On Fri, 28 Sep 2007, David Artman wrote:
So one might have this as a ballot, then:
2) B, C
4) D, E, F, G
And the algorithm will handle it?
Yup. It'd break down to A>B, A>C, A>D, A>E, A>F, A>G, A>H, B>D, B>E, B>F,
B>G. B>H, C>D, C>E, C>F, C>G, C>H, D>H, E>H, F>H, and G>H, with no opinion
on BvC, DvE, DvF, DvG, EvF, EvG, or FvG. And you can pump that right into
the marginal vote table.
I could certainly make it more clear that one can send in rankings like
that, by merely stating something like "you may rank games as tied for a
particular rank, in which case the next rank is lowered for each extra
game in the tie (e.g. if you rank three games as tied for first, the
next game you list is ranked fourth)."
That wording is a bit confusing, but sure (and the numerical ordering
doesn't really matter, just the relative ordering, so it doesn't matter
what number you put next to each group of tied candidates.) (Also,
allowing people to vote more tied contests into their ballot does increase
the chance of a tied final result, if that is a concern.)
(Really, you could think of your ballot as an 8x8 grid, and you can put
either a + or a - in each square, or leave it blank
I could also easily provide such a ballot
Oh, no; listing them in ranked order (even allowing ties) is much, much
easier. But from a conceptual standpoint, the two are equivalent. (And
it'd be only n*(n-1)/2, since the +s and -s have to be matched and
candidates don't compete against themselves.)
Should the judging period be longer, perhaps adjusted for the number of
That's a good idea.