Hybrid ranking/approval ballots
I'm attracted to a couple of Condorcet methods that allow voters to rank
the candidates and also give an approval threshold.
I think an important question is what should be the default placement of
this approval threshold. I've read from one or two
people that all candidates "ranked in any position" should be considered
approved, meaning that a ballot that strictly ranks
all the candidates by putting a number next to all of them should count
as strategically useless approval for all the candidates
while a ballot that does the same thing by putting a number next to all
but one of them counts as approving all but the unmarked
candidate.
I find that silly and unnecessarily unfair to naive or careless voters,
so instead in the past I have gone with "voted above at least one
candidate".
Since the approvals are only used to complete Condorcet it is likely
that most of the time they'll have no effect and so many voters won't
bother giving an explicit approval cutoff.
In that circumstance with either the "voted above at least one
candidate" or the "ranked in any position" rules, the outcome of the
election
could be affected by the addition or removal of candidates that all (or
nearly all) of the voters hate.
I find that also silly and unacceptable. So now I am strongly of the
view that default approval should be only of "voted below no other
candidate".
Condorcet methods with no Push-over incentive should allow above-bottom
ranking.
The ballot rules should allow voters to strictly rank however many
candidates they like and also (if the method has some use for approval
information)
to approve only one candidate or all but one or any number in between.
Hopefully this should all sound obvious. But several people here have
been tolerant or (even supportive) of alternatives.
Someone who proposed a Condorcet completed by Approval method suggested
that a 6-slot grading ballot would be fine and that we would arbitrarily
call the top 3 slots approval and the bottom 3 not.
If there are more than 4 candidates that isn't enough to allow the voter
to strictly rank all the candidates and approve only one or all-but-one.
In principle a grading or multi-slot rating ballot with the top half of
the slots/grades signifying approval is fine as long as there are at
least as many
slots/grades as twice the number of candidates, minus one.
Another abomination is compulsory ranking. This is "GIGO" (garbage in,
garbage out) and I find it analogous to compelled speech.
In Australia, the major parties' "How-to-Vote" cards usually just
advise their supporters to after writing a "1" next to their party's
candidate just to
number all the rest according to the order they appear on the ballot paper.
Chris Benham
This is fair enough. I suggested score or graded ballots, where the scores/grades are purely used for their rank information. But I think ranks are fine as long as they allow for ties and also allow for different interpretations of what people think ranks mean in the case of ties. E.g. someone might rank four candidates 1, 2, 2, 4 and someone else might rank them 1, 2, 2, 3. In fact you might as well open it up fully to simply saying lower is better and everything is fine, in case someone misses out a number. So 1, 2, 3, 4, 6 is the same as 1, 2, 3, 4, 5, or indeed 7, 17, 27, 28, 29. (You could still use scores but have them out of 9, 99, 999 using whatever number of digits you need to encompass all of the candidates. You're unlikely to need more than two digits.)
Toby
On Sunday, 6 August 2023 at 20:13:44 BST, C.Benham <cbenham@adam.com.au> wrote:
The ballot rules should allow voters to strictly rank however many
candidates they like and also (if the method has some use for approval
information)
to approve only one candidate or all but one or any number in between.
Hopefully this should all sound obvious. But several people here have
been tolerant or (even supportive) of alternatives.
Chris Benham
I totally support the conventions suggested by Chris.
For grade ballots I suggest refinement of rankings by using pluses and
minuses like many schools do.
On Sun, Aug 6, 2023, 12:13 PM C.Benham cbenham@adam.com.au wrote:
I'm attracted to a couple of Condorcet methods that allow voters to rank
the candidates and also give an approval threshold.
I think an important question is what should be the default placement of
this approval threshold. I've read from one or two
people that all candidates "ranked in any position" should be considered
approved, meaning that a ballot that strictly ranks
all the candidates by putting a number next to all of them should count
as strategically useless approval for all the candidates
while a ballot that does the same thing by putting a number next to all
but one of them counts as approving all but the unmarked
candidate.
I find that silly and unnecessarily unfair to naive or careless voters,
so instead in the past I have gone with "voted above at least one
candidate".
Since the approvals are only used to complete Condorcet it is likely
that most of the time they'll have no effect and so many voters won't
bother giving an explicit approval cutoff.
In that circumstance with either the "voted above at least one
candidate" or the "ranked in any position" rules, the outcome of the
election
could be affected by the addition or removal of candidates that all (or
nearly all) of the voters hate.
I find that also silly and unacceptable. So now I am strongly of the
view that default approval should be only of "voted below no other
candidate".
Condorcet methods with no Push-over incentive should allow above-bottom
ranking.
The ballot rules should allow voters to strictly rank however many
candidates they like and also (if the method has some use for approval
information)
to approve only one candidate or all but one or any number in between.
Hopefully this should all sound obvious. But several people here have
been tolerant or (even supportive) of alternatives.
Someone who proposed a Condorcet completed by Approval method suggested
that a 6-slot grading ballot would be fine and that we would arbitrarily
call the top 3 slots approval and the bottom 3 not.
If there are more than 4 candidates that isn't enough to allow the voter
to strictly rank all the candidates and approve only one or all-but-one.
In principle a grading or multi-slot rating ballot with the top half of
the slots/grades signifying approval is fine as long as there are at
least as many
slots/grades as twice the number of candidates, minus one.
Another abomination is compulsory ranking. This is "GIGO" (garbage in,
garbage out) and I find it analogous to compelled speech.
In Australia, the major parties' "How-to-Vote" cards usually just
advise their supporters to after writing a "1" next to their party's
candidate just to
number all the rest according to the order they appear on the ballot paper.
Chris Benham
Pluses and minuses cancel out each others information
It is possible to have an exclusion count without cancel of the election
count.
Regards,
Richard Lung.
On 08/08/2023 01:33, Forest Simmons wrote:
I totally support the conventions suggested by Chris.
For grade ballots I suggest refinement of rankings by using pluses and
minuses like many schools do.
On Sun, Aug 6, 2023, 12:13 PM C.Benham cbenham@adam.com.au wrote:
I'm attracted to a couple of Condorcet methods that allow voters
to rank
the candidates and also give an approval threshold.
I think an important question is what should be the default
placement of
this approval threshold. I've read from one or two
people that all candidates "ranked in any position" should be
considered
approved, meaning that a ballot that strictly ranks
all the candidates by putting a number next to all of them should
count
as strategically useless approval for all the candidates
while a ballot that does the same thing by putting a number next
to all
but one of them counts as approving all but the unmarked
candidate.
I find that silly and unnecessarily unfair to naive or careless
voters,
so instead in the past I have gone with "voted above at least one
candidate".
Since the approvals are only used to complete Condorcet it is likely
that most of the time they'll have no effect and so many voters won't
bother giving an explicit approval cutoff.
In that circumstance with either the "voted above at least one
candidate" or the "ranked in any position" rules, the outcome of the
election
could be affected by the addition or removal of candidates that
all (or
nearly all) of the voters hate.
I find that also silly and unacceptable. So now I am strongly of the
view that default approval should be only of "voted below no other
candidate".
Condorcet methods with no Push-over incentive should allow
above-bottom
ranking.
The ballot rules should allow voters to strictly rank however many
candidates they like and also (if the method has some use for
approval
information)
to approve only one candidate or all but one or any number in between.
Hopefully this should all sound obvious. But several people here have
been tolerant or (even supportive) of alternatives.
Someone who proposed a Condorcet completed by Approval method
suggested
that a 6-slot grading ballot would be fine and that we would
arbitrarily
call the top 3 slots approval and the bottom 3 not.
If there are more than 4 candidates that isn't enough to allow the
voter
to strictly rank all the candidates and approve only one or
all-but-one.
In principle a grading or multi-slot rating ballot with the top
half of
the slots/grades signifying approval is fine as long as there are at
least as many
slots/grades as twice the number of candidates, minus one.
Another abomination is compulsory ranking. This is "GIGO" (garbage
in,
garbage out) and I find it analogous to compelled speech.
In Australia, the major parties' "How-to-Vote" cards usually just
advise their supporters to after writing a "1" next to their party's
candidate just to
number all the rest according to the order they appear on the
ballot paper.
Chris Benham
Election-Methods mailing list - seehttps://electorama.com/em for list info