The chart they show from the original paper is approximately just: perceived_ability = 0.5 + (actual_ability / 4) (With the caveat that "ability" is a per

The chart they show from the original paper is approximately just: perceived_ability = 0.5 + (actual_ability / 4) (With the caveat that "ability" is a percentile rather than some absolute measure.) That's not the negative correlation that it's often presented as in discussions. And it could as easily relate to mis-measuring everyone else's skill as to mis-measuring your own skill.