People simply act irrationally, which is a fundamental issue when trying to treat economics as math.
Or think of it like weather and climate: we cannot predict that there will be a thunderstorm on a given day 2 months from now, but we can be fairly confident that in x years, the average global temperature will be y±z°C. Because when you aggregate enough events, statistical effects become dominant.
"But reality is not linear" is not really a gotcha.
At sufficient detail atomic structure has a huge impact on how a metal sheet bends. Metallurgy seriously investigates at this level.
Hand waving details is fine when discussing with friends, it’s not a sound foundation for serious academic research.
Also, the degree to which the weather is unpredictable 2+ weeks out is somewhat overblown. It generally snows in DC several days a year yet the odds it snows in DC 2 weeks from now is essentially zero not ~4/365. Similarly you can more accurately predict thunderstorms than a pure guess 2 months from now. We may call it climate, but a physical model of earths tilt, prevailing winds, CO2, etc is better than just historic data.
It's how physics worked for hundreds of years. Make a handwavy model, find out it works on some cases but breaks down on others, then make a slightly less handwavy model the next time.
Metallurgy considering atomic structure is a very new concept, and was not needed for the first millenias of metalworking.
The issue with economics is not handwaving, is that models are hard to test due to systems not being well isolated.
I know, it’s my job. But as much as we like to obsess about the mechanisms for dislocations climbing and solute interactions, nobody cares when they are designing an aircraft. They have macroscopic laws they put in their finite elements code. These are perfectly adequate to describe most known modes of failures of industrial alloys. Nobody is going to model all the atoms in a macroscopic widget, ever. It’s beyond pointless.
Biology was forced to deal with the insanity of biochemistry because that’s what is actually happening. Economics can’t get away from the innate complexity of actual humans if it wants to be actually useful for more than just propaganda.
Economics chooses to impose assumptions that peoples observed choices are a better way to study their behavior than what they say, and so we look at the observable state of the world for an individual economic agent. You can do analysis of people whose preferences are not rational (in the strict mathematical sense that I described above) but you must choose what kind of irrationality they have. And that gives you the ability to assume any results. That isn't any way to do science. Rationality is the worst option except for all the rest as they say.
You don’t need to read people’s minds to predict their behavior at sufficient granularity to be useful. Economics is blessed with plenty of opportunities to collect high quality real world data without needing to conduct arbitrary experiments.
How much gas will this specific gas station sell on date X. Build whatever models you want at whatever scale is relevant and they face the brutal truth of a knowable answer to judge them. That’s how science progresses not arbitrary assumptions to make modeling easier.
