[EM] Re: MC -1,0,1 ---- IRV hybrid (slightly amended)

[Kevin Venzke]

Chris Benham a écrit:
>I understand that for purely strategic voters with
clear preferences,
>-1,0, 1 is equivalent to Approval. I accept that the
idea of limiting 
>Dissapprovals is a mistake, but the example given by
Kevin Venzke of a 
>2-candidate race being expanded to a 3-candidate race
by the nomination 
>of one of the first two's sister was not in my
opinion good, (assuming 
>the sister only takes votes from her brother) because
any candidate 
>with sufficient proportion of the votes to win a 2
-horse race will 
>always win anyway.

I'd like to defend my example, even if you no longer
limit disapproval.  First, I picked a "2-horse race"
example because it's the simplest, even though it
requires a tie in the first-place rankings for a
candidate to benefit from entering his sister.  If you
were to have, for instance, 8 right-wing candidates
and 3 left-wing ones, the ones on the right would have
an inherent advantage with the disapproval limit in
place.

"Taking votes from her brother" is only an issue when
counting the top rankings.  The supporters could be
instructed not to cast any 1st-place rankings to the
sister.  If there is no majority of 1st-place
rankings, the sister's entrance in the race is
harmless to her brother.

More concretely, with the (candidates-2) disapproval
limit in place:

50: A>S>B, approve A and S, disapprove B
50: B, approve B, disapprove A or S

No candidate has a majority of 1st place rankings or
approval.  Approval minus Disapproval is 0 for B, but
averages 25 between A and S.  If instead of a tie, we
added in third-party candidates, A and S would have a
similar advantage over B, although it would be
weakened.

The system is expressive, but the approval/disapproval
seems to have much more weight than the rankings. 
First-place rankings are always important, of course. 
But later rankings are only considered if no one has a
majority of those, and:
1) >=2 candidates have majority approval, or failing
that,
2) >=2 candidates somehow escape majority disapproval.
I think both scenarios would be pretty rare.

>If all candidates are Disapproved by the majority,
then the winner 
>shall be the candidate with the highest Approvals
minus Disapprovals 
>score (unless you have a rule which states that in
that case noone is 
>elected).

When Dave Ketchum objected to limiting disapprovals, I
think his concern was if the *same* majority rejected
all candidates.  But here we're talking about each
candidate being disapproved by (probably) different
majorities.  You'll be voiding a lot of elections if
no one is elected in that case.

You know that MCA doesn't just add up +1, 0, -1,
right?  The "+1" votes are counted.  If there's no
majority, the "0" votes are counted as equal to "+1"
votes.  ("-1" votes aren't counted as anything.)  It's
because of this behavior that it's not necessarily
silly to vote "0" under MCA, whereas under Approval it
would be inefficient to give somebody half a vote, for
instance.

I only mention that because MCA already acts like a
ranked ballot of sorts...  Just not like IRV.  I'm
curious as to why you prefer IRV for the ranked
portion.

Kevin Venzke
stepjak at yahoo.fr

___________________________________________________________
Do You Yahoo!? -- Une adresse @yahoo.fr gratuite et en français !
Yahoo! Mail : http://fr.mail.yahoo.com