> But the fact that most servers run Linux isn't indicating it's the best choice

True, but server choice is typically made by professionals, while desktop choice typically isn't. So people measure those two by a (imo correct) double standard