# The Relationship Singularity

> A research program to embed human personality at the resolution reality demands and predict relationship outcomes before two people ever meet.

## Researcher

Matthew Fisher
Email: staff@relationshipsingularity.org
Website: https://relationshipsingularity.org
Twitter: https://x.com/relation_sing
YouTube: https://www.youtube.com/@relationshipsingularity

## Core Thesis

The Relationship Singularity is the hypothetical point at which computational models of human personality reach sufficient resolution that relationship outcomes become predictable before two people ever meet. We are pre-CLIP for human personality — the Big Five is a 5-pixel image of a person. The same transition CLIP brought to computer vision is coming for personality: unsupervised, high-dimensional, multimodal representations learned directly from data.

## Research Directions

1. Personality as Embedding Space — Unsupervised, high-dimensional, multimodal personality vectors learned from text, voice, video, behavior, and digital footprints.
2. The Cultural Trope Hypothesis — TV Tropes as unsupervised personality clustering by collective cultural observation. Node2Vec on the character-trope bipartite graph produces personality embeddings.
3. Relationship Outcome Manifolds — Compatibility as a structured probability distribution over relationship dynamics, not a scalar score.
4. Emotion Embeddings as Proof of Concept — Alan Cowen's work at Hume AI proved the embedding paradigm for emotions. Extending to personality.
5. Moneyball Mutual Attraction — Generating the manifold of all possible faces, mapping aesthetic preferences in latent space, finding high mutual attraction in distribution tails.
6. LLM Relationship Simulation — Monte Carlo simulation of relationship dynamics using personality-conditioned language model agents.

## Technical Framework

Deep-ROP (Relationship Outcome Prediction) architecture:
- P3HF: Personality-guided hypergraph transformer disentangling public behavior from private personality
- TempODE: Continuous-time dynamics via Allen-Cahn phase-field potentials modeling relationships as particles in energy landscapes
- RELATE-MACT: Multi-agent LLM simulation for counterfactual turning point scenarios
- MORAL: Multi-Output Ranking Aggregation for Link Fairness addressing the dyadic barrier in algorithmic fairness

## Content Files
- Talk transcript (2hr, Nov 2024): https://relationshipsingularity.org/2024talktranscript.txt
- Roadmap PDF: https://relationshipsingularity.org/roadmap.pdf
- Sitemap: https://relationshipsingularity.org/sitemap.xml

## Institutional Vehicles

- NeurIPS 2026 Workshop: AI for Human Connection (AI4HC) — Sydney, December 2026
  https://relationshipsingularity.org/ai4hc
- Love Symposium — San Francisco conference on technology and human connection
  https://symposium.love
- experiments.dating — Live experiments in AI-mediated connection
  https://experiments.dating

## Press & Media

- New York Times: "Can You Optimize Love?" (January 2026)
  https://www.nytimes.com/2026/01/06/style/love-symposium-artificial-intelligence-keeper.html
- The Daily (NYT Podcast): "Is the Swipe Era Over?"
  https://open.spotify.com/episode/5wJkd1DI5BtpWPXDuVqrLx
- Vectors of Mind Podcast: "AI, Dating Apps, and the Future of Relationships"
  https://open.spotify.com/episode/3r8tb1PZzVtqJIdR4cOiPw
- Full 2-hour research talk (YouTube):
  https://youtu.be/2pQrl_LsjKU

## Key References

- Radford et al. "Learning Transferable Visual Models From Natural Language Supervision." ICML 2021. (CLIP)
- Cowen & Keltner. "Self-report captures 27 distinct categories of emotion." PNAS 2017.
- Grover & Leskovec. "node2vec: Scalable Feature Learning for Networks." KDD 2016.
- Park et al. "Generative Agents: Interactive Simulacra of Human Behavior." UIST 2023.
- Youyou, Kosinski & Stillwell. "Computer-based personality judgments are more accurate than those made by humans." PNAS 2015.
- Gottman et al. "The Mathematics of Marriage: Dynamic Nonlinear Models." MIT Press 2005.