My experience coaching a college sports team for a few years is that even very good players are often pretty bad at assessing the contributions of their teammat

My experience coaching a college sports team for a few years is that even very good players are often pretty bad at assessing the contributions of their teammates. They tend to significantly over-weight the contributions of players who have a similar style of play to themselves, and underweight the contributions of players who play other positions (or play the same position differently). They especially undervalue actions that happen "away from the play" (the software-engineering equivalent might be something like rewarding a teammate who steps in to help you fix an critical block-ship bug, missing the person who had the foresight to write a test case two years ago that prevents a bug from occurring in the first place).