Lex Fridman PodcastPo-Shen Loh: Mathematics, Math Olympiad, Combinatorics & Contact Tracing | Lex Fridman Podcast #183
At a glance
WHAT IT’S REALLY ABOUT
Po-Shen Loh on math, invention, and rethinking pandemic control systems
- Lex Fridman and Po‑Shen Loh discuss how genuine mathematical thinking is about invention and reframing hard problems, not memorizing methods, and how Olympiad-style challenges build that skill. Po explains his teaching philosophy, live-problem-solving approach, and why middle school is a crucial moment to help students experience creating their own solutions. A major portion focuses on NOVID, his privacy-preserving, network-theory-based app that reframes contact tracing as early-warning risk information aligned with personal incentives and freedom. They also touch on combinatorics, distributed algorithms, voting systems, and the broader question of pursuing long-term, high-impact work whose ideas outlast one’s own lifetime.
IDEAS WORTH REMEMBERING
5 ideasTeach math as guided invention, not method rehearsal.
Po argues students should regularly face problems they initially *can’t* do, with hints and dialogue that help them invent methods themselves; this builds a durable skill of creating solutions rather than memorizing procedures they’ll soon forget.
Use positive incentives in epidemic tech by serving self-interest.
NOVID doesn’t just tell you after exposure to quarantine for others’ benefit; it tells you how many ‘hops’ away active cases are in your physical contact network, so your selfish desire to avoid illness naturally drives adoption and cautious behavior.
Respect privacy by modeling *relationships*, not locations.
Instead of GPS, NOVID uses Bluetooth-based proximity snapshots over time to infer strong, recurring contacts, allowing effective network-based risk estimation while avoiding precise geolocation and preserving anonymity.
Think in terms of feedback loops and control, not just rules.
Po frames pandemic policy as a control-theory problem: classical contact tracing fights human incentives (removing people ‘against their will’), whereas informing people of approaching risk creates a feedback loop where individuals voluntarily reduce contacts.
Competitions can be training grounds for general problem-solving.
Math and programming contests, when focused on deep problems and efficient algorithms, train skills—like reframing, abstraction, and back-of-the-envelope complexity analysis—that transfer directly to startups, research, and real-world systems design.
WORDS WORTH SAVING
5 quotesI don’t want to ever just tell somebody, ‘Here’s how you do something.’ I prefer to say, ‘Here’s an interesting question… do you have any ideas?’
— Po‑Shen Loh
We changed the paradigm from ‘what already happened, quick damage control’ to ‘predict the future.’
— Po‑Shen Loh
Free market capitalism was not based on altruism. If you set up the incentives so that everyone maximizing their own situation helps the whole, that’s a game-theoretic solution.
— Po‑Shen Loh
If you solve any one of the six problems at the IMO, you’re a genius.
— Po‑Shen Loh
I wanted to maximize how many person‑years after I’m gone what I did still matters.
— Po‑Shen Loh
High quality AI-generated summary created from speaker-labeled transcript.
Get more out of YouTube videos.
High quality summaries for YouTube videos. Accurate transcripts to search & find moments. Powered by ChatGPT & Claude AI.
Add to Chrome