Summary

Antifragile things get stronger from stress, chaos, and randomness – the opposite of fragile, and fundamentally different from robust. Hormesis demonstrates the principle: a small dose of harm is beneficial. Every form of organic life is antifragile in some way, and stressors are information. Society has become too cautious to exploit this: we deny randomness, over-treat (iatrogenics), and mistake the absence of volatility for stability – like the Thanksgiving turkey that confuses feeding with safety. The key insight is antifragility by layers: economies are antifragile because they're composed of fragile startups (natural selection at scale). Systems beat predictions. Acute stress strengthens; chronic stress kills. Nature is the master of high-dimensional trial and error – and strategic planning is just superstitious babble.

Key Takeaways

Notes

Prologue

• Use randomness, uncertainty, chaos
• Antifragile: gets better with stress, chaos, randomness, uncertainty
• “We’re better at doing than thinking”
• ↑ is a solution to the black swan problem, since sensitivity to harm from volatility (antifragility) is tractable
• Society has become to cautious, risk-averse to take advantage of antifragility
• Life is much more random than we give it credit for

1: The antifragile: an introduction

• Antonym of fragile doesn’t exist
• Hormesis: when a small dose of a harmful substance is actually beneficial for an organism
• Post-traumatic growth: ↑
• Examples
  • Necessity really is the mother of invention
  • We focus better with slight background noise
  • Redundancy / reserves
  • Love, clingy-ness
  • Secrets, banned books
  • Criticism of our identity, ideas, people
  • Reputations of creative people (versus business people inside systems)
• Theory: every form of organic life is in some way antifragile
• Stressors are information (eg. hormones)
• Acute > chronic stress is beneficial
  • “Use it or lose it”
• Complex biological systems are opaque
• We need randomness / volatility: use it or lose it (like a whirlpool needs motion)
• Antifragility by layers / hierarchy:
  • Eg. economy is antifragile because of collective of fragile startups
  • ↑ natural selection
  • The selfish gene
  • Fractal
• ↑ system comes at the cost on the individual (anti-humanist?)

2: Modernity and the denial of antifragility

• “Central illusion in life: randomness is risky”
• Employee’s risks are hidden
• Volatile vs black swan systems
  • Eg. Thanksgiving turkey
• Value of local gov’t > central gov’t
• Randomness in markets is good, otherwise any movement would be over-perceived
• Iatrogenics: over-treatment, harm from the healer (must be factored in)
  • Doctors kill more people than cancer
• Principal agent problem: ↑
  • Eg. newspapers, social media
• Procrastination is a sign you shouldn’t do it at all
• Need to ration your supply of information
• Systems > outcomes: antifragility > prediction

3: A non-predictive view of the world

• Stay robust to how others treat you
• Stoicism
• Mentally write off your belongings
• The domestication (not elimination) of emotions
• Not averse to wealth, just giving with no expectation of return
• Antifragility = more to gain than to lose = more upside than downside = favorable asymmetry
• GDP growth via debt is a fragile system
  • Can all go to zero, thus not a viable strategy long term
• Barbell strategy: high + low risk only
• ↑ “domestication, not elimination, of uncertainty”

4: Optionality, technology, and the intelligence of antifragility

• The teleological fallacy: the illusion that you know where things are going (narrative)
• Option: explicit asymmetry
• Unlike financial options, more “interesting life options” are free
• Progress comes from increasing dispersion / inequality (courage)
• Implementation is harder than invention
• Research → prosperity: causation or just correlation?
• The conflation problem
• Practitioners don’t write, lecturers do. So history is written by “the losers”
• Epistemic base useful in some contexts
  • Eg. physics v cooking (apprenticeship)
• VC > centrally funded research: bet the jockey, not the horse
• “Strategic planning is just superstitious babble”
• Domain specificity: skill at school, games etc. comes at the cost of things outside it

5: The nonlinear and the nonlinear

• Fragility is generally nonlinear
  • Eg. 30-foot fall vs 10x 3-foot falls
• Fragile is hurt a lot more by extreme events
• Convexity effects:
• Squeeze: negative convexity effect of getting too big
  • Eg. banks fire selling with small vs large, market-moving holdings
• Planning fallacy is inherent in nonlinear structure of projects
  • Eg. Airplane early vs significant delay
• Wealth means more headaches
• Concavity / convexity bias:

6: Via negativa

• Disinformation is more rigorous than confirmation
• Negative knowledge is more robust
• 80/20 → 99/1 with winner-take-all effects
• “Less is more”
• ↑ eg. 1 argument against is stronger than multiple / a whole case
• We rely on more old technologies than we give ourselves credit for, eg. fire, knives, chairs, cabs
• The Lindy effect: still relevant objects have 2x the life expectancy (old objects more valuable)
• Survivorship bias
• We notice change, not statics
• Treadmill effect: compulsion for the new
  • Doesn’t apply to antiques
• Can’t be new in a standard way (eg. another Einstein)
• More Iatrogenics in medicine*
• “Nature is the master of high-dimensional trial and error” - evolution vs top-down genetic engineering
• Brain, body, nature is complex and less explainable than we think
• Need to remove steadiness in food intake
• Benefits of autophagy
• “If I had to find the anti-me, with diametrically opposite ideas and lifestyle on the planet, it’d probably be that Ray Kurzweil fellow”

7: The ethics of fragility and antifragility

• “Skin in the game is the only true mitigator of fragility”
• Knowledge work separates decision maker from consequences
  • Eg. president being commander and chief
• Post-dictors (predictors after) have weight of incorrect predictors beared by society
• Iatrogenics of experts ↑
• “Institutions block evolution”
• Decimation: 10% of army put to death in case of retreat so it’s not the dominant strategy
• Soul in the game > skin in the game
• Stock incentives favor volatility > absolute gain (sell high)
• Corporate execs are actors
• Big companies optimize for cost
• Big orgs are so fragile long term that they collapse under the agency problem
  • Eg. no negative manager bonus comp
• Moral hazard:
• Fooled by (big) data, bias
• “All marketed products are inferior”(?)
• Regulators / revolving door agency problem
  • The incentive is to have complex regulation (loopholes)