Reversion to the mean. If it's not someone you know well, you are hiring them based on a limited dataset. You are likely to get the outliers, eg people who happ

Reversion to the mean. If it's not someone you know well, you are hiring them based on a limited dataset. You are likely to get the outliers, eg people who happened to remember how to implement Dijkstra's algorithm. Or people who were above average, plus they guessed right on some of the multiple choice questions. Whatever selection you choose, there's gonna be noise, and some people will benefit from it. Once you hire them, you end up with more information and there's gonna be a tendency for it to be clear that they don't know everything.