Why aren't smart people happier? - Seeds of Science

• IQ correlates with academic/career success but shows zero (or slightly negative) correlation with happiness across multiple large studies
• Spearman's 1904 "general intelligence" theory is wrong: IQ tests only measure well-defined problems (stable rules, clear answers, repeatable) because standardized tests require indisputable solutions
• Life's crucial problems—finding meaning, building relationships, raising kids—are poorly-defined: unstable rules, unclear boundaries, non-repeatable situations that require wisdom, not IQ
• Even genius-level IQ people make catastrophic life decisions (Holocaust denial, conspiracy theories, sexual harassment) because high IQ doesn't help with poorly-defined problems
• Society's century of solving well-defined problems (technology, disease) hasn't increased happiness at all, suggesting we're optimizing the wrong metric
• AI will dominate well-defined problems but remain hopeless at poorly-defined ones—it can predict next words but can't figure out "the moon is a rock, not a god"
• We systematically undervalue wisdom (your grandma who raised a loving family) compared to IQ-smarts, missing the skills that actually matter for living well

Adam Mastroianni challenges the foundational assumption of intelligence research: that there's a general mental ability underlying all problem-solving. While IQ reliably predicts academic and career success, multiple studies show it has zero correlation with happiness—the General Social Survey even found a slight negative correlation (r = -.06) across 50 years and 30,000 people.

The problem traces to Charles Spearman's 1904 observation that students who excel in one subject tend to excel in others. He concluded there must be general intelligence, but Mastroianni argues Spearman only varied the surface features (math vs. French vs. music) while keeping the problem structure identical. All IQ test items share crucial properties: stable relationships between variables, no disagreement about solutions, clear boundaries, and repeatability. These are "well-defined problems"—you can write instructions to solve them, which is why they work on standardized tests.

But life's hardest problems are poorly-defined. "How do I live a good life?" has unstable rules (what makes you happy makes me miserable), unclear boundaries (anything could affect happiness), non-repeatable situations (what worked at 21 fails at 31), and no agreed-upon solutions. Well-defined and poorly-defined problems require completely different cognitive skills—which explains why Christopher Langan (stratospheric IQ) believes 9/11 conspiracies, Bobby Fischer (chess genius) denied the Holocaust, and elite professors commit sexual harassment or fabricate data.

The distinction also explains why a century of solving well-defined problems—eradicating diseases, landing on the moon, gaining 15 IQ points—hasn't budged happiness levels. And it predicts AI's trajectory: it will dominate well-defined problems (predicting next words, matching images) but remain hopeless at poorly-defined ones (recognizing the moon is a rock, not a god). The essay ends by arguing we systematically undervalue poorly-defined problem-solving—calling it "folksy" wisdom when your grandma knows how to raise a loving family and make perfect pie, skills far more important than acing vocabulary tests.