Notes from ‘Skin in the game’ by Nassim Nicholas Taleb

Nassim Nicholas Taleb, the famous author of Fooled by Randomness and The Black Swan, has come up with a new book titled Skin in the Game. I finished it over the weekend and here are my notes.

My sofa, my finger, Taleb’s book

Insights (and points I agree with him on)

Human systems are systems within systems, interacting in complex ways. No article / paper / logical argument can capture all interactions of such systems and their evolution over a non-trivial amount of time. So one shot, large scale interventions (such as invasion, UBI, CRISPR) are extremely dangerous because we now have people in power who can order large scale tech-driven interventions across a large section of society.

– People with fuck you money should call out famous or powerful people when they are wrong

A scientific result takes a slice of the world that’s impossible in reality. Our real world is dynamic, full of interactions and any static slice necessarily misses that

Time is the ultimate test of robustness.

It’s safer to describe the past than to predict the future, but if you must predict it, you must share its downside

– When a thing is described, the limitations imposed by the need to hold that concept in mind and communicate it to others isn’t generally recognised as limitations. In reality, the thing (say voters in Bihar) has a history, embedded in other things, interacts with other things and has goals that evolve that itself isn’t aware of. Try communicating all that to others.

– Our perception of roles that people are playing in a system are usually static and short sighted. We forget. But history doesn’t care. It tends to eliminate bad systems.

Anyone who doesn’t share downside will transfer risks from himself to someone else. Net-net, in systems, you can’t have upside without downside

Understanding and respecting the way people act is more important than telling them how they should act.

Ethics is about  sharing downside if you have a shot at an upside

A deterrent works because actors then take informed risks. If there’s no deterrent for bad behaviour, people will maximise their self interest at the expense of others

Groups change behaviour as they scale. You can’t empathesize in a city just the way you empathsize in a village

– Since reality is multi-dimensional, evidence of achievement (of something) in the real world is always more informative than being able to explain why something would work

Central, top-down planning of any kind (say an intervention) introduces risks in complex systems that are hard to estimate. While bottom-up solutioning and organisation as done by entrepreneurs is safer because it works within existing boundaries of system. Decentralisation > Centralization for most situations

Labels are poor carriers of information. Clubbing twitter and Facebook as social networks misses their details, similarly clubbing Buddhism and wahabishm as religions and saying all religions should be tolerated misses the point. One is a way of life, other is cancer on earth whose stated position is world domination

What people do (revealed preferences) are usually different from what people say (stated preferences). Skin in the game comes only via what people do, not what people say

– Entrepreneurship isn’t about major changes, it’s about tweaking  existing interactions /solutions a little bit to discover something more acceptable within a system. A large change (visionary idea) by definition is fragile in the system

Most people can’t compute probabilities so they rely on well tested, age old heuristics

Rationality can be defined as survival, so whatever promotes survival over long period of time for individuals or collective is rational. If rituals have survived over millennia, they’re surviving a function that helps the population that holds them survive

– Rationality is risk management

– Survival comes first. In cost/benefit analysis, is there’s a chance of you going bust then eventually you will (if you keep playing the game).

Ergodicity means present probabilities continue to remain similar to past probabilities. Non-ergodicity means the system gets ‘stuck’ in a state from which it’s impossible to come out of (say climate has been ergodic – going up and down, but if human actions cause an it to go only in one direction, it is becoming non-ergodic.)

Systems at different levels are ergodic or non-ergodic (my death converts me from ergodic to non-ergodic system but high level system can still be ergodic)

Take a lot of risks that don’t have tail risks but may have tail profits (starting a company might be the best example of it. Downside a job and one year lost, upside $100m usd)

All risks are not equal, Risk and ruin are two different things

Whenever numbers are mentioned, immediately visualise distribution and their possible evolution. Number of people dying from Ebola being lesser than number of people dying on bathtub is an idiotic statement. Ebola mutates and can spread, bathtubs are isolated incidents

– In psychology experiments, we derive results from one or few interactions / actions per individual but in the real world behaviour of individual (trust, loss aversion) is determined by repeated interactions and people want to avoid ruin by repeat exposure



Points that I don’t fully agree with on

For measuring robustness, recent history should be given more weight because the interaction types also change over time. Humans spend vastly more time as hunter gatherers as rural as urban. So as underlying structure changes, so should our assessment of what works and what doesn’t.

Applying simple heuristics. For example, a professional who dresses impeccably should be trusted less (conditional on their achievements) or people who go to Ivy-league vs non-Ivy league (conditional on achievements) should be trusted less. In reality, it depends. Taking his own words, absence of evidence is not evidence of absence. A person can be soft-spoken, well-dressed and authentic. Similarly, a shabbily dressed doctor can really be clumsy. Signals exist because society values them (and gets information from them). A doctor who takes care to dresses well advertises that he is methodical and that’s what the patient may be looking for as it gives him hope.

Where I believe he is inconsistent

– Religion and superstition are acceptable because they serve some societal purpose not directly related to supernatural but same acceptance doesn’t apply to a rich person who goes to Michelin’s rated restaurant not for food (purported function) but to be able to signal wealth to friends (actual function)

Have an opinion on this post? You should then about it (or send your feedback on email). Thank you!