This suggests the following question: Can a set of $n$ dice be numbered in such a way that, given any one of them, there is always another that will beat the first, on average?