlier route?

3. Recovery Is a Claim

“She is back to normal” is a claim.

“The timetable is working again” is a claim.

“The prompts can fade again” is a claim.

“We can restart full-paper timing” is a claim.

Every one of those statements can cause a new forward change.

Therefore each one should be proportionate to evidence.

The Learning Claim already established the principle:

Do not let the language become stronger than the evidence.

The Recovery Check applies that rule to restored states.

4. The Old Routine Is Not the Recovery Target

The tutor can restore the exact old worksheet order, prompt level, study schedule or timing condition.

But the learning target is not the routine.

The target is the capability the routine used to support.

A checking routine exists to support reliable error detection.
A prompt ladder exists to support increasingly independent initiation and reasoning.
A study plan exists to support priority control, completion and recovery.
Moderate timed work exists to support speed without destabilising accuracy.
Blocked practice may exist temporarily to stabilise execution before selection returns.

Recovery therefore asks whether the function returned, not whether the artifact returned.

5. The Three Recovery Questions

Restoration: Did the failed or degraded function return?
Stability: Does it survive more than one favourable moment?
Readiness: Is the restored state strong enough to support the next forward move?

These are related but different.

A learner may restore accuracy immediately but not yet show stable accuracy after delay.

A study week may stabilise again but still not be ready for another new layer of complexity.

Recovery can therefore be real without yet being complete enough for the next transition.

6. Start From the Pre-Change Reference

The strongest recovery reference is usually the verified state immediately before the failed change.

What could the learner do then?

Under what conditions?

With how much support?

At what workload?

With what level of independence?

The reference does not need to be perfect.

It needs to be clear enough that “recovered” means something.

7. Use the Decision Record to Reconstruct the Reference

The Decision Record is valuable here.

Before the failed change, what did the tutor believe?

What was stable?

What remained weak enough to justify trying something new?

That distinction prevents a common error:

Rolling back to a state that was stable in one way and pretending it was complete in every way.

The old route can be known good and still contain the unresolved problem that motivated change.

8. Verify the Function That Actually Failed

Recovery checks should be local.

If timing pressure caused sign accuracy to collapse, verify sign accuracy under the restored timing condition.

If prompt fading caused task initiation to collapse, verify independent or lightly supported initiation.

If a study system displaced actual work, verify work completion, prioritisation and recovery—not merely whether the old planner is being filled in again.

Do not reopen every capability merely because one layer failed.

9. Use One Representative Task Early

Recovery should meet reality quickly.

Do not verify only on the easiest restoration task.

Alicia’s simpler checking routine should be tested on a question family where sign and unit errors genuinely matter.

Faith’s restored prompt level should be tested on an unfamiliar but appropriately difficult task.

Emily’s simpler study plan should meet an ordinary busy school day, not only a quiet Sunday.

Representative evidence keeps recovery from becoming ceremonial.

10. Immediate Rebound Is the First Receipt

After rollback, the tutor often expects one quick change.

Accuracy stops falling.

Backlog stops growing.

Independent starts return.

Planning overhead drops.

Lesson flow becomes manageable.

That immediate rebound is useful evidence.

It is the first receipt, not the final verdict.

11. Delayed Return Strengthens Recovery

A stable state should survive some time.

Return after a lesson.

Return after a school week.

Return after another topic intervenes.

The relevant delay depends on the mechanism.

A recovered concept can be checked quickly and then again after several days.

A recovered study system needs enough calendar variation to show that the plan can survive real life.

12. Support Provenance Still Matters

The old routine may be back.

But is the tutor carrying more of it than before?

Did the learner complete the checking routine only because the tutor reminded them?

Did the study plan work only because a parent rebuilt it every evening?

Did Faith begin only because the tutor supplied a stronger prompt than the historical baseline?

A restored artifact under heavier support is not the same state as the earlier learner-owned routine.

13. The Ownership Check

Recovery should restore appropriate control, not only visible output.

Who starts?
Who chooses?
Who notices the error?
Who decides when to ask for help?
Who replans after disruption?
Who decides the next practice target?

If the learner’s output is back but control has migrated permanently to the tutor or parent, recovery is incomplete.

14. Check for New Defects Left Behind

A failed route can leave residue.

A learner may now over-check because the larger checklist trained excessive caution.

A learner may now rush because heavy timed practice changed pacing habits.

A learner may wait for structure because a rigid writing frame supplied too much of the planning job.

Recovery Check therefore asks not only whether the old defect disappeared, but whether the failed intervention created a new one that the old routine does not automatically remove.

15. Check the Intervention Cost Again

The Intervention Cost does not disappear just because the tutor rolled back.

The restored route may also carry costs.

Denise’s moderate timing may restore accuracy but still leave full-paper completion too weak.

Emily’s simpler plan may restore execution but still lack enough flexibility for peak weeks.

Recovery means the system is stable enough to rebuild—not that the old route suddenly became perfect.

16. Check Workload Feasibility

A recovered learning system must fit inside the learner’s real capacity.

Homework should fit without creating a new rollover cascade.

Revision should fit without borrowing heavily from sleep.

Tutor routines should fit without consuming the entire lesson in process management.

The Learning Budget supplies the capacity boundary.

17. Check the Same Failure Site

Verification is strongest where the failure actually occurred.

If full-paper timing caused collapse, a tiny untimed worksheet is weak recovery evidence.

If a study plan failed under a busy week, a quiet holiday week is weak recovery evidence.

If unsupported starts failed on unfamiliar questions, familiar questions are weak recovery evidence.

Return to the failure site at a tolerable dose.

18. Check a Nearby Unaffected Capability

Rollback should not quietly damage something that was previously stable.

If Denise reduces timing pressure, does her method selection remain stable?

If Alicia compresses the checking routine, does conceptual accuracy remain unchanged?

If Emily simplifies planning, does important retrieval still happen?

One nearby check helps ensure the restored state is not merely trading one problem for another.

19. Recovery Needs a Minimum Evidence Window

One good result can be enough for a small, reversible next step.

It is rarely enough for a large, difficult-to-reverse rebuild.

The more consequential the next move, the more mature the recovery evidence should be.

One clean fresh question may justify a small explanation adjustment.
Several stable attempts may justify reducing a restored prompt.
Two ordinary school weeks may be needed before a study system is called recovered.
Representative timed sections may be needed before full-paper performance pressure returns.

20. Recovery and the Stability Window

The Stability Window asked how long a changed route should be held.

The Recovery Check asks a similar timing question after rollback:

How much evidence must mature before we trust the restored route enough to change it again?

The answer depends on the target mechanism and the risk of moving too early.

21. Recovery Can Be Partial

Not every function returns together.

Alicia’s checking accuracy may recover before paper speed does.

Faith’s willingness to attempt may recover before full self-initiation does.

Emily’s task completion may recover before adaptive replanning does.

Partial recovery should be described precisely.

Do not let one restored variable pull the entire learner model back to “normal”.

22. Recovery Can Be Stronger Than the Old State

Rollback does not always return the learner to exactly the earlier capability.

The failed route may have taught something useful.

Denise may now have better awareness of paper checkpoints.

Beatrice may now understand paragraph structure more consciously.

Faith may now recognise when support is carrying too much of the start.

A Recovery Check should notice positive residue as well as negative residue.

23. Recovery Can Reveal the Old State Was Too Weak

Sometimes rollback restores exactly what it was supposed to restore—and the tutor discovers that the historical state is no longer sufficient.

The learner is older.

The school work is harder.

The examination is closer.

The workload has changed.

This is where verification and validation separate.

Verification says: yes, the old state returned.

Validation asks: is that returned state still good enough for what comes next?

24. If Verification Fails, Do Not Roll Back Forever

The tutor restores the last known good state.

The learner remains unstable.

That tells us something important.

The failure extends deeper than the first rollback target, or the learner’s present environment no longer matches the old state.

Now reopen diagnosis.

Do not keep walking backward through old routines simply because each one is older.

25. The Recovery Receipt

Every Recovery Check should define what evidence would count as restored function.

Accuracy receipt: the old error rate returns or improves under comparable conditions.
Initiation receipt: the learner begins representative tasks with the expected support level.
Selection receipt: the learner chooses the appropriate method without the cue that previously carried selection.
Study receipt: the plan converts into completed priorities without backlog cascade.
Recovery receipt: one ordinary disruption can be absorbed without whole-system redesign.
Ownership receipt: the learner controls the decision the old route was supposed to return.

26. Use Direct Evidence Before Distant Outcomes

A school grade may take weeks to arrive.

It may also be too broad to reveal whether the restored function returned.

If the failed variable was method selection, test method selection.

If the failed variable was task initiation, observe task initiation.

If the failed variable was sign accuracy under time, measure sign accuracy under controlled time.

Distant outcomes can confirm later.

Direct evidence tells the tutor sooner whether recovery is actually happening.

27. Do Not Let the Recovery Test Become Another Intervention

The tutor can accidentally make the recovery check too elaborate.

New worksheets.

New checklists.

New logging.

New parent reporting.

The measurement system can become the next source of load.

A good Recovery Check is discriminating, not administrative.

28. The Recovery Check Should Be Quiet

Many recovery checks can sit inside normal work.

One mixed item inside current homework.

One independent start while the tutor turns to another student.

One delayed retrieval question at the beginning of the next lesson.

One ordinary school week under the restored study plan.

The learner should not have to live inside an audit merely because the tutor is verifying recovery.

29. Class 0 · Homework Helper Recovery Check

Class 0 works nearest to task flow, completion and routine.

After rollback, verify:

the learner can locate the task;
the learner can identify the first action;
the learner begins without unnecessary adult management;
completion does not depend on the tutor remembering every deadline;
the restored routine does not recreate the old dependence.

Class 0 recovery rule: task flow is recovered when the learner can again operate the routine with the intended level of adult support.

30. Class 1 · Explainer Recovery Check

The Explainer restores access to meaning.

After rollback to an earlier representation or explanation level, verify that the learner can reconstruct the idea rather than merely follow the tutor again.

Class 1 recovery rule: clarity is recovered when the learner can generate enough of the relation to use it on a fresh example.

31. Class 2 · Drill Builder Recovery Check

The Drill Builder restores reliable execution.

If practice was rolled back from mixed to blocked, verify the execution weakness directly.

Then include one lightly mixed probe before declaring the learner ready to rebuild the selection layer.

Class 2 recovery rule: execution is recovered when accuracy and fluency return strongly enough that the next variability layer can be reintroduced without immediate collapse.

32. Class 3 · Diagnostic Tutor Recovery Check

The Diagnostic Tutor may roll back from a failed causal model to the last model that still predicted the learner well.

The recovery check asks whether that older model again explains the observed behaviour under fresh conditions.

If not, the learner model itself needs revision.

Class 3 recovery rule: do not call the diagnosis recovered until the restored hypothesis again predicts the pattern it claims to explain.

33. Class 4 · Route Designer Recovery Check

The Route Designer restores sequence.

After rollback, verify that current school participation, prerequisite repair and future route can coexist again without one lane consuming the others.

Class 4 recovery rule: the route is recovered when sequence once again protects both current operability and the highest-leverage next repair.

34. Class 5 · Performance Coach Recovery Check

The Performance Coach restores reliable output under representative pressure.

After rolling back time pressure, verify the core skill at the restored level and then test one moderate performance demand.

Class 5 recovery rule: performance is recovered when the restored skill survives enough realistic pressure to justify the next increase in realism.

35. Class 6 · Learning Architect Recovery Check

The Learning Architect verifies a wider system.

After rollback, ask whether core learner control, school operability, study flow, feedback loops and recovery capacity are functioning together again.

Do not require every optional system to be restored.

Class 6 recovery rule: architecture is recovered when essential functions are reliable enough that optional complexity can remain optional.

36. Learning Recovery

Learning concerns understanding, retrieval, representation and reasoning.

Recovery evidence should show that the learner can again perform the intended cognitive operation under an appropriate support condition.

Then use a fresh example, changed surface or delayed return to strengthen the claim.

37. Studying Recovery

Studying concerns planning, prioritisation, retrieval, resource use and self-management across real days.

A recovered studying system should:

convert plans into work;
survive ordinary disruption;
protect high-value priorities;
avoid uncontrolled backlog;
preserve recovery and sleep;
return planning ownership toward the learner.

38. Training Recovery

Training concerns reliable execution under repetition, variation and pressure.

Recovery first verifies the core operation at the restored load.

Then the tutor reintroduces one challenge variable to test whether the capability is ready to move forward.

Recovered training is therefore not merely comfortable practice.

It is stable capability with enough margin to accept the next training demand.

39. Education Recovery

Education places the learner inside current school stages, subjects, assessments and institutional constraints.

A learner is educationally recovered when the restored route allows them to participate in the present environment again without the recovery process itself consuming all available capacity.

Backlog alone is not the measure.

Functioning is.

40. Improvement Recovery

Improvement is the whole loop from observation to diagnosis, prioritisation, repair, practice, connection, performance and review.

A recovered improvement system can move through the loop again.

The learner is no longer trapped in emergency correction, endless diagnosis or permanent support.

The next decision can once again be made from interpretable evidence.

41. Repair Mode Recovery Check

Repair Mode is recovered when the learner can again engage the weak link productively rather than merely survive a simplified task.

The check should include one step toward reintegration.

Repair recovery rule: the repaired layer should be stable enough to reconnect to the task above it.

42. Alignment Mode Recovery Check

Alignment Mode is recovered when the learner can travel with current school demand again without tuition becoming a second school that pre-carries every task.

Alignment recovery rule: school participation should be stable enough that tuition can return to targeted support rather than continuous rescue.

43. Frontier Mode Recovery Check

Frontier Mode extends a strong learner.

After rollback, frontier work should not restart until core school obligations, recovery capacity and learner ownership are stable.

Frontier recovery rule: extension returns only when the strong core once again has spare capacity.

44. Three-Student Tutorials Are Natural Recovery Tests

The tutor’s attention moves.

That movement creates a useful question:

Does the restored route keep working when the tutor is not immediately available?

Alicia runs the simpler check while the tutor works with Beatrice.

Faith states the task and begins while the tutor checks Ciara.

Emily reorganises one delayed task without asking the tutor to rebuild the whole plan.

These ordinary small-group moments are recovery evidence.

45. Group Recovery Is Not Individual Recovery

The class may feel smoother after rollback.

That does not prove every learner recovered.

One student may now be carrying another.

One may be copying the group’s route.

One may still wait until the tutor returns.

Recovery claims remain learner-specific even when the teaching environment is shared.

46. One-to-One Tuition Needs Deliberate Non-Rescue

One-to-one tutoring can make recovery look faster than it is.

The tutor is always present to smooth the restored routine.

Create deliberate moments where the learner must carry the restored operation without live correction.

Otherwise the tutor may verify their own responsiveness instead of the learner’s recovery.

47. Alicia · Mathematics: The Three-Check Recovery Test

Alicia rolled back from seven checks to three: sign, unit, copied value.

The first recovery test uses a short mixed set containing the exact places where those errors historically appeared.

Accuracy returns.

The second test happens with no tutor reminder.

The third sits inside a timed section.

Now the tutor can say something stronger:

The simpler checking routine has restored the targeted accuracy without recreating the earlier completion cost.

That is a recovery claim worth acting on.

48. Beatrice · English: Structure Without the Cage

Beatrice rolls back from a visible four-line paragraph frame to two planning questions.

The Recovery Check uses two unlike topics.

On the first, Beatrice identifies the paragraph purpose and selects relevant development.

On the second, the topic requires a different paragraph shape.

She adapts instead of forcing the old frame.

Recovery is not “the old writing returned”.

Recovery is that structure and flexibility now coexist again.

49. Ciara · Science: Reconstruct Before Reciting

Ciara rolls back from model-answer dependence to causal reconstruction before wording.

The first check asks her to explain the mechanism in her own words.

The second changes one experimental condition.

The third asks her to use the model language only after the reasoning is complete.

Recovery is verified when language follows mechanism again rather than replacing it.

50. Denise · Additional Mathematics: Accuracy Before Full Pressure

Denise rolls back from full-paper timing to moderate timed sections.

The Recovery Check separates three variables:

method selection;
sign and transcription accuracy;
completion rate.

The restored state is verified when accuracy returns without method selection weakening and completion remains better than the original untimed baseline.

Only then should the tutor design the next full-paper re-entry.

51. Emily · Studying: Does the Simpler Plan Survive Wednesday?

Emily returns to the rolling three-day plan.

Sunday looks fine.

The real Recovery Check arrives on Wednesday when a school task overruns and a project deadline moves.

Can she reallocate without rebuilding the whole week?

Can she preserve the two highest-value priorities?

Can unfinished work remain visible without becoming a panic pile?

If yes, the studying system is functioning again.

52. Faith · Independence: One Prompt Is Not the Same as Full Rescue

Faith restores one prompt after unsupported starts collapsed.

State the task in your own words.

The Recovery Check asks whether that light prompt is enough.

Can she then choose a first move?

Can she continue while the tutor turns away?

Can she ask a specific question if the route fails?

Recovery is not zero prompts.

Recovery is the restored support level producing the intended learner control again.

53. Parent Communication: “Back to Normal” Is Too Vague

A parent-facing recovery explanation should be concrete.

We have restored the earlier routine, but I am not treating that alone as proof of recovery. Over the next week I am checking whether the original accuracy, independent starts and workload stability have returned under the same support conditions as before. If those signals hold, we can rebuild forward from there. If they do not, we will reopen the diagnosis rather than keep walking backward.

This tells the parent that recovery is being verified, not assumed.

54. Learner Communication: What Are We Checking?

The learner should know what recovery means.

We went back one step because the new version was making this part less reliable. Now I want to see whether the stable version is really working again. We are checking your accuracy, whether you can use it without reminders, and whether it still holds when the question changes a little. Once that is steady, we can move forward again.

Recovery becomes a learning process the student can understand rather than something adults do around them.

55. Recovery Check Versus Rebuild

Do not rebuild while the base is still uncertain.

If Denise’s moderate timing is still producing unstable accuracy, adding a new pacing strategy creates another moving variable.

If Emily’s simple plan is not yet surviving real school variation, adding a new dashboard feature is premature.

Recovery Check comes before the next meaningful redesign.

56. Recovery Check Versus Validation

Verification asks:

Did the state we intended to restore actually return?

Validation asks:

Is that restored state sufficient for the next educational demand?

A learner can pass verification and fail validation.

Denise may regain moderate timed accuracy but still not be ready for full-paper timing.

Both conclusions can be true at once.

57. Recovery Check Versus Confirmation

The unpublished Tutor Handbook Volume 0027 is titled The Confirmation and concerns whether a breakthrough has become the learner’s new normal.

The Recovery Check owns a different direction of travel.

Confirmation asks whether a sudden improvement deserves promotion.

Recovery Check asks whether an earlier stable state has genuinely returned after a failed change.

Keeping those jobs separate prevents a useful term from swallowing two different decisions.

58. Recovery Check Versus the New Baseline

Volume 0031 | The New Baseline asks how the learner model should change after the same weak link has required repair twice.

The Recovery Check is narrower and earlier.

It verifies whether rollback restored enough stability to make any later baseline decision trustworthy.

59. Recovery and the Scope Boundary

Recovery evidence can reveal that the problem extends beyond ordinary tutoring.

The tutor restores a stable academic routine, but the learner remains unable to access work under ordinary conditions.

The tutor can document the educational pattern, protect appropriate academic support and communicate with the responsible adult.

The tutor should not keep redesigning academic routines forever if the controlling question belongs elsewhere.

The Scope Boundary remains part of responsible recovery.

60. Recovery and the Drift Check

Rollback can restore the historical route, but ordinary practice may still drift during recovery.

The tutor says the old prompt level has returned, but in practice the prompt arrives sooner.

The study plan looks the same, but the parent now manages the priorities.

The checking routine has three steps again, but the tutor verbally adds four more.

The Drift Check asks whether the restored route being lived is actually the route the tutor thinks was restored.

61. Recovery and the Temporary Override

Recovery periods contain legitimate exceptions.

A difficult school week may require one temporary extra support.

That exception should not invalidate the whole recovery check if it is visible and bounded.

Record the exception so the tutor does not accidentally read supported performance as evidence of full restored independence.

62. Recovery and the Learning Budget

Recovery checks consume capacity.

Keep them proportionate.

A learner should not complete three full papers merely to prove a small timing rollback worked.

A study system should not require an hour of logging to prove that planning overhead decreased.

The best Recovery Check buys a large reduction in uncertainty with a small amount of learner capacity.

63. The Recovery Check Card

Failed change: What newer route became unstable?
Rollback state: What earlier route was restored?
Historical stable function: What could the learner reliably do before the failed change?
Recovery target: Which function must return now?
Immediate receipt: What should improve quickly after rollback?
Representative check: What real task will test the restored function?
Support condition: How much help is present?
Delayed check: What must still hold later?
Collateral check: What nearby capability should remain stable?
Ownership check: What should the learner carry again?
Verification result: restored, partly restored, not restored or unclear?
Validation question: Is the restored state enough for the next demand?
Next action: hold, rebuild, re-enter, re-diagnose or escalate?

64. The Thirty-Second Recovery Check

What failed? What state did we restore? What function should be back? What direct evidence shows it is back under the right support condition? What still does not hold?

That is enough for many everyday tutoring decisions.

65. The Recovery Decision

Restored: the critical function has returned and held enough to support the next planned move.
Partly restored: enough has returned to continue stabilising, but rebuilding forward should wait.
Not restored: the rollback target was insufficient or the earlier state no longer works.
Unclear: evidence conditions were too noisy, support was inconsistent or the wrong function was measured.

All four outcomes are useful when they improve the next question.

66. Research Foundation: Monitor Progress to Adjust Teaching

The Australian Education Research Organisation’s Monitor Progress guidance emphasises using evidence over time to understand what students know and can do and to adjust teaching accordingly.

The Recovery Check applies that principle to a specific transition.

The tutor does not assume that rollback restored the learner.

The tutor gathers enough evidence to decide whether to hold, rebuild or diagnose again.

67. Research Foundation: Scaffolds Should Follow Learner Proficiency

AERO’s Scaffold Practice guide emphasises planned and responsive supports that are adjusted and gradually removed as proficiency develops.

A Recovery Check after restoring a scaffold therefore asks whether the support level now matches the learner’s actual proficiency.

If the learner performs only because the restored scaffold is doing all the control work, recovery may be narrower than it looks.

68. Research Foundation: Variation Reveals Adaptability

AERO’s Vary Practice guidance highlights varied and spaced opportunities that help learners adapt knowledge across examples and contexts.

Recovery verification can use small, discriminating variation to test whether the restored function belongs to the learner rather than to one familiar task surface.

69. Research Foundation: Implementation Must Be Read in Context

The Education Endowment Foundation’s A School’s Guide to Implementation emphasises that implementation is a process enacted in real contexts, not merely an idea adopted on paper.

The same is true after rollback.

The tutor should inspect the routine as actually lived, with real support, real workload and real learner behaviour.

The declared route is not the evidence.

The enacted route is.

70. Research Foundation: Verification and Validation Are Different Jobs

eduKateSG separates Verification from Validation.

This distinction is useful in tutoring because “we restored the previous state” and “the previous state is now sufficient for what the learner faces next” are not the same conclusion.

Recovery Check completes the first job and prepares the second.

71. The Tutor Classification Model Still Governs the Check

The wider Tutor Classification Model by eduKateSG separates Class 0 Homework Helper through Class 6 Learning Architect.

The Recovery Check should verify the capability owned by the active tutor function.

A Class 1 recovery check should not pretend to establish whole-system readiness.

A Class 5 check should not interpret a timing problem as a global knowledge deficit without supporting evidence.

Good verification remains inside the job being verified.

72. The Tutor’s Recovery Checklist

What failed under the newer route?
What state did we roll back to?
What evidence made that state “known good” originally?
Is that reference still relevant to the learner’s current demand?
What exact function should have returned?
What is the smallest direct check of that function?
What support is present during the check?
Has the restored function survived one fresh task?
Has it survived a useful delay?
Has it survived a small change in surface or context?
Has learner ownership returned to the intended level?
Did the failed route leave any new defect behind?
Did rollback create a new cost?
Is workload feasible again?
What nearby stable capability should remain unchanged?
Is the state verified but not yet validated for the next demand?
Should we hold, rebuild, re-enter, re-diagnose or escalate?

73. The Parent’s Recovery Checklist

What exactly was rolled back?
What capability should return because of that rollback?
How will the tutor test that capability directly?
Is the child performing under the same support level as before?
Has the improvement appeared more than once?
Is the study or homework load manageable again?
Has the learner regained control or are adults carrying more?
What new issue, if any, remains from the failed route?
Is the restored state enough for the next school demand?
What is the next step if the recovery check fails?

74. The Learner’s Recovery Checklist

What part of the old route did we bring back?
What should now feel more reliable?
Can I do the important part without extra hidden help?
Can I still do it on a fresh question or task?
Can I do it after a few days?
What part still feels unstable?
What new habit from the failed route do I want to keep?
What new habit do I need to unlearn?
Am I ready for the next harder condition, or do I need more stability first?
What would convince me that I have genuinely recovered this part of my learning system?

75. What the Recovery Check Is Not

It is not a celebration that the old routine has returned.
It is not a demand to retest the whole curriculum.
It is not the same as confirming a breakthrough.
It is not the same as validating readiness for the next stage.
It is not one easy worksheet.
It is not one higher score without process evidence.
It is not proof of independence if support conditions changed.
It is not an excuse to keep the learner in a safe old state forever.
It is not another complex monitoring system.
It is not a reason to ignore the unresolved problem that originally motivated change.
It is not permission to walk backward indefinitely if the old state no longer works.
It is a focused verification that the learner has regained enough stable function to make the next forward decision interpretable.

76. The Ethical Standard

Rollback can feel reassuring.

The routine is familiar.

The adults know what to do again.

The learner may feel relief.

That relief is useful.

It is not evidence by itself.

A professional tutor owes the learner something stronger than comfort with the old route.

The tutor should verify what returned.

Verify under the right support condition.

Verify at the failure site.

Verify enough stability to make the next change fair.

And when the restored state is no longer sufficient, the tutor should say so rather than hiding behind the safety of what once worked.

The Recovery Check is where a tutor proves that rollback did more than recreate yesterday’s routine. It verifies that the learner regained the functions that made yesterday trustworthy, that control has returned to the right owner, and that the next forward move can begin from evidence rather than relief.

Evidence and Connected Reading

Final Compression

Rollback returned the route.

Now verify the learner.

Start from the last verified stable reference.

Name the function that failed.

Test that function directly.

Use a representative task.

Keep the support condition visible.

Look for immediate rebound.

Then look again after delay.

Change the surface just enough to test ownership.

Check one nearby stable capability.

Check the Learning Budget.

Check whether the failed intervention left a new defect behind.

Check whether learner control returned to the right owner.

Do not call a familiar routine recovery merely because adults feel safer again.

Do not demand perfection before moving forward.

Verify enough stability for the consequence of the next move.

If the restored state holds, rebuild deliberately.

If it partly holds, stabilise longer.

If it does not hold, reopen diagnosis.

If it holds but is no longer sufficient for the next demand, validate a stronger route.

The Recovery Check is the discipline of proving that the learner—not merely the old routine—has returned to trustworthy operation. Only then does rebuilding forward become a learning decision instead of another guess placed on top of instability.