The Tutor Handbook Vol No.0067 | The Change Bandwidth

The Tutor Handbook · Volume 0067 · Series ID THB-0067

The tutor has a good plan.

Actually, the tutor has six good plans.

Reduce one prompt.

Add mixed practice.

Increase timing pressure.

Move one topic into maintenance.

Give the learner more planning responsibility.

Introduce a new checking routine.

Every change makes sense.

The learner can still be overwhelmed by the total amount of change.

The Change Bandwidth is the maximum practical rate, size and number of active learning-system changes that can be introduced over a period while the learner can still adapt, the tutor can still interpret what happened, and the route can still stabilise enough for improvement to become usable.

The previous three volumes built the control surface.

Volume 0064 | The Change Queue orders justified future changes.
Volume 0065 | The Concurrency Problem interprets evidence when several systems change anyway.
Volume 0066 | The Configuration Snapshot records what route is actually running now.

This volume asks the next question:

Even when the queue is ordered, the external world is understood and the current route is clear, how much change can the learner and tutoring system absorb before adaptation becomes overload, evidence becomes unreadable and improvement becomes unstable?

Quick Read

Change bandwidth is not the same as learning capacity.
A strong learner can still have low change bandwidth during an unstable week.
The number of active changes matters, but so do their size, novelty, reversibility and interaction.
One large change can consume more bandwidth than several tiny adjustments.
External volatility—school transitions, exam periods, illness, timetable disruption—reduces the amount of tutor-generated change the system can safely absorb.
Evidence has latency. Some changes need time before a useful receipt can appear.
Stacking another major change before the first has produced readable evidence can destroy interpretability.
Bandwidth belongs to the learner, the tutor, the family and the shared timetable—not only to the lesson.
The tutor must distinguish instructional intensity from configuration change. Hard work is not automatically high change.
A stable route can contain demanding practice without consuming much change bandwidth.
A gentle-looking redesign can consume large bandwidth if it changes support, ownership and evidence rules at once.
The Stability Window protects one active route; Change Bandwidth limits how much else is altered around it.
The Learning Budget allocates time and attention; Change Bandwidth limits the rate and complexity of system modification.
The Change Queue decides what waits; Change Bandwidth decides how much can be live.
During high external concurrency, tutor-generated change should usually become narrower.
After successful adoption, bandwidth can recover because the new route becomes ordinary.
Bandwidth should be estimated, not treated as a fixed learner trait.
The learner should increasingly help judge when too many changes are active.
Three-student tutorials have shared classroom bandwidth and separate learner bandwidth.
The ethical standard is simple: do not make the learner carry more change than the system can convert into stable capability.

1. What This Volume Owns

This volume owns the rate and complexity of active change.

It does not decide which future change is more important. That belongs to the Change Queue.

It does not allocate the learner’s total time and attention across subjects. That belongs to The Learning Budget.

It does not decide whether a new route has been held steady long enough to learn from. That belongs to The Stability Window.

Change Bandwidth asks:

How much live modification can this learner-system responsibly carry now?

2. Change Bandwidth Is Not Intelligence

A highly capable learner can have low change bandwidth.

Imagine a strong Secondary 4 student during prelim season.

The learner understands quickly, retrieves well and reasons accurately.

But school deadlines, full papers, oral preparation and sleep pressure already create heavy external volatility.

That may be a poor moment to introduce a new study planner, new checking routine, new feedback code and new timing protocol simultaneously.

Capability is high.

Change bandwidth is temporarily low.

3. Change Bandwidth Is Not Motivation

A motivated learner can enthusiastically agree to five improvements.

Agreement does not prove that all five can be adopted, practised, monitored and maintained at once.

Motivation may raise willingness.

Bandwidth still depends on cognitive load, routine complexity, outside demands, evidence latency and the amount of self-regulation the learner can currently carry.

4. Change Bandwidth Is Not Difficulty

A difficult problem set can be cognitively demanding while leaving the learning configuration unchanged.

A simple-looking new rule can consume large change bandwidth if it changes who chooses the task, when help is allowed and how success is measured.

Do not infer bandwidth from how hard the worksheet looks.

5. Change Bandwidth Is Not the Learning Budget

The learner may have ninety minutes available.

That does not mean the learner can absorb ninety minutes of new routines.

Time capacity and change capacity overlap but are not identical.

A learner may have enough time to practise a stable routine for ninety minutes while having almost no spare bandwidth for a new way of planning, checking and requesting help.

6. Change Has a Size

Not all changes are equal.

Changing one example is small.

Changing from blocked to mixed practice is larger.

Changing from tutor-selected study priorities to learner-selected priorities is larger again because responsibility moves.

Changing tutor, timetable, support rules and performance conditions together is a system transition.

Bandwidth should account for the magnitude of the change, not merely count how many changes exist.

7. Change Has Novelty

A learner who already knows how to use spaced retrieval can apply it to a new subject with relatively little configuration change.

A learner who has never planned a study week may need explicit modelling, guided practice and review before self-directed planning becomes stable.

The same nominal change can consume different bandwidth depending on how novel the operation is to the learner.

8. Change Has Interaction Cost

Two changes can interfere.

Removing a prompt while adding strong timing pressure may make it unclear whether hesitation comes from reduced support or the clock.

Changing task selection while changing marking criteria may force the learner to rebuild two decision systems at once.

Bandwidth falls when active changes interact strongly.

9. Change Has Evidence Latency

Some changes produce useful evidence quickly.

A revised instruction can be tested on the next question.

Other changes need time.

A spacing routine requires delayed returns.

A study-planning change may need several school cycles.

A support fade may need multiple fresh tasks to show whether independence really held.

Do not stack major changes faster than the system can produce interpretable receipts.

10. Change Has Recovery Cost

If a change fails, how costly is recovery?

A short trial of mixed questions may be easy to reverse.

A large timetable redesign may take days to unwind.

A support fade that creates repeated failure may require confidence and routine rebuilding.

High-recovery-cost changes consume more bandwidth because the system needs reserve capacity in case the change does not work.

11. External Volatility Consumes Bandwidth Before Tuition Begins

The learner changes school teacher.

The timetable changes.

Prelims begin.

A new CCA schedule starts.

The household travel routine changes.

These may consume adaptation capacity even if tuition changes nothing.

Volume 0065 called this the Concurrency Problem.

Change Bandwidth turns that observation into a control rule: when external volatility rises, discretionary tutor-generated change usually needs to narrow.

12. The Tutor Has Bandwidth Too

A tutor can only observe so many active hypotheses well.

If prompt fading, timing, mixed practice, planning transfer and feedback changes all occur together, the tutor may not remember which receipt belongs to which change.

A learner can technically carry the interventions while the tutor loses interpretive control.

That is still over-bandwidth.

13. The Family Has Bandwidth Too

Parents may need to change how they prompt, supervise or discuss work.

If home receives four new support rules in one week, implementation fidelity can fall even when the rules are individually sensible.

A route that depends on family coordination should include household bandwidth in the decision.

14. The School Calendar Has Bandwidth

Some weeks simply contain less spare adaptation capacity.

Assessment week.

Transition week.

Heavy project week.

Do not treat every week as an equally suitable time for system redesign.

15. The Learner’s Bandwidth Can Change Quickly

Bandwidth is not a personality trait.

A learner may handle major change comfortably in June and poorly in September.

The difference may come from exam pressure, fatigue, school load, recent success, stronger routines or greater self-regulation.

Estimate bandwidth from the current system, not from a permanent label such as “adaptable” or “resistant to change.”

16. A Conceptual Change-Bandwidth Equation

A practical way to think about change bandwidth is:

Usable Change Bandwidth ≈ Adaptation Capacity × Observability × Recovery Capacity ÷ (Change Magnitude × Interaction × External Volatility)

This is a conceptual model, not a scoring formula.

Its purpose is to remind the tutor that change becomes harder to manage when it is large, highly interactive or introduced into an already unstable environment.

17. Adaptation Capacity

Can the learner understand the new rule, remember it, apply it and coordinate it with existing routines?

A new checking routine that requires seven steps may exceed current adaptation capacity even if each step is easy.

Reduce configuration complexity before concluding the learner lacks discipline.

18. Observability

Can the tutor see what the change is doing?

If receipts are noisy, delayed or hidden inside many concurrent changes, effective bandwidth falls.

A system can tolerate more change when each change has a clear, bounded and timely receipt.

19. Recovery Capacity

If the change fails, can the learner return to a stable route without losing important gains?

High recovery capacity increases safe experimentation.

Low recovery capacity calls for smaller, more reversible changes.

20. Change Magnitude

Magnitude asks how much of the operating system changes at once.

Does one prompt disappear?

Or does the learner now choose all tasks, self-mark all work, manage timing and decide when to seek help?

Both can move toward independence.

The second consumes far more bandwidth.

21. Interaction

Changes that affect different layers may interact unpredictably.

More time pressure may reduce the effectiveness of a newly introduced self-check routine because the learner sacrifices checking first.

Less prompting may interact with a new unfamiliar representation and create a failure that neither change would have produced alone.

Interaction consumes bandwidth because interpretation becomes harder.

22. External Volatility

The same tutoring change can be manageable in a calm term and excessive during a transition.

Bandwidth should therefore be estimated in context.

“This learner handled three simultaneous changes last month” does not prove the learner should handle three during prelim week.

23. The One-Major-Change Heuristic

A useful default is:

One structural change at a time, plus ordinary small adjustments that do not alter the learner’s main job.

This is not a law.

It is a protective heuristic when uncertainty is high.

Tightly coupled implementation steps can still belong together as one coherent intervention.

24. A Major Change Alters the Learner’s Job

Examples include:

moving from explanation to independent retrieval;
moving from blocked to mixed method selection;
introducing real timing pressure;
removing a scaffold the learner previously relied on;
transferring planning responsibility to the learner;
moving from Repair to Frontier mode;
changing the main evidence rule.

These consume more bandwidth than changing examples or adjusting question order.

25. Evidence Latency Should Control Change Rate

If a change needs three delayed returns to judge, do not automatically stack another structural change after the first successful session.

The system has not yet produced the receipt it was designed to produce.

Change rate should respect evidence latency.

26. Stability Window and Change Bandwidth Are Partners

The Stability Window says: hold the new route steady long enough to learn from it.

Change Bandwidth says: do not load so many other changes around that route that the Stability Window becomes meaningless.

One protects duration.

The other protects the amount of simultaneous change.

27. Adoption Restores Bandwidth

A new routine initially consumes attention.

After successful adoption, the routine becomes more ordinary.

The learner no longer has to think about every step.

Bandwidth recovers.

This is one reason not to rush the next structural change before the current one has become operationally stable.

28. Failure Consumes Bandwidth Too

A failed intervention can leave:

confusion;
extra work;
reduced confidence;
stale materials;
uncertain support rules;
parent concern;
the need for rollback.

Recovery uses capacity.

After a failed change, reduce the rate of new change until a stable baseline returns.

29. Rollback Should Usually Lower the Change Rate

Rollback is already a transition.

Do not combine rollback with a fresh redesign unless immediate learner need demands it.

Return to stability, verify recovery, then reopen the queue.

30. Cutover Should Usually Lower the Change Rate

During Cutover, the learner is moving from one normal to another.

That transition consumes bandwidth.

Let the new normal become normal before launching a second major structural change.

31. Handoffs Lower Bandwidth Temporarily

A new tutor changes language, pacing, trust, expectations and observation patterns even if the route is preserved.

This is why The First Change argued for narrow modification after a Receiving Check.

The handoff itself uses bandwidth.

32. Examination Periods Often Lower Bandwidth

Near a major examination, the learner may need high practice intensity and low configuration novelty.

This is not a universal rule.

But it is often wiser to strengthen known-good routines than to redesign the entire studying system shortly before performance is required.

33. Post-Exam Periods Can Restore Bandwidth

After the examination, urgency falls and evidence becomes available.

This can be a good time to redesign deeper study routines, transfer more learner ownership or rebuild a route whose weaknesses were tolerated temporarily for examination alignment.

34. Bandwidth Can Be Increased

Change bandwidth is not only something to ration.

It can grow when the learner becomes better at:

understanding why a change is being made;
monitoring the result;
following a stable review routine;
asking for clarification;
distinguishing current from old support;
recovering after a failed attempt;
managing one active change independently.

Self-regulation increases the amount of change the learner can participate in without being passively reorganised by adults.

35. Explain the Change to Increase Bandwidth

An unexplained change consumes more coordination because the learner has to infer what changed and why.

A brief explanation can reduce that burden:

Your method choice is stable now. We are keeping the mixed questions, but I am adding moderate timing. The job is not to rush. It is to keep the same accuracy while making the route more efficient.

Clarity frees bandwidth.

36. Reduce Administrative Overhead to Increase Bandwidth

A new change may require tracking.

Do not build a complex dashboard if one short note captures the needed receipt.

Every coordination layer competes with learning.

37. Preserve Familiar Anchors During Change

When one important element changes, keep other parts familiar where possible.

New timing can use familiar question structures.

New learner ownership can use a familiar weekly planner.

New mixed practice can preserve familiar checking rules.

Stable anchors reduce total novelty.

38. Use the Configuration Snapshot to Protect Bandwidth

The Configuration Snapshot shows what is currently live.

Before adding a structural change, read the snapshot.

How many other active changes are already running?

Has one of them produced enough evidence yet?

Is the learner still adapting to a recent transfer of responsibility?

Bandwidth decisions improve when current configuration is visible.

39. Use the Change Queue to Protect Bandwidth

When a new good idea appears, it does not need immediate implementation.

Place it in the Change Queue.

Queueing preserves the idea without consuming current bandwidth.

40. The Change-Bandwidth Card

Active structural changes: What major changes are already live?
Magnitude: How much does each change alter the learner’s job?
Novelty: How unfamiliar are the new operations?
Interaction: Do the active changes affect one another?
Evidence latency: How long before a useful receipt can appear?
External volatility: What else is changing in school, home or timetable?
Learner adaptation: Can the learner explain and operate the current change?
Tutor observability: Can the tutor still tell what is producing what?
Family coordination: Can home implement the relevant support rule?
Recovery reserve: Is there capacity to recover if the change fails?
Decision: Add, hold, shrink, queue or reverse the proposed change?

41. A Simple Bandwidth Decision

Before activating a new structural change, ask:

What major change is already live, has it produced readable evidence yet, and what will become harder to interpret or operate if I add this change now?

If the answer is unclear, queue the change rather than forcing it into the current route.

42. Class 0 · Homework Helper Bandwidth

Routine changes can look small but accumulate quickly.

New homework log.

New packing rule.

New start cue.

New parent check.

Class 0 should install routines in an order the learner can actually automate.

43. Class 1 · Explainer Bandwidth

Changing explanation style usually consumes less bandwidth than changing the learning route.

But multiple new representations, terminology and note structures can still overload a novice.

Keep explanatory novelty subordinate to conceptual clarity.

44. Class 2 · Drill Builder Bandwidth

The Drill Builder can change spacing, variation, cueing, timing and mixture.

Do not change all five together unless they form one tightly controlled progression.

Practice design should become more demanding at a rate that preserves the target capability.

45. Class 3 · Diagnostic Tutor Bandwidth

Diagnostic change can be small and high-information.

One discriminating question may consume little learner bandwidth while preventing a large wrong intervention.

When bandwidth is low, prefer small probes over large programme redesign.

46. Class 4 · Route Designer Bandwidth

The Route Designer controls sequence and therefore controls change rate.

Class 4 should ask not only “what comes next?” but “is the learner-system ready for another next?”

47. Class 5 · Performance Coach Bandwidth

Performance changes often add pressure.

Time limits, longer papers, reduced support, unfamiliar ordering and stricter checking expectations can all be useful.

Add pressure in layers so the tutor can see which performance component breaks first.

48. Class 6 · Learning Architect Bandwidth

The Learning Architect sees many levers.

That creates the highest risk of sophisticated over-intervention.

System visibility should increase restraint, not the number of levers pulled at once.

49. Repair Mode Bandwidth

Repair Mode can justify concentrated change around one weak link.

But concentration is different from breadth.

Change several supports only when they serve the same repair mechanism and can still be interpreted as one coherent intervention.

50. Alignment Mode Bandwidth

Alignment Mode often experiences the highest external volatility because school demand moves quickly.

During busy periods, tutor-generated configuration change may need to shrink even while practice intensity rises.

51. Frontier Mode Bandwidth

Frontier Mode can create the illusion of unlimited bandwidth because the learner is strong.

Extension should still be introduced at a rate that preserves core stability and learner ownership.

High ability earns more possible routes.

It does not abolish the cost of constant redesign.

52. Three-Student Tutorials Have Shared and Individual Bandwidth

The tutor may introduce a shared classroom routine.

That consumes group bandwidth.

Each learner may simultaneously have an individual route change.

That consumes personal bandwidth.

A well-run three-student class therefore separates:

changes everyone must learn;
changes one learner must carry;
ordinary instructional variation that does not count as a structural change.

53. Do Not Use One Learner’s Bandwidth for Another Learner’s Change

Alicia needs timed work.

Timing the whole group may force Beatrice and Ciara to adapt to a configuration change they did not need.

Shared delivery should not create unnecessary shared change.

54. Alicia · Mathematics: Do Not Add Timing and New Representation Together Yet

Alicia is ready for moderate timing.

She is also ready for more varied algebraic representations.

Both changes are good.

The tutor activates timing first and leaves representation variation in the Change Queue.

After timing becomes stable, the second change can enter without making any drop in accuracy impossible to interpret.

55. Beatrice · English: One Evidence Change, Not Three Writing Changes

Beatrice can improve evidence precision, vocabulary precision and paragraph compression.

The tutor chooses evidence precision as the active change because it is the current bottleneck.

Vocabulary enrichment continues through ordinary reading, but it does not become a second structural intervention.

Paragraph compression waits.

56. Ciara · Science: Scaffold Fade Before New Answer Architecture

Ciara is fading model language.

The tutor also wants to introduce a more advanced answer-planning frame.

That would remove one support while adding another.

The tutor waits.

First prove that causal explanation survives the fade. Then decide whether the new planning frame solves a real next problem.

57. Denise · Additional Mathematics: Pressure Has Bandwidth

Denise can handle longer timed sections.

The tutor could also reduce checking time, increase question difficulty and move toward full papers.

Doing all three would make the first performance collapse unreadable.

Change one pressure dimension, observe, stabilise, then add the next.

58. Emily · Studying: Busy Week Means Low Change Bandwidth

Emily’s family schedule changes and school assigns two major projects.

This is not the week to introduce a completely new digital study dashboard.

Keep the familiar planner and make one bounded modification: protect a small retrieval floor.

Low bandwidth calls for narrow change.

59. Faith · Independence: Transfer One Decision at a Time

Faith is ready for more independence.

The tutor could transfer task start, task choice, help-seeking, checking and weekly planning at once.

Instead, transfer the next decision that the evidence says she can carry.

Once that becomes ordinary, transfer another.

Independence grows through successful ownership, not through sudden adult withdrawal.

60. Parent Communication: “Good Ideas Can Still Wait”

Parents may ask why a known problem is not being addressed immediately.

A useful explanation is:

I agree this is worth improving. We already have one important route change running, and I want enough stable evidence to know whether it works before we add another. I am keeping this item in the queue and will activate it when the current change has produced the receipt we need.

Waiting becomes an evidence decision rather than neglect.

61. Parent Communication: Watch Home-Generated Change

Home can accidentally exceed change bandwidth.

New tutor rule.

New parent rule.

New revision app.

New weekend schedule.

Each addition may be reasonable.

Coordinate major changes so the learner is not carrying several adult redesigns without anyone seeing the combined load.

62. Learner Communication: Name the One Big Change

The learner should be able to say:

The big thing changing now is that I choose the method without a topic label. Everything else is mostly the same until that becomes stable.

That sentence is evidence that the configuration is understandable.

63. Learner Voice Can Detect Over-Bandwidth

A learner may say:

I know what each new rule means, but I keep forgetting which one I am supposed to use first.

That is valuable system evidence.

The problem may not be motivation or memory of subject content.

The configuration itself may be too change-dense.

64. Research Foundation: Implementation Takes Time and Context

The Education Endowment Foundation’s A School’s Guide to Implementation treats implementation as a staged process influenced by behaviours, context and how an approach is enacted rather than as a one-off decision to adopt an idea.

Change Bandwidth translates that implementation insight to tutoring: selecting the right change is not enough. The route must have enough time and operating capacity for the change to become real practice.

65. Research Foundation: Monitoring Should Guide Adjustment

The Australian Education Research Organisation’s Monitor Progress guide emphasises checking what students understand and can apply, identifying gaps and adjusting instruction, guidance and feedback.

Change Bandwidth adds a rate limit: adjustment should respond to evidence, but not so quickly that the system never becomes stable enough to generate evidence worth trusting.

66. Research Foundation: Scaffolds Should Be Responsive and Fade

AERO’s Scaffold Practice guidance describes planned and responsive supports that are gradually removed as proficiency develops.

Fading support is itself a change. Doing it while simultaneously changing task type, timing and responsibility can make failure hard to interpret. Bandwidth discipline helps scaffolding remain responsive rather than chaotic.

67. Research Foundation: Independent Learning Develops Through Guided Transfer

EEF’s 2026 guidance on metacognition and self-regulation emphasises planning, monitoring and evaluating learning, supported by modelling and guided practice before independent use.

This supports a central bandwidth principle: learner ownership is transferred progressively, not dumped all at once in the name of independence.

68. Research Foundation: Adaptive Expertise Is Not Constant Change

The OECD Teaching Compass describes teachers as adaptive experts and co-designers of learning while emphasising student agency.

Adaptive expertise does not mean changing everything whenever new information appears. It means selecting proportionate responses while preserving enough coherence for learners to act with increasing agency.

69. Research Foundation: Effective Tutoring Depends on Implementation

The EEF’s Making a Difference with Effective Tutoring stresses the importance of careful implementation, monitoring and evaluation alongside the positive average evidence for tutoring.

A tutor can therefore have the right intervention and still weaken its effect by implementing too many changes for the learner-system to absorb coherently.

70. Common Failure: Fast Tutor, Slow System

The tutor diagnoses quickly and sees the next five moves.

The learner still needs time to make move one ordinary.

Expert speed should not become learner change pressure.

71. Common Failure: Change Is Added Before the Receipt Arrives

The first new route feels promising after one lesson.

The tutor adds another.

Then another.

When performance changes later, the original intervention can no longer be read cleanly.

Respect evidence latency.

72. Common Failure: Strong Learner Means Unlimited Change

High-performing students often receive the most redesign because adults believe they can handle it.

They may be able to complete the work.

That does not mean the system remains interpretable, efficient or learner-owned.

73. Common Failure: Low Marks Trigger Maximum Change

A bad result creates urgency.

Adults change tuition, homework, revision, screen time, sleep rules and study materials simultaneously.

The system becomes less readable exactly when diagnosis needs to become more precise.

Low marks can justify action without justifying uncontrolled breadth.

74. Common Failure: New Tutor Proves Value Through Novelty

A receiving tutor changes several things to demonstrate expertise.

The learner experiences the tutor’s professional identity as configuration churn.

Value is not measured by how many visible differences the new adult creates.

75. Common Failure: Parent and Tutor Spend the Same Bandwidth Twice

Tutor installs a new self-check routine.

Parent simultaneously installs a new homework supervision routine.

Both ask the learner to coordinate new behaviour.

Neither adult sees the combined adaptation load.

Coordinate structural changes across adult systems when possible.

76. Common Failure: Bandwidth Becomes an Excuse for Stagnation

“Too much change” can become a reason never to progress.

That is not the purpose.

Bandwidth is a rate limit, not a prohibition.

Once a route stabilises and receipts support it, activate the next justified change.

77. The Tutor’s Change-Bandwidth Audit

What structural changes are already live?
How large is each one?
How novel is each one to the learner?
Do any of them interact strongly?
What receipt are we waiting for?
How long should that receipt take to appear?
What external changes are already consuming adaptation capacity?
Can the learner explain the current change?
Can the tutor still interpret the result?
Can the family implement the relevant home rule?
Is there enough recovery reserve if the change fails?
Would the proposed new change alter the learner’s job?
Can it wait in the Change Queue?
Can it be shrunk into a smaller trial?
Should current change be held until evidence arrives?
Has an adopted routine become ordinary enough to release bandwidth?

78. The Parent’s Change-Bandwidth Audit

How many new learning routines is my child currently carrying?
What changed at school this week?
Are we adding another home rule at the same time?
Which current change is most important?
What evidence are we waiting for?
Could the next good idea wait?
Is my child’s difficulty coming from the subject or from too many new routines?
Are we restoring old support because the new route feels unfamiliar?
Has a recent successful change become stable enough for the next one?

79. The Learner’s Change-Bandwidth Audit

What is the biggest thing changing in my learning right now?
What parts are staying the same?
Can I remember the new rule without being reminded?
Can I explain why the change exists?
Am I learning the subject, or mostly learning new routines?
Which new rule still takes a lot of effort to remember?
What result are we waiting for before changing something else?
Do I think another change would help now, or would it make the route harder to manage?
What decision am I ready to own next?

80. What the Change Bandwidth Is Not

It is not a fixed trait.
It is not intelligence.
It is not motivation.
It is not the Learning Budget.
It is not the Stability Window.
It is not the Change Queue.
It is not a reason to avoid necessary support.
It is not a reason to freeze a failing route.
It is not a demand for only one tiny change forever.
It is not a numerical score that can be measured precisely.
It is a practical limit on the rate and complexity of live system change so adaptation, interpretation and stable improvement remain possible.

81. The Ethical Standard

A tutor can be right about every individual improvement and still build a bad learning system by activating too many of them at once.

The learner pays for that mistake.

They carry the new rules.

They absorb the contradictions.

They experience the uncertainty.

They are blamed when the system becomes too complex to operate.

Professional tutoring does not ask how many useful changes the tutor can imagine. It asks how many changes the learner-system can responsibly convert into stable capability now—and it leaves the rest waiting until the system has earned more bandwidth.

Evidence and Connected Reading

Final Compression

Good change still has a cost.

Count the structural changes already live.

Estimate their size.

Estimate their novelty.

Notice their interactions.