Final Unknown System

A previously unseen inherited system, no instructor framing of what matters, and a short live defence of the entire evidence trail.

Due: 21 May 2027 · Weight: 35%

Here is a system you have never seen, and a report of what a user observed. Investigate it, fix what needs fixing, and be ready to explain every step.

This is the course’s integrative assessment. You receive a system consistent with Weeks 10–11’s conditions — multi-component, incompletely documented, and containing at least one source of operational uncertainty — with a failure report but no framing of which subsystem, evidence class, or tool is likely to matter. It is not Broken Service Investigation at a larger scale: the difference is the absence of cueing, not the size of the repository.

What you submit#

An early checkpoint of your initial observations and planned next step; a full evidence trail; an explicit account of any evidence source you judged unreliable and why; a diagnosis that addresses alternatives; a scoped intervention with a verification argument tied to the diagnosed cause; a statement of remaining uncertainty; and a short live defence in which a questioner probes your observations, hypotheses, evidence choices, exclusions, intervention scope, and verification — the questions are not supplied in advance.

How it’s marked#

Reaching a working final state through undirected changes, without being able to explain why alternatives were excluded or why the verification demonstrates the diagnosed cause is resolved, does not earn a high mark on that basis alone. A rigorous investigation with a justified, safely-scoped intervention can still score well if genuine system complexity prevents a complete repair — provided the remaining uncertainty is stated plainly rather than hidden. Defence carries the single largest weight, because it tests whether the investigation is an argument you understand, not a narrative assembled after the fact.