Book Summary: Antifragile by Nassim Taleb

Book Summary of Antifragile: Things that Gain From Disorder by Nassim Nicholas Taleb


The wind energizes the fire and extinguishes the candle.  In the same way, we want not just to survive uncertainty, but be strengthened by it.  The opposite of fragility is not robustness, because robustness only indicates the ability to withstand uncertainty.  Anti-fragility is the ability to be strengthened by unpredictability, not just to survive it.


Anything that has more upside than downside in the face of a random events is anti-fragile.


The tragedy of modernity is that the systems put in place to protect us from uncertainty actually end up harming us the most.


Our minds make history look linear and smooth in hindsight. This gives us the illusion of control.  Looking forward, risk and probability are very difficult to calculate. Fragility, however, can be observed and calculated.


Fragility can be defined as that which does not like volatility, randomness, chaos, non-linearity, and time.


The barbell strategy: a system of tinkering and trial and error allows you to learn from mistakes very quickly. This is an anti-fragile system.


Debt always makes you more fragile.


The myth of Damocles describes fragility. He attends a banquet with a sword hanging above him by a hair from a horse’s tail.  Fragility can be measured by the strength of the string.  Robustness can be represented by the Phoenix, who rises from the ashes after being destroyed.  Anti-fragility can be represented by the Heidra, who grows another head every time one is cut off.


Depriving people and systems of stressors is not necessarily a good thing.


Domain dependence is a cognitive bias that prevents us from observing reoccurring patterns in different contexts.


Absence of challenge often results in underperformance. This can be observed in horseracing, the office, and academics. For example, the adage that if you want something done, give it to the busiest person.  People tend to squander their free time.


Mental effort gets us into higher gear, engaging our mind’s System 2. This is why the most powerful person in the room often speaks with the softest voice.


Redundancy is nature’s insurance policy.  For example, the body has two lungs, two kidneys, etc.  This is a sign of robustness.  Man made systems, on the other hand, often contain single points of failure.  For example, debt is the opposite of a redundancy.


Overcompensation is nature’s way of managing risk.  Our bodies assess risk better than our intellects do much of the time (e.g. vaccines, exercise, etc.).  Examples include hormesis (e.g. tanning skin after sun exposure, muscle growth after working out) and mithridatization (injecting small doses of toxins to develop immunity).


Risk management techniques such as stress testing use past information to determine future risk, ignoring the possibility that the future could be different than the past.  This exercise commits the logical fallacy that the worst case event by definition exceeded the worst case event at that time.


Information is anti-fragile. The more ideas are attacked, the more widespread they become. Likewise, criticism is also anti-fragile. This is due to an inherent survivorship bias—why criticize one idea instead of another?


Everything that has life is somewhat anti-fragile.  This is a feature of the natural that is not the case with man made objects.  Synthetic creations may be complicated, but they are not complex.  They are not complex because they do not have interdependencies that complex objects possess.  Complex synthetic systems such as ideas, businesses, and markets can exhibit anti-fragile behavior.


Complex organisms receive information about stressors via a variety of sources. This kind of ambiguity is known as causal opacity.  This means it is not always easy to distinguish correlation from causality, making it difficult to predict future events, especially in the presence of non-linearities.


Machines: use it and lose it (fragile), organisms: use it or lose it (anti-fragile).


Touristification: the simplification and scheduling of rigid activities in order to avoid unpredictability and randomness. A set routine makes one a little more dead. Like the difference between the rush of running from a wild animal to going to a gym to get on a treadmill.


Parts of a system may be required to be fragile in order to make the system anti-fragile.  For example, restaurants by themselves are very fragile, and for that reason have to maintain a certain level of quality. This fragility of the parts makes the restaurant scene anti-fragile as a whole.  The key to benefiting from this phenomenon as an individual is to use trial and error and rapid iteration to improve when one venture fails.


Evolution is an anti-fragile process.  There are two mechanisms of this.  The first is that species will improve over time due to natural selection.  The second is that when a dominant species receives a shock so strong it is wiped out, other species benefit as a result, and the system is improved.


Harm to an organism is beneficial to a population.  Fractal similarity—parts tend to resemble bigger parts of a fragmented system.  Harm to cells can help organisms and harm to organisms can help populations. For example, when someone fasts, the body eliminates the bad proteins first. This makes room for newer, stronger proteins.


Error can be an advantage when used in a trial and error process. For example, without the Titanic, then we would have tried to keep building bigger and bigger ships, making the eventual disaster even worse.


The airline industry is anti-fragile because it is able to learn from plane crashes.  That is because of many trial and error situations which are independent and frequent, and allow interpretation of failures.  The same is not true for the economic system, because of the infrequency and interconnectedness of failures on a global scale.  This phenomenon makes errors spread and compound.


The anti-fragility of the higher level (the economy) often requires the fragility of the lower pieces (individual businesses).


Economies require entrepreneurs to be overconfident and underestimate risks.  The fragility of the parts leads to overall strength of the economy.


Neitsche’s “what doesn’t kill me makes me stronger” might be confused for anti-fragility in the form of mithridatization or hormesis, but may simply be survivorship bias (only strong people survive volatility, and therefore appear to become stronger as a result).


There is no such thing as a failed entrepreneur or scientific researcher.  Their risk taking infertility are necessary for the growth of (or even survival of) the economy.


The Procrustean bed: in Ancient Greece, there was an innkeeper who cut off the limbs of people that were too tall to fit in his bed, and stretched people who were too small.  The term can be applied to any arbitrary condition that everyone is forced to conform to.  This is what happens when people force linear predictions on non-linear systems.


Artisanal careers like taxi driving or prostitution lend themselves to rapid feedback loops, which allow them to adapt and make changes quickly.  Normal desk jobs may appear robust due to the lack of volatility in income, but they are not—they just mask the volatility and randomness of the underlying industry.  For the artisan, fluctuations in earnings provide valuable information.  These minor fluctuations reduce the risk of major disruption, like being laid off.


Nature loves small errors. Without them, evolution is not possible. Avoiding these small mistakes makes the large ones even bigger.


Switzerland is the most anti-fragile place on earth.  Global volatility makes Switzerland stronger, as more assets are moved into the country.  It is so because it has no top down government.  Rather, it is ruled by a series of small city states.  Day-to-day quibbles and small conflicts of the municipalities prevent large-scale disruptive volatility.  Small is more anti-fragile than the large.  The nonlinearities of extrapolating large numbers of people, values, and opinions result in large-scale volatility.


An analogy to this is the sadness we feel when a baby is crying near us compared to that of large-scale genocide in a faraway country.  The former is relatable and real, while the latter is a statistic in our minds.


It is common to mistake absence of evidence for evidence of absence (Black Swan Problem).  For example, just because we can’t conclude that fossil fuels cause climate change doesn’t mean that there is evidence of no harm.


Systems composed of many independent small units do not lend themselves to large-scale failure, because because risk is distributed, not concentrated.


Risk is in the future, not in the past (the Turkey Problem).


The adverse effect of artificial stability: the more a process is stabilized and volatility suppressed via arbitrary controls, the greater the volatility will be when there is a small shock.


Randomness can lead to significantly better results.  One example is an SOS signal, in which background noise pushes the signal above the detectable limit.  In another example, Nathan Neirvald discovered that oil exploration benefited from random selection of drilling locations.  Another example is called anealing, in which metal is heated and cooled off in order to improve the material strength of it through randomness.


Buridan’s donkey illustrates this phenomenon—an equally thirsty and starving donkey midway between food and water will die due to indecision.  Any degree of randomness would allow the donkey to decide.


Theories are fragile.  Outside of physics, they come and go.  A phenomenon is robust.


Over-intervention in some areas can lead to under-intervention in other (often more important) areas due to signal and noise.  The true hero of a Black Swan system is one who does not intervene and thereby averts a crisis.


The Latin saying festina lonte, or “make haste slowly,” suggests that using procrastination can be an effective way to mitigate over-intervention because it gives nature a chance to work things out.  It can be used as a source as a form of decision-making and as a way to filter information. Over-intervention stems from the inability to distinguish signal from noise.


Over-reliance on data can lead to naïve interventionalism. In medicine, this is known as iatrogenics, in which a patient is actually worse off after the treatment.


Sometimes the catalysts of a Black Swan event are mistaken for its causes.  In order to overcome this, the system of fragility must be studied, rather than trying to predict individual events within it.  Long-tail events are unpredictable.


Those who are robust or anti-fragile do not have to rely on forecasting to determine their future success and risk profile.  On the other hand, those who are fragile (e.g. those who have debt) need to try to predict the next event that could wipe them out. Not seeing the next major uprising or tsunami is excusable, but building something that is fragile to these events is not.


Cultural, social, and economic life lie in the black swan domain. Unpredictable things will always be unpredictable. The increasing trend of mass communication and winner-take-all companies, events, and situations results in an increasing number of Black Swans.


Success results in asymmetry—you have more to lose than to gain.  Constant worry about downside is a form of punishment.  Mentally adjusting to the worst case scenario is how Stoics deal with the possibility of loss.  Invest in good actions rather than possessions.  Things can be taken away, but good deeds cannot.


The purest form of anti-fragility is asymmetry with upside.  If you have nothing to lose, you are anti-fragile.  Fragility has an irreversible tendency—the irreversibility of damage.


GDP growth is a Procrustean bed that masks fragility.


A barbell strategy uses a combination of extremely low risk and extremely high risk elements to protect against negative black swans and capitalize on the upside from positive black swans.  The strategy avoids medium risk, which can lend itself to high amounts of volatility.  Due to the allocation of resources into each category, the maximum loss is quantifiable.


The first step toward applying a barbell strategy is to remove the risk of ruin.

  • Exercise barbell: short periods of high intensity wait or cardio training complemented by long, effortless walks.
  • Reading barbell: trashy gossip magazines and literary classics.
  • Lifestyle barbell: do crazy things every so often.  Break furniture.  For the majority of the time, be rational.
  • Conversation barbell: either talk to undergraduate students and taxi drivers or scholars and the highest members of society.  Avoid middlebrow conversation.

Just as stoicism is the domestication, not elimination, of emotion, the barbell strategy the domestication, not elimination of risk.


Options are an agent of anti-fragility.  An option gives the buyer the right (but not the obligation) to participate.  The seller has the obligation (but not the right) to provide something.


Options like dispersion of outcomes.  They don’t care about the average.  An option equals asymmetry plus rationality.  Rationality is required (as in evolution) to select the upside from an asymmetric option.


Gamma is a measurement of the non-linearity (volatility) of an option. Conventional option valuation can underprice an option due to the lack of consideration of a rare event.


Progress requires innovation and implementation.  Both require some degree of randomness.  Implementation may occur long after the initial innovation.  Example: wheels on luggage.


The Soviet-Harvard illusion: the idea that lecturing birds on how to fly leads to their being able to fly.  The arrow of causality is not always clear.  By the same logic, you could say that gender is determined by having short or long hair, or that wearing a tie makes one a business person.  Likewise, there is a tendency to believe that wealth is created through academic research (Baconian loop).


The Granger cause allows us to identify causality by applying a sequencing technique.  This introduces an element of time.


Poverty makes experiences.  Necessity is the mother of all innovation.  Likewise, wealth and wisdom cannot be bought by attending an elite university.  People only do well when shaken.  That is what anti-fragility is all about.


The green lumber phenomenon: domain-specific knowledge is not always as important as one might think.  You don’t have to know what green lumber is to make money off of a trade on it. This is the opposite of the halo effect.  People with complex knowledge, views, and models often miss the most elementary things.


When you are fragile, you need to know more to avoid failure.  When you rely on knowledge more, you become more fragile.  When you are anti-fragile, you are indifferent to knowledge and the need for forecasting to avoid failure.


Humans are different than animals in our ability to collaborate.  Because collaboration cannot be predicted, the future is quite unpredictable.  The best thing we can do is create environments conducive to collaboration and tinkering.


Corporate planning is overrated. Constraining a company’s future path can obscure valuable optionality.  Consider that Coca-Cola started out as a pharmaceutical company, Avon products started out as a door-to-door sales company, and Nokia started out as a papermill.  Refer to the book “The Management Myth.”


In anti-fragile systems, the qualities, not the defects, will be hidden. This is the result of positive asymmetry.  Fragile systems are the opposite.  Rare events are rare and give us a skewed view of reality.  Negative “turkey scenarios” have limited downside and almost unlimited upside.  Take biotech for example.  Unlike banking, the potential in biotech is nearly unlimited compared to the downside.


Rules of turkey scenarios:

  1. Look for optionality, and rank things according to their optionality.
  2. Invest in things with open ended payoffs, not closed-ended ones.
  3. Invest in people, not in ideas. The people you invest in should be capable of changing several times throughout their careers.
  4. Make sure you are barbelled.


Things that are fragile are hurt disproportionately more by rare large events then by the cumulative effect of small events.  Anti-fragile is the opposite (marathon running, weight lifting), up to a point.


Convex patterns (anti-fragile) resemble a smile, concave behavior (fragile) resembles a frown.  Convexity likes volatility, as small negative payouts can become large positive ones.  Concavity requires stability, as small positive payouts can quickly turn into huge losses.


Larger organizations are more fragile than smaller ones because they are exposed to large-scale phenomena.  By the same token, larger projects are much more prone to overruns than our smaller ones.  Incentive schemes, the use of computers, and the drive for efficiency (usually from business schools) has increased this.  Size gives the illusion of strength, but is more prone to agency problems.


The convexity bias: getting hit by a 10-pound stone causes more than 10 times as much damage as a single one-pound stone).   Someone with a linear payoff needs to be right 50% of the time.  Someone with a convex payoff, much less so.


The increasing number of winner-take-all situations in today’s society are converting Pareto’s 80–20 rule to 99–1. For example, 99% of Internet traffic is attributable to 1% of websites. This is exacerbated by technology and social media.


Less is more. More data can mask what is truly important.  If you require more than one reason to do something, then you probably shouldn’t do it (via negativa).


How to understand the future: respect the past, be curious about history, seek out the wisdom of elders, and acknowledge the presence of heuristics. The past is a better teacher about the future then the present.


We notice what changes and varies more than the total and more than what really matters.  We need water more than cell phones, but because water doesn’t change and sell phones do, we mistake them for being a more important.  We rapidly tire of what we have, and constantly seek out better versions.


The impulse to buy new things that will eventually lose their value is called the treadmill effect. When people buy something new, they feel an initial boost of satisfaction, then quickly revert back to their original state of happiness. These are known as Hedonic states, as characterized by Kanheman and Tversky.


The treadmill effect is characterized by a focus on differences rather than qualities or similarities. This also happens primarily with things in the technological realm (things with an on/off switch), and does not apply to artisanal or industrial goods.


Wealth of details and fractal jaggedness (as found in nature) in things around us lead to inner peace.  Sterile modern design leads to greater anxiety for the author.


No evidence of harm vs. evidence of benefits—the un-natural needs to prove its benefits.  Things in nature should be innocent until proven guilty, while science and technology should be guilty until proven innocent.  There is an innate robustness built into nature—owing to millennia of experimentation—that humans cannot easily replicate.


Real people take risks for their opinions.  Real people put their souls in the game.


If you take risks and face your fate with dignity, there is nothing that can make you small.


Predictions without skin in the game are dangerous as well as meaningless.  To avoid problems arising from the transfer of fragility, modern professionals should abide by Hammurabi’s code.  This ancient room states that if a building collapses and kills the owner, the builder should be put to death.  This is not just for sake of punishment, but rather to provide an upfront incentive to do good work.  In a similar way, you can’t really have an opinion about something without having some exposure to risk surrounding it.


The cure to ex post narratives: ask people what they have in their portfolio, not what their forecast is or what they believe will happen. Your portfolio should agree with your opinions.


A man is only as heroic as the amount of downside he is willing to expose himself to for the sake of others.  The agency problem is the opposite of heroism.  To be a hero is to sacrifice oneself, and the root of “sacrifice” is “sacred.”  The Socratic Plato definition of courage is standing up for one’s own beliefs.  This is the definition of the modern man.  The privilege of dying for one’s ideas or truth is the highest form of honor in life.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s