There is some logic behind always hiring the best you can find and having a process to cull the low performers. It's not like the company's actually shrinking by 10% every year. The problem only shows up when you hit a point of actually having all high-performers, and you're only throwing people away to meet a quota that's become counterproductive.
Even if hires were assigned teams at random, attrition doesn't occur evenly. Bad teams lose good people very fast. Teams with bad managers will have huge rates of churn (people rarely quit their job, they quit their manager). So the end result is to naturally divide into multimodal distributions: Rich get richer, poor get poorer. Since ranking teams is, from an organizational perspective, extremely political, you'll rarely get management to point fingers to the bad teams.
Also, splitting a high performing team rarely works: A team is not a collection of individuals, but also a culture. Dump a great performer on a bad culture, and even if you put him in charge, you do not get a good culture: Instead, you get less out of that person.
If there's anything I've learned, is that we are better off ranking teams, not people. The one reason to look inside a team is for voting people off the island, and that is something that should be started from within a team.