I like this idea! I could definitely see running an ML algo to cluster the Reasonings by keywords and then hide the shorter ones, with the assumption that the longer explanations are more in-depth.