[EM] IRV mailing party adventures

[LAYTON Craig]

Tom wrote (in part):

>I would assume a voter ALWAYS wants a ranked candidate to get a vote over
an
>unranked candidate, but I suppose this may not be the case. He claimed the
>system shouldn't make assumptions about what it means to leave candidates
>unranked.

The first thing to remember is that this is not specific to Condorcet - it
is a matter of interpreting ANY ranked ballots.  In IRV, a vote that numbers
candidate A first, is considered to vote A above all other candidates,
including those which are not ranked.  It is a decision that must be made
when implementing preferential voting.

We had discussed this on the list once before.  There are a number of
options available.  The simplest is just to interpret the unranked
candidates as below the ranked candidates - most people understand this, and
it is the practice in all preferential systems I'm aware of that allow
truncated ballots.  To avoid any objection, you may include an instruction
on the ballot or in the polling booth letting voters know that this is how
the ballots will be counted.  There is also a possibility of allowing the
voter to pick which of the two ways she wants her vote to be counted.  Mike
originally proposed this, but then withdrew the idea because of complexity
and the fact that you would have to have a default choice in any case (not
to mention that voters who don't number all candidates aren't very likely to
specify a preference between two alternative ways of counting their unranked
options)

>I wish I could think better on my feet! Damn philosopher tricked me!

Philosophers aren't that bad! :-)