Episode Summary
Executive Summary: Tim Harford’s Cautionary Tales examines the long history of using science and technology to “solve” romance, from 1920s compatibility tests to modern dating algorithms. Through Chris McKinley’s OKCupid hacking and the limits of choice overload, the episode argues that online dating often optimizes shallow matching signals, not lasting love—and that romance remains stubbornly human and unpredictable.
Main Topics: Early scientific matchmaking fantasies (Priority: 5/5): The episode opens with Hugo Gernsback’s 1924 proposal to use electrical measurements, odor tests, and stress tests to determine marital compatibility, showing that techno-solutionism in dating is nearly a century old. Chris McKinley’s OKCupid hack (Priority: 5/5): A UCLA PhD student frustrated by poor matches reverse-engineers OKCupid using bots and clustering to identify women most likely to match his preferences. The limits of algorithmic compatibility (Priority: 5/5): The story shows that matching systems often rely on superficial variables and can mislead users into believing compatibility is scientific when it may be mostly random or placebo-like. Computer dating as a business model (Priority: 4/5): The episode highlights how dating platforms profit from ongoing search rather than successful closure, since subscriptions depend on users staying in the market. Choice overload and dissatisfaction (Priority: 5/5): Using a psychology experiment about photographs, Harford argues that endless choice can reduce satisfaction, which helps explain why dating apps may make romance feel harder, not easier. Chris and Christine’s real-world outcome (Priority: 5/5): Despite the algorithm’s failures, Chris meets Christine Tian Wang through a less-engineered path, and the relationship succeeds—suggesting that genuine chemistry and accident matter more than optimization.
Key Arguments: Scientific dating systems have a long history, but their methods often feel absurd in hindsight, suggesting modern apps may age just as poorly. OKCupid’s matching could be reverse-engineered because the platform’s most useful signals were visible through behavior and self-report patterns, not deep knowledge of love. Compatibility scores can act as a placebo: telling users they are a good match increases engagement more than the algorithm’s true estimate does. Dating apps may worsen dissatisfaction by keeping users aware that endless alternative partners are always one swipe away. Matching people on measurable traits like age, height, religion, or location is far easier than predicting long-term relational happiness. Chris’s eventual success came not from perfect algorithmic prediction but from a real mutual spark and a nearly accidental match outside the algorithm’s top predictions.
Data Points: Year of Hugo Gernsback compatibility article: 1924 - Science and Invention magazine feature on scientific marital matching Chris McKinley’s education level: Degree in Chinese, master’s in mathematics, nearly completed PhD in data science - Introduced as a highly qualified but lonely UCLA graduate student Age range when story begins: Early 30s - Chris is ready to settle down and commit to a serious relationship Number of women in Los Angeles mentioned: 2 million - Narrator notes the size of the dating pool relative to Chris’s situation Compatibility Research Inc. fee: $3 - 1960s Harvard student-run computer dating service charge Approximate modern equivalent of fee: About $30 - Inflation-adjusted comparison for Compatibility Research Inc. OKCupid data collected by Chris’s bots: 6 million answers from 20,000 women - Dataset used to infer matching clusters Number of distinct women types found: 7 clusters - K-modes clustering output from Chris’s reverse-engineering First-date total Chris went on: 87 first dates - Outcome of his hacked dating strategy before meeting Christine Second dates Chris obtained: 3 - Only three first dates progressed to second dates Third dates Chris obtained: 1 - Only one relationship advanced to a third date Compatibility shown in a successful match: 99% with dozens of local women; 90% with more than 10,000 Los Angeles users - Messages and compatibility scores produced by Chris’s profile experiments Sex recession comparison: American adults had sex nine fewer times per year in the early 2010s than in the late 1990s - Narrator uses this to frame modern dating dissatisfaction Dating app experiment year: 2014 - OKCupid published placebo-compatibility experiment results Photography experiment sample: Harvard students - Gilbert and Ebert’s study on choice and satisfaction Marital history of Hugo Gernsback: Second of three marriages at the time - Used to underscore the irony of his compatibility claims
Pivotal Quotes: "How much would the average man or woman give to know beforehand if his or her prospective married life is to be success or failure?" — Hugo Gernsback: His 1924 argument for scientific marital testing "What mattered was that users were told they were a good match. That assurance was perfectly effective at promoting an extended online chat." — Tim Harford: Summarizing OKCupid’s placebo-compatibility findings "You didn't find me. I found you." — Christine Tian Wang: Christine’s reflection on how she and Chris actually met
Implications: Dating tech can help surface options, but it cannot reliably predict love. Users and platforms should treat matching scores as rough filters, not truths, and remember that chemistry, timing, and human agency still decide outcomes.