THE DYNAMICS OF ERGODICITY
(thread)
in which I explain the link between ergodicity, the Fence Paradox and human approach to risk management.
Also available in paper form at luca-dellanna.com/ergodicity-dyn…
(thread)
in which I explain the link between ergodicity, the Fence Paradox and human approach to risk management.
Also available in paper form at luca-dellanna.com/ergodicity-dyn…
1/ People tend to maximize overall ergodicity across all stressors they are exposed to
2/ Risk homeostasis (aka Fence Paradox) is the reaction of rationally-maximizing-ergodicity actors reacting to having been exposed to a non-representative subset of the environment.
2/ Risk homeostasis (aka Fence Paradox) is the reaction of rationally-maximizing-ergodicity actors reacting to having been exposed to a non-representative subset of the environment.
3/ First, let me define ergodicity. I have high ergodicity if in my life I am not much exposed to "game-over risks", such as death, bankruptcy, social ostracization, etc.
For a more precise definition of ergodicity, see below
For a more precise definition of ergodicity, see below
4/ (Ergodicity purists, you'll find a justification for my use here of ergodicity as a spectrum in tweet #14 below.)
5/ Ergodicity is a function of an actor, an activity and the exposure of the actor to the activity.
For example, Russian Roulette is a non-ergodic activity for a player (he _will_ eventually "game-over") but not for a company hiring Russian Roulette players.
For example, Russian Roulette is a non-ergodic activity for a player (he _will_ eventually "game-over") but not for a company hiring Russian Roulette players.
6/ Ergodicity depends from exposure to high-magnitude stressors. Driving a car, what you want to avoid is not a 10 mph crash, but a 100mph one.
However, high-magnitude stressors are rare. Therefore, people use low-magnitude ones to estimate the ergodicity of an activity.
However, high-magnitude stressors are rare. Therefore, people use low-magnitude ones to estimate the ergodicity of an activity.
7/ Example: driving is a risky activity, where (say) on average one experiences one low-speed crash every 5 years of driving and one high-speed crash every 100 years.
On average, after 30 years, a driver experiences few low-speed crashes & estimates driving as low-ergodicity.
On average, after 30 years, a driver experiences few low-speed crashes & estimates driving as low-ergodicity.
8/ The ABS (anti-braking system) avoids low-speed crashes.
On average, after 30 years, a driver would experience no crashes. He would erroneously estimate driving as a high-ergodicity activity and take more risks, exposing himself even more to the rare high-speed crashes.
On average, after 30 years, a driver would experience no crashes. He would erroneously estimate driving as a high-ergodicity activity and take more risks, exposing himself even more to the rare high-speed crashes.
9/ The Fence Paradox arises when a policy or technology prevents people from experiencing low-magnitude stressors.
They end-up being exposed to high-magnitude ones only, but because these are rare, they underestimate their exposure and take more risks.
They end-up being exposed to high-magnitude ones only, but because these are rare, they underestimate their exposure and take more risks.
10/ Translating in ergodicity terms: when a policy or technology prevents people from experiencing low-magnitude stressors, people will overestimate the ergodicity of an activity and will take more risks in order to increase the ergodicity of other contexts they are exposed to.
11/ The Fence Paradox arises from the tendency of humans to maximize their ergodicity across all domains of their life; this means selectively decreasing their ergodicity respect to a domain if this would increase their ergodicity respect to another one by a larger amount.
12/ In fact, often, increasing risk-taking in an activity yields to an abundance of resources which can then be used to mitigate other risks.
Driving faster to avoid missing a plane saves me money which I can use to invite a date to dinner, mitigating the risk of being childless
Driving faster to avoid missing a plane saves me money which I can use to invite a date to dinner, mitigating the risk of being childless
13/ Summing it up so far: rational actors exposed to a non-representative subset of the environment react taking excessive risks and thus decreasing their overall ergodicity.
That's why policies & technologies which "protect from small harm" make us take risks ("Fence Paradox")
That's why policies & technologies which "protect from small harm" make us take risks ("Fence Paradox")
14/ Purists of ergodicity might have noticed that, in this thread, I use the term ergodicity as if it represented a spectrum, whereas it is originally a binary concept.
I do so because, for the human mind, ergodicity is on a spectrum (justification in the next tweet)
I do so because, for the human mind, ergodicity is on a spectrum (justification in the next tweet)
15/ The reason is that people have to juggle multiple risks in their life: bankruptcy, illnesses, social ostracization, childlessness, etc. Spending too much time & resources on avoiding one makes us vulnerable to the others.
16/ It is risky for people to completely negate exposure to a risk, just as relegating yourself inside a house would cause you to lose any form of social support. Therefore, the need to evaluate ergodicity as a spectrum.
17/17 You can read more about the dynamics of ergodicity in my latest paper: luca-dellanna.com/ergodicity-dyn… 
