Why Algorithms Can’t Predict Your Love Life with Dr. Paul Eastwick

Summary of Why Algorithms Can’t Predict Your Love Life with Dr. Paul Eastwick

by Pushkin Industries

41mFebruary 23, 2026

Overview of Why Algorithms Can’t Predict Your Love Life (The Happiness Lab)

This episode of The Happiness Lab (host Dr. Laurie Santos) interviews social psychologist Dr. Paul Eastwick about his new book, Bonded by Evolution. Eastwick challenges the popular "EvoScript" — the idea that attraction is mostly fixed mate-value (you’re a 3 or a 9) determined by evolved cues — and explains research showing attraction is far more about dyadic compatibility that is built over time. He critiques how online dating and algorithms amplify misleading myths about who we will pair with and offers practical, research-backed advice for modern dating.

Key takeaways

  • The classic mate-value/EvoScript view (people occupy a fixed desirability rank and pair assortatively) is overstated and demoralizing.
  • Attraction is better modeled as three components: popularity (consensus desirability), selectivity (how picky someone is), and compatibility (unique, pair-specific fit). Compatibility often explains most attraction—even early on.
  • Stated preferences (surveys) overstate gender differences. Revealed preferences (e.g., speed-dating behavior) show men and women respond similarly to traits like attractiveness and ambition.
  • Short-term sexual desirability (the “alpha” versus “beta” idea) predicts sexual opportunity but does not predict long-term relationship success.
  • Matching algorithms and large-profile questionnaires do a poor job predicting which specific pairs will click — much of compatibility is constructed via repeated interactions and serendipity.
  • Online dating promotes “resume dating,” quick judgments, option overload, and premature filtering, which undermines the relational processes that build compatibility.
  • Practical dating advice: expand your aperture, give people multiple chances, cultivate mixed-gender social networks and community-based interactions, and prioritize in-person, repeated contact.

Main arguments and evidence

The three-part model of attraction

  • Popularity/consensus: some people earn more likes (especially in photo-based contexts), but consensus on desirability is weaker than assumed.
  • Selectivity: individuals vary in how open vs. picky they are.
  • Compatibility: the unique fit between two people (shared moments, coordination, interdependence). Eastwick finds compatibility explains a large share of attraction, and it grows in importance as people get to know each other.

Consensus on attractiveness is limited

  • Studies show agreement exists when evaluating photos, but also large disagreement: for most faces, different raters put them in different halves of attractiveness distribution. Universal “top-half” consensus is rare.

Revealed preferences vs surveys

  • Survey responses reproduce gender-difference stereotypes (men claim to prioritize attractiveness more; women claim to prioritize resources more).
  • Speed-dating / revealed-preference studies show both sexes actually prefer attractiveness and ambition similarly when choosing among real people — indicating surveys misrepresent real behavior.

Short-term vs long-term desirability myths

  • Traits linked to short-term sexual success (e.g., perceived sex appeal, confidence) relate to more sexual partners but do not reliably predict long-term relationship quality or stability.
  • The “alpha/beta” and “sliding scale” narratives oversimplify and fuel toxic online subcultures.

Algorithms and big-data matching fail at predicting pair-fit

  • Studies simulating full-profile algorithmic matching (e.g., Samantha Joel’s work) could predict who is selective or popular but failed to predict which pairs would click — matching requires interaction history, not static trait alignment.

Why compatibility is “creative chaos”

  • Compatibility is constructed: it emerges across sequences of interactions (shared jokes, small serendipities, conversational hooks) and is therefore hard to predict from profiles.
  • Chaotic element: luck and timing matter (you might click on the 8th interaction, not the 1st).
  • Similarity alone (demographics, deal-breakers, stated preferences) does poorly explaining who becomes compatible. People often inhabit milieus of similar others, so similarity appears predictive but doesn’t explain why a few pairings work while most don’t.

Practical dating advice from the episode

  • Date from a wider pool: open your criteria rather than narrowing them prematurely.
  • Give people multiple chances: prioritize second or third interactions before ruling someone out.
  • Favor community-based, repeated, in-person interactions (friends-of-friends, classes, sports, group activities) over one-off resume-style app dates.
  • Relearn “hanging out”: prioritize shared experiences and slow acquaintance-building rather than expecting immediate sparks.
  • Don’t over-rely on algorithmic matching or rigid filters (education, job title, narrow deal-breakers).
  • Embrace cross-gender friendships: they are valuable networks for introductions and are not inherently “traps.”

Notable quotes / concise formulations

  • “Mate value is out and the creative chaos of compatibility is in.”
  • Compatibility = the unique connection between two people that grows through interaction and is hard to predict from profiles alone.
  • “Speed-dating reveals what people actually do is often different from what they say they want.”

Actionable to-do list (quick)

  • When using apps: expand your search criteria, try dates beyond the resume exchange, and allow at least a second meet-up before deciding.
  • Build mixed social networks: join clubs, classes, or groups where you’ll meet recurring acquaintances.
  • Practice patience: expect compatibility to sometimes grow over multiple interactions.
  • Reduce reliance on rigid deal-breakers; prioritize curiosity and small compatibilities.
  • Treat online dating as a supplement, not a substitute, for real-world socializing.

Research caveats and nuance

  • Consensus-based traits (attractiveness, status) do matter, especially for initial sexual interest and in photo-limited contexts.
  • Much of the evidence comes from speed-dating, lab studies, and social-network observations; different cultures and contexts may vary.
  • Eastwick doesn’t claim biology is irrelevant — rather, he reframes evolutionary theory toward bonding and coordination rather than fixed market ranks.

Further resources

  • Paul Eastwick — Bonded by Evolution: The New Science of Love and Connection (book)
  • Paul Eastwick — Podcast: Love Factually
  • The Happiness Lab with Dr. Laurie Santos — episode featuring Paul Eastwick

If you want the one-sentence takeaway: attraction is less a fixed marketplace value and more an emergent, buildable property between two people — which means patience, community, and repeated interaction matter far more than polished profiles or algorithmic matches.