Week 1: Introduction to Algorithmic Romance
Meet Alex, check two invitations and distinguish observations from hidden state. Use supplied facts to test each claim in a fictional bio.
22 February 2027
Open the slides. Use A/D or scroll to turn pages; Esc returns here.
Alex’s story · Alex opens a blank profile#
Fictional course story · Week 1.
Alex, our fictional CS student on Null Island, starts with a blank profile and a modest specification: board games, terrible puns, perhaps a relationship. The first draft says “Seeking synergistic engagement.” It is grammatical. It could also advertise a printer. Alex replaces it with concrete details and sketches a Friday invitation. Before the profile goes any further, someone needs to check what those words actually promise.
Investigate this week: Use Alex’s supplied facts to audit the bio and two invitations. Which claims can you check, and what remains unknown?
The syllabus drop#
The opening provocation is a claim to test: “Love is an inefficient market problem solvable through data structures.” A system can store availability and reject an invitation that conflicts with it. Whether a recipient feels interested is a different question. This week we work through what a small model can establish and where its evidence runs out.
A recommendation, a reply and a relationship are three different outputs. HTTP 200 means the request succeeded; it is not evidence of affection.
This course is for students comfortable with probability and pseudocode. Work on Null Island, our fictional market, using supplied synthetic data. No dating account, partner or disclosure of your own life is required.
By the end of today#
- Identify supplied inputs and constraints.
- Separate an observation from an inference about hidden state.
- Justify a profile claim using a case fact.
The examples below practise those three tasks during the lecture. They are not a separate tutorial or assessed submission.
Scheduled teaching and independent work#
60-minute lecture#
Monday, 11 am–12 pm. Use the complete 16-slide deck in the following order.
| Lecture time | Slides | Focus |
|---|---|---|
| 0–10 min | 1–4 | Introduce the course question and distinguish recommendations, replies and relationships. |
| 10–25 min | 5–7 | Work through Alex’s constraints and the two invitation checks. |
| 25–40 min | 8–9 | Classify supplied facts, observations and hidden state; discuss the boundary example. |
| 40–55 min | 10–13 | Check bio claims against case evidence and explain participation and tool-use rules. |
| 55–60 min | 14–16 | Connect the semester sequence, questions and Week 2 preparation. |
Tutorial and independent work#
No tutorial in week 1. The examples are completed during the lecture; there is no separate deliverable. Optional independent review: revisit the boundary example and course policies. Week 2’s preparation is allocated in its own teaching plan. All 16 slides and the worked notes remain available as the complete Week 1 resource.
Meet Alex#
The toolkit’s fixed case describes Alex as an adult CS student who likes board games and terrible puns. Alex is free Friday 5–7 pm, has a $20 outing budget, travels by bus, and wants to meet someone with the possibility of a relationship. These facts are supplied inputs. Do not invent an extra interest, skill or possession to make a profile sound better.
Test two invitations#
For this exercise, assume that each proposed outing costs $12 in total. These are invented options, not real venue recommendations; bus times are unknown.
| Option | Proposed time | Time and budget checks |
|---|---|---|
| A | Friday 6–6.30 pm | Passes: inside 5–7 pm and $12 ≤ $20 |
| B | Friday 8–8.30 pm | Fails: outside 5–7 pm |
The rule keeps an option only if its proposed time falls entirely inside Alex’s availability and its total cost is at most $20. Option A passes both checks. Option B fails even though its assumed cost is within budget.
What is still missing? Option A remains provisional because travel to and from the outing must also fit the available time. Knowing that Alex uses the bus does not supply a route or timetable. Passing the two checks also gives no evidence about whether a recipient will reply or want a relationship.
A boundary example from the lecture#
Classify these statements before reading the explanation:
- Alex is free Friday 5–7 pm.
- A reply is recorded in a synthetic message log.
- “The recipient wants a relationship with Alex.”
Worked answer: the first is a supplied input. The second is an observation within the fictional exercise. The third is an inference about hidden state: the recipient’s actual motivation. The log records that a reply occurred, but does not distinguish a polite reply from an interested one. Keep the recorded event and the explanation you might assign to it separate.
Week 2 develops this distinction into a platform audit: add retrieval, ranking and exposure to the map, and identify which parts are supplied, observed or unknown.
Key concept: the Romantic Turing Test#
This course exercise asks whether a fictional bio describes Alex in natural language while remaining faithful to the supplied case. It does not measure romantic appeal or predict replies.
A: “Seeking synergistic engagement.”
B: “Board games, bad puns. Free Friday 5–7 pm to meet someone.”
Your check: quote two phrases in B and match each to a case fact. Then identify what a concrete invitation would still need.
Worked answer: “board games” and “bad puns” match Alex’s supplied interests; “Friday 5–7 pm” matches the availability; “meet someone” matches the stated aim. A supplies no concrete case detail to check. B gives a reader specific information, but an actual invitation would still need a place, total cost and workable bus travel. Natural wording earns no exemption from evidence.
One system, twelve weeks#
Start with platform architecture and photo assets. Move through bio experiments, mutual-selection probability, opening messages and asynchronous communication. Then examine threats, offline logistics, Markov first dates, follow-up decisions and maintenance. From week 2, each Thursday tutorial adds one artefact to the same fictional profile release.
No tutorial in week 1. This week is lecture-only. The first tutorial is Thursday 4 March, 2–3.30 pm, in week 2.
The data report is 20%, due 19 March. The midterm is 30%, on 23 April. The final profile deployment is 50%, due 28 May. Evidence earns marks; romantic outcomes and benchmark rank do not.
The strict “No Feelings” policy#
A dataset may contain “reply observed.” It may not silently convert that into “likes Alex.” This is a restriction on unsupported labels, not on having feelings. People can change their minds without filing a bug report.
Our scraping ethics begin with provenance, permission, purpose and minimisation. Public visibility is not permission to harvest a neighbourhood. Parse the course’s synthetic CSVs; do not scrape apps, private messages or classmates. The policies define the boundary.
The lecture deck presents the case, questions, worked answers and semester map. Continue to week 2’s lecture and the first tutorial. Bring the worked example; build the boundary map during that tutorial.
What Alex discovers#
Alex leaves with a bio that says something checkable. The invitation still needs a bus route, and no line of text establishes another person’s interest. Then a more basic gap appears: a profile can be ready without anyone seeing it. Alex’s next task is to follow the draft into the platform’s retrieval, ranking and exposure pipeline.
Follow Alex into the ranking queue in Week 2.
Teaching team
- Casimir Beng — convenor