Should theories that no conceivable experiment can confirm be rejected outright?