It’s no fun implementing QuickSort unless you can force it out of its blister-fast, O(n log n) speed and humiliate it with its worst-case, O(n^2) runtime. So that’s what I set out to do.

My naive partition simply pivots around the low item, but my randomized partition defeats the sucky inputs by choosing a random pivot. (If you’re interested in checking out a QuickSort which naively partitions until it hits an attempt to get it to run in quadratic time, check out IntroSort — which simply fails over to Merge Sort when it exceeds its optimal recursion depth.)