Comment by casey2

Perhaps for data points in similar extremes, but there are definitely a good many theorems that show there is reason to weight extreme outliers more heavily. And practical experience using them as a signal to test for measurement errors etc.