Exams are designed to be accessible, but also have grades that
distinguish.
Inaccessible to the dumb.
What is that paradox? Applies to many metrics in general.
The main problem is that zero-sum means that even if average society gets smarter over time, the dumb will fail.
So inevitably it's biased toward the good, but there is a dilemma about how much it will affect the bad. Is that the natural nature of competition?