[EM] One more thing about "unknown middle"

[LAYTON Craig]

Mike wrote (in part):

>For one thing, even if he doesn't get on the ballot, he can register
>as a write-in, and I'll include him in my ranking if it's to my
>advantage to do so.

Write-in? Do you mean writing the name directly on the ballot?  This would
require special rules.  What happens when there are spelling errors?  Are
they considered as different people, or would there be some kind of 'person
is reasonably identified by writing on the ballot' type rule.  What happens
if there is more than one person with that name?  What if voters write stuff
like 'the guy with the beard'.  If there is only one candidate with a beard,
is this a valid vote?  I think you're talking about major headaches for vote
counters and legislators.

>I believe that voters shouldn't have to rank any more candidates than
>they want to. I'm sure that nearly all here agree with me on that.

Ah, I really wasn't aware that this was the case.  It is perhaps a flaw that
your (and I don't mean you in particular) examples always assume
non-truncated rankings.  There should, perhaps, be additional rules for
this.  I think that pairwise contests where no candidate gets an absolute
majority should be considered differently from pairwise contests where a
candidate does.

>Why tell people that they must rank more candidates than they want to?
>Just say "Rank as many candidates as you wish to in order of preference."

I basically agree.  I was offering a counter-argument because I assumed that
there would be alot of support for compulsory preferences.

I imagine that the justification would be that no-one likes two candidates
exactly the same.  Encouraging as many people as possible to use preferences
means that the results more accurately reflect what the voters think
(without invalidating voters who do not rank all candidates where their
preference is clear).  Please note that I don't necessarily support this
view.

>But there is a rank-balloting issue that I'm not sure of the answer to.
>Obviously, if you just put an "x" next to 1 candidate, then you
>intend to vote him over everyone. But say you rank several. Maybe
>you mean to vote all of those over everyone whom you didn't rank.
>Or maybe there are other candidates who belong inserted at various
>places in your ranking, but you didn't bother to include them because
>they don't seem winnable, and you don't want to take the time to
>rank everyone. You just rank the more winnnable candidates, to save
>time. In that case, you aren't saying that your ranked candidates are
>better than the others.
>
>I suggest that there should be a box that you can mark on the ballot,
>to indicate which of those 2 interpretations you want. But which
>should be the default assumption? Maybe, if the voter doesn't indicate
>a preference about that interpretation, the least possible should be
>assumed. In any case, I feel that there should be a box where a voter
>can indicate whether or not he votes all of his ranked candidates over
>all the others.

I think that this is worse than requiring voters to rank all candidates.  I
doubt that it could be phrased clearly enough so as not to confuse voters (I
hear that some American voters have difficulty identifying arrows pointing
at circles; I think that your two options might be a bit much for them).  I
would assume that the default interpretation is that the voter intends all
numbered candidates to be preferred over all unnumbered candidates.  This is
the interpretation used in all preferential systems that I'm aware of, and
any other seems rather counter-intuitive.  You can include an instruction on
the ballot that this is how the votes will be counted.