I think that the problem is where we assume everything has an equal chance of appearing. Apply a weighting factor and the %ages get screwed. It's still random. Just not as we interpret it.