five live experiments
The book's research is blunt: abstract warnings barely work — experience does. So don't take the chapters' word for it. Fall into each trap yourself, then watch the math explain what just happened to you.
Behind one door, a car. Behind two, goats. Pick a door; Monty — who knows where the car is — opens a goat door and offers you a switch. In 1990 Marilyn vos Savant said switch, and ~10,000 readers, including nearly 1,000 PhDs, wrote in to tell the world's highest recorded IQ she was wrong. She wasn't. Are your instincts better than theirs?
Pick a door.
TOO SLOW? LET THE MACHINE PLAY 1,000 GAMES OF EACH STRATEGY.
The edge: you win by switching exactly when your first pick was a goat — which happens ⅔ of the time. Monty's reveal isn't random; it's information from someone who knows. Even Paul Erdős only believed it after a simulation. You just ran one.
You'll play 10 rounds against another prisoner. Each round: stay silent (cooperate) or betray (defect). Betrayal is the dominant strategy every single round — and if you both follow it, you both lose. In Axelrod's legendary 1980 tournament, the simplest program entered — Tit for Tat — beat every clever strategy. It's behind the glass, waiting.
Round 1 of 10 — your move.
THE DEAL ON THE TABLE (YEARS IN PRISON — LOWER IS BETTER)
| They stay silent | They betray | |
|---|---|---|
| You stay silent | 1 / 1 | 3 / 0 |
| You betray | 0 / 3 | 2 / 2 |
Whatever they do, betraying shaves a year off your sentence. That's a dominant strategy. Follow it together for 10 rounds and you'll each serve 20 years instead of 10. Welcome to the foundation of game theory.
Tit for Tat's whole source code: cooperate first; then copy their last move. Nice, retaliatory, forgiving, clear.
The edge: in one-shot games, betrayal rules. In repeated games with memory — supply chains, marriages, WWI trenches where enemies developed "live and let live" truces — cooperation becomes the winning strategy. The shadow of the future changes everything.
Martin Shubik auctioned dollar bills in his Yale classroom for decades and claimed he never lost money. One rule makes it a trap: the second-highest bidder also pays — and gets nothing. Bids rise in 5¢ steps. Your rival has already invested and hates losing. So will you, in about forty-five seconds.
The auctioneer holds up a crisp $1 bill. Opening bid is 5¢. In?
The edge: every individual bid is "rational" — it costs 5¢ to convert a total loss into a smaller one. The sequence of rational steps is collectively insane: Shubik's record topped $20. Vietnam escalation, takeover battles and strikes follow the same script. The optimal strategies are brutal: never enter, or pre-commit so overwhelmingly that nobody bids against you.
How many people must walk into a room before two of them probably share a birthday? Most people guess somewhere near 183. The truth is 23 — because the question isn't about your birthday, it's about all n(n−1)/2 pairs in the room. Fill the room and see.
The edge: 23 people make 253 pairs — each a lottery ticket for a match. The same "pair explosion" let researchers break SHA-1 cryptography with a birthday attack, and it's why eerie coincidences in your life are statistically mandatory, not mystical.
Keynes said markets aren't about picking the prettiest face — they're about picking the face others will pick, knowing they're doing the same. The classic lab version: everyone picks a number from 0–100; the winner is whoever lands closest to ⅔ of the average guess. You're up against a simulated crowd of 100 players with realistic levels of reasoning. Choose.
The edge: if everyone were perfectly rational, the logic spirals — 50 → 33 → 22 → … → 0, the Nash equilibrium. Real crowds average around 22–33, because most humans run one or two steps of reasoning, not infinity. Winning means being exactly one step smarter than the crowd — not infinitely smarter. That gap between equilibrium and reality is where bubbles live.