The idea here is that by taking an approach similar to [BOGO Sort](https://en.wikipedia.org/wiki/Bogosort) we can have a roughly randomized starting point, and adjust off that. Basically BOGO randomize once, find all the bad items, and then assumed/forward fill off that.
