• Classify any system, career or decision as fragile, robust or antifragile — and re-run the map per domain, because the classification does not travel.
  • Measure fragility instead of forecasting risk, using a property of the thing in front of you rather than a prediction about the future.
  • Apply the Barbell Strategy to make ruin impossible on one end so the speculative end can be genuinely aggressive.
  • Detect the Turkey Problem in your own track record, and respond by changing your exposure rather than by forecasting better.
  • Identify iatrogenics — net harm from someone trying to help — and separate it from intervention that is actually warranted.
  • Decide when to intervene and when to abstain, using harm that is systemic and fast as the trigger and the smallest unit as the actor.
  • Apply via negativa: improve health, work and decisions by removing what harms before adding what helps.
  • Distinguish Mediocristan from Extremistan, and stop applying averaging logic in the domain where one event moves the total.
  • Build optionality out of asymmetry plus rationality, so you gain without needing to know what happens next.
  • Spot a fragile system by its nonlinearity — harm that accelerates faster than the shock that caused it.
  • Evaluate any adviser, institution or recommendation with the skin-in-the-game test: who carries the downside if it is wrong.
  • State the limit of every concept you use — the dose, the survivability condition, the domain — so you stop applying it where it fails.