Skip to main content

Quiz Calibration

Attempt the quiz before reading this page. Preserve the original answer, correction and help received. These explanations calibrate retrieval; successful recall alone does not complete the module or its practical evidence gates. For context such as shared schemas, use the question page.

Question 1: Archetypes​

A Principal engineer's calendar shows: 60% sprint ceremonies, 20% 1:1s with team members, 15% PR review, 5% cross-team design discussion. Which Staff+ archetype is this person actually operating as? Why?

Calibration: Tech Lead. Sprint ceremonies and team 1:1s are the Tech Lead's work. This is not Architect work (almost no cross-team design), Solver work (no single focused investigation), or Right-Hand work (no exec partnership).

If You Got This Wrong:

Question 2: Altitude Judgment (Scenario)​

Your team's release pipeline is 90 minutes long and flaky. You have 2 hours this week to work on it. Which move is the best use of Staff+ time?

a) Open the 3 flakiest tests, fix them, reduce runtime by ~20 minutes. b) Write an RFC proposing to split fast unit tests from slow integration tests across the entire monorepo. c) Schedule a strategy conversation with platform leadership about whether the company's one-pipeline topology matches the team structure. d) Pair with the most junior engineer on the team and teach them to fix one of the flaky tests.

Calibration: There is no universally best choice from role title and two hours alone. Pairing can build future capability if the learner is available and the fault is teachable; direct repair can be appropriate for urgent release harm. A bounded investigation or proposal can also be useful. State impact, urgency, uncertainty, feasible output and follow-up; grade the reasoning, not a predetermined archetype.

Question 3: Diagnosis vs Symptoms​

Which of the following is a diagnosis, and which is a symptom list?

A. "The platform team ships slowly because they spend 60% of their time on unplanned consulting from product teams. The crux is the intake pattern, not staffing." B. "The platform team has too much work. Backlog is growing. People are tired. Velocity is down."

Calibration: A is a diagnosis (one theory, names the crux, implies a specific class of action). B is a symptom list (several true statements, no theory, does not imply any particular response).

If You Got This Wrong: Concept 4: Diagnosis, Guiding Policy, Coherent Action.

Question 4: Guiding Policy (Scenario)​

Which of the following is a guiding policy, and which is a slogan?

a) "We will be customer-obsessed and move fast." b) "We will not build custom integrations for teams below a $2M ARR threshold." c) "Quality is our top priority this quarter." d) "Paved road over consulting: we build self-service paths, not bespoke integrations."

Calibration: (b) and (d) are guiding policies. Each rules something out. (a) and (c) are slogans - rewrite for a different company in the same sentence and they still make sense. If you cannot state what a policy rules out, it is a slogan.

If You Got This Wrong: Concept 4.

Question 5: Roadmap Capacity​

A team of 5 engineers, 13-week quarter, 60% project time (rest is on-call + toil). What is the realistic engineer-week capacity for Now? If current Now is packed with 45 engineer-weeks of commitments, what is the correct response?

Calibration: 5 × 13 × 0.60 = 39 engineer-weeks. If Now is 45 eng-weeks, we are over by 6. Correct response: move ~6 engineer-weeks of work out of Now and into Next. Warning: any plan that relies on "working harder" to close the gap is not a plan.

If You Got This Wrong: Concept 6: Roadmaps.

Question 6: Anti-Scope​

Why is anti-scope often more load-bearing than scope in an engineering strategy document?

Calibration: Anti-scope names things a reasonable reviewer could have expected to see in scope but that are deliberately excluded. It prevents the strategy from being cited to justify work it was not written for, and prevents scope creep during execution. Scope without anti-scope drifts; anti-scope makes the strategy defensible under pressure.

Question 7: RACI Judgment (Scenario)​

A decision has two people in "Accountable" because "both their orgs are affected." What is wrong with this setup, and how do you fix it?

Calibration: Accountability cannot be shared; if two people are accountable, no one is. Nothing ships because each assumes the other is deciding. Fix: pick exactly one Accountable. The other person becomes Consulted (with veto if appropriate). If neither can be the single Accountable, the decision is the wrong size - split it into two decisions, each with one A.

If You Got This Wrong: Concept 8: Stakeholder Mapping.

Question 8: Disagree-and-Commit (Scenario)​

Your VP has chosen Go over Rust for a service rewrite. You disagree but executed anyway. Six weeks later, the rewrite is behind schedule. In the retrospective, you say: "I always had concerns about the language choice." What did you do wrong?

a) Nothing; you executed in good faith. b) You should not have agreed to execute. c) You should have stated the disagreement in writing before the decision, with an observable trigger for reopening. Surfacing concerns only in retrospect is silent dissent. d) You should have executed Go faster.

Calibration: (c). Disagree-and-commit requires the disagreement to be stated before the decision, in writing, with a named trigger. Surfacing concerns after the outcome is known is the anti-pattern - it damages trust regardless of whether you were right.

If You Got This Wrong: Concept 9: Disagree-and-Commit, Managing Up.

Question 9: Written-First Culture​

Name two specific gains a written-first culture produces that a spoken-first culture does not.

Calibration: Two of these: (a) parallel alignment - ten reviewers can comment in parallel where a meeting of ten serializes everyone; (b) decisions persist - the written decision survives the meeting while verbal consensus evaporates; (c) junior participation - async writing gives junior reviewers time to engage where spoken fast-loops silence them; (d) better arguments - writing forces completion of the thought before defense.

Question 10: Audience Mismatch (Scenario)​

A Staff engineer sends a 10-page database-migration design doc to the VP with the subject "Approve migration?" The VP does not respond for two weeks. What is the most likely cause?

a) The VP is busy. b) The doc was sent to the wrong VP. c) The doc was the wrong artifact for an exec audience; the engineer should have sent a 3-5 sentence exec summary with a specific ask. d) The design is technically flawed.

Calibration: (c) is the most likely and most actionable cause. Execs do not read 10-page engineering documents to find a decision request. The engineer's job was to produce the exec-audience artifact separately. (a) may also be true but is not actionable.

If You Got This Wrong: Concept 10: Audience-Aware Explanation and Concept 11: Executive Summary.

Question 11: BLUF​

Rewrite this executive summary as BLUF (under 5 sentences):

"We've been investigating the orders service performance for the past quarter. After multiple attempts at optimization, we now believe the database is the bottleneck. We've considered several alternatives. We'd like to propose a migration."

Calibration: > Decision requested: fund a bounded investigation of the Orders latency bottleneck and migration alternatives. Current evidence: repeated optimization attempts have not resolved the issue; the database is a hypothesis requiring workload and plan evidence. Next step: produce a measured comparison, migration/rollback risks and cost estimate before approving a migration.

Lead with the ask. The prompt supplies no incident rate, deadline, budget or performance target; inventing those numbers would weaken the decision record. Label any illustrative assumptions explicitly.

Question 12: Storytelling Slots (Scenario)​

A design presentation hits "Problem," "Choice," and "Consequences" but skips "Constraint" and "Options." What is the most likely flaw in the design?

Calibration: The choice is probably arbitrary. Without a named constraint, the designer cannot explain what ruled out the obvious alternatives. Without named options, reviewers cannot tell if alternatives were seriously considered. A design with no constraint slot is usually a design that was picked first and justified afterward.

Question 13: Feedback (Scenario)​

"You were really unhelpful in the review yesterday." Rewrite this as SBI feedback.

Calibration: Situation: "In yesterday's architecture review for the payment-gateway RFC, around the 20-minute mark." Behavior: "You interrupted the author three times in five minutes and rewrote their proposal on the whiteboard before they finished describing it." Impact: "The author stopped presenting and the two junior engineers didn't speak again. I think we lost their input on the consistency tradeoff." Intent ask: "That probably wasn't your goal - what were you going for?"

Key moves: named time and place, observable verbs, named impact on a specific group, an intent question that invites conversation instead of defense.

Question 14: Mentorship vs Sponsorship​

Give one concrete example of a mentorship move and one concrete example of a sponsorship move. Explain why they are different.

Calibration: Mentorship: weekly 1:1 with a junior engineer walking them through their first design doc. Low political cost; advice is private. Sponsorship: nominating that engineer to lead the upcoming paved-road RFC, publicly, with your name attached to the recommendation. Sponsor spends credibility; if the engineer underperforms, the sponsor is on the hook. Mentorship grows a person; sponsorship unlocks them.

Question 15: Career Anchor Fit (Scenario)​

A Staff engineer with a "technical/functional competence" anchor is offered a management promotion. Their current role is depth-focused. The new role is people-focused, with +20% comp. The offer is presented as a career step. What is the right response?

a) Accept - promotions should always be accepted. b) Decline reflexively - management is bad. c) Evaluate against the named anchor: does management serve their technical/functional anchor, or contradict it? If it contradicts, declining is the right move even at +20% comp. d) Accept conditionally, with a written option to return to IC in 18 months.

Calibration: (c) is the right evaluation process, and (d) may be the right operational move after that evaluation concludes the fit is uncertain. Accepting or declining without running the anchor check is the failure mode. Named anchors protect engineers from career drift driven by comp or social pressure.

If You Got This Wrong: Concept 15: Personal Strategy.

Question 16: Accountable Decision (Scenario)​

A team says an automated account-suspension feature is ethically acceptable because it reduces fraud overall and passed legal review. Name four pieces of evidence or control still required before that claim is defensible.

Calibration: Any four of: differentiated affected stakeholder groups; false-positive and subgroup evidence; a viable manual or narrower alternative; an accessible notice and appeal/recovery path; safeguards tied to specific harms; an accountable owner for each safeguard; explicit unknowns; monitoring bounds; or an observable pause, rollback, or remediation trigger. Aggregate benefit and legal approval alone do not show who bears mistakes or how the system recovers.

If You Got This Wrong: Concept 16: Ethical Impact Decisions and the Ethical Impact Decision Lab.

Prior Module Question 1 (Semester 7, Module 5 - ADRs)​

What are the three required sections of an ADR, and what fails if any is missing?

Calibration: Context, Decision, Consequences. Missing Context: future readers cannot argue with the reasoning. Missing Decision: the ADR records a discussion, not a choice. Missing Consequences: the decision looks costless and cannot be evaluated in hindsight.

Prior Module Question 2 (Semester 8, Module 1)​

Why is "latency" usually a poor first-level SLI and what is better?

Calibration: Latency needs a population, window and measurement definition; it is not inherently an average. One useful SLI is the fraction of eligible requests completing below a threshold, alongside success/error accounting. Percentiles can also help, but their population and aggregation must be explicit; success-only percentiles can conceal dropped requests.

Prior Module Question 3 (Semester 8, Module 2)​

What is the difference between a service boundary and a data boundary?

Calibration: A service boundary is a deployment and ownership unit. A data boundary is about authority over a piece of state. They are often the same in healthy architectures but are not the same concept; many microservices failures come from splitting the service boundary while keeping the data boundary shared.

Prior Module Question 4 (Semester 8, Module 3)​

Why does event-driven architecture reduce coupling in one direction while increasing it in another?

Calibration: It reduces synchronous coupling (services do not wait on each other) but increases temporal and schema coupling (producers and consumers must agree on event schemas over time, and subscribers must tolerate ordering and replay semantics).

Prior Module Question 5 (Semester 8, Module 4)​

What is an SLO and how does it constrain engineering priorities?

Calibration: A Service Level Objective is an agreed-upon target for an SLI (e.g., 99.9% of requests under 500 ms). It constrains priorities by creating an error budget: when the SLO is met with margin, the team can take more reliability risk (new features); when the budget is spent, the team must stop feature work and invest in reliability.