When first principle reasoning fails
I admire first principle thinkers. They ace physics tests and master engineering classes quickly. At MIT, mastery of first principle thinking comes with a huge academic advantage.
However, the real world is full of imperfect information and uncertainty. Information is always changing, and the most valuable data is the most current data. The real world is a chaotic system that is intractable to solve analytically, especially any system with people involved.
This challenges first principle thinking, which builds mental models on top of axioms considered to be true with certainty, and assumes tractable solutions.
I think empirism, aka the scientific method, can be a very useful alternative approach in chaotic systems. This process looks a lot more like 1) form hypothesis, 2) gather information, 3) synthesize and revise hypothesis. And repeat.
First principle reasoning can be useful for forming and revising hypotheses that are then tested in the real world following the scientific method. However, it can be misleading to form conclusions from axiomatic reasoning alone because information can become outdated and hidden variables are everywhere.