Christian Rudder, co-founder of OK Cupid, explains how dating data reveals people say one thing but do another—e.g., men claim they want partners their age but actually chase younger women. He says this 'say-do' gap also applies to investing, where fund managers claim long-term focus but trade often. He advocates for 'high-variance' strategies—polarizing opportunities some love and others hate—over average ones. Key holdings: OK Cupid (founder exited, profit grew after acquisition); Keybase (co-founder's new crypto startup); online vision test company (advisor, helps users skip eye doctors).
Christian Rudder (co-founder of OK Cupid and New York Times bestselling author) discussed with host Patrick O'Shaughnessy the application of quantified self and data science in investing. Core insight: By analyzing OK Cupid's dating data, quantifiable patterns in human behavior are revealed, which c
Christian Rudder (OK Cupid co-founder, New York Times bestselling author) discusses with host Patrick O'Shaughnessy the application of quantified self and data science in investing. Core thesis: By analyzing OK Cupid dating data, Rudder reveals quantifiable patterns in human behavior that are analogous to irrational behavior in markets. Key takeaway: Data-driven insights (e.g., user preferences, matching algorithms) can uncover hidden patterns, similar to how AI and NSA technology analyze massive datasets. Rudder emphasizes that understanding these data trends helps investors identify behavioral biases, such as how preferences for specific traits influence decision-making. The discussion also extends to books on the American Civil War, suggesting historical patterns offer lessons for investment strategies.
Rudder points out that OK Cupid data repeatedly proves a systematic divergence between what users verbally claim to prefer and their actual behavior. For example, male users claim they want partners of a similar age (e.g., a 30-year-old man says he wants women aged 30-34), but actual behavior shows that men of all ages tend to pursue women around age 20 — the "Wooderson Law" (from the movie Dazed and Confused: "I get older, they stay the same age"). Rudder explains: "A 35-year-old man will tell us he wants women aged 30-32, but his actual behavior is all about women in their early 20s." Women, in contrast, are consistent: a 25-year-old woman finds 25-year-old men most attractive, and a 30-year-old woman finds 30-year-old men most attractive, aligning with their stated preferences.
This "say-do" gap also applies to racial preferences. Users generally claim "race is not a factor," but data shows that black men, Asian men, and black women receive about 25% fewer likes, messages, and replies on OK Cupid compared to others (i.e., a 75% experience). Latino users behave almost identically to white users, neither suffering from bias nor holding similar biases against other groups. Rudder emphasizes: "This is data about direct human interaction, not third-party metrics like SAT scores or income, so it offers a rare perspective."
Implications for investors: Market participants also exhibit a "say-do" gap — fund managers claim to be long-term investors yet trade frequently, investors claim to be value-oriented yet chase momentum. Data can reveal true behavioral patterns, rather than relying on self-reports.
Rudder finds that among people with the same attractiveness level (e.g., both rated 5), those with the highest rating variance receive far more attention than those with the same average score. Specifically, if half the people rate someone a 9 and the other half a 1, that person gets attention equivalent to a traditional 7 or 8. Rudder explains: "It's like activation energy — you need a strong enough positive reaction to prompt someone to take action." Extreme traits (e.g., tattoos, nose rings, blue hair) effectively filter for a "highly favorable" group while leaving others uninterested.
This pattern holds in business as well: Bestsellers often have a large number of both 1-star and 5-star reviews, rather than a uniform 4-star rating. Rudder notes: "If you offer an iterative, generic product, it's hard to generate interest. But if you do something new and different, some people will hate it, and some will love it — that's actually more effective." He even finds that being strongly disliked (a 1-star rating) can slightly boost attention, possibly due to a "white knight effect" — users think, "Many people won't like her, so my message will be the one that discovers a hidden gem."
Implications for investors: Investment strategies should also pursue "high variance" — opportunities that clearly diverge from market consensus, which may be questioned by the majority but highly favored by a minority, often offer greater excess return potential than "middle-of-the-road" strategies. However, this requires investors to have the psychological resilience to withstand negative feedback.
Rudder emphasizes that choosing the wrong metric can misdirect an entire business. Early on, OK Cupid considered using "page views" or "time spent" as success metrics, but quickly realized these would be disastrous for a dating site — users visiting day after day without finding a partner is good for Twitter but a failure for a dating site. They ultimately chose "four-way conversations" (four ways) as the core metric: measuring the number of deep, ongoing conversations on the platform. Rudder explains: "Raw message count doesn't work either, because users might send a flood of messages to a few attractive people while ignoring others — the sender is unhappy, and the receiver is unhappy too."
This choice stemmed from a deep understanding of user needs: People come to OK Cupid to meet, and conversations are the best proxy for meetings. Rudder notes: "If we had kept staring at page views or click-through rates, our site would have failed." He extends this principle to writing: "The hardest part of writing Dataclysm wasn't analyzing the data, but asking the right questions — which questions can be answered? Which questions do people want to read the answers to?"
Implications for investors: Portfolio management also requires choosing the right metrics. Traditional metrics like short-term returns, volatility, and Sharpe ratios may not reflect true risk-return characteristics. Investors should think about "what we truly want" (e.g., long-term compounding, downside protection, liquidity) and then design metrics that measure those goals.
Rudder distinguishes between the nature of corporate data collection and government surveillance. Companies (e.g., OK Cupid, Facebook, Twitter) collect data for advertising, but "no company pulls up an individual's profile and asks, 'What did this person do, what do they like, what will they do next, who are they talking to?' — that never happens." Advertising targets groups (e.g., "males aged 18-25"), not specific individuals. Rudder emphasizes: "It's not worth the time, and there are all kinds of rules preventing it."
Government surveillance is the opposite: The NSA focuses on specific individuals — "a specific 25-year-old male, who he's talking to, what he's doing, what he likes, what he might do next — with a name, address, and family." Rudder considers this targeting more "harmful and invasive." He warns: "When people talk about privacy, it's crucial to distinguish between the actual use cases of data and their impact on individuals."
Implications for investors: When analyzing data, investors must distinguish between "aggregate trends" and "individual signals." Market data (e.g., index returns, sector valuations) reflects group behavior, while company-specific data (e.g., management conference calls, insider trading) may contain individual signals. Investors should clarify which type of data they need and understand its limitations.
Rudder believes that artificial general intelligence (AGI) will emerge within 20 years, and its risk is far from the Hollywood "Terminator" scenario. Citing Nick Bostrom's Superintelligence, he argues that the real risk lies in AGI's unpredictability: "Any superintelligent entity is a roll of the dice — just like you never know what a child will become, even if you know the genes and upbringing." He warns that the most dangerous outcome is not AGI "killing people with karate or shotguns," but "making everyone die more painlessly."
Rudder proposes a solution: A global, coordinated effort akin to the "Manhattan Project," developing AGI openly and transparently. He opposes secret development by a single government (e.g., NSA, China, Russia) or company (e.g., Twitter programmers): "I don't want Twitter programmers to set the rules for the whole world — that sounds terrifying." He further notes that even without AGI, automation (e.g., self-driving cars) will cause massive unemployment, requiring "a tax on AI usage, with the revenue redistributed to those displaced."
Implications for investors: Long-term investors need to consider AI/automation's impact on economic structure — which industries will be disrupted, which will benefit, and how social inequality may worsen. Rudder's view suggests that investing in companies that "create positive real-world outcomes" (rather than purely pursuing profit) may offer long-term resilience.
| Position | Guest's Stance | Key Data |
|---|---|---|
| OK Cupid | Founder's perspective (exited) | Founded 2003, EBITDA ~$6M when acquired by Match in 2011, EBITDA ~$40M when left in 2015; ~32-35 employees |
| Keybase | Investor (position not disclosed) | Founded by OK Cupid co-founders Chris Coyne and Max Krohn, focused on making encryption accessible to ordinary users |
| Online Vision Test Company | Advisor/Partner (position not disclosed) | Aims to allow users to get a prescription without visiting an eye doctor |
1. "Wooderson Law" (named by Rudder): Men of all ages consistently pursue women around age 20 in dating behavior, despite claiming they want peers — women are consistent in word and deed. Evidence: OK Cupid data shows 35-year-old men claim to want women aged 30-34, but actual behavior points entirely to women in their early 20s.
2. High-variance strategies outperform average strategies: At the same attractiveness level, those with the highest rating variance receive attention equivalent to a traditional 7 or 8 (out of 10). Evidence: Extreme traits like tattoos and nose rings filter for a "highly favorable" group, similar to bestsellers having both many 1-star and 5-star reviews.
3. "Four-Way Conversation" metric (named by Rudder): Choosing the wrong metric can doom a business — OK Cupid chose "number of deep conversations" over "page views" or "time spent." Evidence: Users visiting day after day without finding a partner is good for Twitter but a disaster for a dating site.
4. The fundamental difference between corporate and government data: Companies collect data for group advertising (never targeting individuals), while government surveillance targets specific individuals (with names, addresses, families). Evidence: "No company pulls up an individual's profile and asks, 'What will this person do next?' — that never happens."
5. The risk of superintelligence is not a "Terminator": AGI could make everyone "die more painlessly," rather than through violent means. Evidence: Requires a global "Manhattan Project"-style coordination, opposing secret development by a single government or company.
6. The role of luck in startup success is underestimated: Rudder admits, "We were smart and worked hard, but things could easily have gone the wrong way." Evidence: OK Cupid took from its founding in 2003 until 2008-2009 to truly take off, enduring years of "lean times."
7. The core of data science is not math, but asking the right questions: "The math isn't hard; the hard part is what to do with the math results." Evidence: The hardest part of writing Dataclysm was deciding "which questions can be answered? Which questions do people want to read the answers to?"
8. Automation will cause massive unemployment, requiring a tax on AI: Self-driving cars will put millions of drivers out of work, while Uber shareholders get richer. Evidence: Requires "a tax on AI usage, with the revenue redistributed to those displaced," or society will face upheaval.