8 The Flight Simulator: Bounded Conversation Practice

A simulation is useful when its limits are visible and its feedback can be checked.

8.1 Brief, practise, debrief

A conversation simulation gives students a chance to try questions, encounter resistance and revise their approach. It can supplement peer role-play and observed practice. It is not a faithful model of every workplace or a substitute for disciplinary teaching.

Use three phases: establish the case and boundaries, practise the interaction, then review the decisions against evidence. A model can play a role or propose feedback, but the educator retains responsibility for the scenario and criteria.

The critique toolkit comes first because learners need a way to evaluate feedback before receiving it. A confident critique is itself an output requiring review.

8.2 Prepare the scenario

Use P1-D1 through P1-D5 in the pilot pack. Alex’s manager has proposed a performance plan. The junior adviser is gathering information, not issuing an approved decision. The pack supplies the dates, records and limits of authority.

Explain the capability: plan and justify a fair first response. If your unit also assesses live communication, add an appropriate observed interaction. Do not infer that skill solely from a typed conversation.

Give advance notice of the employment topic. Students can stop, take a break or use an agreed neutral scenario. The task does not require personal disclosure. The model or peer actor should not pressure the learner, introduce abuse or reward extraction of private details.

8.2.1 Setup for a peer or approved model

Use the fictional PDP-2026.1 P1 source cards.
You are Alex. Use P1-D4 as your opening position.
Do not invent evidence or legal obligations.
If asked about missing facts, say they are not
provided. You may disagree or ask a question.
You need not disclose private information.
Stop after six exchanges or when I say STOP.
Do not score the adviser or decide the case.

In the paper route, the fixed exchange below provides the role’s responses. In a paired route, the partner reads the same cards. With a model, check any new claim against the pack and discard inventions. Similar instructions do not guarantee similar difficulty across products or runs.

8.3 A worked exchange

This dialogue is an authored fictional example, not a captured student performance or evidence that a model behaves consistently.

Adviser: Alex, the delivery extract lists three late milestones. Before I recommend a next step, I want to understand the dependencies and your account. The formal plan is a proposal at this stage.

Alex: I know some work was late. Two projects depended on reviews I could not complete myself. I also need to discuss starting later on two mornings.

Adviser: Which delivery records would help us check the dependencies? We can discuss the work adjustment without collecting private family or medical details. I can refer the request, but cannot approve it myself.

Alex: I do not want to explain the appointments. I also want to know whether this meeting means the plan is already decided.

Adviser: You do not need to explain those details to me. The decision is not made. I propose checking the records and the earlier contact note, then arranging a follow-up after the people manager reviews the request.

Alex: Who will receive the notes?

Adviser: I will explain the authorised sharing arrangements before sharing the relevant work-related record. I should not promise complete confidentiality without checking the process. We can pause if that needs clarification first.

The exchange ends with work to do. That is appropriate: the missing evidence has not become available merely because the conversation went well.

8.4 Inspect the decisions

The opening distinguishes the delivery record from a conclusion about fault. The adviser does not invent the earlier warning, refers the support request within authority and, in the final answer, avoids a confidentiality guarantee.

There are also limits. No dependency record has been examined, no support arrangement approved and no finding reached. A reader should not label the case resolved or legally compliant.

Try a different opening: “Your manager warned you in April, so the plan is now required.” The error is not just tone. P1-D3 does not supply an April warning, and P1-D1/D5 do not permit the adviser to announce the decision. Rephrasing it more empathetically would leave the substantive defect intact.

8.5 Debrief before scoring

Ask the learner to select two decisions and explain:

What evidence was available.
What they chose to do.
Why that was within the task’s criteria.
What they would check next or change after new information.

A useful model-feedback request is deliberately bounded:

Review the selected response using the supplied
P1 cards and five rubric criteria.
Quote the response accurately.
For each comment, identify a card or criterion.
Separate observed wording from inferred outcomes.
Flag insufficient evidence.
Do not give legal assurances or predict Alex's
real-world behaviour. Your feedback is provisional.

The learner checks the feedback rather than submitting it as the final word. A rubric prompt does not ensure accurate application of the rubric.

The pack includes P1-F, an intentionally flawed simulated critique awarding 90% to weak work. The educator correction scores that response 4/20 against the supplied criteria. The traceable correction matters more than a claim that one human score is universally correct. Marking teams should calibrate disagreements using the evidence.

8.6 Rehearsal and transfer are different

Repeating P1 can help a learner practise an improved response. It can also teach the learner to reproduce a familiar script. Follow it with P2, where the manager’s alleged retaliation changes the referral decision.

Compare the explanations, not the number of attempts or the model’s willingness to cooperate. A respectful question may still receive resistance. A flawed question may receive a friendly response. Persona behaviour is not a scoring key.

For solo practice, write a response to Alex’s question about the notes, compare it with D1/D5, and explain a limitation. For paired practice, swap adviser and actor roles, then compare decision notes. Both can prepare the same target capability without a model.

8.7 Three ways to use the activity

Practice only: students rehearse, identify one checked improvement and retain their notes privately. No model score enters the grade.

Guided formative task: students submit two selected decision extracts and a short revision explanation. The educator gives feedback against the rubric.

Assessment component: use published criteria, appropriate data handling, an independent demonstration and moderation. See the assessment chapter. Do not convert a practice tool into a high-stakes assessment merely by assigning a percentage.

Do not require all private attempts “to prevent gaming”. Collect only the evidence necessary for the stated assessment purpose under the institution’s retention rules. Process records can be fabricated and should not be treated as conclusive proof of independent performance.

8.8 Adapt the challenge, not just the persona name

For a supplier conversation, give the actor the S1 capacity limit and a clear authority boundary. A confident promise to supply 200 units is not supported when the offer says 160. Practise asking for clarification and explaining a constraint.

For a marketing discussion, use M1’s audience and cost limitations. The actor can ask why the learner refuses to call the campaign more profitable. The learner should distinguish a calculable rate from an unsupported business claim.

Those adaptations change the evidence and criteria. They are more instructive than replacing “employee” with “customer” in an unchanged HR script.

8.9 Activity: debrief an unresolved exchange

Allow 12 minutes; paper or paired route available. Use the worked exchange, stop before the final adviser response and write your own. Explain what you would record, what permission or process you must check and why the outcome remains open.

A sound response stays within authority, avoids unnecessary disclosure and identifies a next step. It need not secure Alex’s agreement or close the case.

# The Flight Simulator: Bounded Conversation Practice {#sec-simulator} > A simulation is useful when its limits are visible and its feedback can be checked. ## Brief, practise, debrief A conversation simulation gives students a chance to try questions, encounter resistance and revise their approach. It can supplement peer role-play and observed practice. It is not a faithful model of every workplace or a substitute for disciplinary teaching. Use three phases: establish the case and boundaries, practise the interaction, then review the decisions against evidence. A model can play a role or propose feedback, but the educator retains responsibility for the scenario and criteria. The [critique toolkit](critique-toolkit.qmd) comes first because learners need a way to evaluate feedback before receiving it. A confident critique is itself an output requiring review. ## Prepare the scenario Use P1-D1 through P1-D5 in the [pilot pack](../appendices/pilot-pack.qmd). Alex's manager has proposed a performance plan. The junior adviser is gathering information, not issuing an approved decision. The pack supplies the dates, records and limits of authority. Explain the capability: plan and justify a fair first response. If your unit also assesses live communication, add an appropriate observed interaction. Do not infer that skill solely from a typed conversation. Give advance notice of the employment topic. Students can stop, take a break or use an agreed neutral scenario. The task does not require personal disclosure. The model or peer actor should not pressure the learner, introduce abuse or reward extraction of private details. ### Setup for a peer or approved model ```text Use the fictional PDP-2026.1 P1 source cards. You are Alex. Use P1-D4 as your opening position. Do not invent evidence or legal obligations. If asked about missing facts, say they are not provided. You may disagree or ask a question. You need not disclose private information. Stop after six exchanges or when I say STOP. Do not score the adviser or decide the case. ``` In the paper route, the fixed exchange below provides the role's responses. In a paired route, the partner reads the same cards. With a model, check any new claim against the pack and discard inventions. Similar instructions do not guarantee similar difficulty across products or runs. ## A worked exchange This dialogue is an **authored fictional example**, not a captured student performance or evidence that a model behaves consistently. **Adviser:** Alex, the delivery extract lists three late milestones. Before I recommend a next step, I want to understand the dependencies and your account. The formal plan is a proposal at this stage. **Alex:** I know some work was late. Two projects depended on reviews I could not complete myself. I also need to discuss starting later on two mornings. **Adviser:** Which delivery records would help us check the dependencies? We can discuss the work adjustment without collecting private family or medical details. I can refer the request, but cannot approve it myself. **Alex:** I do not want to explain the appointments. I also want to know whether this meeting means the plan is already decided. **Adviser:** You do not need to explain those details to me. The decision is not made. I propose checking the records and the earlier contact note, then arranging a follow-up after the people manager reviews the request. **Alex:** Who will receive the notes? **Adviser:** I will explain the authorised sharing arrangements before sharing the relevant work-related record. I should not promise complete confidentiality without checking the process. We can pause if that needs clarification first. The exchange ends with work to do. That is appropriate: the missing evidence has not become available merely because the conversation went well. ## Inspect the decisions The opening distinguishes the delivery record from a conclusion about fault. The adviser does not invent the earlier warning, refers the support request within authority and, in the final answer, avoids a confidentiality guarantee. There are also limits. No dependency record has been examined, no support arrangement approved and no finding reached. A reader should not label the case resolved or legally compliant. Try a different opening: “Your manager warned you in April, so the plan is now required.” The error is not just tone. P1-D3 does not supply an April warning, and P1-D1/D5 do not permit the adviser to announce the decision. Rephrasing it more empathetically would leave the substantive defect intact. ## Debrief before scoring Ask the learner to select two decisions and explain: - What evidence was available. - What they chose to do. - Why that was within the task's criteria. - What they would check next or change after new information. A useful model-feedback request is deliberately bounded: ```text Review the selected response using the supplied P1 cards and five rubric criteria. Quote the response accurately. For each comment, identify a card or criterion. Separate observed wording from inferred outcomes. Flag insufficient evidence. Do not give legal assurances or predict Alex's real-world behaviour. Your feedback is provisional. ``` The learner checks the feedback rather than submitting it as the final word. A rubric prompt does not ensure accurate application of the rubric. The pack includes P1-F, an intentionally flawed simulated critique awarding 90% to weak work. The educator correction scores that response 4/20 against the supplied criteria. The traceable correction matters more than a claim that one human score is universally correct. Marking teams should calibrate disagreements using the evidence. ## Rehearsal and transfer are different Repeating P1 can help a learner practise an improved response. It can also teach the learner to reproduce a familiar script. Follow it with P2, where the manager's alleged retaliation changes the referral decision. Compare the explanations, not the number of attempts or the model's willingness to cooperate. A respectful question may still receive resistance. A flawed question may receive a friendly response. Persona behaviour is not a scoring key. For solo practice, write a response to Alex's question about the notes, compare it with D1/D5, and explain a limitation. For paired practice, swap adviser and actor roles, then compare decision notes. Both can prepare the same target capability without a model. ## Three ways to use the activity **Practice only:** students rehearse, identify one checked improvement and retain their notes privately. No model score enters the grade. **Guided formative task:** students submit two selected decision extracts and a short revision explanation. The educator gives feedback against the rubric. **Assessment component:** use published criteria, appropriate data handling, an independent demonstration and moderation. See the [assessment chapter](assessment.qmd). Do not convert a practice tool into a high-stakes assessment merely by assigning a percentage. Do not require all private attempts “to prevent gaming”. Collect only the evidence necessary for the stated assessment purpose under the institution's retention rules. Process records can be fabricated and should not be treated as conclusive proof of independent performance. ## Adapt the challenge, not just the persona name For a supplier conversation, give the actor the S1 capacity limit and a clear authority boundary. A confident promise to supply 200 units is not supported when the offer says 160. Practise asking for clarification and explaining a constraint. For a marketing discussion, use M1's audience and cost limitations. The actor can ask why the learner refuses to call the campaign more profitable. The learner should distinguish a calculable rate from an unsupported business claim. Those adaptations change the evidence and criteria. They are more instructive than replacing “employee” with “customer” in an unchanged HR script. ## Activity: debrief an unresolved exchange **Allow 12 minutes; paper or paired route available.** Use the worked exchange, stop before the final adviser response and write your own. Explain what you would record, what permission or process you must check and why the outcome remains open. A sound response stays within authority, avoids unnecessary disclosure and identifies a next step. It need not secure Alex's agreement or close the case.