Sessions
The twelve-week teaching schedule: what happens in each session, what to prepare, and what leaves the room.
Twelve weeks, each opening from a specific failure rather than a topic. Scaffolding is heaviest in Weeks 1–2 and withdraws steadily from there; Week 12 has none. Titles below are questions, not chapter headings — the setting changes week to week, but the six-stage investigative cycle does not. Incident details are deliberately withheld here; seeCourse Overview for the method itself.
- Weeks 1–2Modelled cycle
- Weeks 3–5Wider evidence
- Weeks 6–7Competing hypotheses
- Weeks 8–9Intervention & verification
- Weeks 10–11Independent investigation
- Week 12Defence
Naming What You See
What is the difference between an observation and an assumption? The semester's first session works on a system small enough to read in one sitting, so the discipline — not the difficulty — is the point.
Week 1 · 22 February 2027
A Hypothesis That Could Be Wrong
What makes a hypothesis useful, as opposed to merely plausible? A reproducible failure with two similar-looking causes forces a real choice, and the choice has to be committed to before it's tested.
Week 2 · 1 March 2027
Reading a System You Weren't Given a Map To
How do you build a working model of a system's intended behaviour when no one is going to explain it to you? Documentation stops being a reliable evidence source from this session onward.
Week 3 · 8 March 2027
Evidence That Isn't Static Text
What can runtime evidence tell you that reading the code cannot? The fault this week only shows up under specific runtime conditions, so evidence has to be generated, not just read.
Week 4 · 15 March 2027
What the System Used to Be
Can a fault be explained by something that is no longer in the code at all? This session treats version history and prior configuration as first-class evidence, not background.
Week 5 · 22 March 2027
Two Explanations, One Fault
How do you choose between two hypotheses that both fit the evidence you have so far? This session's failure has two genuinely plausible causes, and stopping at the first one that fits is the mistake being taught against.
Week 6 · 29 March 2027
When the Evidence Disagrees
What do you do when two pieces of evidence appear to contradict each other? This session's evidence itself is unreliable in places, which is a different problem from an unresolved hypothesis.
Week 7 · 12 April 2027
The Fix That Isn't Safe
How do you decide whether an available fix is acceptable, not just whether it makes the symptom go away? The diagnosis this week is deliberately easy to reach; the risk is entirely in the fix.
Week 8 · 19 April 2027
Verifying You Didn't Just Move the Problem
How do you know an intervention actually resolved the cause, rather than relocating or masking the symptom? This session's fix looks like it worked — closer verification is what the week is actually testing.
Week 9 · 26 April 2027
No One Left to Ask
How do you investigate a system when the people who built it, and much of its context, are simply gone? Scaffolding drops substantially from this session onward.
Week 10 · 3 May 2027
The System That Fights Back
How do you keep the investigation disciplined when the system itself resists being understood — through unreliable documentation, inconsistent reproduction, or an environment where a careless intervention makes things worse?
Week 11 · 10 May 2027
Defending the Trail
Can every step of your investigation withstand informed challenge? The final session is a live defence of a complete, unassisted investigation of a system you have never seen before.
Week 12 · 17 May 2027