[EM] IRV in action

Adam Tarr atarr at purdue.edu
Fri Apr 4 07:32:13 PST 2003

James Gilmour wrote:

>The problem with this "horse trading" is that the decisions are all made 
>by the
>parties and by the candidates, not by the voters.  With preferential 
>ballots, the
>voters decide.  Our problem is putting that into effect in an acceptable way
>through the counting part of the voting system.

I wasn't really referring to the horse trading at the party level - rather 
I was referring to the chance voters have to evaluate the support 
candidates have, and decide if they should dump their top surviving choice 
in favor of a compromise candidate.  I'm not claiming I think this is a big 
factor; I'm only saying that some people have mentioned it as a possibility.

> > The winner does have "the support of half or more of those who are voting",
> > but only over the remaining candidates.  You can't overstate the
> > significance of that.  Good candidates, candidates who have more support by
> > any reasonable measure, can be discarded by IRV before a final decision is
> > made.
>This is all true, but look VERY carefully at the words I actually wrote: "the
>winner has the support of half or more of those who are voting at the 
>point when
>the final decision is made".   It is understood that IRV is an elimination 
>so any comment about the result assumes it is with regard to the state of
>elimination when the winner achieved half or more of the available votes.
>If some voters opt out part-way through by not marking all preferences (so 
>their ballot papers become non-transferable), the eventual winner will 
>still have
>the support of half or more of those who are voting at the point when the 
>decision is made.  Given the democratic right of voters to say "If it is a 
>among these other candidates, I have no further preference I wish to 
>express", I
>think that is a reasonable result.  And it is not a distortion of the truth.

I'm not really concerned with the attrition of the voters; I'm more 
concerned with the attrition of the candidates.  That "support" could 
consist mostly of voters choosing their second least favorite over their 
least favorite.

> >  Here's just such a "nightmare" example:
> > 10% FarRight>Right>Centrist>Left>FarLeft
> > 10% Right>FarRight>Centrist>Left>FarLeft
> > 15% Right>Centrist>FarRight>Left>FarLeft
> > 16% Centrist>Right>Left>FarRight>FarLeft
> > 15% Centrist>Left>Right>FarLeft>FarRight
> > 13% Left>Centrist>FarLeft>Right>FarRight
> > 11% Left>FarLeft>Centrist>Right>FarRight
> > 10% FarLeft>Left>Centrist>Right>FarRight
> >
> > "Centrist" has the most first place votes, the most second place votes, and
> > the most third place votes.  Also, centrist is never ranked fourth or fifth
> > on any ballot.  In any non-probabalistic election method except for IRV,
> > centrist wins in a landslide.  But in IRV, centrist is eliminated in the
> > third runoff, and Right beats left 51-49.
> >
> > The results won't usually be that blatantly undemocratic, of course, but
> > the fact remains that the best candidate by most reasonable measures can be
> > eliminated early in an IRV election.
>I recognise the problem very well, but what is the practical solution?  I 
>am very
>sympathetic to Condorcet, but there must be serious questions about the public
>acceptability of some of the results it is likely produce.

I suppose the "nightmare" scenario for Condorcet (in your mind) looks 
something like this:

47% Right>Centrist>Left
5% Centrist>Right>Left
5% Centrist>Left>Right
43% Left>Centrist>Right

So a candidate with only 10% first place support wins, over two candidates 
with over 40% first place support each.  A couple thoughts here:

1)  I think it's hard to argue this is _worse_ than the IRV nightmare 
example above.  Even the layman can recognize that centrist was a 
compromise choice.

2)  After the election results are announced, nobody has any reason to 
regret their first place vote.  The same is not true in IRV, in the case 
above and in many others.  Winning votes Condorcet, in general, minimizes 
the tendency to regret your vote.

3)  If we're just talking about the public perception of a Condorcet result 
like the above one, then you can take Forest Simmons' advice and use a 
graded ballot.  It's hard to argue viscerally against the candidate that 
got mostly B's and a few A's, when the other candidates got a mix of A's 
and C's.


More information about the Election-Methods mailing list