what metrics would you use to assign points? What would a point mean? While I understand and sympathize with the desire to quantify 'good teaching', it's a 'do

what metrics would you use to assign points? What would a point mean? While I understand and sympathize with the desire to quantify 'good teaching', it's a 'domain' that has a whole lot of noise. Abstractly, agents will optimize for the metrics, and not what the metrics approximate. Now if an agent in the system optimizes for the metrics (and not what they're measuring), but the metrics themselves are solid, then you'll come out with a largely okay system. I think the point of this article is that maybe we don't have good metrics to measure solid pedagogical performance, and so, in that case, when agents optimize for the metrics, and not what the metrics measure, then the system provides no value. So maybe the crazier idea is back-off, and try a different approach that doesn't involve rewarding or punishing the teachers themselves.