If a quantum computer outperforms normal ones, can you tell if it's right?

Three approaches to the issue of quantum results that can't be verified classically.