He means to reason using axioms, or the basic building blocks of understanding, rather than to use analogy or argument as basic building blocks. For every day life it is slower to do such laborious reasoning, but it is often the only way to untangle a difficult question. A programming equivalent would be like how being able to program in C or assembly, better yet hardware, really helps to create new algorithms and use high level languages.
Thank you for this response, it's a nice addition to the responses from reddit which were:
{
\His answer is essentially "systems engineering" which is multidisciplinary or "holistic" applied physics. This then combined with awareness of what is broadly possible and a certain confidence that what you are doing is important and will be the right way of doing things.
Basically the entire world becomes you own Lego/Tinkertoy/Erector set without any of the artificial boundaries defined by academia.
Nature doesn't have teams called "physics", "chemistry", "biology", or "science" vs. "engineering". Those are human social inventions to address bounded rationality of humans, not qualities of nature itself.
permalinkreportgive goldreply
\He is saying that when you are looking at a creating a solution to a problem it is best to understand the root problem from first principles - otherwise you end up doing a derivation of what is already known/being done. He is saying that is the only way to do something truly new and unique without resorting to luck.
}
It is not special to physics in my opinion.