Book summary: Antifragile by Nassim Nicolas Taleb
Another great read in the Incerto series by Taleb. The core idea is that certain systems benefit from uncertainty. And our goal should be to make all systems antifragile, so that, they can benefit from uncertainty.
Fragile breaks under stress. Robust is unimpacted by stress. Antifragile benefits from stress.
- Fragile breaks under stress. Robust is unimpacted by stress. Antifragile benefits from stress. Glass is fragile, wood is robust while human muscles are antifragile (up to certain limits). Anything that has more upside than downside from a random event is antifragile; the reverse is fragile.
- It is far easier to figure out if something is fragile than to predict the occurrence of an event that may harm it. Fragility is a rachet, damage is irreversible. a broken package does not repair itself.
- Nature likes to overinsure itself against complete unknowns. Nature assumes that never-seen-before harm is possible.
- The secret of life is antifragility. Everything that has life is to some extent antifragile (but not the reverse). Relying on predictions makes you fragile to prediction errors.
- One should not have antifragility at the expense of the fragility of others. Modern-day “too big to fail” banks have antifragility while they transfer fragility to the customers and society at large.
- Phenomenology is the observation of an empirical regularity without a visible theory to it. Theories are fragile, phenomenology is robust. Our understanding of biological processes has led to declining in pharmaceutical discoveries over time.
- Social sciences are fragile; physical sciences aren’t. During the cold war, the Physics department of the University of Chicago and Moscow agreed with each other while the economics department taught the exact opposite of each other.
- 100% medium-risk securities have a wipe-out risk. 90% cash + 10% high-risk securities don’t, they are antifragile. This barbell approach is safer.
- For the antifragile, the good news is missing from the past data. For the fragile, bad news does not show up easily. The reinsurance business is antifragile. One single episode of the Asbestos liabilities bankrupted the Lloyd family while Harvard professor Kenneth Froot thought that the reinsurer made too much profit.
- Philosophers talk about truth and falsehood. Real-life cares about payoffs. They are not necessarily the same thing.
- Human lungs when put on mechanical ventilators with constant pressure fail. Dispensing high pressure on occasions and low pressure at other times is much better. It also opens up collapsed alveoli. Humans are antifragile to lung pressure.
- A complex regulation is fragile. The regulators who know the loops and glitches would use it later as his differential edge for unethical but legal gains.
- Systems that emerge over time make small errors. Designs, however, make large errors.
- Lindy Effect – A 1000-year-old book in publication will last another 1000 years. A 10-year-old will last another 10. The older something survives, has a reason for its survival, the longer it will last.
Wealth leads to the rise of education and not vice-versa.
- Technology is the result of antifragility, exploited by risk-takers, in the form of tinkering and trial and error, with nerd-driven design confined to the backstage.
- An overconfident entrepreneur takes a heroic risk that might result in a benefit for others. An overconfident scientist whose risk computations lead to the Fukushima disaster leads to negative Black Swans. The former overconfidence is good, the latter is bad.
- Growth in society may not come from raising the average but increasing the number of people in the tails. The very small number are crazy enough to have ideas of their own, imagination, and courage to make things happen. Life is long gamma (positive/negative changes into inputs lead to exponentially positive/negative outcomes).
- Greeks had steam-based turbines, an aeolipile, for amusement. But not till the industrial revolution, it was used for industrial work. Mayans had wheels, for children’s toys while they used human labor for agriculture. Practical inventions create theoretical ancestry. The significance can only be revealed through practice though. The naive rationalism though encourages academic knowledge over uncodifiable, more complex, intuitive, and experience-based knowledge.
- Like Britain, in the industrial revolution, America’s asset is risk-taking with the use of optionality without the fear of failing.
- Wealth leads to the rise of education and not vice-versa. In 1960, Taiwan had a much lower literacy rate than the Philippines and half the per-capita income. Today, it has Taiwan has 10 times the income of the Philippines. Korea had a much lower literacy rate than Argentina and one-fifth of the per-capita income, today it has three times as much. Over the same period, sub-Saharan Africa saw a markedly increased rate of literacy all the while with a decrease in their standards of living. Universities prosper as a consequence of the national wealth and not vice-versa.
- It is better to make the world greed-proof and make society benefit from the greed and other perceived defects of the human race than to eliminate it.
- Human learning is context-dependent. We learn in a situation and then fail to apply the same learning outside that situation.
- Humans are bad at filtering short-term information. Procrastination is a way of filtering.
- Excess wealth, if you don’t need it, is a heavy burden. When you become rich, the pain of losing your fortune exceeds the emotional gain, so, you start living under continuous emotional threat. Seneca’s approach was to mentally write off such possessions, so that, when losses occur, he won’t feel the sting.
- Wisdom is more important than knowledge in decision-making.
- Teleological fallacy – you know where you are going, you know where you were going, and others have succeeded in the past by knowing where they are going. Flaneur is not a prisoner of the plan. He decides on the flay.
- Coca-cola began as a pharmaceutical product.
- Tiffany, the jewelry company as a stationary store.
- Raytheon, the missile maker, was a refrigerator maker.
- Nokia as a paper mill.
- DuPont was an explosives company.
- Does the scientific researcher whose ideas apply to the real world, applied to his daily life? If so, take him seriously. An academic claimed making beyond $50,000 does not bring additional happiness. However, he was always looking for paid speaking engagements despite his salary being twice that. This alone concludes that his research was theoretical.
- Champagne Socialists live a luxurious life while preaching austerity for others.
- Someone speaking against their self-interest should be given weightage. For example, if a big pharma executive says that Diabetes can be cured with starvation.
- Never ask someone for their opinion or recommendation. Ask what’s in their portfolio.
Narratives are dangerous
- Optionality is fore-thinking while narratives are after-thinking (cherry-picking to fitting theories into past events after the events). We fall for narratives and overestimate the role of good-sounding ideas.
- Cherry-picking or Fallacy of confirmation – The one telling and publishing the story has the advantage of showing confirmatory examples and ignoring the rest.
- People who rely on really complicated tricks miss elementary things. During the Iraq war in 2003, many fund managers had a war-room setup and missed out on the basic fact that a pre-announced war has already been priced into the oil prices. The oil prices, instead of rising, crashed.
- A child does not need to solve aerodynamics equations to ride a bicycle. A Chicago pit trader does not need to know the Girsanov theorem either.
- Most technologies, especially complex ones, are like cooking than Physics. Computer technology advanced unpredictably and in a self-directed way, the “science” helped but it never set the direction, rather it was a slave to it.
- History has been written by those who want you to believe that reasoning has a (near) monopoly on the production of knowledge.
- Schools deprive people of erudition by squeezing people into a narrow set of authors.
- Obvious decisions require no more than a single reason.
- What is being marketed is necessarily inferior or it won’t require marketing. That’s why Coca-cola has to create a campaign to associate itself with happiness.
It is illusory to believe that randomness is risky.
- A taxi driver’s income is random, some good days and some bad days. While an employee’s income is fixed but the risks are hidden. An employee’s income can go to zero after a single call from the HR department. The centralized state resembles an employee; the city-states resemble a taxi driver.
- Every plane crash makes the next one less likely. However, every bank crash makes the next one more likely due to the interconnectedness of the global economic systems. We need to eliminate the second type of error. The restaurant industry is anti-fragile as the failure of one does not lead to another while in the banking industry it does. When randomness is distributed across a large number of small units, we get benign Mediocristan. When it concentrates, we get the sneaky Extremistan.
- When some systems are stuck in a dangerous impasse, the only randomness can unlock and set them free.
- Adding a certain number of randomly selected politicians can improve the political process.
- In complex systems, there is no stability without volatility. Reducing American intervention can make other countries more stable.
- Data is toxic in moderate to large quantities. Unfortunately, it is ample in the modern world.
- Artists, authors, and even philosophers are much better off having a small number of fanatic supporters than a large number of those who appreciate the work.
- Averages are of no significance if one is fragile to variations.
Size invariance – things that don’t scale with the size
- Some transformations are not scale-invariant, they behave marked differently at a 100x scale. A large state is not a giant municipality, just like a baby does not resemble a small adult. We, humans, have traditionally lived in small tribes and know how to manage the small units better than the large ones. That’s why a federal system like Switzerland where the cantons are relatively independent in decision-making work so well. Stalin could not have existed in a municipality. By creating large nation-states, we put civil servants in a position to make decisions based on abstract and theoretical matters, with the illusion that they will be making them in a rational accountable way.
- The thinker lacking a word/narrative is handicapped, the doer isn’t. A small company can follow the profits with or without a good-sounding story, a big company needs can only do things that fit a narrative.
- Central planning leads to famines as large states are inflexible with procurement. China killed 30 million between 1959 and 1961, the famine was more severe in the areas with higher food production before 1959, meaning that the government policy of food distribution caused this problem.
Intervention in complex systems
If nature ran the economy, it would not continuously bail out its living members to make them live forever.
- A complicated system has a hard-to-explain behavior. A complex system, however, has interdependencies. The stock market is a complex system. A washing machine is a complicated system. It is a mistake to confuse a complex system with a complicated one. An economy, unlike a car, does not require regular maintenance.
- For a system to be antifragile, some parts of it have to be fragile. Fragile human cells impart antifragility to the human body. If nature ran the economy, it would not continuously bail out its living members to make them live forever. Paradoxically, many government bailouts end up hurting the weak and consolidating the established. Anti-fragility of a higher level may require fragility and sacrifice of the lower one.
- Light controls work while close controls lead to overreaction, sometimes causing the machinery to break into pieces.
- It is much easy to sell “Look what I did for you” than to sell “what I avoided for you”. No one becomes a hero for a non-action. And that’s why policymakers are prone to interventionism. A doctor that says no to a complex expensive surgery as oppose to self-healing will not be rewarded as opposed to the one that makes it look indispensable.
- Complex systems are hard to analyze from a cause-effect perspective. The fat-free movement happened because when carbohydrates and fats were consumed, only fats showed the responsibility. Similarly, the link between salt consumption and blood pressure rise does not exist. (ashishb’s note: it is still heavily debated, see this detailed article)
- Iatrogenic, literally, harm caused by a healer is the term used in the medical community for the harm that doctors cause. When you medicate a child for an invented disease like ADHD or depression, the long-term iatrogenic harm is largely unaccounted for. (ashishb’s note: Iatrogenic is the third leading cause of death in the US).
- Thalidomide reduced nausea in pregnant women, an immediate benefit, it led to birth defects, a delayed adverse effect.
- Diethylstilbestrol harmed the fetus and lead to cancer in daughters.
- In the 1940s and 1950s, many kids received radiation therapy for acne, birthmarks, etc. 7% suffered thyroid cancer 20-40 years later.
- Metric lowering drugs like Statins are worse. If the doctor does not prescribe then they will be sued for negligence. And if they do, then it fixes the metrics without fixing the underlying problem. The side effects will show up much later.
- Back surgery to cure sciatica, spinal cord compression, is mostly useless after 6 years.
- Mammograms to look for breast cancer do not lead to an increase in life expectancy. It might even decrease it. The doctor, seeing the tumor, has to act with radiation/chemotherapy that has iatrogenic associated. While all malignant tumors do not lead to death.
- Religion might have saved people by keeping them away from the doctor while the body repaired itself.
- Capitalism is an inverse-iatrogenic system where selfish aims are converted into the benefit for the collective.
- The non-natural need to prove its benefits and not the natural approach. In a complex system, a long time of survival is the only evidence. Many people will benefit by removing things that didn’t exist in the ancestral habitat: sugars, what products, milk (for non-European origin), Sodas, and wines (for Asians), vitamin pills, and painkillers.
- It’s not just the calories in the food that counts, the frequency and its variation also matter. Even if we agree on a balanced diet, it is wrong to assume that every meal should be balanced. The order matters. Deprivation is a stressor and that allows for adequate recovery. In fact, our ancestors used to consume proteins in lumps and other nutrients in a more distributed fashion.
- Historically, humans didn’t eat in the morning without doing physical work first.
- A loser is someone who does not introspect after a loss. He considers himself a victim of some large plot, a bad boss, or bad weather.
- Mithridatization – small doses of poison make you immune to the large doses.
- Hormesis – when small doses of the harmful substances are beneficial. Harm is dose-dependent. A balanced meal is bad, it deprives you of hormesis.
- When a country doesn’t have debt then it doesn’t care about its reputation in the economic circles. And it is only when it doesn’t care about its debt that it tends to have a good one.
- Central banks can print money with no effect and then “unexpectedly” there is a jump in inflation. That’s how nonlinearity operates.
- Never trust the words of a man who is not free. A free man is someone free of his opinions as a side effect of being free with his time.