[EM] Hybrid/generalized ranked/approval ballots

[Kevin Venzke]

Hi Peter,

You are right. Your treatment has been discussed for, I suppose, 15+ years
on this list, and your reasoning for it is a major factor.

--- En date de : Jeu 26.5.11, Peter Zbornik <pzbornik at gmail.com> a écrit :
Kevin Venzke wrote in his mail below (May 9th 2010):
35 A>B
25 B
40 C
A will win. This is only acceptable when you assume that the B and C
voters meant to say that A is just as good as the other candidate that
they didn't rank. I don't think this is likely to be what voters expect.
It seems misleading to even allow truncation as an option if it's treated
like this.
End of quote

you wrote:
Well I think think that as a voter I would indeed be pleased if A would win and not C.
If the completion system above would be used (i.e. A=B would be counted as 0.5 win for A vs B and 0.5 win for B vs A), then there the winner would always be the same disregarding which of the following winning criteria was used: winning votes, losing votes, margins or quotas. 

[end quote]

Can you explain why as a voter you would be pleased if A won and not C?
What reason do these ballots give us to suppose that A is a better
candidate than C? Does the fact that A voters have a second preference
make A a better candidate?

In other words: Electing A violates Woodall's Plurality criterion.

I don't find the election of C to be very good, but it is better than
electing A.


[resume quote] 
Let us analyse the example in your mail below. We apply the Schulze beatpath with different criteria (biggest win, margins and ratios):
[end quote]

I think you forgot Schulze as it is usually done: Weakest biggest loss. 
In that case you elect B, which is my preference. This minimizes the
number of voters who feel the outcome was spoiled by one of the 
candidates.


[resume quote]
What are the pros and cons of the approach above?
Prima facie it seems that the treatment of hybrid ballots above could solve the problem of bullet voting, but I am far from sure.
[end quote]

I still stand by what I originally said, that I don't believe voters will
expect or like this treatment of truncation. Voters will expect that if
more than half of the voters say X was better than Y, then that's the
important contest, and Y should certainly not win. The presence of some
weak candidate Z should not cause the method to be confused about X vs.
Y.

You will tell them you're only doing this for their own good. But I
think there is an incurable disconnect between what truncation seems to
mean and what it actually does mean in your scheme. I think that when
you use this scheme, truncation should not be allowed at all: You should
have to explicitly rank all the worst candidates equal, if that's how
it's going to count.

Otherwise, you are going to sometimes be electing candidates on the basis
of equal rankings that were cast in the form of voters completely ignoring
a candidate. That seems like an opportunity to criticize the legitimacy
of the winner.

Experimentally, in simulations: When you treat equal-ranking as split 
votes, voters will have to compromise more often, instead of just
compressing the top ranks. This suggests weaker, non-frontrunner 
candidates are more likely to be best advised to drop out of the race,
because their presence is more likely to harm the voters that support
them.

Kevin