The problem with the dull-witted "fire the bottom 10%" approach is that it assumes a standard Gaussian distribution. But if a company's hiring and training policies are good, then what it will have is a very skewed distribution. And that totally invalidates the simple-minded "fire the bottom 10%" idea. Exxon learned this the hard way many years ago and it's depressing to know the same stupid idea is still in circulation. Especially in Silicon Valley.