Read on Twitter
Some highlights / takeaways / thoughts / comments from #RWRI 14, Day 6:
1. Chaotic systems occur when tiny differences in initial conditions result in wildly different results (e.g. changing a variable from 10 to 10.0001 gives a completely different result).
1. Chaotic systems occur when tiny differences in initial conditions result in wildly different results (e.g. changing a variable from 10 to 10.0001 gives a completely different result).
To predict a system like this, you would need infinite precision, which you don't have. These systems are "unpredictable but characterizable," as you can predict the possible states but not where it will be at any given point.
Example: You can't predict the weather 1 year from now, but you can know the range of outcomes of what the weather could be 1 year from now.
2. "Central banks do not decrease volatility."
The volatility is just hidden / shifted elsewhere in the system.
The volatility is just hidden / shifted elsewhere in the system.
Volatility is information; suppressing volatility suppresses information. Would you rather get a little information daily and adjust? Or wait and receive all of the information from the past decade at once?
3. Process vs Plan.
Plans are fragile based on the planner, quality of plan, assumptions, etc.
Processes are antifragile - they do not rely on a central party to steer or build them; they use information / feedback loops to adapt / improve
Plans are fragile based on the planner, quality of plan, assumptions, etc.
Processes are antifragile - they do not rely on a central party to steer or build them; they use information / feedback loops to adapt / improve
4. Highly complex, effective systems can result from shockingly simple rules. Much of nature comes from simple rules that, when played out over time, exhibit complex behaviors / structures.
For example, we might think that ants behaving in a collective manner is a result of highly complicated rules or some intricacy that could never be understood.
But this complex behavior can be re-created with very simple rules in a simulation.
But this complex behavior can be re-created with very simple rules in a simulation.
5. An act that is seen as 'irrational' for the individual or small group is likely 'rational' when viewed through the lens of the entire system.
This is why much of behavioral economics is dubious: you can't just view a decision / action in a vacuum and declare it irrational.
This is why much of behavioral economics is dubious: you can't just view a decision / action in a vacuum and declare it irrational.
6. Whether permanent climate change is occurring or not is beside the point. You'll never be able to prove 100% yes or no until it's too late.
The point is that if there is *any* possibility of climate change, the effects will be nonlinear / unpredictable (and thus potentially catastrophic for humanity). Thus, extreme precaution is a must for the survival of humanity.
7. "It's better to be approximately right than precisely wrong."
Single-point predictions provide a false sense of security and will very often be wrong. Better to spend your time understanding the properties so you can be prepared to adapt as you gain more information.
Single-point predictions provide a false sense of security and will very often be wrong. Better to spend your time understanding the properties so you can be prepared to adapt as you gain more information.
Sound risk management is concerned with identifying the potential sources of risk. This is far more important than quantifying risk.
8. Rather than trying to predict an outcome, stress test your system by asking how it can break. A business owner can ask: What could cause the business to go bust?
Some answers might be unactionable (e.g. an asteroid hits earth, not much you can do), but others may prompt you to realize that you're highly fragile to one key employee / supplier / sales channel.
Questions like these can help you identify your tail risk exposures / fragilities. Once identified, you need to bound them (i.e. cut the tail) as much as possible.
9. Insurance companies are necessarily experienced at cutting the tail. Without the 'physical damage' clause that is present in many contracts, the entire industry would have gone bust due to COVID.
10. Something that is not usually correlated (my probability of vehicle damage vs. yours) can become highly correlated due to a catastrophe (e.g. an earthquake destroys all cars in the area).
An insurance company without an understanding of fat tails might base their policies on a complicated study that finds the correlation between car damage of two random people to be zero. Then, a massive earthquake hits, causing correlation to spike to near 1.
Complication / precision / prediction-making is the enemy of understanding real world risks. You don't need a PhD to figure out where your fragilities lie.
