Mathematically, a random sample of a few thousand is better than a non-random, bias sample of a few million. A larger -N of a biased sample just leads to a more pronounced bias. Additionally, WoTC's choice to count characters instead of players is, well, interesting. It keeps any of us from...