Algorithmic Romance#
The syllabus drop.
SLOP1276 · Week 01 · Alex’s story
Fictional Alex’s first bio says “Seeking synergistic engagement.” It could advertise a printer. A Friday invitation needs checking too.
Investigate: Which promises do the supplied facts support?
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.
We will practise all three with one fictional person: Alex.
The opening provocation#
Claim to test: “Love is an inefficient market problem solvable through data structures.”
A system can store availability and reject plans that conflict with it.
Does passing those checks establish romantic interest?
Let’s work through what the evidence actually supports.
Three different outputs#
A recommendation is not a reply.
A reply is not a relationship.
“Reply observed” is data.
“They like Alex” is an inference.
HTTP 200: the request succeeded. That is not evidence of affection.
Meet Alex#
Fictional case · adult CS student.
Likes board games and terrible puns.
Free Friday 5–7 pm · $20 outing budget · travels by bus.
Wants to meet someone, with the possibility of a relationship.
These are supplied facts. Add no biography to improve the score.
Test two invitations#
Exercise assumptions: each outing costs $12 in total. Bus times are unknown.
A: Friday, 6–6.30 pm.
B: Friday, 8–8.30 pm.
Alex is free 5–7 pm and has $20. Which option passes those checks?
What must be checked before calling it feasible?
Apply the rule#
Keep an option only if it is inside 5–7 pm and costs at most $20.
A passes: 6–6.30 pm is inside the window; $12 is within budget.
B fails: 8–8.30 pm is outside the window.
A is still provisional: bus travel must fit the available time too.
Passing a scheduling check does not predict a reply.
Classify the evidence#
1. Alex is free Friday 5–7 pm.
2. A reply is recorded in a synthetic message log.
3. “The recipient wants a relationship with Alex.”
Classify each as an input, an observation, or an inference about hidden state.
Worked answer#
Input: Alex’s supplied availability.
Observation: a reply recorded in the synthetic log.
Inference: “They want a relationship with Alex.”
Hidden state: the recipient’s actual motivation remains unknown.
The log does not distinguish a polite reply from an interested one.
Romantic Turing Test#
A: “Seeking synergistic engagement.”
B: “Board games, bad puns. Free Friday 5–7 pm to meet someone.”
Quote two phrases in B and match each to a supplied fact.
Which details would an actual invitation still need?
Check the bio claims#
Board games / bad puns → supplied interests.
Friday 5–7 pm → supplied availability.
Meet someone → supplied aim.
An invitation still needs a place, total cost and workable bus travel.
Natural wording earns no exemption from evidence.
Strictly: “No Feelings”#
Do not invent emotion labels in a dataset.
People still get to have feelings.
They can decline, leave or change their minds.
A refusal ends the interaction; it is not a temporary error to retry.
Acquisition ethics#
Use the supplied synthetic data.
Check provenance, permission, purpose and minimum fields.
No app scraping, private messages or real-user experiments.
Public visibility is not a harvesting licence.
Twelve weeks, one system#
1–3 · Measure: introduction, platforms, photos.
4–7 · Model: bios, matches, messages, latency.
8–12 · Ship: threats, offline plans, first dates, follow-up, maintenance.
From week 2, each tutorial hands an artefact to the next.
What earns a mark?#
Data report · 20% · 19 March
Midterm exam · 30% · 23 April
Fictional profile release · 50% · 28 May
Engineering evidence, not romantic outcomes or a required rank.
Take the boundary into week 2#
Alex’s bio now has evidence. Whether anyone sees it is still unknown. Week 2 follows the ranking queue.
No tutorial in week 1. First tutorial: 4 March, week 2.
Bring this worked example; build the map during that tutorial.