Well, in his defense, type theory is supposed to make software engineering simpler, not more difficult. So if even he doesn't understand it, then how can we expect a random programmer to?
Well software engineering is the only engineering field that produces half baked unreliable crappy solutions continuously. Other fields cannot afford such attitude towards products. I think a simple (inferred) type system is pretty useful to increase correctness and it helps you to increase reliability (even though it does not avoid all of the mistakes you can make in software).