The method is the same in all three contexts. The entry point differs.
You are given 45 minutes to design or analyse a system you have not seen before. The pressure is not urgency — it is completeness.
Apply the method: 1. Identify the archetype (30 seconds). State it out loud: “This is primarily a Social & Communication system with elements of Search & Discovery.” 2. Run the review questions in order. Say the question before answering it. “Q1 — Scale: how many users, what QPS for reads vs writes?” 3. Name tradeoffs as you make decisions. “I’m choosing eventual consistency here — Consistency vs Availability (AT1) — because the use case can tolerate stale feeds.” 4. Name failure modes as you finish the design. “The main risks are FM7 (Thundering Herd) on cache expiry for popular celebrities and FM6 (Hotspotting) on write-heavy celebrity shards.”
This approach demonstrates systems thinking, not memorised answers.
You are reviewing a proposed design before it is built. The goal is to find gaps before they become incidents.
Apply the method in reverse: 1. Classify the archetype. Confirm the reviewer’s classification matches the designer’s intent. 2. Run the review questions and mark which are answered, which are incomplete, and which are absent. Absent answers are the primary deliverable — they are the decisions that need to be made before the design is approved. 3. For each design decision, confirm the tradeoff has been named. An unnamed tradeoff is an implicit decision that may surprise future maintainers. 4. Run F3 as a pre-mortem. For each component, ask: if this component behaves incorrectly, what failure mode does it create?
You are joining a team or organisation and need to get productive on a system you did not build.
Apply the method as a structured onboarding exercise over three days: - Day 1: Classify the archetype. Draw all five architecture diagrams from the existing documentation. Fill in what you can; note gaps. Running the metrics dashboard for 30 minutes usually fills in D1 (Request Flow) and D2 (Data Storage) with a clear sense of what to look for. - Day 2: Run the review questions against the diagrams. For each gap, ask a current team member to fill it in. This is also an efficient way to identify which team members know which parts of the system. - Day 3: Name the top three failure modes you have identified. Present them in a team meeting. If you are wrong, you learn. If you are right, you have produced a pre-mortem before your first production incident.