Mr. Marks (https://www.oaktreecapital.com/insights/memo/the-illusion-of…) makes some statements about modeling that I believe should not be taken at face value. Specifically he seems to confuse models as having to represent the real world in its infinite detail, which is not possible.
A real simulation of the U.S. economy…
Simulations by definition are not “real,” they are characterizations of the real world.
This is yet another example of why a model simply can’t replicate something as complex as an economy.
Of course a model cannot replicate reality, that is why it is called a “model” - by definition a simplification of the real world (like a “model” car for example).
Can a model replicate reality? Can it describe the millions of participants and their interactions?
No to the first question and maybe to the second. As computing speed and storage increase, models with increasing complexity will be available for use.
Here is a famous quote I prefer: “All models are wrong, but some are useful.”
Models can represent our state of understanding of a system, which is obviously incomplete. Studying how and why models deviate from reality, including forecasts, is one important approach to improve our understanding of a system.
An interesting and I think successful application of some fairly complex modeling is the forecasting of the paths of tropical storms. While not perfect, I think they do pretty well and you can bet they’ve been working at it diligently for a long time and have been plenty wrong in the past. But they’ve kept at it, and they are getting better all the time.
Thank you, hurricane track forecasters.