That's why I explicitly noted that the sample size was very small. Sure this is not a scientic exercise. But the thread reflects impressions from recent premades, so it is still interesting to look at this data, even if it is not representative and thus should not influence class revamps (which will not be coming anyway).