I finally finished Antifragile: Things That Gain From Disorder, by Nassim Taleb, and I devoured every page of it, word for word. It’s been some time I enjoyed diving deep into a book. Hence, would like to really share what I took away from it, and later try to apply the key lessons into my work. 

Sharing them here as reference for myself, and for anyone who might find it useful. This is not a book review, just raw notes lifted directly from the book, with some minor edits, interpretations and categorisations of my own. This is part of my reading list for a new season. 

Read past notes:

• (1) Prologue
• (2) Book 1: Intro to antifragility
• (3) Book 2: Antifragility on a systems level and interventionism ?

-–

Antifragile, Book 2

Banker vs taxi driver: which job has more job security

Variability of taxi driver income doesn’t means less security. Central illusion in life is that randomness is risky. Artisans like taxi drivers, prostitutes, carpenters, tailors have some volatility in income but rather robust to minor black swans (that would bring income to a halt). Their risks are visible, but not so with employees, who can see their income become zero from the HR dept. Employee risks are hidden. 

Continuous stressors

Small variations make artisans adapt and change continuously by learning and under pressure to be fit. Continuous supply of stressors (which are info) makes them adjust opportunistically. They are also open to gifts, positive surprises, free options - the hallmark of antifragility. The longer one goes without market trauma, the more hidden vulnerabilities accumulate silently under the surface, and the worse the damage when it occurs.

Avoidance of small mistakes make the large ones more severe

The more variability you observe in a system, the less Black Swan-prone it is. Man-made smoothing of randomness makes things smooth, steady and fragile. Natural randomness smaller chance of large shocks, but daily variability. Mediocristan vs Extremistan - plenty of variations, but tend to cancel out in aggregate, vs a few variations but when it does, are extreme. One fluctuates, the other jumps. We don’t look at evidence for risks in Extremistan - evidence comes too late. Risk for Extremistan is in the future, at potential damage, not in the past. No stability without volatility.

Bottom-up variations

People handle local affairs in vastly different ways compared to large, abstract public expenditures. Humans scorn what’s not concrete. Small is beautiful in many ways, more antifragile than the large, which have a mathematical property dooming them to breaking. Manage noise, let it run its own course, instead of minimizing it. “Prefer war to prison.”

Turkey problem

A turkey is fed everyday for months, each day confirming that butchers love turkeys with increasing statistical confidence. Until Thanksgiving. Mistaking absence of evidence for evidence of absence, a mistake that tends to prevail in intellectual circles. Nuclear bombs are not safer because they explode less often.

Adding randomness to a antifragile system hungry for it

When systems are stuck in an impasse, randomness and only randomness can unlock them and set them free. Often absence of randomness equals guaranteed death. Ironically, chaotic systems can be stabilized by adding randomness to them. Adding randomly selected politicians can improve the functioning of the parliament. Athenian assemblies were chosen by lot, a method meant to protect the system from degeneracy. They exploit randomness.

Modernity is

Human’s large-scale domination of the environment, systematic smoothing of the world’s jaggedness, and stifling of volatility and stressors. Indeed in the past, we were not fully aware of antifragility and self-organization, and denied things can take care of themselves without agency, but we managed to respect these properties by constructing beliefs that managed survived uncertainty, imparting improvements to the agency of gods. The transfer of agency and control to humans created modernity and its fragility via concentration and magnification of human errors.

Naive, fragilizing interventionism and the principal-agent problem

Long term harm (“iatrogenics” or harmful unintentional side effects, or causing harm while trying to help) done by over intervention/treatment in hospitals is largely unaccounted for. Iatrogenics is where there’s low competence, high intervention and disrespect for spontaneous operation/healing. “Agency problem” / “Principal-agent problem” is when one party (agent) has personal interests that are divorced from those who are using his services (principal). A stockbroker/doctor’s ultimate interest is their own bank account, not your financial/medical health, and they will give you advice geared to benefit themselves.

Inverse-iatrogenics is antifragile

Someone who ends up helping while trying cause harm - banned books, hackers. Fragilizing, naive interventionism are pervasive in many industries, and depletes mental and economic resources such that it becomes rarely available when it’s needed most - eg editors over-editing but missing out real typos. Over-intervention in areas with minimal benefits (and large risks) usually paired with under-intervention in the areas that matter.

Examples:

- medicine: overtreatment, pharmaceutical addition not subtraction

- ecology: micromanaging forest fires

- politics: central planning, supporting rotten regimes for sake of stability

- economics: state buy-outs, optimization, illusion of economies of scale, ignoring 2nd order effects

- business: positive advice, focus on return not risk

- forecasting: forecasting black swans

- literature: copy editors changing your text

- parenting: tiger moms, removing random elements from child’s life

- education: pure interventionism

- media: high frequency sterile info

How to reduce interventionism

Intervening to limit size (eg of companies), concentration and speed are beneficial in reducing black swan risks. Attention is antifragile - less street signs and deregulation led to increased road safety (“Drachten effect”), more automation in planes makes pilots less alert. Have a systematic protocol to determine when to intervene and when to leave systems alone. There’s some ecological/naturalistic wisdom, benefits to nonaction, procrastination, voluntary omission, passive achievement (“wu wei” in Tao Te Ching), letting things take care of themselves and exercise their antifragility. Too much noise/data leads to too much neuroticism and therefore interventionism. So ration supply of info - the more data you get, the less you know what’s going on, and more iatrogenics you cause. 

Confusing catalyst as cause

When constrained systems fail due to fragility, failure is never seen as the result of fragility, but of poor forecasting, eg attribute a bridge collapse to the last truck that crossed it, or an intelligence failure for not foreseeing Egypt revolution. Hard to know which catalyst will produce which effect in interdependent systems.

Protection against prediction

What makes life simple is that the robust and antifragile don’t have to have as accurate a comprehension of the world as the fragile—and they don’t need forecasting. Redundancy is nonpredictive, or less predictive mode of action, eg extra cash in the bank means you don’t need to know with precision which event will cause difficulties like war, recession, unlike those in debt. Switch from blame from inability to see an event coming to failure to understand fragility: from “How did we not see it coming?” to “Why did we build something so fragile to these types of events?”