First principle thinking is a problem solving strategy where you break problems down. Then build the solution back up from facts. At MIT, tests are often designed to be solved this way, where there's one right answer.

However, in the real world, information is always changing. Also, there are often hidden variables that affect the outcomes. For example, whenever people are involved, behavior can be unpredictable. Finally, you are often looking at a situation where you have incomplete information.

I think applying the scientific method can be very useful here. This looks like:

1. Form a hypothesis
2. Run an experiment
3. Evaluate the results

Then repeat. This is very different from first principle thinking because you are not trying to derive the right answer from facts. Instead, you form a good enough guess, take action, then iterate.