Why I prefer ranked-choice voting to approval voting
(Replying farther down--I don't have a way to delete text.)
On Oct 15, 2016 11:49 AM, "robert bristow-johnson" rbj@audioimagination.com
wrote:
---------------------------- Original Message ----------------------------
Subject: Re: [EM] Why I prefer ranked-choice voting to approval voting
From: "Michael Ossipoff" email9648742@gmail.com
Date: Sat, October 15, 2016 2:32 pm
To: "Jeff O'Neill" jeff.oneill@opavote.com
election-methods@electorama.com
Jeff:
My reply was mostly to your justification of your lesser-evil voting.
As for IRV, it can be said more simply:
IRV was repealed in Burlington because it violated majority wishes. The
Democrat it eliminated was someone who was majority-preferred to
everyone
else.
A majority consisting of Democrats & Republicans wanted the Democrat
instead of the Progressive.
That's what the Republicans thought that they were ensuring when they
ranked the Democrat in 2nd place:
(You wrote) :
which essentially punished the GOP Prog-haters for voting sincerely.
despite the (false) promise that they could vote for their favorite
candidate without fear of helping elect their least favorite candidate.
(Endquote)
Exactly.
...
(I'd said) :
No voting-system can guarantee that the CWs will always win.
(You wrote) :
well, if there is a CW, any Condorcet method will guarantee that the
CW wins.
(endquote)
Yes, a voted CW (CWv) always wins in Condorcet.
But I'm talking about a sincere CW (CWs).
A sincere CW, a CWs, is a candidate who'd be CWv under sincere voting.
...a candidate who, compared to each of the other candidate, is preferred
to hir by more people than vice-versa.
(You wrote) :
all of this differentiation between sincere CW and just CW is really too
difficult because we cannot open people's cranium and peer inside to see
what their sincere vote is.
(endquote)
"CW" is used with 2 completely different meanings.
Sometimes, by CW, people refer to a CWv...& sometimes to the CWs.
It's necessary to distinguish between those 2 meanings, so I use CWs & CWv.
(You wrote) :
i think the only reasonable assumption is that, with sufficient credible
assurance (that tactical voting will not help their political interest any
more than sincere voting) that all ballots are sincerely marked.
(endquote)
With most pairwise-count methods, that assurance definitely isn't there.
Say there are 3 candidates: Favorite, Middle, & Worst. W, M, & F.
The Worst voters do the offensive strategy. Middle is the CWs.
By themselves, the letters W, M, & F represent the faction-sizes.
For example, with margins Condorcet, if M's voters rank W 2nd, truncation
by W's voters takes the win away from the CWs if:
- W > 2M
and
- F > (M+W)/2
With winning-votes Condorcet, that truncation couldn't possibly succeed. M
wins.
With margins:
If the W voters bury M, then, even if all of M's voters plump, the burial
takes away M's win and give it to W if:
- W > M
and
- W > M/2 + F
With winning-votes Condorcet, if the M voters plump, then burial of M by
W's voters reliably elects F.
(I'd said) :
But the better methods, like Approval, Score & Bucklin, gjve hir a
better
chance than IRV did in Burlington.
(You wrote) :
of course, a Condorcet method will do better than any of those. why bother
with something non-Condorcet if the target is electing the CW? i don't get
it.
(endquote)
wv RP vs Bucklin:
Both meet MMC, and so if you're majority-favored, then you have the luxury
of being assured of electing from your top-set, merely by you and your
mutual majority ranking sincerely. ...the luxury of being able to choose
among your top-set, assured of electing one of them. ( ...but equal top
ranking , when it departs from sincerity, can violate MMC's premise and
make it inapplicable.)
With both methods, the CWs's voters can protect hir win by plumping. In
Bucklin, it would additionally be necessary for other voters to not rank
past the CWs.
The difference in WV is that, if the threat of defensive plumping has
deterred burial, then a failure to plump is harmless.
The problem with wv (and all Condorcet methods) is the possibility of the
"perpetual burial fiasco":
If the CWs is in your bottom-set, then, by the definitions of
the top & bottom sets, you don't much care which bottom-set candidate wins,
as long as you try to make the winner come from your top-set.
So you aren't deterred from burying a CWs if s/he's in your bottom-set.
So, burial against the CWs could happen all the time, with the result that
the CWs never wins.
The word "cluster-f*ck" suggests itself.
Bucklin is less deluxe, but it's more solid & reliable. And therefore more
proposable.
What about Bucklin vs Approval?:
Not a lot of difference.
Of course Bucklin is stepwise Approval.
Difference:
Bucklin meets MMC, and so you get the above mentioned luxury if you're
majority-favored.
Problem: Only some voters are majority-favored.
-
Often you might not have information to be sure that you are.
-
A benefit for only some isn't so appealing. That lessens its democratic
Importance.
Additional problem:
I'd like to be able to vote equal top ranking in Bucklin, especially if I'm
not majority-favored (MF).
...and seriously gullible & cowardly overcompromisers, like the ones who
are going to vote for Hillary, badly need an equal top-ranking option.
...but the use of it can spoil Bucklin's MMC.
Oops, there goes the advantage.
So wv beats margins, wv's advantage over Bucklin is at least a little in
doubt, and so is Bucklin's advantage over Approval.
The advantage of the rank methods over Approval is questionable.
(You wrote) :
and direct comparisons cannot be made. Approval collects too little data
compared to a Ranked Ballot and Score forces the voter to concoct and yield
too much data. it can only be compared if the ballot format is the same.
so different RCV methods can be direct compared. but comparing methods
with different ballots requires assumptions to be made.
(endquote)
No, they can be compared strategically.
...and Uformation-gathering isn't the complete goal.
(Replying farther down) :
IRV gave the win elsewhere in a way that the people in Burlington
rightly
perceived as arbitrary.
not "arbitrary" but incorrect.
For one thing, Approval, Score, & Bucklin allow the CWs's preferrers a
better chance to protect hir win, to not give it away.
and Ranked Pairs of Schulze allow an even better chance.
(endquote)
Sure if the perpetual burial fiasco doesn't happen, and if it's
winning-votes instead of margjns.
Margins' protection of the CWs isn't as good as that of Bucklin.
In Bucklin, hir voters can protect hir by defensive plumping. ...but not
in margins Condorcet.
Michael Ossipoff
JFS Ross followed Laplacein favoring Borda method to Condorcet pairing.
Borda suffers from "later harm:" later preferences count against earlier
preferences. And there is no way of telling, which weighting by
mathematical series best approximates the relative importance to which
voters value their greatest to lesser choices. The harmonic series is
the happy medium.
Condorcet pairing can be weighted to give accurate relative importance
but only between two candidates at a time (binary choice).
Weighted Condorcet pairing perhaps can be about as informative as Borda
method. I once refuted a claim that theorem Arrow was substantiated by
five different single-member election methods giving five different
answers, when I substituted weighted Condorcet pairing, for unweighted
Condorcet pairing, which then agreed with method Borda.(Lol!)
It was not until the coming of method Gregory, for the transfer of
quota-surplus votes, that a more rational assessment, of relative
importance of ranked choices than Borda, could be made, in a count, by
stages, which abides by "later no harm."
Preference voting or ranked choice are essential to direct counts, which
cannot be conducted in only one stage.
Traditional Single Transferable Vote still suffers from premature
exclusion of candidates, as do all the worlds official elections,
especially in single-member elections, which don't allow the Gregory
method of rationally electing prefered candidates to further seats.
To remedy premature exclusion, my method of Binomial STV complements
rational election counts with rational exclusion counts (by extending
Meek method use of keep values, amongst other things).
From
Richard Lung.
On 15/10/2016 17:33, robert bristow-johnson wrote:
---------------------------- Original Message ----------------------------
Subject: Re: [EM] Why I prefer ranked-choice voting to approval voting
From: "C.Benham" cbenham@adam.com.au
Date: Sat, October 15, 2016 11:01 am
To: election-methods@lists.electorama.com
I want to briefly address another form of ranked voting called
Condorcet voting. Condorcet voting also uses a ranked ballot, but the
votes are counted in a different way.
While Condorcet voting is a great voting method, ...
"Condorcet voting" isn't decisive enough to qualify as a "voting
method". Condorcet is just a criterion (or a category of methods that
meet the criterion).
Min-Max Margins, Schulze (Winning Votes), Smith//Approval, "Benham"
(check for a CW among remaining candidates before each IRV-style
elimination)
are all very different methods that happen to meet the Condorcet
criterion.
forgot Tideman (ranked-pairs).
regarding semantics: i think that it is loosely appropriate, in
discussion here and in discussion among the laity to use the term
"Condorcet method" or "Condorcet voting" as any of the
Condorcet-compliant methods for the sake of discussion to
differentiate ranked-choice methods from each other. i.e. Condorcet
vs. IRV or Condorcet vs. Bucklin or Condorcet vs. Borda . the
different Condorcet-compliant methods potentially differ in outcome
only when there is a cycle, which in real elections continues to
appear to be a rare occurrence. not only that, at least when margins
are considered and there is only three candidates in the cycle, the
outcome of Schulze and Ranked Pairs and Min-Max appear to the same.
so, as rare as cycles are, even rarer in the real world are cycles
with a Smith set of more than 3 candidates. so i think, as long as we
may leave the details about what to do about cycles to a future
debate, it's an acceptable semantic to use the term "Condorcet method"
or "Condorcet voting" in the discussion of RCV.
that said, i think any of these Condorcet methods are far better
than IRV because when IRV actually fails to elect the CW (which has
happened in reality in a governmental election in 2009), even if
voters don't know or understand exactly what went wrong, they know
something did and Ranked-Choice Voting (in general) is discredited
along with IRV. so also are other voting reforms. Fairvote doesn't
get it and even after the experience of 2009, they push IRV portraying
it as the only way of reform realizing RCV, which is explicitly
dishonest. and they don't get it that somewhere, sometime again, they
will get a skeptical jurisdiction to adopt IRV and when it fails again
(which may well happen when there are three equally viable
candidates), IRV (and, by dishonest association, RCV) will be again
discredited.
still seems that Condorcet -> Ranked Pairs -> Margins is the simplest,
mostly fair (doesn't deviate from Schulze-Margins) method to simply
adopt this principle: "When more voters mark their ballots that they
prefer Candidate A over Candidate B than the number of voters who
prefer the contrary, then Candidate B is not elected." Who can argue
with that?? Whenever Candidate B is elected when more of us wanted
Candidate A or Candidate C or someone else, i mean WTF? how can
anyone who believes in majority rule and one-person-one-vote disagree
with that?
--
r b-j rbj@audioimagination.com
"Imagination is more important than knowledge."
Election-Methods mailing list - see http://electorama.com/em for list info
--
Richard Lung.
http://www.voting.ukscientists.com
Democracy Science series 3 free e-books in pdf:
https://plus.google.com/106191200795605365085
E-books in epub format:
https://www.smashwords.com/profile/view/democracyscience
The French presidential election isn't IRV though is it? It's just a two-round system - https://en.wikipedia.org/wiki/French_presidential_election,_2012 In such an election it's much clearer that you might not want to vote for your favourite in the second round than in IRV.
But Jameson, I think your centre squeeze illustration is a good one. I've seen many people (mainly on the Center for Election Science forum) saying that IRV is bad because there are situations where voting for your favourite results in a worse outcome for you, but without an explanation for the mechanism, and when and why anyone might know when not to rank their favourite top.
So as I understand it, you've got three candidates: L, C and R - C is the Condorcet winner. Someone's preference order might be L>C>R, but by submitting that ranking, C gets eliminated first and L could then lose to R in the run-off. But by ranking C first, C would be a sure winner over R in a run-off, so it's a safer vote, especially if your main priority is for R not to get elected.
From: Jameson Quinn <jameson.quinn@gmail.com>
To: Jeff O'Neill jeff.oneill@opavote.com
Cc: election-methods election-methods@electorama.com
Sent: Saturday, 15 October 2016, 14:29
Subject: Re: [EM] Why I prefer ranked-choice voting to approval voting
But there's an important, and predictable, class of election scenarios where it's strategically crucial NOT to rank your favorite first: center squeeze scenarios. If you have two candidates at ideological opposite points, with a third candidate in the middle near the median voter, it is actually quite common for the center to have the lowest first-choice support and get prematurely eliminated. This kind of thing happened in Burlington 2009; in multiple recent French elections; tragically, in Egypt 2011; and would have happened in the US 2000 if Nader had gotten over 25%. In this case, the correct strategy for one group of voters is to rank their true first choice in second place. Understanding this, and correctly seeing when it applies, is a HUGE cognitive burden for IRV voters.
When you say that "it is actually quite common for the center to have the lowest first-choice support and get prematurely eliminated," we need to clarify "quite common." If you mean 5-10% of the time, then I could believe that, but if you mean 50% of the time, then I would disagree. It would be really interesting if a statistician could collect the data and present results.
Depends on various factors, but generally I'd put it in the 10%-50% range. It's happened in 2 of the last 4 French presidential elections so 50% is not a crazy estimate.
On 10/16/2016 3:03 AM, robert bristow-johnson wrote:
still seems that Condorcet -> Ranked Pairs -> Margins is the simplest,
mostly fair (doesn't deviate from Schulze-Margins) method to simply
adopt this principle: "When more voters mark their ballots that they
prefer Candidate A over Candidate B than the number of voters who
prefer the contrary, then Candidate B is not elected."
Robert,
Take this example:
40: C
35: A>B
25: B
There is no Condorcet winner. Margins (using Ranked Pairs, Schulze,
River or MinMax) elects A.
Please take off your Pairwise-Margins glasses for a moment and put on
your Common-sense Positional ones, and then tell us
how you would feel as a C supporter about C losing to A.
Electing A here is a failure of the "Plurality criterion" and a very bad
failure of Later-no-Help (assuming that we don't elect A if the 35 A>B
had just voted A).
All Condorcet methods fail Later-no-Help, but in my view no acceptable
method fails Plurality. (IRV meets both Plurality and Later-no-Help).
Who can argue with that?? Whenever Candidate B is elected when more
of us wanted Candidate A or Candidate C or someone else, i mean WTF?
how can anyone who believes in majority rule and one-person-one-vote
disagree with that?
Some people think that it is more important that the method meets
Later-no-Help, and so isn't vulnerable to Burial strategy; or that the
method
meets the Favorite Betrayal Criterion and so has no order-reversal
Compromise incentive.
But I nonetheless like compliance with the Condorcet criterion.
so i think, as long as we may leave the details about what to do
about cycles to a future debate, it's an acceptable semantic to use
the term "Condorcet method" or "Condorcet voting" in the discussion of
RCV.
That isn't very fair on IRV (aka RCV, aka Alternative Vote) because then
the comparison is between (on the one hand) an actual method with some
problems and
(on the other hand) a motherhood criterion (with a presumption that at
some time in the future we'll discover some way of resolving cycles
without any problems).
Chris Benham
---------------------------- Original Message ----------------------------
Subject: Re: [EM] Why I prefer ranked-choice voting to approval voting
From: "C.Benham" cbenham@adam.com.au
Date: Sat, October 15, 2016 11:01 am
To: election-methods@lists.electorama.com
I want to briefly address another form of ranked voting called
Condorcet voting. Condorcet voting also uses a ranked ballot, but the
votes are counted in a different way.
While Condorcet voting is a great voting method, ...
"Condorcet voting" isn't decisive enough to qualify as a "voting
method". Condorcet is just a criterion (or a category of methods that
meet the criterion).
Min-Max Margins, Schulze (Winning Votes), Smith//Approval, "Benham"
(check for a CW among remaining candidates before each IRV-style
elimination)
are all very different methods that happen to meet the Condorcet
criterion.
forgot Tideman (ranked-pairs).
regarding semantics: i think that it is loosely appropriate, in
discussion here and in discussion among the laity to use the term
"Condorcet method" or "Condorcet voting" as any of the
Condorcet-compliant methods for the sake of discussion to
differentiate ranked-choice methods from each other. i.e. Condorcet
vs. IRV or Condorcet vs. Bucklin or Condorcet vs. Borda . the
different Condorcet-compliant methods potentially differ in outcome
only when there is a cycle, which in real elections continues to
appear to be a rare occurrence. not only that, at least when margins
are considered and there is only three candidates in the cycle, the
outcome of Schulze and Ranked Pairs and Min-Max appear to the same.
so, as rare as cycles are, even rarer in the real world are cycles
with a Smith set of more than 3 candidates. so i think, as long as we
may leave the details about what to do about cycles to a future
debate, it's an acceptable semantic to use the term "Condorcet method"
or "Condorcet voting" in the discussion of RCV.
that said, i think any of these Condorcet methods are far better
than IRV because when IRV actually fails to elect the CW (which has
happened in reality in a governmental election in 2009), even if
voters don't know or understand exactly what went wrong, they know
something did and Ranked-Choice Voting (in general) is discredited
along with IRV. so also are other voting reforms. Fairvote doesn't
get it and even after the experience of 2009, they push IRV portraying
it as the only way of reform realizing RCV, which is explicitly
dishonest. and they don't get it that somewhere, sometime again, they
will get a skeptical jurisdiction to adopt IRV and when it fails again
(which may well happen when there are three equally viable
candidates), IRV (and, by dishonest association, RCV) will be again
discredited.
still seems that Condorcet -> Ranked Pairs -> Margins is the simplest,
mostly fair (doesn't deviate from Schulze-Margins) method to simply
adopt this principle: "When more voters mark their ballots that they
prefer Candidate A over Candidate B than the number of voters who
prefer the contrary, then Candidate B is not elected." Who can argue
with that?? Whenever Candidate B is elected when more of us wanted
Candidate A or Candidate C or someone else, i mean WTF? how can
anyone who believes in majority rule and one-person-one-vote disagree
with that?
--
r b-j rbj@audioimagination.com
"Imagination is more important than knowledge."
Election-Methods mailing list - see http://electorama.com/em for list info
Sorry - in a two-round system, it's clear that it might be advantageous to vote for a candidate other than your favourite in the first round. It's not so intuitively clear that it might be to your advantage not to rank your favourite candidate top in IRV.
From: Toby Pereira <tdp201b@yahoo.co.uk>
To: Jameson Quinn jameson.quinn@gmail.com; Jeff O'Neill jeff.oneill@opavote.com
Cc: election-methods election-methods@electorama.com
Sent: Sunday, 16 October 2016, 11:29
Subject: Re: [EM] Why I prefer ranked-choice voting to approval voting
The French presidential election isn't IRV though is it? It's just a two-round system - https://en.wikipedia.org/wiki/French_presidential_election,_2012 In such an election it's much clearer that you might not want to vote for your favourite in the second round than in IRV.
But Jameson, I think your centre squeeze illustration is a good one. I've seen many people (mainly on the Center for Election Science forum) saying that IRV is bad because there are situations where voting for your favourite results in a worse outcome for you, but without an explanation for the mechanism, and when and why anyone might know when not to rank their favourite top.
So as I understand it, you've got three candidates: L, C and R - C is the Condorcet winner. Someone's preference order might be L>C>R, but by submitting that ranking, C gets eliminated first and L could then lose to R in the run-off. But by ranking C first, C would be a sure winner over R in a run-off, so it's a safer vote, especially if your main priority is for R not to get elected.
From: Jameson Quinn <jameson.quinn@gmail.com>
To: Jeff O'Neill jeff.oneill@opavote.com
Cc: election-methods election-methods@electorama.com
Sent: Saturday, 15 October 2016, 14:29
Subject: Re: [EM] Why I prefer ranked-choice voting to approval voting
But there's an important, and predictable, class of election scenarios where it's strategically crucial NOT to rank your favorite first: center squeeze scenarios. If you have two candidates at ideological opposite points, with a third candidate in the middle near the median voter, it is actually quite common for the center to have the lowest first-choice support and get prematurely eliminated. This kind of thing happened in Burlington 2009; in multiple recent French elections; tragically, in Egypt 2011; and would have happened in the US 2000 if Nader had gotten over 25%. In this case, the correct strategy for one group of voters is to rank their true first choice in second place. Understanding this, and correctly seeing when it applies, is a HUGE cognitive burden for IRV voters.
When you say that "it is actually quite common for the center to have the lowest first-choice support and get prematurely eliminated," we need to clarify "quite common." If you mean 5-10% of the time, then I could believe that, but if you mean 50% of the time, then I would disagree. It would be really interesting if a statistician could collect the data and present results.
Depends on various factors, but generally I'd put it in the 10%-50% range. It's happened in 2 of the last 4 French presidential elections so 50% is not a crazy estimate.
Robert--
- First, Burlington has special importance, as a city that showed interest
in a better voting system, and then experienced IRV's inadequacy.
So, Burlington is a place where, more than anywhere else, proposal of
genuinely good methods is particularly promising and called-for.
So, is there an already-existing
Organization or committee that is discussing voting system proposals for
Burlington?
How would one contact them?
- All here agree that IRV isn't what you want. It's been shown that all of
the better methods we're discussing would have elected the Democratic, the
apparent CWs.
Therefore it isn't necessary to rule any of them out yet.
Approval, Score & Bucklin are simple, solid, and they all have
use-precedence. Approval in medieval Venice for a long time, Score in
Sparta, Approval again in electing the U.N. secretary-general, & Bucklin
in at least 39 cities during the Progressive Era.
And I doubt that those are their only precedence instances.
Approval or Score was used for electing the Pope for quite a while.
Any Condorcet method has more complexity, & no use-precedence. Even if you
prefer Condorcet, why not first get some good method, some improvement.
Approval doesn't even need any new balloting equipment or count software.
Expense of changing to Approval: Zero.
Something completely new, more controversial & complicated could be
proposed later.
Michael Ossipoff
.
On Oct 15, 2016 11:49 AM, "robert bristow-johnson" rbj@audioimagination.com
wrote:
---------------------------- Original Message ----------------------------
Subject: Re: [EM] Why I prefer ranked-choice voting to approval voting
From: "Michael Ossipoff" email9648742@gmail.com
Date: Sat, October 15, 2016 2:32 pm
To: "Jeff O'Neill" jeff.oneill@opavote.com
election-methods@electorama.com
Jeff:
My reply was mostly to your justification of your lesser-evil voting.
As for IRV, it can be said more simply:
IRV was repealed in Burlington because it violated majority wishes. The
Democrat it eliminated was someone who was majority-preferred to everyone
else.
A majority consisting of Democrats & Republicans wanted the Democrat
instead of the Progressive.
That's what the Republicans thought that they were ensuring when they
ranked the Democrat in 2nd place:
which essentially punished the GOP Prog-haters for voting sincerely.
despite the (false) promise that they could vote for their favorite
candidate without fear of helping elect their least favorite candidate.
...
No voting-system can guarantee that the CWs will always win.
well, if there is a CW, any Condorcet method will guarantee that the
CW wins.
all of this differentiation between sincere CW and just CW is really too
difficult because we cannot open people's cranium and peer inside to see
what their sincere vote is. i think the only reasonable assumption is
that, with sufficient credible assurance (that tactical voting will not
help their political interest any more than sincere voting) that all
ballots are sincerely marked.
But the better methods, like Approval, Score & Bucklin, gjve hir a better
chance than IRV did in Burlington.
of course, a Condorcet method will do better than any of those. why
bother with something non-Condorcet if the target is electing the CW? i
don't get it.
and direct comparisons cannot be made. Approval collects too little data
compared to a Ranked Ballot and Score forces the voter to concoct and yield
too much data. it can only be compared if the ballot format is the same.
so different RCV methods can be direct compared. but comparing methods
with different ballots requires assumptions to be made.
IRV gave the win elsewhere in a way that the people in Burlington rightly
perceived as arbitrary.
not "arbitrary" but incorrect.
For one thing, Approval, Score, & Bucklin allow the CWs's preferrers a
better chance to protect hir win, to not give it away.
and Ranked Pairs of Schulze allow an even better chance.
--
r b-j rbj@audioimagination.com
"Imagination is more important than knowledge."
2016-10-16 16:29 GMT-04:00 Michael Ossipoff email9648742@gmail.com:
Robert--
- First, Burlington has special importance, as a city that showed
interest in a better voting system, and then experienced IRV's inadequacy.
So, Burlington is a place where, more than anywhere else, proposal of
genuinely good methods is particularly promising and called-for.
So, is there an already-existing
Organization or committee that is discussing voting system proposals for
Burlington?
How would one contact them?
- All here agree that IRV isn't what you want. It's been shown that all
of the better methods we're discussing would have elected the Democratic,
the apparent CWs.
Therefore it isn't necessary to rule any of them out yet.
Approval, Score & Bucklin are simple, solid, and they all have
use-precedence. Approval in medieval Venice for a long time, Score in
Sparta, Approval again in electing the U.N. secretary-general, & Bucklin
in at least 39 cities during the Progressive Era.
And I doubt that those are their only precedence instances.
You're missing Approval's use in Greece from 1860-1917, and Bucklin's use
(at Condorcet's posthumous suggestion) in pre-Napoleonic Geneva.
Robert is looking for a method with a low cognitive burden. MAS is the
answer. It's based on Bucklin, but simplified as much as possible. Simple
voting heuristics would give strategically optimal ballots in most
scenarios. It's expressive enough to distinguish a favorite from an
acceptable compromise from an unacceptable candidate: the main messages
that any voter should be able to send.
As a stepping-stone to MAS, approval is the first step. Unambiguously
better than plurality, it gets most of the benefit with literally no extra
complexity. Yes, it has some cognitive burden; but, as spoiled ballots and
scenarios like Burlington show, actually less than IRV.
Jameson
Approval or Score was used for electing the Pope for quite a while.
Any Condorcet method has more complexity, & no use-precedence. Even if you
prefer Condorcet, why not first get some good method, some improvement.
Approval doesn't even need any new balloting equipment or count software.
Expense of changing to Approval: Zero.
Something completely new, more controversial & complicated could be
proposed later.
Michael Ossipoff
.
On Oct 15, 2016 11:49 AM, "robert bristow-johnson" <
rbj@audioimagination.com> wrote:
---------------------------- Original Message ----------------------------
Subject: Re: [EM] Why I prefer ranked-choice voting to approval voting
From: "Michael Ossipoff" email9648742@gmail.com
Date: Sat, October 15, 2016 2:32 pm
To: "Jeff O'Neill" jeff.oneill@opavote.com
election-methods@electorama.com
Jeff:
My reply was mostly to your justification of your lesser-evil voting.
As for IRV, it can be said more simply:
IRV was repealed in Burlington because it violated majority wishes. The
Democrat it eliminated was someone who was majority-preferred to
everyone
else.
A majority consisting of Democrats & Republicans wanted the Democrat
instead of the Progressive.
That's what the Republicans thought that they were ensuring when they
ranked the Democrat in 2nd place:
which essentially punished the GOP Prog-haters for voting sincerely.
despite the (false) promise that they could vote for their favorite
candidate without fear of helping elect their least favorite candidate.
...
No voting-system can guarantee that the CWs will always win.
well, if there is a CW, any Condorcet method will guarantee that the
CW wins.
all of this differentiation between sincere CW and just CW is really too
difficult because we cannot open people's cranium and peer inside to see
what their sincere vote is. i think the only reasonable assumption is
that, with sufficient credible assurance (that tactical voting will not
help their political interest any more than sincere voting) that all
ballots are sincerely marked.
But the better methods, like Approval, Score & Bucklin, gjve hir a
better
chance than IRV did in Burlington.
of course, a Condorcet method will do better than any of those. why
bother with something non-Condorcet if the target is electing the CW? i
don't get it.
and direct comparisons cannot be made. Approval collects too little data
compared to a Ranked Ballot and Score forces the voter to concoct and yield
too much data. it can only be compared if the ballot format is the same.
so different RCV methods can be direct compared. but comparing methods
with different ballots requires assumptions to be made.
IRV gave the win elsewhere in a way that the people in Burlington
rightly
perceived as arbitrary.
not "arbitrary" but incorrect.
For one thing, Approval, Score, & Bucklin allow the CWs's preferrers a
better chance to protect hir win, to not give it away.
and Ranked Pairs of Schulze allow an even better chance.
--
r b-j rbj@audioimagination.com
"Imagination is more important than knowledge."
Election-Methods mailing list - see http://electorama.com/em for list info
(Replying farther down)
On Oct 16, 2016 6:34 PM, "Jameson Quinn" jameson.quinn@gmail.com wrote:
2016-10-16 16:29 GMT-04:00 Michael Ossipoff email9648742@gmail.com:
Robert--
- First, Burlington has special importance, as a city that showed
interest in a better voting system, and then experienced IRV's inadequacy.
So, Burlington is a place where, more than anywhere else, proposal of
genuinely good methods is particularly promising and called-for.
So, is there an already-existing
Organization or committee that is discussing voting system proposals for
Burlington?
How would one contact them?
- All here agree that IRV isn't what you want. It's been shown that all
of the better methods we're discussing would have elected the Democratic,
the apparent CWs.
Therefore it isn't necessary to rule any of them out yet.
Approval, Score & Bucklin are simple, solid, and they all have
use-precedence. Approval in medieval Venice for a long time, Score in
Sparta, Approval again in electing the U.N. secretary-general, & Bucklin
in at least 39 cities during the Progressive Era.
And I doubt that those are their only precedence instances.
(You wrote) :
You're missing Approval's use in Greece from 1860-1917, and Bucklin's use
(at Condorcet's posthumous suggestion) in pre-Napoleonic Geneva.
Thanks--I didn't know about those. Most likely, the first reforms would
have to be methods with use-precedent.
New methods probably would have to wait until people are used to not using
Plurality.
Robert is looking for a method with a low cognitive burden.
(endquote)
But, as a participant at CES, you know that improvements on Approval &
Score are usually slight, difficult, or illusory.
Approval needn't be complicated to use or explain.
Approve your top-set.
In our official elections, for progressives, there certainly is a top-set.
The difference between progressive vs Republocrat candidates is obvious,
stark, drastic & dramatic.
Honest, humane & in tune with the changes that people want, vs corrupt,
obviously dishonest & bought.
That lineup makes Approval pretty easy.
A 3rd category, conservative populists, seem to oppose corruption, but
offfer nothing different from the status quo.
If there weren't a top-set, people could just use an approval cutoff at
the merit-midrange of the candidates, not counting the unwinnable
undesirables.
The merit-mean is too dependent on the numbers of candidates in the various
categories, & its false assumption that they have equal win-probabillty.
(You wrote) :
MAS is the answer. It's based on Bucklin, but simplified as much as
possible. Simple voting heuristics would give strategically optimal ballots
in most scenarios.
(endquote)
What incentive is there to not just approval-vote, given a top-set?
Bucklin's MMC gives some such incentive, if you're pretty sure you're
majority-favored.
But 3-slot methods fail MMC.
Approval-voting maximizes Pt, Probability of electing from top-set.
3-Slot ICT is offered as an Approval version.
Michael Ossipoff
It's expressive enough to distinguish a favorite from an acceptable
compromise from an unacceptable candidate: the main messages that any voter
should be able to send.
As a stepping-stone to MAS, approval is the first step. Unambiguously
better than plurality, it gets most of the benefit with literally no extra
complexity. Yes, it has some cognitive burden; but, as spoiled ballots and
scenarios like Burlington show, actually less than IRV.
Jameson
Approval or Score was used for electing the Pope for quite a while.
Any Condorcet method has more complexity, & no use-precedence. Even if
you prefer Condorcet, why not first get some good method, some
improvement.
Approval doesn't even need any new balloting equipment or count software.
Expense of changing to Approval: Zero.
Something completely new, more controversial & complicated could be
proposed later.
Michael Ossipoff
.
On Oct 15, 2016 11:49 AM, "robert bristow-johnson" <
rbj@audioimagination.com> wrote:
---------------------------- Original Message
Subject: Re: [EM] Why I prefer ranked-choice voting to approval voting
From: "Michael Ossipoff" email9648742@gmail.com
Date: Sat, October 15, 2016 2:32 pm
To: "Jeff O'Neill" jeff.oneill@opavote.com
election-methods@electorama.com
Jeff:
My reply was mostly to your justification of your lesser-evil voting.
As for IRV, it can be said more simply:
IRV was repealed in Burlington because it violated majority wishes.
The
Democrat it eliminated was someone who was majority-preferred to
everyone
else.
A majority consisting of Democrats & Republicans wanted the Democrat
instead of the Progressive.
That's what the Republicans thought that they were ensuring when they
ranked the Democrat in 2nd place:
which essentially punished the GOP Prog-haters for voting sincerely.
despite the (false) promise that they could vote for their favorite
candidate without fear of helping elect their least favorite candidate.
...
No voting-system can guarantee that the CWs will always win.
well, if there is a CW, any Condorcet method will guarantee that
the CW wins.
all of this differentiation between sincere CW and just CW is really
too difficult because we cannot open people's cranium and peer inside to
see what their sincere vote is. i think the only reasonable assumption is
that, with sufficient credible assurance (that tactical voting will not
help their political interest any more than sincere voting) that all
ballots are sincerely marked.
But the better methods, like Approval, Score & Bucklin, gjve hir a
better
chance than IRV did in Burlington.
of course, a Condorcet method will do better than any of those. why
bother with something non-Condorcet if the target is electing the CW? i
don't get it.
and direct comparisons cannot be made. Approval collects too little
data compared to a Ranked Ballot and Score forces the voter to concoct and
yield too much data. it can only be compared if the ballot format is the
same. so different RCV methods can be direct compared. but comparing
methods with different ballots requires assumptions to be made.
IRV gave the win elsewhere in a way that the people in Burlington
rightly
perceived as arbitrary.
not "arbitrary" but incorrect.
For one thing, Approval, Score, & Bucklin allow the CWs's preferrers a
better chance to protect hir win, to not give it away.
and Ranked Pairs of Schulze allow an even better chance.
--
r b-j rbj@audioimagination.com
"Imagination is more important than knowledge."
Election-Methods mailing list - see http://electorama.com/em for list
info