That assumes "rigging" is the only non-random component. There's also unintentional bias; when asked to name a random number between 1-100, multiples of ten are underrepresented due to our bias.

It solves the rigging problem, it doesn't necessarily make claims as to whether that is the only problem that exists.