DS-A · Library entry
A fictional college is testing a card-based library entry system. The system reads a card ID, retrieves its stored similarity score from a classroom test file, and accepts entry when score ≥ threshold. The score is a made-up diagnostic value, not the probability that a person has a valid card. Staff can manually check rejected visitors.
The table’s known status was assigned by the test designer. This tiny balanced sample does not estimate real-world accuracy or represent a population.
| Test ID | Known status | Score |
|---|---|---|
| A1 | Valid card | 0.91 |
| A2 | Valid card | 0.78 |
| A3 | Valid card | 0.64 |
| A4 | Invalid card | 0.83 |
| A5 | Invalid card | 0.67 |
| A6 | Invalid card | 0.3 |
Source A · Developer note
Fictional internal pilot note, 10 September 2026: “The threshold can be changed. Higher thresholds reject more attempts. Our six-case test is a starting diagnostic and does not show performance for the whole campus.”
Author: development team. Purpose: explain the prototype. Evidence: the supplied small test. Interest: demonstrating a workable system.
Source B · Access committee note
Fictional meeting note, 11 September 2026: “Rejected legitimate users may wait for staff. The pilot should record queue time and manual-check outcomes before any wider rollout.”
Author: access committee. Purpose: examine access costs. Evidence: anticipated risks, with no measured queue data yet.
Your individual response
- Trace the input, processing, output and human decision at a threshold of 0.75. Identify which valid and invalid cards are accepted or rejected.
- Compare the consequences at thresholds 0.60 and 0.85. Explain one cost to a legitimate user and one concern for library staff.
- Use both sources. Identify one useful contribution and one limitation of each. Do not describe the synthetic notes as independent research findings.
- Recommend a provisional threshold or a pilot condition. Support the judgment, identify a trade-off, and name the evidence that would change your recommendation.
Answer in clear prose, an annotated diagram, or an agreed oral/recorded explanation that preserves the same reasoning demand.
Teacher follow-up
“Suppose the score of A3 changes from 0.64 to 0.76. What changes at threshold 0.75, and what does not change?”
DS-B · Equipment checkout
Use after targeted practice. A fictional equipment-loan desk uses the same kind of scored classroom rule to approve a checkout. A staff member can review rejected requests. Use the same question structure, timings and rubric, but use this table and threshold 0.70. Compare 0.55 and 0.80.
| Test ID | Known status | Score |
|---|---|---|
| B1 | Eligible request | 0.88 |
| B2 | Eligible request | 0.73 |
| B3 | Eligible request | 0.59 |
| B4 | Ineligible request | 0.78 |
| B5 | Ineligible request | 0.62 |
| B6 | Ineligible request | 0.25 |
Source A: a fictional checkout developer says the six-case sample demonstrates how the adjustable rule works, not population performance. Source B: a fictional student equipment committee asks the pilot to measure waiting time, missed access and manual-review outcomes. Each source has the same role and evidential limitation as in DS-A. Follow-up: change B3 to 0.71 and ask what changes at 0.70.
Scoring and timing
5 minutes to establish access; 10 to model an unrelated simple rule; 10 guided tracing; 25 individual response; 12 minutes of rolling teacher follow-ups; 8 minutes to record the next action. Finish the observation rotation in subsequent lessons. Prior subject knowledge and reading speed are context, not bonus points.
| Criterion | 1 | 2 | 3 | 4 |
|---|---|---|---|---|
| Mechanism | The input-to-decision explanation is missing or reverses the stated rule. | Some steps are correct but a causal link requires a prompt. | The input, rule, output and human decision point are explained accurately. | A changed rule is traced accurately and the explanation identifies the model limit. |
| Source use | A claim is repeated without identifying supporting source information. | A source is identified but credibility or limitation is generic. | A specific source feature supports the claim and a relevant limitation is identified. | The sources’ different contributions and limitations are weighed, and a further evidence need is justified. |
| Stakeholder reasoning | An effect is named without connecting it to the digital decision. | A simple consequence is partly connected to the mechanism. | A mechanism feature is linked to different consequences for two stakeholders. | A changed threshold is used to explain redistribution of benefit and error costs. |
| Qualified judgement | A preference is asserted without evidence or a reason. | A reason is present but little case evidence or qualification is used. | The judgement uses case evidence and a condition that could change it. | Competing priorities are weighed and a plausible alternative is tested under a changed condition. |
Use the same missing-evidence, support and teacher-verification rules as the Photoshop diagnostic. The four local ratings total 4–16 when complete. They do not convert to an official IB grade.