Freakonomics

nerds like to measure things, but the game is knowing what to measure — measure wrong and you optimize something that shouldn't exist: Goodhart's law: "when a measure becomes a target, it stops being a good measure." teachers (altering answer sheets under high-stakes tests) vs sumo (7-7 guy beats the safe 8-6 guy way more than chance — match's sold, repaid next tournament). same disease, different arena.

incentives reward cheaters when payoff is big AND detection is low (even when high): that's the model — see how promotion works in most companies (ask yourself). counterweight: Feldman's bagel data (honor-system, near-zero detection, people still pay). incentives predict cheating, but honesty survives anyway. keep both.

distrust the single-cause story: the 90s crime drop is the case study - everyone grabbed an obvious cause, Levitt floated a clever non-obvious one... the Roe v. Wade (1973) → fewer unwanted kids → ~18 yrs later that cohort never hits peak crime age. but his version got contested too (Foote & Goetz found a coding error; real cause is probably multi — policing, crack, lead, economy). the move isn't believing him, it's refusing any single-cause explanation.

mental model: long-lag causality — one event, effect 18 yrs later; that gap is why the story's easy to get wrong either way.

information asymmetry leverage: experts exploit what they know and you don't. real-estate agents sell their own houses for more; KKK gutted once Stetson Kennedy leaked their secrets. when an "expert" advises you, ask whose incentive their info serves. a good exercise here as well is to ask "what's the limit time to look for an slot while looking for parking"?

parenting + emotion vs rational argument: emotion usually wins, but should it? gun vs pool — pool kid is ~100x more likely to die, but parents sort of fixate on the vivid rare risk (gun) over the boring likely one. risk assessment matters EVERYWHERE (read elon's book from isaacson). i think it's worth to also become a sort of "regression analysis" expert here + check Jesse Zhang's post but more importantly the genealogy and what they did?