One can learn to think like Fermi. Perhaps not at his level of intelligence (this is to some extent genetically fixed, after all) but with the same qualities.

One can learn to think like Fermi. Perhaps not at his level of intelligence (this is to some extent genetically fixed, after all) but with the same qualities. A good start is asking what the fundamental uncertainties are and which ways there are to resolve them. Try to formulate these uncertainties in terms of physical, mathematical, or otherwise experimentally established laws. From what I understand, this is one thing Elon Musk is decent at, answering "is there any world in which this would even be physically possible, given no constraints on the innovation we can do"? In a more fluffy example, I started out on a project recently and realised that if I improve on the state of the art along only one dimension, I don't think the product will be good enough to gain traction. So I need to outperform the state of the art along at least two dimensions -- this is significantly harder and I now doubt my ability to do this with the resourcew I'm willing to spend on it. That reasoning requires no technical knowledge, just thinking about the fundamentals of market adoption and what the end product could look like under no constraints on innovation.