A couple weeks ago Noah Smith wrote a very sharp piece for the Atlantic about the limits of macroeconomic models that nevertheless, IMHO, fails to reach the most important conclusion of all.

Let’s say I devise a macroeconomic model that can accurately predict the economy. Whoohoo! Now, what to do about it? There are two broad categories of action I could proceed to take, both which end in the same confounding paradox – either publish the model, or proceed to make a massive killing in the markets.

Either way, someone is going to make a massive killing, either immediately or once the value of the model becomes clear. So much so that model-adherent traders and investors will quickly grow to the point where they are a substantial share of the market itself. So the model will make a prediction and traders and investors will act in such a magnitude as to alter the economy’s path. But the model will then consume all the new data and spit out a new prediction, which will then spur traders into a new path-altering round of trading, and is the paradox clear yet? Essentially, the very existence of this model would send markets into a Gödelian-loop of accelerating paroxysms of wild trading. Imagine this model connected to HFT algorithms and you get the idea – volatility without end, markets seizing.

The moral of the story is that all finite formal systems are limited by their inability to account for themselves (ie, can never be complete, consistent, and decidable) and that more broadly human behavior is in part determined by human knowledge about human behavior and thus as the latter changes so does the former and thus can never be complete or settled.

Or to simply quote Ambrose Bierce:

Mind, n. A mysterious form of matter secreted by the brain. Its chief activity consists in the endeavor to ascertain its own nature, the futility of the attempt being due to the fact that it has nothing but itself to know itself with.