Rubric — Assignment 1 (20 marks)

Applies to Assignment 1. Marks are per criterion; use the band descriptors, not gut feeling. Award half-bands where genuinely between.

Task 1 — Decomposition (5)

Band	Descriptor
5	Tree covers the whole task; every leaf is genuinely small; assumptions are stated, specific, and marked for impact; the marked assumption is defended.
4	Tree complete but one leaf is really two subproblems, or one assumption is vague ("assume normal things").
2–3	Major leaf missing (e.g. the bonus decision) or assumptions absent; the shape is visible but not executable.
0–1	A single undecomposed step, or the tree restates the scenario in other words.

Task 2 — Rules as pseudocode (6)

Band	Descriptor
5–6	All rules present and ordered correctly (discount applies to the whole purchase; bonus requires ≥ 2 purchased); boundary values right (5 credits, 8 attendances); outputs named; every purchase-size answer traceable to the rules.
3–4	Rules present but one boundary wrong, or the discount applied per-credit instead of per-purchase, or bonus tested before the purchase size is known.
1–2	Rules ambiguous where the scenario was precise, or Python used but doing the thinking invisibly.
0	The scenario restated; no executable precision.

Task 3 — Decision and flowchart (5)

Band	Descriptor
5	Flowchart reaches one recommendation; every branch is reachable; the contested branch is annotated with a genuine two-sided disagreement.
4	Correct chart; the "contested" branch annotation is decorative (no real alternative given).
2–3	Chart has an unreachable branch or a missing exit path, or the recommendation contradicts the task-2 rules.
0–1	No chart, or arrows into nowhere throughout.

Task 4 — The hostile friend (4)

Band	Descriptor
4	A concrete input where intuition and rules diverge; both answers derived; the rule change that would flip the verdict identified.
3	Concrete divergence found but the rule change is vague.
1–2	Asserts the friend is wrong/right without a concrete input.
0	Not attempted.

Cross-cutting adjustments

−1 (once) if assumptions appear after results that depend on them.
+0 marks for extra Python: code is accepted as an appendix, never as a substitute for pseudocode precision.
Submissions violating the length guide by more than double are returned for one resubmission at 90% cap.