This is the crucial bit: "In a study of engineers at two major Silicon Valley companies conducted some years ago, I found that nearly 40% perceived themselves as performing within the top 5% of their peers. Ninety-two percent felt they were in the top quartile, and only one engineer felt his or her performance was below average."
Say you're the manager. Let's start off with the optimistic case: you have a good relationship with your team, are using objective but not necessarily easy to measure (to prevent gaming) criteria to evaluate your employees, and you are pretty accurate in your assessment of how valuable each person is. They aren't going to believe you. Almost everyone thinks they should be near the top. You have no way to win this unless you judge purely based on lines of code or other such measures, in which case you've actually already lost for other reasons.
And then consider how hard this is for a less experienced manager new to the team, who might not have a full picture of everything but is going to get hammered by people complaining about not being at the top of their peer pay grade.
You can go away from pay for performance, which the article notes, but do you really want to? Is that the situation you'd prefer to work in, where you know what someone makes purely because you know how much anybody with that many years tenure would make?
