The shape of the semester

Twelve weeks, three acts, and the chain of dependencies that connects them. Every week names the earlier week it stands on, and a check in spec/ fails the build if that chain ever breaks.

A course is one idea explored for twelve weeks, not twelve topics that happen to share a subject. This page is where SLOP2710 shows its working.

Each week below declares the single new idea it introduces, the real system it argues from, and the earlier week it builds on. Those declarations are not written here — they live in each lecture's frontmatter, andspec/course-arc.test.ts reads the same four fields. If a week ever pointed forwards, pointed nowhere, or reused another week's headline system, the build would fail rather than quietly producing a prettier version of this diagram.

Act IAct IIAct III12345678910111210%15%15%40%
What to look for: most weeks lean on the one before them, but two reach further back — week 8 onto week 4 and week 10 onto week 2. Those are the course's two callbacks: price can only sort a queue once time has been given a price, and a network of queues is variability from week 2 propagating. The gold chips below the line are the four pieces of assessment, at the week they fall due — measure one, change one, test one, design one.

How the grade adds up

The four gold chips on the timeline above are 80% of the mark: 10% for measuring a real queue in week 3, 15% for changing who goes first in week 6, 15% for simulating that change in week 9, and 40% for the design brief the whole semester has been loading for. The remaining20% is not a fifth brief — it is a short quiz at the start of every lecture.

In the room

Two minutes on the clock, and you may talk it over with whoever is next to you. Turning up without the reading is entirely recoverable here — that is what the two minutes and the room are for.

Watching online

About thirty seconds — long enough to answer if you already know it, not long enough to go and look it up. Doing well here means the week's page was read before the lecture started, not during it.

Same question, same week, two different clocks — and which one a student gets is decided by where they happened to be sitting. Best 7 of 12 scores count, so the other five are free to lose, to a bad morning or a genuinely hard question, with nothing owed by way of explanation.

It is a small, low-stakes version of the course's opening claim: a rule decides who goes first, and it is rarely neutral about who it favours.

Every lecture is paired with a lab in the same week — the link sits beside each week's title below. Labs are not marked and carry no percentage of their own; they exist so that pooling two queues, running shortest-job-first, or boarding a plane by rows is something you have built and broken yourself, not just watched happen on a slide.Participating in 6 of the 12 is a condition of passing the course rather than a mark, which is why labs draw no chip on the timeline above.

Act I · weeks 1–4Why waiting happens

The physics. Learn to see a queue, explain why it forms, measure it from outside, and price what it costs.

  1. Week 1The Queue You Didn't NoticeSlidesLab

    A queue is a designed object, and somebody chose the rule it runs on.

    Argued from: Campus coffee shopStarts the course

  2. Week 2Variability, Not BusynessSlidesLab

    Variability creates queues, and congestion grows non-linearly as load approaches capacity.

    Argued from: Drive-through windowBuilds on: week 1 (same act)

  3. Week 3Little's LawSlidesLab

    L equals lambda times W, so any two of length, throughput and wait give you the third.

    Due this week: Queue Census (10%)

    Argued from: Airport security screeningBuilds on: week 2 (same act)

  4. Week 4The Cost of WaitingSlidesLab

    Waiting is an unpriced cost borne by the queue and invisible to the organisation causing it.

    Argued from: Passport officeBuilds on: week 3 (same act)

Act II · weeks 5–9Who goes first

The rules. Five ways to order the same people, and what each one buys and charges.

  1. Week 5First Come, First ServedSlidesLab

    FCFS is a choice with a specific fairness claim and a measurable efficiency price.

    Argued from: Bank branchBuilds on: week 4 (Act I)

  2. Week 6Shortest Job FirstSlidesLab

    Ordering by job length provably minimises mean wait, and starves the long jobs to do it.

    Due this week: The Ordering Rule (15%)

    Argued from: Express checkout laneBuilds on: week 5 (same act)

  3. Week 7Triage: Order by NeedSlidesLab

    Ordering by urgency saves lives and depends entirely on the quality of a ninety-second judgement.

    Argued from: Emergency departmentBuilds on: week 6 (same act)

  4. Week 8Pay to Skip: Order by PriceSlidesLab

    Prices sort by willingness to pay while queues sort by willingness to wait, and the two select different people.

    Argued from: Theme park priority passBuilds on: week 4 (Act I)

  5. Week 9Don't Queue At AllSlidesLab

    Appointment and virtual systems relocate waiting rather than eliminating it, and the relocation has its own winners.

    Due this week: Simulation Report (15%)

    Argued from: Restaurant reservation bookBuilds on: week 8 (same act)

Act III · weeks 10–12Systems that fight back

The complications. Queues joined to other queues, and queues full of people who have worked out the rule.

  1. Week 10Queues in NetworksSlidesLab

    Queues connected to other queues behave unlike any of them alone, and local optimisation can make the whole worse.

    Argued from: Aircraft boardingBuilds on: week 2 (Act I)

  2. Week 11Queues That Know You're WaitingSlidesLab

    Queues made of people are feedback systems, and information about the wait alters the wait.

    Argued from: Call centreBuilds on: week 10 (same act)

  3. Week 12There Is No Neutral QueueSlidesLab

    Every queue optimises something, the objectives conflict, and choosing between them is a political act.

    Due this week: Final Design Brief (40%)

    Argued from: Organ transplant waiting listBuilds on: week 11 (same act)