Talk is cheap.
To build fundamentally valuable skills, we need fundamentally reliable mechanisms of error correction.
And unfortunately, the world is full of misleading signals.
For instance: imagine I play you a song on the guitar (I don't know how to play the guitar).
Out of politeness, you might tell me I'm not that bad.
But if I asked you to buy a recording of my guitar songs for $20, I doubt you'd be so lenient.
Which brings us to the two forcing functions for truth: free markets (which teach us communication) and the laws of physics (which teach us to think from principles).
"Markets" can include any kind of voluntary transaction, not just monetary, and "physics" can include anything grounded in physical laws, not just rocket science.
The less diluted these are, the better.
It's worth trying to tighten these feedback loops by avoiding memorisation and bureaucracy (easier said than done).
Then, once we have a strong enough foundation, it becomes much easier to move on to prong two.
P.S. Due to emergent properties of systems, we can't explain everything in terms of physics and human nature. In practice, it may sometimes be helpful to learn more abstract concepts, but it's useful to avoid stuff that's completely disconnected from reality's feedback (think: macroeconomic news).