Assignment 1: Design a Signal

Build a fictional dating persona and defend which of its claims are costly signals versus cheap talk

Due: 20 March 2028 · Weight: 15%

The brief

Build a fictional dating profile, then prove you know which parts of it are actually worth believing.

A profile is a small bundle of claims made under an incentive to look good. This assignment asks you to build one deliberately — persona, photo description, bio — and then step outside it and audit your own work: for each claim, say whether it’s cheap talk or a costly signal, and what would have to be true for that classification to hold. Weak defenses lose more marks than weak claims do.

The persona must be invented. No real person’s real profile, including your own, is submitted, referenced, or based closely enough on someone identifiable to be recognized — the exercise is about signal design, not about anyone’s actual dating life.

What you submit

A one-page fictional profile (persona name, a described photo, a short bio) and a second page auditing it: each claim, its classification, and a short defense. Evidence of your process (drafts, what you changed and why) strengthens the second page but isn’t required on its own.

The sorting exercise this builds on

Tutorial 3’s “Signal or Noise?” exercise runs on a fixed set of claims, the same ones for everyone. Here’s what that exercise looks like — this is a demonstration, not a real submission form.

Persona 07 (fictional, assigned to everyone in this example)

Runs marathons (photo attached, bib number visible). Says they 'love to cook' with no photo of any food. Lists 'looking for someone low-maintenance.'

This page is a demonstration, not a submission form. Nothing you type here is saved, sent, or marked — the real exercise runs on a fixed set of claims for every student, in tutorial 3, before Assignment 1 is due.

The spec

The brief poses the problem and leaves room for a response. The spec is the fixed part: what a marker checks. Some lines can be checked mechanically; the rest call for judgement.

How it is marked

CriterionWeightPerfect (HD)Excellent (HD)Very Good (D)Good (C)Satisfactory (P)Borderline (P)Unsatisfactory (F)Poor (F)Not Shown (0)
Signal design50%Every claim is a deliberately engineered costly signal or a deliberately obvious cheap-talk foil; the profile reads as a genuine artefact, not an exercise.Signal design is sophisticated — costly signals are varied, specific, and clearly hard to fake.Strong signal design with at least one genuinely convincing costly signal.Thoughtful profile with a workable mix of costly signals and cheap talk.Clear attempt at costly signals, though some are weak or easy to fake.Basic profile; most claims are unclassified cheap talk dressed up as signals.Superficial profile with little evidence of deliberate signal design.Profile reads as generic, with no discernible signal/cheap-talk distinction attempted.No profile submitted.
Defense of claims40%Every claim's classification is defended with a specific, testable reason that would survive a determined challenge.Sophisticated defenses that anticipate and pre-empt likely challenges.Strong defenses for most claims, with clear reasoning.Thoughtful defenses that classify each claim correctly and explain why.Clear classification of each claim, with a defense that mostly holds.Basic classification, but defenses are asserted rather than argued.Superficial or inconsistent classification of claims.Claims are unclassified, or the classification contradicts the profile itself.No audit submitted.
The profile is charming, not just accurate10%The profile would genuinely get a match — every line is engaging as well as classifiable, no filler.Charming and specific throughout; reads like a real (if fictional) person, not a case study.Mostly charming, with a clear voice and at least one memorable line.Pleasant to read, with some personality beyond the minimum.Functional but flat — the claims classify fine, but nobody's swiping.Reads like a checklist of cheap talk and costly signals rather than a person.Dry to the point of being hard to follow as a profile at all.No attempt at voice or personality — pure taxonomy.No profile submitted.