#28 Michael Mauboussin: A Decision Making Jedi

Episode Summary

Executive Summary: Michael Mauboussin discusses decision-making under uncertainty, arguing that good judgment comes from blending base rates with inside views, using checklists, and understanding when intuition is valid. He also explores noise in organizations, AI’s promise and risks, complexity science, scaling laws, reading, sleep, parenting, and how to build a life and career around strengths, process, and continuous learning.

Main Topics: Base rates, outside view, and regression to the mean (Priority: 5/5): Mauboussin explains Kahneman’s inside-versus-outside view and argues that effective forecasting starts with base rates, then blends in experience depending on the amount of luck versus skill in the domain. Decision-making systems, checklists, and noise (Priority: 5/5): He says organizations should formalize decision processes, because humans introduce large inconsistency ('noise') even when rules are clear. Checklists and algorithmic defaults can improve quality while still allowing limited judgment. Intuition, expertise, and deliberate practice (Priority: 4/5): Intuition is powerful only in stable environments with quality feedback. Expertise requires deliberate practice and cannot be acquired casually by observing others or relying on tenure alone. Complexity, scaling laws, and changing environments (Priority: 4/5): Mauboussin discusses the Santa Fe Institute, increasing returns, and how complex systems exhibit scaling behavior. He uses these ideas to reason about cities, corporations, innovation, and environmental change. AI, algorithms, data concentration, and incentives (Priority: 4/5): He is optimistic about algorithms but concerned about non-stationarity, overfitting, and the concentration of data/resources in big tech companies whose incentives may not align with societal good. Reading, learning, sleep, and life balance (Priority: 3/5): He describes a reading-centered life, the value of synthesis and curiosity, and the importance of sleep, exercise, and diet for productivity and well-being. Parenting, humility, and self-knowledge (Priority: 3/5): He updates his views on parenting, emphasizing peer groups, letting children solve their own problems, and learning from failure by finding work aligned with one’s strengths.

Key Arguments: Base rates improve forecasts because most people overweight their own perspective; the right mix of outside and inside views depends on how much skill versus luck exists in the domain. Regression to the mean can be operationalized statistically by comparing observed variance to what would be expected under pure randomness. Organizations are noisier than leaders think; systematic processes can still produce wildly inconsistent outcomes because humans deviate from the script. Checklists are powerful because they force comprehensive thinking and reduce omitted steps, especially in high-stakes decisions. Intuition should be trusted mainly in stable domains with repeated patterns and strong feedback; otherwise, it is likely to mislead. Deliberate practice is necessary for real expertise; you cannot simply absorb domain knowledge by proximity or exposure. As one moves up organizations, decisions generally become fewer, more consequential, and more luck-laden, making process and feedback more important than raw results alone. AI and algorithms can be extremely effective in predictable environments, but their value is constrained by changing systems, overfitting, and misaligned incentives. Data scale may already create durable advantages for large technology firms, raising questions about competition and whether older data should be treated more like a public good. Complex systems research is useful because many real-world outcomes emerge from interactions among agents rather than isolated linear causes. Reading widely and writing are forms of synthesis that deepen understanding; time allocation should reflect opportunity cost and personal strengths. Sleep, exercise, and diet are foundational inputs that amplify performance across all other domains. Parenting should shift from solving problems for children to helping them solve problems themselves, while recognizing the strong role of peers during adolescence.

Data Points: Podcast guest history: First podcast guest almost 2.5 years earlier - Shane Parrish notes Mauboussin was the first guest on The Knowledge Project, roughly two and a half years before this episode. NBA variance example: Teams can be analyzed by observed win-loss variance versus a pure-luck binomial model - Used to infer how much of league outcomes come from skill versus randomness. Noise study variance: 40% to 60% variance - Kahneman-style insurance claim study found far more outcome dispersion than managers expected, who predicted only 5% to 10%. Manager expectation of noise: 5% to 10% variance - Insurance managers estimated only small differences among similarly trained adjusters before the study results were shown. Sleep target: 8 hours - Mauboussin says he is religious about getting about eight hours of sleep per night. Alcohol intake: Rarely more than one drink in a day - He says low alcohol consumption helps sleep and overall well-being. Book collection size: 3,000 to 4,000 books - He keeps books in two offices and describes them as a source of comfort and ongoing learning. Reading pace: 25 to 30 books per year - He says he is around 25 or 30 books in 2017 when discussing his reading habits. Reading split: 90% new, 10% rereading - Most reading is new material, though research often requires revisiting older books and articles. Five children: 5 - He mentions having five kids, with most now out of the house. Energy use benchmark: 100 watts/day - Human metabolic energy use should be about 100 watts per day in the scaling-law discussion. U.S. average energy use: 11,000 watts/day - He says average U.S. energy usage far exceeds biological needs due to technology. Global average energy use: 3,000 watts/day - He contrasts U.S. usage with global averages. Energy footprint multiple: 30x body mass energy footprint - He summarizes that modern humans have an energetic footprint about 30 times their biological requirement. Career failure timeline: Early 1988 - He describes his failed post-college stockbroker role as beginning in early 1988. Training program duration: 1.5 years - He spent a year and a half in investment-bank training before starting his first job.

Pivotal Quotes: "If there's a lot of luck, a lot of randomness, you should rely exclusively or almost exclusively on the base rate." — Michael Mauboussin: Explaining when to use the outside view versus the inside view in forecasting and decision-making. "The key is to try to be as systematic about that as possible." — Michael Mauboussin: On designing organizational decision processes and evaluating the quality of decisions. "People were just completely inconsistent in applying the basic rules and the basic algorithms." — Michael Mauboussin: Describing the insurance-company study that illustrates noise in human judgment.

Implications: Listeners should build decisions around base rates, process, and feedback rather than intuition alone. For firms, the big challenge is reducing noise, adapting to changing environments, and using algorithms wisely without overfitting or ignoring incentives.