A refutation of the doomsday argument