5 ms·

I find it hard to believe that there could ever really be a good economic model. Predicting the weather works with a supercomputer since all the processes (temp

by mihaic 5y ago

I find it hard to believe that there could ever really be a good economic model. Predicting the weather works with a supercomputer since all the processes (temperature, pressure, etc) can be modeled as continuous. A very successful Spiderman movie saving Sony's quarterly profits hangs often on a coin-toss. Our lives are governed by relatively simple rules of physics but the world is in an incredibly complex state. Economics takes the output of those rules of physics and tries to wrangle the state of all the atoms in the world into some simple values, and then uses some simple equations to make predictions. How is this not worse than spherical cows?

joe_the_user 5y ago
Lots of system consisting of discreet pieces can be modeled fairly exactly. The thing that makes economies very hard to predict is they're a combination of people acting according to quantifiable economic incentives and people acting accord to a collection of ideas, fashions and emotions that can switch unpredictably or simply aren't known.
JumpCrisscross 5y ago
> Predicting the weather works with a supercomputer since all the processes (temperature, pressure, etc) can be modeled as continuous Except it's not continuous. That assumption works most of the time. But sometimes--often--it doesn't. Particle interactions chaotically manifest systemic effects in unpredictable, dramatic ways. The limitations on our current models of fluid dynamics and economics are uncannily symmetric. (The latter fails more unexpectedly.)
- mihaic 5y ago
  Well, yes, my main point was that the weather is much closer to a perfect continuous model than the economy, and we get weather predictions wrong all the time. I doubt we'll have a change with the economy in the next 50 years.
MuffinFlavored 5y ago
> Predicting the weather works with a supercomputer since all the processes (temperature, pressure, etc) can be modeled as continuous. Why can't the flow of money be modeled? Don't we have enough data on human behavior historically? Can't we trace where money generally ends up? Using the recently printed money isn't a super great example because as far as I understand, it's mainly held up in the banking system. Maybe the stimulus money given to American citizens recently would be a better example. We know some of it went to savings, some of it went to bills, some of it went to frivolous purchases, etc. If AI can detect fraud / objects on a road and make decisions, why can't a few of the most common economic possibilities be fed into some kind of model?
- saurik 5y ago
  The argument was that discontinuities exist in things like, starting with the specific example, whether a movie turns out to be particularly interesting enough to drive people back into theatres.

by mihaic 5y ago

joe_the_user 5y ago

Lots of system consisting of discreet pieces can be modeled fairly exactly. The thing that makes economies very hard to predict is they're a combination of people acting according to quantifiable economic incentives and people acting accord to a collection of ideas, fashions and emotions that can switch unpredictably or simply aren't known.

JumpCrisscross 5y ago

> Predicting the weather works with a supercomputer since all the processes (temperature, pressure, etc) can be modeled as continuous Except it's not continuous. That assumption works most of the time. But sometimes--often--it doesn't. Particle interactions chaotically manifest systemic effects in unpredictable, dramatic ways. The limitations on our current models of fluid dynamics and economics are uncannily symmetric. (The latter fails more unexpectedly.)

mihaic 5y ago
Well, yes, my main point was that the weather is much closer to a perfect continuous model than the economy, and we get weather predictions wrong all the time. I doubt we'll have a change with the economy in the next 50 years.

MuffinFlavored 5y ago

> Predicting the weather works with a supercomputer since all the processes (temperature, pressure, etc) can be modeled as continuous. Why can't the flow of money be modeled? Don't we have enough data on human behavior historically? Can't we trace where money generally ends up? Using the recently printed money isn't a super great example because as far as I understand, it's mainly held up in the banking system. Maybe the stimulus money given to American citizens recently would be a better example. We know some of it went to savings, some of it went to bills, some of it went to frivolous purchases, etc. If AI can detect fraud / objects on a road and make decisions, why can't a few of the most common economic possibilities be fed into some kind of model?

saurik 5y ago
The argument was that discontinuities exist in things like, starting with the specific example, whether a movie turns out to be particularly interesting enough to drive people back into theatres.