> the 4th point around context is almost universally true - to put it another way, you cannot make a bell curve with a single data point. It becomes difficult to accurately evaluate how well a single person is doing.

I've always been confused about the terrible reputation of "stack ranking" for roughly this reason. What other kind of ranking is there?

You can rate people according to metrics that have been established for the task they're doing, but the nature of software is that when a task has been done once it doesn't need to be done again.