,
5 tweets,
2 min read
Read on Twitter
@jben0 Yes. Unknowingly, of course. Laplace and vNM succeeded, Bernoulli failed.
‘Hunting ergodic observables’ is not a bad definition of ‘doing science.’
‘Hunting ergodic observables’ is not a bad definition of ‘doing science.’
@jben0 These psychological arguments apparently don't lead to a consistent theory. If people really are less risk averse for large losses, let's ask why.
The ergodicity framework will tell us why it's there, what it means, what it's good for, when it's counterproductive etc.
The ergodicity framework will tell us why it's there, what it means, what it's good for, when it's counterproductive etc.
@jben0 Example: for 20 years, I'm pretty happy and have $100. One day I gain $20 in the morning and lose them at night. Now I'm concerned, less happy.
Why? Possibly because I've had to update the model of my wealth dynamics. I thought my wealth was safe, but now I know it's not.
Why? Possibly because I've had to update the model of my wealth dynamics. I thought my wealth was safe, but now I know it's not.
@jben0 Fluctuations reduce the ergodic growth rate, so this response - a preference for no fluctuations - is predicted, rational etc.
It doesn't require estimating pleasure and pain of gains and losses - it's contained in the mathematics of growth rates. No psychology, just "physics."
It doesn't require estimating pleasure and pain of gains and losses - it's contained in the mathematics of growth rates. No psychology, just "physics."
@jben0 So now I wonder: are there any reliable experiments that show this effect? Can we design one?
If utility is physically interpreted, as an ergodicity transformation, dynamically updating stochastic wealth model parameters (a la Bayes) would lead to path dependence.
If utility is physically interpreted, as an ergodicity transformation, dynamically updating stochastic wealth model parameters (a la Bayes) would lead to path dependence.
