I agree with Gerbrand Ceder. The more advanced our knowledge becomes, the less we can permit ‘trial and error’ experiments. If researching properties of materials is possible by computation, why would you keep experimenting? We all know that computers are more efficient than humans, thinking of a new material yourself is probably not that easy. This doesn’t mean that experiments can never be done, I would encourage them as a second step, to see how everything works out in real life.