[icebraining]

I don't think that's what he's saying at all. It's closer to following a scientific approach; what you "know" is simply a more or less wrong model of the reality. You may study something "magical" until you understand it completely, but you still must be prepared to revise your knowledge if suddenly you learn something new that goes against it.

[emptytheory]

Programming systems can be treated scientifically, but they can also be treated deductively (like math). It's possible to have full knowledge of what's going on if you treat it like a deductive system. If your primary way of learning is through experiment, then yes, you will be wrong a lot of the time. But then I would never claim you had a good reason for believing anything.

I can't view the stackoverflow link right now, but the top answer basically claimed "I feel strongly about my reasoning, but nevertheless, I could be wrong". How could that be unless you made a mistake? The system is logical!