uncertainty

Judge a decision by the distribution of outcomes it could have produced, not the one that happened to occur.#alternative-histories

Comparisons to past technologies establish the range of possible outcomes, not a probability over them.#analogies-are-rangefinders

Robust survives shocks; antifragile gains from them — so ask not whether a thing will break but which way volatility moves it.#antifragility-vs-robustness

History is driven by a few unpredictable outliers that we explain confidently only after they happen.#black-swans-rule-history

Every path away from foundation models becoming commodity infrastructure requires something to change; commoditisation is the default.#commodity-or-monopoly

When the reference class does not exist yet, ask which outcome is the default and what would have to change to get another.#default-vs-what-must-change

A small chance of ruin is not a small risk — once you are out of the sequence, the expected value you would have collected goes with you.#ergodicity-and-ruin

Real uncertainty is not a casino — dice have known odds, life does not, and models built on games mislead exactly where it matters.#ludic-fallacy

Probability theory works on repeatable events; the decisions that matter most are usually one-offs it was never built for.#measurable-vs-unknowable

Some domains are bounded and some are scalable — using bell-curve intuitions in the second is the central error of modern risk management.#mediocristan-vs-extremistan

We compress the past into stories that feel explanatory and predict nothing — coherence is not evidence.#narrative-fallacy

Quantifying a risk changes how it feels, not how likely it is — the number’s real work is often psychological.#numbers-domesticate-fear

Regression to the mean is real but useless as a timing device — knowing the average pulls you back says nothing about when.#regression-is-not-destiny

Risk is a modern invention — for most of history the future was fate, and the idea it could be measured is only a few centuries old.#risk-was-invented

The record was written by what survived — the failures that would refute the thesis left no documents.#silent-evidence

You only ever see the winners — the graveyard of identical strategies that failed is silent and uncounted.#survivorship-bias

A tiny number of events produce most of the outcome — so most of what you do is supposed to not matter.#tails-drive-everything

Given enough participants, spectacular track records appear by chance alone — success is evidence of skill only after you count the entrants.#track-records-are-noisy