The 5-days workshop held by Nassim Taleb in NYC on risk modeling and taking is over. It has been a rewarding experience that will take weeks to fully digest. I’ve been writing a personal experience log every day (Feb 5, 6, 7 and 8) on Avanscoperta’s blog: here today is my last one. Enjoy!
Raphael Douady opened the day with a session on the instability and fragility of financial markets. After that Taleb got back to lecturing for the remainder of the day. As usual, I’ve got out of those sessions plenty of notes on a very diverse set of risk-related topics, but here I’ll just highlight a few.
Lack of trust is worse than overconfidence
One of the key takeaways of the whole RWRI workshop is that risk, whatever its source, is only fatal if you model it in a way that blindly expose you to negative effects when it’s too late. Failure is not a problem if it’s frequent and small enough.
Unfortunately, instead, the most common measures to reduce risk — in banking, in public policies as much as in knowledge work, as we will see here below — are based on keeping negative stress away of the systems, underestimating the impact of truly rare events and, in the end, increasing the chance of severe blowups.
Does it mean that we have to live in a constant feeling of fear and inability to act, frozen by our blindness? No, it does not.
Cap maximum risk, not maximum trust
If we become obsessed with safety and only express our will to survive through rules, we usually end up making the cost of each of our actions too high. This cost reduce the willingness to act and, in the end, we stop acting. This is not good.
The problem, one could say, is that we never know how risky every move we make can be. What else can we do to make our chances to succeed higher? The answer, pointed out by Taleb, is very simple though not easy: “simple” is almost never “easy”.
If we focus on the impact of events more than their likelyhood to happen, we are able to shape our risk landscape in a better way, less dependent on time and more on the aggregate composition of our actions. If we play this game correctly we may even get to a point where the prediction of the performance of our investments (we are not necessarily talking about finance here, but finance lends its terms easily here) becomes irrelevant to decide which investment to pursue or not. If danger becomes irrelevant, trust is boosted: you become able to iteratively validate your trust assumptions.
Don’t be smart, be barbelled
On a very simple basis, this means accepting only two classes of risk: very low volatility, low performance ones, on one side, and very high volatility, very convex performance ones, on the other. As long as you have a diverse mix of ongoing investments, a small part of these may be spread on the highest possible number of opportunities as long as each one of these opportunities may asymmetrycally explode on the upside, no matter the exact probability of this happening.
Been there, done that
It (doesn’t actually) strikes me that among the many heuristics discussed during the 5-days workshop, so many have a deep reflection in the agile way of managing software development and knowledge work in general. Whenever you try to extract value from a complex environment and there is no way to estimate the final end-to-end value of every action you do, it becomes meaningful to perform small steps and probe reality to assess its behaviour. This will never bring you to be sure about what works, beware, but may provide stable answers on what won’t work.
The whole rationale behind iterative and incremental delivery, so much promoted in the lean and agile communities, is
- to cap the maximum exposure to risk: if you don’t deliver early you become fragile
- to get frequent and early feedback about your assumptions on the value of what you are working on: if you deliver rarely it’s easier for you to blow-up
Nassim Taleb closed the day with a nice list of suggestions to take risks. Among its almost 20 points, here I want to share a few:
- Better be convex rather than smart. It doesn’t matter if you don’t understand why and how something is going to be profitable, as long as you analyze it enough to understand whether it is going to be more profitable than loss-making.
- Don’t be too precise. Models are always wrong. Creating models with too precision thickens our comfort zone and doesn’t provide any increment in safety. KISS!
- Always have an exit clause. If you have the option to get out of anything, you are capping your losses. If you cap your losses, you are convex. If you are convex, you are antifragile.
In 2011, after 12 years as en entrepreneur in software development, I started identifying the root-cause of many problems in that market in the agreements we used to make with suppliers and customers. Since then I have been doing my own experiments, with my skin in the game as an entrepreneur, and after a few years I came out with a few principles that now I apply to all my professional agreements.
In 2017 I collected these 8 principles under the name Extreme Contracts. Here they are:
- Skin in the game. The cheapest way to make everybody collaborate is to have everybody’s skin in the game: don’t trust agreements that may see you failing alone.
- In their shoes. Understand your counterparts real needs. If you can address them, rules will fade a bit leaving room for value creation.
- Talk to the organ grinder, not the monkey. Never trust the agreement of a person who is not free. Negotiate your agreements with those who’ll pay in the end. That has deep implications on skin in the game and most applied procurement policies.
- Value centred. Don’t care about metrics and KPIs. They are models and often they don’t take into account the real impact of unexpected events, both on the upside and the downside. A knowledge worker has to solve a problem for someone, not make an Excel file be even.
- Ethics over rules. It is way better to sign a simple agreement that allows all the counterparts to frequently validate and re-asses its validity than a complicated one that only describes ways for all the parts to sue each other when things blow up.
- Chaos in small doses. Don’t think monolithically. Complexity will always backfire eventually. Chop your actions into bites, make agreements for one or a few bites at most. Act and re-validate. If needed, exercise your exit options.
- Optionality. Options are the currency of antifragility. Let yourself — and even your counterpart — free to go. As Taleb said the morning of the fourth day: ” we just want to rent each other, nothing more. We are free to go.”
- Customer channel. Never forget: the agreements you propose are part of your branding and, furthermore, they are among the first touchpoints with your audience. Shape them to communicate and reflect your culture. This will attracts like-minded customers and collaborators, reducing the friction, helping to cap your losses.
All of these 8 principles have reflections in the suggestions explicitely or implicitely given by Nassim Taleb & co. during this week. This motivates me to keep on elaborating and refining Extreme Contracts in the next months, bringing the conversation to a broader audience.
This is the end — and the beginning
So, this is over. I enjoyed these 5 days a lot.
I want to thank Nassim Taleb for his welcoming attitude. I witnessed him lending hundreds of bucks to an attendant who lost her wallet, just saying.
I want to thank all Taleb’s friends who joined him in making these days more interesting. The gags between Taleb and Douady were part of the show!
I truly sincerely thank all the attendees. They were such a diverse audience, going from guatemalan entrepreneurs to saudi CFOs, from financial advisers born in italian german language islands to basque machine learning specialists… I really felt the dumbest in the room and it felt so good. Thank you friends!
Last but definitely not least, I thank Avanscoperta for contributing in this small adventure of mine. They made it easier for me to accomplish a goal I had for 2018 and showed once again they care about spreading knowledge and good practices, with their skin in the game.
It’s been fun and growth. What else? See you next time. Ad maiora!