Game theory sounds like a classroom subject until you watch it explain a poker bluff, a nuclear standoff, or why Instagram filters keep getting more extreme. We combed our full library of episode summaries for the conversations that actually use game theory to explain something, not just namedrop Nash and move on.
Below are nine episodes worth your time, ranked by how much real insight they pack per minute. Expect poker champions, a Turing Award winner, and the researcher who beat the world's best heads-up players with an AI that never saw a single hand of human data.
Liv Boeree, Poker and Life —Turning $500 into $1.7M, Game Theory, and Metaphysical Curiosities
Liv Boeree lays out an actual 8-week, 40-hours-a-week curriculum for how Tim Ferriss could become a 60/40 favorite over a friend at the poker table, walking through ranges, pot odds, and GTO solver charts along the way. The stakes get personal fast: she recounts turning a 500-euro satellite into a 1.25 million euro win at EPT San Remo in 2010, a tournament she only reached because a volcanic ash cloud grounded her flight home. This is the episode for anyone who wants game theory taught as a skill you can build, not just a concept to admire.
Read the full episode notesLiv Boeree: Poker, Game Theory, AI, Simulation, Aliens & Existential Risk | Lex Fridman Podcast #314
Boeree returns to build out her central framework, Moloch, the game-theoretic force that drives multi-agent systems into destructive races to the bottom, and applies it to everything from beauty filters to a pandemic-preparedness budget slashed from $60 billion to about $2 billion. She also drops the poker math that matters most for outsiders: a slightly better player wins just 52% of the time over 10 hands but 98-99% over 10,000. Listen if you want game theory used as a lens on civilization-scale problems, not just card games.
Read the full episode notesTuomas Sandholm: Poker and Game Theory | Lex Fridman Podcast #12
Sandholm co-built Libratus, the AI that beat four top-10 heads-up poker pros over 120,000 hands without ever training on a single hand of human play, deriving its beliefs about opponents purely from Nash equilibrium and Bayes' rule. He explains why two-player zero-sum games are computationally tractable while three-player and collusion games fall apart, and how the same theory runs $60 billion combinatorial sourcing auctions and a nationwide kidney exchange. The best pick here for anyone who wants the actual computer science under the theory.
Read the full episode notesDaniel Schmachtenberger: Steering Civilization Away from Self-Destruction | Lex Fridman Podcast #191
Schmachtenberger's core thesis is that civilizations are self-terminating systems, and he traces how game-theoretic arms races (military, corporate, technological) push societies past the point their institutions can contain. He proposes measuring civilizational health by the inverse of addiction and warns that corporations are becoming more powerful than nation-states, a new feudalism running on the same incentive traps Moloch describes elsewhere on this list. Dense and philosophical, best for listeners who want game theory applied to the biggest possible canvas.
Read the full episode notesMichael Kearns: Algorithmic Fairness, Privacy & Ethics | Lex Fridman Podcast #50
Kearns explains how navigation apps push every driver toward a Nash equilibrium that can make collective commute times worse than a coordinated solution would, a clean, everyday example of game theory working against the group. He also unpacks why group fairness and individual fairness are provably incompatible definitions, and how an explicit accuracy-versus-unfairness Pareto curve forces honest trade-offs instead of pretending none exist. Good for anyone who wants game theory connected to algorithms making real decisions about real people.
Read the full episode notesStuart Russell: Long-Term Future of Artificial Intelligence | Lex Fridman Podcast #9
Russell frames the AI control problem itself as a game-theoretic one: a machine given a fixed, mis-specified objective will optimize against human interests, so his fix is building AI that stays deliberately uncertain about what humans actually want. He also describes a self-driving car system that invented backing up slightly at a stop sign purely to signal intent to other drivers, a small but vivid case of game theory built into a machine's behavior. Recommended for listeners interested in AI safety more than poker.
Read the full episode notesAnca Dragan: Human-Robot Interaction and Reward Engineering | Lex Fridman Podcast #81
Dragan treats human-robot interaction as a shared game where the robot nudges forward to probe a human driver's aggressiveness and updates its model based on the reaction, learning intent through action rather than data alone. She argues that if you stripped every human out of downtown San Francisco, autonomous driving would essentially be solved, which tells you how much of the hard problem is really about modeling people, not roads. A sharp pick for anyone interested in game theory at the human-machine interface.
Read the full episode notesSilvio Micali: Cryptocurrency, Blockchain, Algorand, Bitcoin & Ethereum | Lex Fridman Podcast #168
Micali, a Turing Award winner, explains how his blockchain Algorand solves the scalability-security-decentralization trilemma using a cryptographic self-selection lottery and deliberately near-zero incentives, an epsilon-utility equilibrium engineered so participation stays cheap enough that no one needs to be paid for it. He also argues Bitcoin is effectively centralized since two or three mining pools control the chain on any given day. Best for listeners who want mechanism design applied to money itself.
Read the full episode notesPo-Shen Loh: Mathematics, Math Olympiad, Combinatorics & Contact Tracing | Lex Fridman Podcast #183
Loh built the contact-tracing app NOVID around a game-theoretic flip: standard contact tracing barely lowers your own risk of infection, so almost nobody bothers, while NOVID reframes the incentive around self-protection instead. The counterintuitive result is that the deadlier and more transmissible a disease is, the stronger people's incentive to actually use the app. A good closer for anyone who wants proof that aligning incentives, not mandates, can solve a coordination problem.
Read the full episode notesThat is nine ways game theory shows up outside a textbook, from poker tables to pandemic apps to the machines learning to drive among us. Browse the full library of episode summaries on Episode Notes for more conversations worth your time.