Goodhart's Law
Any metric that starts as a rough proxy for something you care about degrades once you tie rewards to it. People optimize for the number, not the thing. Test scores, LOC written, closed tickets, likes — all Goodhart-vulnerable.
Card 1 of 8: Goodhart's Law
Any metric that starts as a rough proxy for something you care about degrades once you tie rewards to it. People optimize for the number, not the thing. Test scores, LOC written, closed tickets, likes — all Goodhart-vulnerable.
You encounter a rule, a process, or a fence in a field with no obvious purpose. The naive move is to remove it. Chesterton's rule: first understand why it's there, and only then decide whether it's safe to remove.
A principal hires an agent to act on their behalf — shareholders hire executives, homeowners hire contractors, citizens elect officials. The agent has more information and different incentives, so their decisions systematically diverge from what the principal would choose. Every institution runs on layers of this.
The British in colonial Delhi paid a bounty for dead cobras to reduce cobra populations. People started breeding cobras. When the government cancelled the program, the breeders released their now-worthless snakes — and the cobra population increased. The name stuck.
If you run a school cafeteria, the order in which you arrange the food determines what the kids will eat. If the apples are at eye level and the cake is hidden in the back, the kids will eat more apples. The cafeteria director is a 'Choice Architect.'
A 'Nudge' makes it easier for you to make a good decision (like automatically enrolling you in a retirement plan). 'Sludge' makes it intentionally difficult for you to do something (like a 15-page form required to apply for financial aid, or having to call a phone number during business hours to cancel a digital subscription).
Versailles was designed on paper before anyone visited the site — its geometry imposed on the land, the land's own contours flattened to fit. The English landscape tradition did the opposite: walk the ground first, then ask what it's already trying to become.
128 to 192 hours to build one meaningful friendship from scratch, a roughly 70% chance it fails within two years anyway, versus investing that same time in friendships that already cleared that bar.