[amsha]

> I think that in studying economic growth, we (and especially we in the Silicon Valley) focus way too much on gadgets, and too little on the simple fact of human knowledge of how to do things.

This is a major component of software's power – it's a tool for formalizing knowledge and skills permanently and distributing it rapidly at ~0 marginal cost. If a piece of software is both correct and performant, it can last a very, very long time, possibly outliving its creators. And the best software composes, so one program can build on top of another to build up a massive skillset.