For a perpetual motion machine to exist, you don't have to be able to draw energy out of it - it is sufficient that the motion be perpetual. Unless there is a simpler explanation , I would say that Eurox's suggestion would fail because a perfect vacuum is a theoretical thing that cannot be achieved in reality, so collisions with gas molecules would slow the spinning magnet down. Dawkins may have something, but we could imagine doing the experiment on a cold planet, where the superconductor would not need cooling. Have you heard the First Law of Thermodynamics stated as "You can't get something for nothing"? And the Second law stated as "Not only that, but you can't break even"?