[EM] Pairwise Margins

[Blake Cretney]

Blake Cretney wrote:

> I don't know how many are needed by the optimal procedure, but I do 
> know how many are needed by my procedure, 2*(n^2-n)
>
> So, for 5X5, that is 40

Blake makes some good points, but I have to quibble with that statement. 
 You only need two distinct factions for every pair of candidates (not 
every cell).  So, the formula is actually n^2-n.
So, for 5X5, that gives 20.

---
Blake Cretney