Fwd: (3) MJ -- The easiest method to 'tolerate'
Sorry; I sent a premature version of this response to Chris alone. This is
the same as that, except for this paragraph. The one thing I wanted to
reemphasize here is that I'm not trying to make this into a battle between
good systems. IBIFA, Smith//Approval, Schulze, and Approval do not belong
to the class of single-winner methods I call "best" (though IBIFA almost
does); but I still consider all of those methods extremely good, and would
enthusiastically join a campaign for their use in some jurisdiction.
2016-09-06 8:56 GMT-04:00 C.Benham cbenham@adam.com.au:
On 9/6/2016 6:00 AM, Jameson Quinn wrote:
It's really hard to respond point-by-point as a third party in a
discussion like this. However, I'd like to say in general that I believe
that Majority Judgment, and more-generally, the class of "median" or
"graded Bucklin" systems which includes MJ, MCA, GMJ, ERB, DA, etc., are
the best non-delegated single-winner systems for a potentially-strategic
electorate, in terms of outcome.
C: How are they better than IBIFA? Or, say, Smith//Approval?
First off: IBIFA is similar to MCA, and thus almost a member of this class.
So I definitely consider it to be a very good system.
Second, I think that DA may be better than IBIFA, not in outcomes, but in
ease of explanation. This may, in practice, impact outcomes; something
that's easier to understand may make people marginally more likely to vote.
In terms of outcomes: I think that IBIFA does the wrong thing when it fails
to find a "top-rating" winner. Falling back to "most above-bottom" rankings
brings the chicken dilemma into the choice of whether to rank above bottom.
That's why, in DA, in the case where all candidates are majority
disqualified, the winner is the one with the most top ratings, not the most
non-disqualified (above-bottom) ratings.
As to Smith//Approval: also a very good method. But it does include
incentives for non-semi-honest strategy, which rated Bucklin systems don't.
To me, the toughest realistic election scenario is the chicken dilemma.
For instance, consider the following 900-voter scenario
300: A>B>>C
200: B>>A>C
400: C>>B>A
(where ">>" indicates universal agreement, and ">" at bottom indicates
90% agreement and 10% reversal)
C: How is this a "chicken dilemma" scenario? In this context, I don't
understand what the words "agreement" and "reversal" mean.
What I'm saying there is that 200:B>>A>C is actually shorthand for
180:B>>A>C, 20:B>>C>A.
DA: "Double Approval" or "Disqualify/Approve" voting. Voters can rate each
candidate preferred, neutral, or disqualified. (Both preferred and
disqualified is also legal and counted, though it's strategically
nonsensical.) Winner is the most-preferred among those not majority
disqualified. If all candidates are majority-disqualified, winner is simply
most-preferred. Any candidate who is majority-disqualified is prohibited
from appearing on the ballot for the same office in the following election.
Lately, I favor DA,
C: Why is rating a candidate both approved and "disqualified" "legal and
counted"?
Two reasons:
- Even though it doesn't make any sense strategically, it avoids throwing
away the ballot as spoiled. In general, for most candidates, only one of
the two kinds of votes is going to matter; and insofar as the voter meant
either of the marks, it seems more likely that they mean the one that
matters. - This makes the system meet the Frohnmayer balance criterion.
What is the default rating?
There is a complex rule for this.
There are three check mark spaces per ballot, for approve, explicit
neutral, and disqualify. Explicit neutral, when it's in combination with
another mark or marks, has no meaning. For blank lines, use the following
rules, in order:
- If a voter has used any explicit neutral rankings, any blank rankings
they include are considered "semi-explicit" disqualify ratings. (This is a
case where I think intent is pretty clear.) - If the top two candidates by approvals both have a majority of explicit
or semi-explicit disqualify ratings, then neutral rankings are all counted
as implicit disqualify ratings. (This helps prevent unintentional "dark
horse" wins. It's also a statistical view of intent: if voters tend to be
negative with their explicit votes, then implicit votes should be
considered negative too.) - If neither of the above hold, then blank lines are considered as neutral
grades. (As above: if voters aren't too negative with explicit votes, then
I think their intent is most likely not negative with implicit ones.)
The rule regarding "the following election" I consider arbitrary
and very undemocratic.
This is the case where a candidate wins despite the fact that the majority
of the electorate put them at bottom. If they are to win a second term,
they must do so as a write-in.
"NOTA" is a popular proposal. I consider it wrong-headed as part of
plurality, but I understand the sentiment, and so I think the above is a
reasonable compromise.
In all these Bucklin/MJ/MCA methods the strategic incentive for the voters
to just cast approval-type votes (i.e. only use the top and bottom
slots/grades) is in my view
far too strong. The least bad of them is 3-slot MTA (where if more than
one candidate exceeds the majority threshold in the second round the winner
is the one of those with the
most Top Ratings), which is simple and slightly reduces the truncation
incentive. It didn't make your list.
That was not intentional. Note that AD is pretty similar, and arguably
farther in the direction you're favoring.
All them absurdly fail Irrelevant Ballots independence and (unless you
have a fetish for strict compliance with Later-no-Help) are completely
dominated by IBIFA.
The IBIFA winner will always pairwise beat any different winner that any
of these methods come up with.
Let's take a case where IBIFA and DA disagree.
40: D>...>A,B
35: A>B>D
25: B>...>A,D
IBIFA elects A; DA elects B.
Obviously, one could tweak these numbers to try to make one or the other
system look better or worse. But I think that this scenario is more or less
a neutral example of its class.
The main question is: are the B voters being honest? If they truly feel
that A is unacceptable, then I think that B should win. If they are using a
chicken strategy and hiding their true preference for A over D, then A
should win.
My feeling is that this kind of situation, where the strategic symmetry
between the A and B factions is badly broken, is more likely to occur
through some honest difference rather than through strategy. After all, if
their sizes are similar, their strategic incentives are relatively
symmetric; and if the A group is highly dominant, then it's likely that A
voters would disqualify B, given that even a minority of A voters would be
enough to do so.
So, while I think that this is a legitimately hard kind of scenario and
there is some argument for the IBIFA answer, I still feel that DA is not
only undominated by IBIFA, it's actually better here.
Chris Benham
On 9/7/2016 1:21 AM, Jameson Quinn wrote:
Let's take a case where IBIFA and DA disagree.
40: D>...>A,B
35: A>B>D
25: B>...>A,D
IBIFA elects A; DA elects B.
C:I assume comma means = and ... means one skipped grade. I suppose,
since "DA" uses 3-slot ballots. that the ballots explicitly mention the
unapproved (bottom-rated)
candidates. In that case I would translate this as:
40: D
35: A>B
25: B
But then IBIFA elects B. No candidate X's Top-Rating score is higher
than any candidate Y's approval (i.e. above-bottom rating) score on
ballots that don't top-rate X, so
IBIFA elects the most approved candidate, B.
Obviously, one could tweak these numbers to try to make one or the
other system look better or worse.
C: Well then I am very curious to see any example where IBIFA "looks
worse" than DA.
What I'm saying there is that 200:B>>A>C is actually shorthand for
180:B>>A>C, 20:B>>C>A.
C: That is very useful and clarifying "shorthand".
What is the default rating?
There is a complex rule for this.
C:I greatly prefer a very simple rule: the default rating is Bottom
(i.e. the lowest possible).
The rule regarding "the following election" I consider arbitrary and
very undemocratic.
This is the case where a candidate wins despite the fact that the
majority of the electorate put them at bottom. If they are to win a
second term, they must do so as a write-in.
C:That is just ridiculous and unfair and arbitrary. The rule as I read
it didn't just refer to the winner (not that I can accept either
version). A lot can happen in the years between
elections. Why should a candidate be severely disadvantaged because of
how s/he went in the last election?
Note that AD is pretty similar [to MTA], and arguably farther in the
direction you're favoring.
I assume "AD" is a typo and means DA. Compared with MTA, DA is
needlessly complicated and too (IMO) too focused on the bottoms of the
ballots.
Chris Benham
Sorry; I sent a premature version of this response to Chris alone.
This is the same as that, except for this paragraph. The one thing I
wanted to reemphasize here is that I'm not trying to make this into a
battle between good systems. IBIFA, Smith//Approval, Schulze, and
Approval do not belong to the class of single-winner methods I call
"best" (though IBIFA almost does); but I still consider all of those
methods extremely good, and would enthusiastically join a campaign for
their use in some jurisdiction.
2016-09-06 8:56 GMT-04:00 C.Benham <cbenham@adam.com.au
mailto:cbenham@adam.com.au>:
On 9/6/2016 6:00 AM, Jameson Quinn wrote:
It's really hard to respond point-by-point as a third party in
a discussion like this. However, I'd like to say in general
that I believe that Majority Judgment, and more-generally, the
class of "median" or "graded Bucklin" systems which includes
MJ, MCA, GMJ, ERB, DA, etc., are the best non-delegated
single-winner systems for a potentially-strategic electorate,
in terms of outcome.
C: How are they better than IBIFA? Or, say, Smith//Approval?
http://wiki.electorama.com/wiki/IBIFA
<http://wiki.electorama.com/wiki/IBIFA>
First off: IBIFA is similar to MCA, and thus almost a member of this
class. So I definitely consider it to be a very good system.
Second, I think that DA may be better than IBIFA, not in outcomes, but
in ease of explanation. This may, in practice, impact outcomes;
something that's easier to understand may make people marginally more
likely to vote.
In terms of outcomes: I think that IBIFA does the wrong thing when it
fails to find a "top-rating" winner. Falling back to "most
above-bottom" rankings brings the chicken dilemma into the choice of
whether to rank above bottom. That's why, in DA, in the case where all
candidates are majority disqualified, the winner is the one with the
most top ratings, not the most non-disqualified (above-bottom) ratings.
As to Smith//Approval: also a very good method. But it does include
incentives for non-semi-honest strategy, which rated Bucklin systems
don't.
To me, the toughest realistic election scenario is the chicken
dilemma. For instance, consider the following 900-voter scenario
300: A>B>>C
200: B>>A>C
400: C>>B>A
(where ">>" indicates universal agreement, and ">" at bottom
indicates 90% agreement and 10% reversal)
C: How is this a "chicken dilemma" scenario? In this context, I
don't understand what the words "agreement" and "reversal" mean.
What I'm saying there is that 200:B>>A>C is actually shorthand for
180:B>>A>C, 20:B>>C>A.
DA: "Double Approval" or "Disqualify/Approve" voting. Voters
can rate each candidate preferred, neutral, or disqualified.
(Both preferred and disqualified is also legal and counted,
though it's strategically nonsensical.) Winner is the
most-preferred among those not majority disqualified. If all
candidates are majority-disqualified, winner is simply
most-preferred. Any candidate who is majority-disqualified is
prohibited from appearing on the ballot for the same office in
the following election.
Lately, I favor DA,
C: Why is rating a candidate both approved and "disqualified"
"legal and counted"?
Two reasons:
-
Even though it doesn't make any sense strategically, it avoids
throwing away the ballot as spoiled. In general, for most candidates,
only one of the two kinds of votes is going to matter; and insofar as
the voter meant either of the marks, it seems more likely that they
mean the one that matters. -
This makes the system meet the Frohnmayer balance criterion.
What is the default rating?
There is a complex rule for this.
There are three check mark spaces per ballot, for approve, explicit
neutral, and disqualify. Explicit neutral, when it's in combination
with another mark or marks, has no meaning. For blank lines, use the
following rules, in order:
-
If a voter has used any explicit neutral rankings, any blank
rankings they include are considered "semi-explicit" disqualify
ratings. (This is a case where I think intent is pretty clear.) -
If the top two candidates by approvals both have a majority of
explicit or semi-explicit disqualify ratings, then neutral rankings
are all counted as implicit disqualify ratings. (This helps prevent
unintentional "dark horse" wins. It's also a statistical view of
intent: if voters tend to be negative with their explicit votes, then
implicit votes should be considered negative too.) -
If neither of the above hold, then blank lines are considered as
neutral grades. (As above: if voters aren't too negative with explicit
votes, then I think their intent is most likely not negative with
implicit ones.)The rule regarding "the following election" I consider arbitrary
and very undemocratic.
This is the case where a candidate wins despite the fact that the
majority of the electorate put them at bottom. If they are to win a
second term, they must do so as a write-in.
"NOTA" is a popular proposal. I consider it wrong-headed as part of
plurality, but I understand the sentiment, and so I think the above is
a reasonable compromise.
In all these Bucklin/MJ/MCA methods the strategic incentive for
the voters to just cast approval-type votes (i.e. only use the top
and bottom slots/grades) is in my view
far too strong. The least bad of them is 3-slot MTA (where if
more than one candidate exceeds the majority threshold in the
second round the winner is the one of those with the
most Top Ratings), which is simple and slightly reduces the
truncation incentive. It didn't make your list.
That was not intentional. Note that AD is pretty similar, and arguably
farther in the direction you're favoring.
All them absurdly fail Irrelevant Ballots independence and (unless
you have a fetish for strict compliance with Later-no-Help) are
completely dominated by IBIFA.
The IBIFA winner will always pairwise beat any different winner
that any of these methods come up with.
Let's take a case where IBIFA and DA disagree.
40: D>...>A,B
35: A>B>D
25: B>...>A,D
IBIFA elects A; DA elects B.
Obviously, one could tweak these numbers to try to make one or the
other system look better or worse. But I think that this scenario is
more or less a neutral example of its class.
The main question is: are the B voters being honest? If they truly
feel that A is unacceptable, then I think that B should win. If they
are using a chicken strategy and hiding their true preference for A
over D, then A should win.
My feeling is that this kind of situation, where the strategic
symmetry between the A and B factions is badly broken, is more likely
to occur through some honest difference rather than through strategy.
After all, if their sizes are similar, their strategic incentives are
relatively symmetric; and if the A group is highly dominant, then it's
likely that A voters would disqualify B, given that even a minority of
A voters would be enough to do so.
So, while I think that this is a legitimately hard kind of scenario
and there is some argument for the IBIFA answer, I still feel that DA
is not only undominated by IBIFA, it's actually better here.
Chris Benham
Election-Methods mailing list - see http://electorama.com/em for list info
No virus found in this message.
Checked by AVG - www.avg.com http://www.avg.com
Version: 2016.0.7752 / Virus Database: 4649/12956 - Release Date: 09/06/16