The question I would like to see investigated is the hypothesis that the best posts are those with a large value of (#points)/(1+#comments). I've used that for years as my measure of when an article that would otherwise have been missed should be read.