How Much Does Google Know About Me? - Seth Stephens-Davidowitz - #134

Episode Summary

Executive Summary: The conversation explores Seth Stephens-Davidowitz’s work using anonymous digital traces—especially Google, Pornhub, and other platforms—to reveal what people actually think, want, and fear, often contradicting surveys and social norms. Topics include sexuality, dating, suicide risk, voting behavior, parenting, and happiness, with a recurring theme that big data exposes hidden human behavior and can inform more effective, socially useful decisions.

Main Topics: Big data as a truer window into human behavior (Priority: 5/5): Seth explains that anonymous search and platform data often reveals behavior more accurately than surveys, because people lie or shade answers in public and even in anonymous questionnaires. Pornography and sexual fantasy analysis (Priority: 5/5): A large portion of the discussion focuses on Pornhub and Google search data as a revealing source for studying sexual desires, taboos, and regional differences in fantasy. Dating, attraction, and relationship signals (Priority: 4/5): The podcast covers data-driven findings about first dates, including speech patterns that predict mutual interest, and challenges common beliefs like ‘opposites attract.’ Mental health, stigma, and suicide risk (Priority: 5/5): Seth discusses how search patterns around suicide can reveal triggers such as herpes diagnoses and how people seek validation through celebrities with similar conditions. Parenting, neighborhoods, and role models (Priority: 4/5): He argues that where children are raised and who they see around them may matter more than direct parental lecturing, because peers and neighborhood role models shape aspirations. Politics and election prediction (Priority: 3/5): The conversation touches on using search data to anticipate voter behavior and election outcomes, including the idea that search order may signal subconscious preference. Using data for social good and future research (Priority: 4/5): They conclude by noting that data science is expanding beyond profit-making into purpose-driven work, with potential applications in public health, welfare, and policy.

Key Arguments: People lie in face-to-face settings and even in surveys, so digital behavior is often a more honest signal of intent and desire. Anonymous search data can predict real-world actions, such as voting, relationship problems, or sexual interests, better than self-reporting. Pornography data offers an unprecedented view into sexual fantasy because it captures what people actually choose privately, not what they say publicly. Some sexual preferences appear surprisingly stable across cultures and may be tied to childhood exposure, environment, or imprinting rather than simple politics or gender equality. Dating success can be inferred from conversational patterns; for example, women laughing at jokes, using fewer hedge words, and talking more about themselves are associated with better outcomes. Suicidal ideation in young people can be driven by social shame and insecurity; search data suggests stigmatized diagnoses like herpes may be a major trigger. Role models outside the family—especially neighborhood peers and admired adults—may shape children more strongly than direct parental instruction. Big data can confirm common sense, but its value is often in overturning intuition and revealing counterintuitive patterns. The same methodology that helps understand taboo or stigmatized topics may also improve policy design, public health messaging, and personal decision-making.

Data Points: Women’s porn viewing share of lesbian content: about 20% to 25% - Seth says roughly one-fifth to one-quarter of pornography views from women are for lesbian porn. Male gay porn viewing share: about 5% - He estimates gay porn makes up around 5% of male porn viewing, varying slightly by site. Self-reported gay identity in conservative states: about 1% to 2% - He cites Mississippi, Alabama, and Tennessee as places where only 1–2% of men say they are gay. Self-reported gay identity in liberal states: about 4% - He contrasts that with New York and California, where about 4% of men say they are gay. Happiness boost from alcohol: 3 to 4 points on a 100-point scale - He references happiness studies showing drinking can raise reported happiness modestly. Facebook fandom gap for Black NBA players: much more popular; especially large among African Americans - He says Black players gain fans across groups, with the biggest boost among Black users. Election prediction for Bloomberg: 1% in Nate Silver’s model; Seth suggests 5% to 10% - He argues Bloomberg’s unprecedented resources may mean the model underestimates him. Age group vulnerability: young people, especially 18–19 year olds - He repeatedly notes that adolescents and young adults are especially prone to paranoia and stigma-related distress.

Pivotal Quotes: "People kind of put on a very presentable front, but in the privacy of their own home on their Google search engine or the websites they visit, they kind of just show a different side of themselves." — Seth Stephens-Davidowitz: Explaining why anonymous digital data can reveal more truthful human behavior than surveys. "I think creativity is a huge part of data science. Otherwise, you kind of just drown in the data and you don't really know what's interesting." — Seth Stephens-Davidowitz: Describing the interpretive skill required to find meaningful insights in large datasets. "I think the way I read the evidence is you can't mold your kids too much because they can rebel. You have to do it more subtly." — Seth Stephens-Davidowitz: Discussing parenting, neighborhood effects, and the influence of role models outside the home.

Implications: The episode suggests big data can reshape social science, marketing, health policy, and personal choices by exposing hidden behavior and stigma. It also warns that interpretation matters: raw data is powerful, but only when used creatively and responsibly.