I really love this book by Nassim Nicholas Taleb. A lengthy book full of fierce arguments, that might not be pleasant for some people. I think he just enjoys being an arrogant antagonistic asshole at times. But I can’t help but enjoy his arguments and ideas so much that I bare with a lot of negatives of the book and kept on reading.
Note: This post is my attempt of summarizing.
Hello Antifragile
Antifragility is a property of complex/biological systems 🧬. That was an ‘aha!’ moment for me, particularly after being introduced to a class of problems in systems thinking (from another book) that can’t be solved without first learning how the system behaves.
TL;DR; while the synthetic/artificial needs maintenance and eventually wears down, the antifragile has the ability self-repair and grow 🪴🌷.
Rather than explaining what it is, it’s easier to describe its properties. Antifragility:
- Learn from stressors:
Complex system share information through stressors (flow/feedback). Naturally, pain or learning by doing, is a better teacher than learning by instructions.
- Likes and needs randomness (or uncertainty, chaos, errors):
They not only resist shock, but benefits from them.
Event dispersion (asymmetry) creates more upside (learning discoveries) than downside (risk).
Deal with unknowns without the need to understand.
Prefer fluctuating but steady in long term, over something stable in short term but jumps.
- Overcompensates and needs redundancy:
Being redundant is a natural risk-management (2 kidneys, extra capacity..). To sustain, the system needs to invest on diversity and redundancy to prepare for the worse.
Innovation/growth comes from a place of chaos, discomfort, and necessity. It comes from the excess energy of overreacting and overcompensating for what you need.
E.g. - Need tiny dose of noise to keep us focused. (ie. Stochastic resonance) - Additional motivation from setbacks. - Saving extra money/supplies - Body/muscle response to exercise - Repressing hatred through force, stirs more hatred (manipulation works better) - Impediment leads to obsession - Information spreads wildly as secrets - Language/communication is better learned outside the classroom. (healthy amount of fear of making mistakes)
- Requires time and the right dose:
Hormesis: Harm/cure is dose dependent.
Unlike machines, we (complex systems) don’t need a ‘fix’, but time to recover.
- Needs sacrifice at the micro-scale
Like evolution’s selection process, a failure makes the whole system stronger. Our death and diversity serves to make nature more antifragile.
Good failures are ones designed to be independent, and doesn’t drag the whole system. You want individuals to take lots of risks for the sake of the system. Thus, Decentralizing and distributing randomness works well to constrain the problem. (as opposed to centralizing and concentrated randomness).
- Likes Polarity. Use Barbell strategy to achieve more upside than downside
Keep the offense (risk-loving, growth-focused) completely separate from the defense (risk-aversion, survival-focused) strategies
Avoid the middle (having offense + defense in the same basket), since its subject to miscalculated risks.
Success lies on the tail/extremes/good-black-swans, so it’s better to have everyone being good at different things, rather than raising everyone to an average.
E.g. - Do low-risk job to maintain security, and high-risk job on spare time. - Prefer short burst of intense hours and switch off, rather than having constant mildly-intense job all the time (these harmful stressors include taxes, exams, commutes, emails, much of things that run civilization). - Drink freely and recover on some days, rather than drink everyday.
- Likes to have more options = asymmetry + rationality
One doesn’t need to know where they’re headed. They simply need to make decisions that open more options, which puts them in a better position to find things that fit future desires.
More option = more to gain. It’s not lottery because life has no limit to upside.
E.g. Invest based on people’s ability to aggressively tinker and capture future options, rather than their idea/plan.
- Survives time:
Time is a natural disorder, and a good cleanser of noise.
History is best indicator of the future. (Old will live longer than the young, perhaps because it correspond to something deep in our nature.)
I’ll try not to be Fragile
In contrast to being anti-fragile, some warnings on excessive fragility includes: - Leading you to underestimate rare black swan events. - Much of the side-effects are buried. People focus more on visible, but tiny and unnecessary improvements. - Absence of evidence isn’t evidence of absence. We should assess risk based on its pay-off rather than probability of occurring. (i.e. airport check) - Procrustean bed: force fitting systems into a mental model. Locking itself to a non-agile program. - Risk is in the future, you can’t forecast risk nor assume a complete vision. - Touristification / soccer-mom: getting rid of small stressors/volatility and making things predictable in their small details makes large mistakes more severe. - You panic and caught off-guard when you’re not used to small errors. - Converts activities to scripts, and complexity to fragile machines. - Absence of randomness limit our imagination. - Efficiency kills self-organization (discussed in team-of-teams). - Bad asymmetry. Those that don’t take risk (more to loose than to gain) - E.g. A prisoner that protect reputation by all means, is fragile to information against them. (politicians, corporation, governments). - Harm from wealth: Possessions make us worry about downside, avoid opportunities, avert risk, thus fragile. - A method is to counter this is to ‘mentally write-off of your possessions’. Avoid giving your all, commit with less than what you have, so you can take risk without feeling much downside. - Having no skin in the game - Society is fragile when human acts on their self-interest over the collective. - What we need is incentivising people to expose themselves to risks, sacrifice heroically for others (e.g. rewarded by respect or honor for trying/failing since it helps others learn). - But society is doing the opposite. Failed enterpreuneurs are unheard are criticized. Or worse, granting mere talkers a free option without any down-sides (ie. Politicians, journalists, fund managers). - It’s dangerous for the decision-maker to expose others except himself, to risk. It’s unethical to express opinion without being liable to it. - Trust word of mouth. It’s unethical to self-promote (agency problem) especially when you hide defects or play with a person’s cognitive bias (ie. Relating smoking with romanticism). - Only trust opinions from those that is exposed to risk. E.g. Ask the doctor not what you should do, but what he would do in your place.
‘Wealth is the slave of the wise man and master of the fool’ - Seneca
Lastly, the author talks about nonlinearity as a measure of fragility: - Measure fragility with concave graph: the bigger the shock (more concentrated), the more it harms you. Steeper graph means more fragile. - Measure antifragility with convex graph: the bigger the shock, the more it benefits you (up to a point).
My thoughts/mixed feelings with his ideas:
- Being redundant, capping the downside, barbell strategy, and having lots of options can only be leveraged if you’re the privileged few. Otherwise, it’s hard to adapt and or even do anything outside of the making-a-living grind.
- Touristification is necessary to focus on harder problems.
- Predictions: I agree we shouldn’t be dependent on forecasts (especially in black swan domains like social, economic, cultural). But it does save resources, and is necessary since we can’t always afford to have redundancy to manage risk.
I like his arguments on education. Has some truth but is also debatable: - Practice/experimenting create theories, not the other way. - Wealth creates education, not the other way. - Learn by doing helps you stick with simplicity. Don’t try to optimize the world based on theories. - Against curriculum and structured learning - School blinds you from seeing heuristics as a form of knowledge. There are things you can only learn through apprenticeship (ie. Cooking) - Unstructured learning creates autodidacts. They’re comfortable with ambiguity and can exist in domains outside the organized constructs. - School limits the infinite options, to a narrow set of authors. When in fact you should move on to a new book if you’re bored, rather than hating reading altogether.
Takeaways
I found it’s important to watch out that antifragility is a spectrum (everything likes and hates volatility to a point. Everything is both antifragile and fragile).
Here’s a bit of digression. I personally dislike the use of ‘anti-’ in a word. This resonates a lot with something that I read long ago, about constructively focusing on what you want to see more of, rather than defining something and unifying people based on what you’re against - making us over-sensitive and creating the witch-hunting culture. Similarly, using anti- sets a passive voice that focus on what not to do, rather than what we should do. In this case, it has the connotation that one should avert fragility. I don’t think that’s true. We need some amount of fragility and it exist in all of us. I think NTT should’ve named it something else.
Otherwise, a lot of ideas resonated with me. Examples include being starved of social interaction during the pandemic, which makes me a bit more social, a bit more independent, a bit stronger. Only after graduating formal education do I realize I can learn most things myself (through books, online). Most of the good part of university is the people, joining activities, internships - much of which are random events. Events where you talk to people, being open to surprises and opportunities along the way. It’s scary but rewarding taking the non-planned routes.
So a reminder I have for myself:
Don’t plan too much. Just focus on your options and what’s ahead.
Beware of naïve intervention early on (premature optimization). You can justify inaction if you suspect a delayed/hidden cost. Don’t feel like you have to fix everything - inaction is often preferred (via negativa), and tech is best when it’s invisible and ‘feels natural’. Focus on the signal and don’t overreact on noise (idea: perhaps procrastination is a natural instinct to avoid naïve intervention unless necessary).
It’s fine to fix things but make sure the risks are constrained and have your skin-in-the-game.