[Election-Methods] RE : Re: Borda-elimination, a Condorcet method for public elections?

[Jonathan Lundell]

On Dec 22, 2007, at 11:55 AM, Kevin Venzke wrote:

> Jonathan,
>
> --- Jonathan Lundell <jlundell at pobox.com> a écrit :
>> On Dec 22, 2007, at 6:45 AM, James Gilmour wrote:
>>> If you wish to utilise in some way all the information that could be
>>> recorded on a preferential ballot, that is a completely
>>> different voting system from IRV, with different objectives.  The
>>> preferences are no longer 'contingency choices', but take on a new
>>> function depending on the detail of the voting system.  It is almost
>>> certain that the voters would mark their ballots in a different
>>> way in an election by such a voting system from how they would mark
>>> their contingency choices in an election by IRV.
>>
>> This seems plausible enough (and certainly IRV voters should be
>> instructed along contingency lines). WRT marking ballots differently,
>> setting manipulation aside, and considering only contingency vs
>> preferential ranking, do you have an example or two of how and why a
>> voter might end up with different ballots in the two contexts?
>
> I imagine James Gilmour is hinting at the fact that specifying lower
> preferences under Condorcet methods can damage higher-ranked  
> candidates. So
> the Condorcet voter has to think about how far down to rank while  
> the IRV
> voter should not have to.
>
> Not just voter incentives but candidate incentives will vary by  
> method,
> too. Clearly certain types of candidates will not be viable under  
> IRV but
> could be under Condorcet. It would probably be more safe to give a  
> high
> ranking to this sort of candidate under Condorcet than under IRV.

I can see that, more or less. STV and Condorcet are subject to  
different kinds of manipulation, and that could influence a voter's  
ranking.

What I'm asking, though, is something else, and I concede in advance  
that it's a fuzzy question, or perhaps one with an obvious answer.

Is there a case in which a voter's sincere ballot might differ between  
an IRV (contingent choice) election, and (say) a Condorcet-compliant  
election? I take James Gilmour to suggest that there might be, and  
while it seems plausible to me that there might, I'm not able to think  
of a clear example. It depends, I suppose, on what we mean by "sincere  
ballot", and here the literature is not very helpful; it relies on the  
intuitive idea that each voter has an ideal ranking of candidates,  
possibly with ties, possibly with don't-cares.

It's easy to produce a counterexample for a multiple-seat STV  
election. Here the voter must deal with two different contingencies:  
my first choice might be eliminated, or my first choice might be  
elected with a surplus. My sincere second choice might well be  
different in those two cases; I might prefer A if only one of my top  
choices is elected, but prefer B and C if two are to be elected (let's  
say, for example, that B and C have complementary virtues, but that  
either alone might be less effective than A in representing my views).

This problem doesn't arise in a single-seat STV (IRV/AV) election, of  
course.

At the moment, my guess is that in fact the a difference arises only  
because of the different opportunities for manipulation presented by  
the different voting systems, but the question of whether a voter's  
sincere ranking might be different is intriguing.