This is the sinister part to me, that the right way is so hard to find, if it even exists. Try something -> get error -> fix error -> working might be ok for most uses, but this probably isn't going to yield the best way of doing things.

It's the equivalent of building a bridge, watching it fall apart, then figuring out how to artfully tie a rope around the bridge to keep it upright, rather than just designing a bridge that can stand on it's own. Most practical knowledge is therefore about artfully wrapping different ropes in clever ways around various problems, rather than fundamental theory on how to build the best design.

As a result of these thoughts, I doubt my code, I doubt everyone's code, and I've lost that naive optimism about technology. Good enough is good enough to ship, after all.