Ergodicity
all our decisions are underpinned by an implicit assumption of a time horizon to optimize for.
performance is subordinate to survival. It is the fastest racer of those who survive that wins races, it is the highest performing employee who doesn’t burn out that becomes the most successful, and so on.
If we make decisions based on what happens over narrow intervals and forget about these “phantom consequences,” we will make bad decisions. Ergodicity is the study of these phantom consequences.
Your individual expected outcome if you play Russian Roulette once is also 1/6 of a death and 0 and a hole in your head.
A system is ergodic21 if its population outcome coincides with the lifetime outcome of each of its components. Otherwise, it is non-ergodic.
in ergodic systems, you can use the population outcome to make optimal decisions. In non-ergodic systems, you cannot.22 For example, Russian Roulette is non-ergodic because the lifetime outcome differs from the population outcome. If you keep playing it because you fail to grasp its non-ergodicity and mistakenly believe that your lifetime outcome equals your population one, you will end up dead instead of rich.
something working on average doesn’t guarantee it works.
we cannot rely on averages whenever there is a possibility for irreversible damage35. This is because the possibility of irreversible damage makes a context non-ergodic.
Fully relaying on averages and expected outcomes makes sense only in ergodic contexts. In non-ergodic ones, you want to ensure that you make decisions based on an estimate of your lifetime outcome.
Ergodicity is verified if, more or less, every time that you compute a statistical measurement (across space or time), you find the same result.
An injury can cause a skier’s championship to be over, nullifying any future wins he could have collected in the remaining races.
most practical contexts are non-ergodic. Work, sports, relationships, and real life are all non-ergodic.