HCD.6 - Design Representative Practice Tasks

Type: Human Capability Development practitioner practice-task design pattern Status: Stable

HCD.6:0 - Use This When

Use this pattern when a named human contribution and a capability-development Method have been selected, but the person still needs a task in which to recognize the situation, act, obtain an inspectable result, receive feedback and improve the action. Begin with the target action and target-domain truth; do not begin by changing names, adding difficulty, or copying a production task without a learning purpose.

The first useful result is a representative practice set: a sufficiently whole task family with a worked demonstration or explanation where useful, a first attempt, repetitions or variations selected for the practice purpose, target-domain success and critical-error criteria, permitted help, feedback and correction opportunities, and progression or return conditions. The set states what each task is intended to expose. It does not establish that the tasks were performed or that capability changed.

The practical gain is an interpretable attempt. A practitioner can tell whether the person failed to calculate, to recognize the applicable Method, to challenge an unsupported source, to coordinate a specialist contribution, or to act when the supplied evidence is sufficient. A warranted refusal and a supported continuation can both be correct in different cases.

Do not use HCD.6 to invent target-domain correctness, turn a worked answer into learner evidence, schedule all repetitions, provide unavailable teachers or tools, or assess later capability. HCD.5 supplies the selected development Method; HCD.7–HCD.8 supply needed support; HCD.9–HCD.10 enact and vary practice; HCD.11–HCD.13 interpret performance, transfer, and retention.

HCD.6:0.1 - Working Terms

Name in this pattern	What it denotes
target human action	The recognition, judgement, operation, coordination, or result-producing contribution the person is intended to practise under stated conditions.
representative practice task	A bounded task that preserves the later-Work relations needed to elicit the target action and interpret its result. Representation is relative to the contribution and claim; visual realism or a workplace label is not enough.
sufficiently whole task	A task containing the purpose, result recipient, decision-bearing conditions, required contributions, interfaces, and stopping or critical-error boundaries needed for the target action. It may omit detail that does not change that action.
decision-bearing condition	A fact whose value can change which Method applies, which action is warranted, or which result can be supported.
worked explanation	A demonstration that exposes the intermediate observations, comparisons, reasons, checks, and uncertainty needed for the target action rather than showing only the final answer.
answer-bearing cue	Information that supplies the Method choice, decisive relation, calculation, or conclusion the current attempt is meant to elicit from the person. A legitimate reference or tool is not answer-bearing merely because it helps.
permitted help	The references, tools, AI actions, prompts, peers, specialists, teacher moves, and other support allowed for one stated practice purpose. It remains distinct from help actually available and help actually received.
representative practice set	The task descriptions, conditions, materials, permitted help, criteria, variations, feedback opportunities, progression or return rules, and evidence-purpose limits needed for a coherent practice use.

A task description, learning product, performed attempt, work product, feedback event, and capability claim remain different objects. HCD.6 designs the task set; later Methods establish what was provided, performed, observed, and inferred.

HCD.6:1 - Problem Frame

A practice task can look realistic while testing the wrong contribution. A pump story can reduce to arithmetic when the target is causal discrimination. A ticket-queue story can repeat the same calculation even though the real difficulty is choosing the unit of account. An engineering release case can reward generic caution when a complete compatible report actually supports a bounded action.

Support changes the task. A worked example may help a novice see the reasoning, but leaving the same intermediate prompts in the next attempt can turn recognition into copying. Removing every source or tool can make the task unlike later Work and test memory that was never required. AI assistance can generate a plausible answer while hiding whether the person noticed a configuration mismatch.

The practitioner therefore has to preserve the relations that make the target action necessary and choose help for the current practice purpose. When recognition or adaptation is the target, vary a condition that changes the warranted action rather than just the scenery.

HCD.6:2 - Problem

The common shortcut is surface variation: change the industry, numbers, names, or diagram while leaving the same Method and answer obvious. Repetition may improve familiarity without showing whether the person can recognize applicability or adapt the Method.

The opposite shortcut maximizes authenticity. A full production case includes so many uncontrolled dependencies, missing inputs, risks, and specialist decisions that neither learner nor practitioner can tell which contribution changed. Complexity becomes noise rather than representation.

A third shortcut treats refusal as the safest universal answer. It can hide a real capability gap when sufficient evidence is present and action is required. The task set must include both missing-evidence and sufficient-evidence cases when the target contribution depends on that distinction.

HCD.6:3 - Forces

Whole tasks expose coordination and applicability, while irrelevant complexity can make the target action uninterpretable.
Worked explanations can reveal reasoning, while answer-bearing cues can conceal recognition and selection.
Legitimate references, AI, tools, and specialists belong to later Work, while their contributions must not be attributed to the person.
Familiar cases lower entry burden, while meaningful transfer requires changes that affect the decision rather than only the story.
Stable conditions help isolate one action, while varied conditions reveal whether the Method is selected and adapted.
A correct refusal protects against unsupported action, while habitual caution is wrong when the supplied evidence supports continuation.
Immediate correction supports another attempt, while feedback can teach, prime, fatigue, or otherwise contaminate later evidence.
Target-domain realism improves relevance, while professional correctness and hazards must come from qualified domain sources or explicitly stipulated teaching data.
A fixed challenge level is easy to schedule, while prior knowledge, task complexity, and support determine usable difficulty.

HCD.6:4 - Solution

Design backward from the target human action and the later-Work result it supports. Build one sufficiently whole task, expose relevant reasoning when useful, and choose the help, repetitions and variations that serve the present practice purpose. To observe independent recognition, selection or adaptation, withhold help that supplies the contribution being judged and change the case enough to distinguish that contribution from remembering the earlier answer.

HCD.6:4.1 - Bind the Practice Use and Domain Truth

Name:

the exact human holder or declared future-audience branch, participation basis, and practice recipient;
the later-Work family, target human action, selected HCD.5 Method, and intended practice result;
the target-domain source or stipulated teaching oracle that supplies correct facts, calculations, quality criteria, critical errors, hazards, and specialist boundaries;
current performance and prior knowledge relevant to task complexity and support;
permitted references, tools, AI, prompts, peers, teachers, specialists, time, environment, data, access, privacy, authority, and safety conditions; and
whether the current task is for explanation, first practice, correction, changed-condition practice, or a later observation.

If domain correctness, a necessary Method description, safe conditions, or a required provider contribution is absent, return the exact missing result. Do not repair the gap by simplifying the answer key, accepting generic caution, or treating fluent output as correct. For a future audience, keep task branches conditional on the entry evidence they are meant to distinguish.

HCD.6:4.2 - Build a Sufficiently Whole Task

Give the task a real purpose. State the situation, result recipient, decision or operation, intended work product, and what happens next when the result is acceptable.
Include the relations that elicit the target action. Supply the facts, comparisons, interfaces, alternative explanations, configuration identifiers, constraints, or specialist-return conditions that make the selected Method relevant.
Make the human contribution visible. State what the person must notice, choose, calculate, explain, adapt, coordinate, check, or return. Preserve what references, AI, tools, peers, and specialists may contribute.
Declare success and critical error before the attempt. Use target-domain criteria. Include warranted uncertainty and stopping conditions; a polished answer cannot compensate for a critical mismatch.
Remove irrelevant burden. Omit production detail that changes neither the target action nor its interpretation. When realism introduces another independently limiting contribution, prepare it, support it, or split the practice purpose.
Keep the result feasible. Fit the case, feedback, correction, and retry inside the declared learner and provider resources. A case too large for the available observation window returns for redesign rather than silently extending the programme.

Show what encompassing work is being performed through the target action during the attempt. A practice task can isolate a weak intermediate contribution, but it must also prepare the combination in which that contribution is needed. For a dancer whose strength and figure recall are already adequate, suitable reduced-demand practice can develop balance during rotation. Returning to the figure then tests coordination with its timing and other movements. Select variations that change that coordination; listing more figure names supplies no such test. State which assistance is available during practice and which will be available during later performance. FPF B.1.5.EW helps recover this vertical; the dance Method, not the framework example, determines the exercise and acceptable result.

When the task presupposes a distinction the learner cannot yet notice, HCD.6.1 helps construct an accessible encounter, vary a misleading cue and connect the noticed difference to action. Incorporate that attempt into the task and return to the whole contribution. A preparatory encounter need not remain a step of later performance.

HCD.6:4.3 - Separate Demonstration, First Attempt, and Help

A demonstration or worked explanation should expose the intermediate action the learner will later need: the cue noticed, Method selected, competing alternatives, calculation or transformation, check, remaining uncertainty, and next move. It should not rely on a final answer whose reasoning remains hidden.

For a first practice attempt, retain or supply worked support when it helps the person learn the target action. The learner may complete a partially worked task or repeat an operation before a new condition is useful. State what the learner performs and what the support supplies.

When the purpose is to observe independent recognition, selection or adaptation, change a condition that can expose that contribution. Remove the Method name, recommended option, decisive calculation or next move when the person must supply that value. Retain references, calculators, AI and specialist results compatible with the intended contribution; independence does not require removing all support.

Record permitted help before the attempt. Intended help in a programme, available help in the environment, and help actually received are separate. If a teacher or AI supplies an answer-bearing cue, keep the attempt as supported practice and do not reuse it as evidence of independent recognition.

HCD.6:4.4 - Design Decision-Bearing Variations

Choose variations from the relation the person must discriminate:

change a measurement or its uncertainty when competing explanations predict different observations;
change the unit of account when duplicate events, batching, or aggregation alters the workload calculation;
change a configuration identifier when evidence is valid for one configuration but not another;
supply a complete compatible report after an earlier missing-evidence case so the learner must act rather than repeat a refusal;
change available time, support, task similarity, or authority when those values change the warranted continuation; and
change the cue that signals which Method applies when recognition and selection are the target.

Select a variation when its intended contribution is needed. Use the smallest change that can distinguish recognition or adaptation from answer recall or generic caution; ordinary acquisition or correction can first use a repeated task. HCD.10 later chooses how variation, spacing, interleaving, progression, or support change should be distributed across practice episodes.

HCD.6:4.5 - Provide Feedback and Correction Opportunities

Link feedback to the criterion or action difference. State what remains correct, what must change, why it matters to the result, and what the learner should inspect or try next. A score or praise can summarize but cannot replace actionable information.

Let the learner use the feedback to correct the affected action, on the same case when that makes the correction learnable. For a subsequent attempt intended to distinguish recognition or adaptation from remembering the answer, change enough of the case to expose that difference. Recheck only the affected claims unless the new response gives reason to reopen others. If feedback revealed a different limiting target, return to HCD.5 or HCD.3 rather than adding repetitions under the old explanation.

Practice feedback and later assessment serve different uses. HCD.9 records the feedback and next attempt as practice history. HCD.11 decides what a condition-qualified performance attempt can support, including the effect of help and contamination.

HCD.6:4.6 - Record the Representative Practice Set

Result position	Required content
practice boundary	Human holder or future-audience branch, later-Work family, target action, selected Method, practice purpose, participation, conditions, horizon, and receiving use.
domain basis	Qualified source or stipulated oracle, required result, correct facts, quality criteria, critical errors, hazards, and outside-owner boundaries.
task family	Demonstration where used, first attempt, correction task, and repetitions or variations selected for the practice purpose; the action or relation each task is meant to expose.
human and support allocation	Human contribution, permitted references, tools, AI, prompts, peers, teachers, specialists, authority, and the required human check.
materials and work products	Data, descriptions, models, reports, calculations, decisions, explanations, or other artifacts needed and returned by each task.
criteria and feedback opportunity	Success, warranted uncertainty, critical error, stop conditions, feedback source and content, and retry condition.
progression or return	What observation permits less help, greater complexity, another variation, or later assessment; what returns to Method selection, diagnosis, support, domain truth, or task redesign.
evidence-purpose limit	Which attempt is demonstration, supported practice, independent practice, correction, or later observation, and which stronger learning, transfer, retention, or effectiveness claim remains unestablished.

The result is complete when another competent practitioner can present the intended task and help, identify the target action in the work product, apply the domain criteria, and choose the stated feedback, correction, progression, or return without reconstructing the design from private notes.

HCD.6:4.7 - Recognition and Assurance

Recognition. Use HCD.6 when practice feels busy or realistic but the practitioner cannot say which human action the task elicits, which condition changes the correct move, or what the learner must do without an answer-bearing cue.

Assurance. Stronger reliance needs a qualified target-domain oracle, safe and accessible conditions, explicit support allocation, criteria that discriminate warranted action from critical error, and a task relation suitable for the intended evidence claim. Domain experts retain correctness and hazard decisions; performed-attempt and capability claims require their own evidence.

HCD.6:4.8 - What Changes in Practice

A practitioner stops asking, “Can we make another realistic exercise?” and asks, “Which relation must this person notice and act on, what help is legitimate, which condition would change the warranted answer, and what result will let us give focused feedback?” The next case can require a justified refusal, a supported continuation, or a different Method rather than the same answer in new scenery.

HCD.6:5 - Archetypal Grounding

HCD.6:5.1 - Pump Explanation and a Pressure Change

The stipulated demonstration begins with a pump delivering 100 m³/h at 400 kPa and 20 kW. After a controller change, measured flow remains 100 m³/h and power becomes 22.4 kW, while comparable pressure data are absent. The worked explanation calculates the 12% rise in energy per volume and keeps resistance, regulation efficiency, and measurement error as rivals.

The task supplies the domain comparison: at the same flow and previous efficiency, increased pressure requires greater power; reduced regulation efficiency can also increase electrical demand. A pressure value near 448 kPa would support increased resistance under those conditions, while a value near 400 kPa would direct attention toward regulation efficiency. Neither value establishes a unique cause by itself.

The controller log and four specialist-hours are available. The supplied complete measurement contributions cost one hour for independent power, one and a half for flow, and one and a half for pressure; they include safe preparation, measurement and result return. A day-long shutdown is outside the allowed arrangement. The learner has ninety minutes to justify the observations and next decision, including explanation and comparison. For the selected pressure branch, changing the receiving engineer’s next diagnostic question warrants the remaining specialist contribution and the learner’s effort, including displaced work. Selection does not supply a pressure result.

The first attempt receives independent measurements of 22.3 kW and 99.5 m³/h, with uncertainties of 0.2 kW and 1 m³/h respectively. Ask the person to compare like quantities and use both uncertainty bounds. Even the lowest new ratio, 22.1/100.5, exceeds the stipulated baseline 20/100; this weakens the measurement-error explanation under the supplied conditions. After the power and flow contributions, one and a half specialist-hours remain for pressure. The learner can select that observation and explain how its possible values change the resistance/efficiency comparison while keeping the cause unresolved.

The target action is conditional explanation and a present response suited to the available basis, not arithmetic alone. In a later variant without a worthwhile obtainable observation, the learner can finish the qualified explanation with its live rivals and return that account while leaving a resistance-specific remedy unselected. A unique cause remains unproved. A renamed machine with the same numbers would add no useful variation; neither does a generic request to collect more data.

HCD.6:5.2 - Ticket Load after Duplicate Removal

Four people each have six active-processing hours per day. Incoming records rise from 50 to 80 and each distinct request takes 18 minutes. The initial worked case shows that 50 and 80 distinct requests require 15 and 24 person-hours; the observed queue alone does not establish lower individual productivity.

In a stipulated variation, the 80 records include 20 duplicate messages attached to existing requests, leaving 60 distinct requests. The target action is to recognize that the unit of account changed before calculating 18 person-hours of active processing.

Now leave the duplicate count unknown and ask whether the day’s requests fit the stated active-processing capacity. There can be no more than 80 distinct requests: at 18 minutes each, they require at most 24 person-hours, equal to the four people’s supplied capacity. Return that sufficient upper bound and keep the exact count unknown. The answer concerns active effort under the stipulated processing assumptions; it establishes no queue-clearance, response-time or safety guarantee. Obtaining the count is unnecessary for this answer.

Change the receiving use to an allocation that needs the exact active-processing load of these requests. The upper bound no longer supplies that quantity. Stipulate an authorized existing request view and five minutes of available coordinator time to open it, resolve the duplicate labels and check the total. Correcting the allocation warrants that obtainable burden. Obtain the count for this use; until it is available, retain the upper bound with the exact allocation unresolved. If the view supplies 60 distinct requests, the processing load is 18 person-hours. The five minutes of coordination are an additional cost, not request-processing time or free work.

These cases distinguish a changed unit, a sufficient bound and a receiving use that needs another quantity. Merely replacing the support team with a warehouse would not test those distinctions.

HCD.6:5.3 - FW7 Refusal and FW8 Continuation

The first engineering case supplies a vendor report for FW7 while the intended release configuration requires HW3/FW8/SCHED2. The correct action is to identify the mismatch and withhold that use; a correct financial calculation cannot repair incompatible evidence.

The varied case supplies a report for HW3/FW8/SCHED2, confirms mandatory checks, and leaves 24 test hours. Ten changes in three homogeneous batches require 23 hours. The person should propose the bounded ten-change use and preserve the continuing queue when demand is twelve. Repeating the earlier refusal is now wrong because compatible evidence is present. When available time falls to 22 hours, promising ten changes becomes unsupported again.

This task family makes evidence applicability and the resource calculation jointly visible. It tests a warranted refusal, a supported continuation, and a changed constraint rather than rewarding universal caution.

HCD.6:5.4 - AI-Supported Practice without Attribution Drift

AI may propose sources, calculations, alternatives, or draft wording in a practice task. The task description states which actions the person still performs: checking decisive references, configuration, assumptions and calculations; selecting the Method; explaining the result; and returning unsupported claims. A later attempt may remove AI for a narrow independent contribution while retaining supplied tables and a calculator. The supported and changed-support attempts remain separate in the practice history; neither is silently upgraded to transfer or retention evidence.

HCD.6:5.5 - Correct the Same Action before Adding a New Condition

A trainee must enter a measurement and its unit in an inspection record. The teaching data give 2.4 mV; the criterion is to preserve both the number and unit. Design the first task with a worked row available if the trainee needs it. If the trainee writes 2.4 V, feedback points to the unit on the supplied reading and asks the trainee to correct the entry. The same reading is suitable for this correction: changing it would add a different demand before the present action has been repaired.

That corrected entry is a useful supported practice result. When the later question is whether the trainee independently preserves units on a new reading, supply another reading and withhold the answer-bearing worked row. If later work also requires unit conversion, a separate variant can ask for 0.0024 V to be entered in a column marked mV, whose correct value is 2.4. Select that variant only for the conversion target. Each task keeps its own help and conclusion; the design alone establishes no learner performance.

HCD.6:6 - Bias-Annotation

Lenses: Gov, Arch, Onto/Epist, Prag, Did. Scope: representative practice-task design for a named human contribution or an explicitly conditional future-audience branch.

Recurring bias	Likely drift	Repair
surface-realism bias	A workplace story or visual detail is treated as representation.	Preserve the relations and conditions that elicit and discriminate the target action.
difficulty bias	More complexity is assumed to create more learning.	Remove irrelevant burden and vary only what changes the Method, action, or interpretation.
cue blindness	Method names, intermediate prompts, or AI output supply the answer being tested.	Declare permitted help; remove the answer-bearing cue when the attempt must elicit that contribution independently.
tool-removal bias	All references and tools are removed in the name of independence.	Retain legitimate later-Work supports and isolate only the human contribution the current task targets.
caution bias	Refusal is accepted in every uncertain case.	Include sufficient-evidence cases that require bounded action as well as genuine missing-evidence stops.
answer-key bias	The final answer is checked while the reasoning, source applicability, or critical error remains hidden.	Require the decision-bearing intermediate action and use domain criteria.
production-copy bias	A real case is copied without controlling risk, access, or confounding contributions.	Build a sufficiently whole bounded task and return professional facts to qualified owners.
assessment leakage	A supported or corrected practice attempt is reported as independent capability.	State each task’s evidence purpose and preserve help and contamination history.

HCD.6:7 - Conformance Checklist

The human holder or future-audience branch, later-Work family, target action, selected Method, practice purpose, and receiving use are explicit.
A qualified target-domain source or stipulated teaching oracle supplies facts, calculations, result criteria, critical errors, hazards, and specialist boundaries.
The task is sufficiently whole for the target action but omits complexity that changes neither action nor interpretation.
The result recipient, decision or operation, required work product, and next move are visible.
Human, reference, tool, AI, peer, teacher, specialist, environment, permission, and authority contributions remain distinct.
Permitted help is stated before each attempt, and answer-bearing cues are removed only where the target requires it.
A demonstration, when used, exposes intermediate reasoning rather than only the final answer.
Help and repetition fit the current practice purpose. When recognition, selection or adaptation is the target, the set includes a condition change that can expose that contribution.
The set includes warranted refusal and supported continuation when that distinction belongs to the target.
Feedback names the affected criterion and next action. Same-case correction is distinguished from a later changed-case observation of recognition or adaptation.
Learner and provider resources include task, feedback, correction, and retry; an oversized case returns for redesign.
The result states progression or return conditions and distinguishes demonstration, supported practice, independent practice, correction, transfer, retention, and effectiveness uses.

HCD.6:8 - Common Anti-Patterns and How to Avoid Them

Anti-pattern	Why it fails	Repair
“Change the pump to a ticket queue”	New nouns can preserve the same cue and answer.	Change a relation or condition that alters the warranted Method or result.
“Use a real project”	Production realism can add uncontrolled risk, access limits, and other human or specialist contributions.	Bound the task around the target action and obtain qualified domain and safety inputs.
“Repeating the shown answer demonstrates adaptation”	The repeated answer may depend on memory of that case.	Keep its legitimate practice use; test adaptation through a changed condition without supplying the answer being judged.
“No help means independent”	Removing legitimate references or tools can test a different contribution from later Work.	Specify the exact human contribution and retain compatible support conditions.
“Always refuse if evidence is uncertain”	Generic caution fails when sufficient compatible evidence supports bounded action.	Pair missing-evidence and sufficient-evidence variants with explicit criteria.
“The learner got it after feedback”	Corrected practice is contaminated by the feedback event.	Record help and test the affected action on a changed task before stronger inference.
“Harder is more representative”	Difficulty without target relevance raises burden and obscures the result.	Scale complexity to current knowledge and the action being elicited.
“The answer key defines professional truth”	An invented key can encode unsafe or wrong domain guidance.	Obtain a qualified domain source or label stipulated teaching facts and their limits.

HCD.6:9 - Consequences

The practitioner gains tasks that make a selected human action observable and correctable without pretending to be full production Work. Meaningful variations expose applicability, source compatibility, resource limits, and the difference between warranted refusal and supported continuation.

The cost is domain and provider work. Someone must qualify the facts and hazards, prepare cases, expose intermediate reasoning, review attempts, and prepare the variations needed to distinguish recognition or adaptation from answer recall. A task may return unfinished when correctness, access, safe conditions, or feedback capacity is absent.

HCD.6:10 - Rationale

A practice task is representative relative to the target contribution and intended evidence claim. It preserves the relations needed for that contribution and use, not every detail of the workplace. This permits bounded safe practice while protecting domain truth.

Supported acquisition and correction can use the same task; a changed-condition attempt answers a different question when recognition or adaptation is required. Separating those uses keeps the learning contribution of help visible while preserving the limits of the resulting evidence. Designing both refusal and continuation cases prevents a safety-oriented boundary from becoming inert caution. The resulting work products give HCD.9 a focused practice object and HCD.11 an interpretable attempt without making their later conclusions in advance.

HCD.6:11 - SoTA-Echoing

The working question is: How should a practitioner design practice tasks that elicit a target human contribution and support meaningful correction and variation without mistaking surface realism for representation?

The selected best-known line combines sufficiently whole complex tasks, expertise-sensitive worked support, and decision-bearing variation. The maintained 4C/ID model and its fourth-edition presentation supply the serious whole-task design candidate: learning tasks, support for non-routine reasoning, timely procedural information, and additional practice for recurrent parts. Paas and van Merriënboer’s 2020 review (COG-05) supports worked examples and integrated information for novices while warning that unchanged guidance can become redundant as task knowledge grows. HCD.6 adapts these contributions into §§4.2–4.3 and the separation of whole-task and focused-part practice. The amount of help and the use of a repeated or varied task follow the learning purpose and starting capability; the sources do not prescribe a changed condition for every correction.

The serious ordinary alternative is a topic exercise or surface-isomorphic scenario: present the rule, vary the nouns and numbers, and score the final answer. It is cheaper to produce, but it can preserve answer-bearing cues, omit the later-Work purpose and interfaces, and reward either calculation without judgement or generic caution. HCD.6 rejects surface change as representation and adds the decision-bearing variation, human/support allocation, domain criteria, and correction task in §§4.1–4.6.

Brunmair and Richter’s 2019 meta-analysis (COG-03) supplies limit evidence that interleaving varies by material and similarity; it does not prescribe one mixed task set. Czyż, Wójcik, and Solarská’s 2024 motor-learning review (COG-04) supports testing delayed transfer while its applied estimate and cross-domain reach remain weak. Bastani and colleagues’ 2025 school-mathematics trial (AI-04) supports separating assisted practice from a later unassisted attempt; it does not supply an adult engineering effect or a universal AI ban. These sources change the task contrasts and evidence-purpose boundary rather than selecting one universal schedule.

The selected line costs more domain and case-design effort than renaming exercises, but it preserves action interpretability, legitimate support, critical errors, and the opportunity to learn from a changed result. Reopen it when stronger sources alter the whole-task, worked-support, variation, AI-help, or transfer boundary, or when representative use shows a lower-effort task design that preserves the same decision-bearing relations and evidence value.

HCD.6:12 - Relations

HCD.1, HCD.3, and HCD.4: can supply the later-Work demand, qualified target, current evidence, protected strengths, and priority conditions.
HCD.2: can expose a programme component and task closure needing elaboration; HCD.6 does not select the programme.
HCD.5: supplies the selected development Method, proposed mechanism, learner/task fit, support need, and evidence plan.
HCD.7: establishes whether the required provider, source, tool, AI, access, environment, and human-check contributions are available.
HCD.8: builds a missing task-production, provider, assessment, or practice-environment capability only when the required result is explicit.
HCD.9: enacts the tasks, feedback, correction, and retries while recording the practice history.
HCD.10: selects variation, spacing, interleaving, progression, or support change across practice episodes; HCD.6 supplies the meaningful task conditions.
HCD.11–HCD.13: interpret condition-qualified performance, transfer, retention, and tool dependence; a task design or answer key is not those observations.
HCD.14: can revise the affected task, criterion, help, or progression condition from compatible evidence.
LG.05 and other learning-product practices: can explain and package a selected task; pedagogical presentation does not select the target relation or establish professional correctness.
Target-domain, safety, clinical, legal, employment, provider, privacy, and specialist practices: retain correctness, hazard, participation, authority, and consequence decisions.