[EM] Multiple Same Choices

[Mike Ossipoff]

When a ballot ranks 2 or more candidates at the same rank position, 
the voter isn't expressing any preference between them. If you rank
A & B equally, then you aren't one of the voters ranking A over B,
& you aren't one of the voters ranking B over A. So it wouldn't
be right to count you as saying A is 1/2 better than B, & B is 1/2
better than A. That isn't what you're saying.

Purely on principle, then, the voter should be able to decline to
express a preference between A & B, and should be able to not be
counted as expressing preference between them.

***

Aside from that, there's the results to consider. As Steve said,
it matters how circular ties are solved. There are circular tie
solutions that don't care whether we count that shared rank position
as no preference, or 2 equal & opposite preferences. Copeland's method
is one such. Another is Lucien's suggestion to solve circular ties 
randomly.

But Condorcet's method _does_ care which way we count shared rankings.
Condorcet counts votes-against. I've talked about why that's desirable,
in order to achieve what we want from a single-winner method. Do
you wanted to be counted against A & B both, because you didn't
vote for either over the other? 

***

To summarize:

1. If you don't express a prefrence between A & B that means that
you should be counted as voting neither over the other. Counting
you as voting each over the other is both self-contradictory &
contrary to your expressed wishes.

2. Condorcet's method requires that voters not be counted as voting
a preference that they didn't vote. Its valuable properties depend
on that.

***

Mike


--