[EM] Legacy IRV limitations
Richard, the VoteFair guy
electionmethods at votefair.org
Tue Dec 19 18:23:55 PST 2023
On 12/18/2023 11:17 AM, Greg Dennis wrote:
> On Mon, Dec 18, 2023 at 1:36 PM Richard, the VoteFair guy ... wrote:
[... as an example of a misrepresentation from FairVote ...]
>> Ranked choice voting software must be backwards compatible with
>> prior elections in Australia.
> I do not know the source of this claim and have never heard or read
> anyone from FairVote state this. If you have a source, I'd be glad to
> read it.
Here's the wording that was adopted for Portand (Oregon) using ranked
choice voting:
https://efiles.portlandoregon.gov/Record/16114624/File/Document
Here's the quote in this document that specifies how overvotes are counted:
"Overvote. If a ballot contains an overvote, the voter’s vote is
transferred to the next highest-ranked active candidate on the voter’s
ballot, if any."
This choice makes sense in Australia where two handwritten numbers being
the same is a reason to suspect the voter lost track of their numbering.
With that perspective it's reasonable to skip the duplicate numbers.
However, here in Oregon everyone votes at home using paper ballots and a
pen. So the adopted interpretation (above) is exactly the opposite from
what the voter intends, which is to rank the same-ranked candidate
above, not below, the "next highest-ranked active candidate."
The recommendation to choose this option came from the Ranked Choice
Voting Resource Center, not from FairVote. However, I've seen wordings
from FairVote, and their wordings are very similar.
The other better option that certified software can handle is to dismiss
the remaining marks on the ballot.
Even better, I'm promoting awareness of this detail so that in the
future a city or state will have the option to correctly count the
so-called "overvotes."
> I have not seen any FairVote claim that overvotes cannot in theory be
> counted. ...
Yes, FairVote does not claim that overvotes cannot "in theory" be counted.
However, two years earlier, the FairVote organization tried to push
through an Oregon bill that handled overvotes in a similar way.
Fortunately the Oregon attorney who co-wrote the recently adopted state
referendum -- and who has received payments from FairVote -- paid
attention to feedback from me and other election-method reformers. He
recognized that counting overvotes is a "counting detail" that did not
need to be included. So overvotes are not mentioned in the bill that
passed in the Oregon state legislature. This allows for a software
upgrade to correctly count so-called "overvotes" without changing the
legal wording.
FWIW I pointed out the overvote counting flaw while the Portland
election officials were taking feedback, but my feedback about this
point was ignored. I also gave verbal testimony to the Portland city
council but they couldn't change the wording. (Later one of the council
members did try to switch city council elections from STV to IRV but
backlash from lots of us ended that attempt.) The mayoral election will
use IRV. I'm not as concerned about the overvote-counting mistake for
the STV city-council elections because a close STV election will mostly
only affect the order in which the 3 seats (per district) are filled.
As for the other responses from Greg Dennis, I'll repeat my main point,
which is that the fans of STAR voting are exploiting misrepresentations
in what FairVote and RCVRC (Ranked Choice Voting Resource Center) promote.
Keep in mind the fans of STAR voting are trying to block the statewide
RCV reform (already on the November 2024 ballot) because they are
pushing a statewide petition to adopt STAR voting statewide. And
because of financial backing they are collecting lots of signatures,
although not likely enough to get it on the ballot. Yet their
misrepresentation of RCV is undermining the chances of voters approving
the state-legislature-approved referendum.
Multiple STAR fans expressed strong opposition when the Oregon state
legislature considered the RCV bill that they passed. That opposition
was both in writing and in verbal testimony. That too undermines
support for the referendum.
Again, this "counting detail" would be much less important if FairVote
and RCVRC would communicate with Oregon election officials about the
fact that these weaknesses (overvote counting and Burlington/Alaska
mistakes) are easy to remedy without abandoning ranked choice ballots --
particularly without needing to switch to STAR ballots.
>> The winner in Burlington VT was the correct winner.
> There being no universally agreed upon meaning of "correct winner," that
> statement cannot be a "misrepresentation," only an opinion. If there
> were such a meaning, this list, whose purpose it is to understand and
> balance the various criteria, would have no need to exist.
The fans of STAR voting promote the Burlington VT election result as a
failure of the "center squeeze effect."
Their success in promoting STAR voting reveals that lots of voters do
regard the Burlington election as giving the wrong result.
Of course this issue becomes complex to explain because Condorcet fans
regard the Burlington election as a Condorcet failure. And the failure
also fits within the bigger category of an IIA (independence of
irrelevant alternatives) failure.
Personally I portray the Burlington/Alaska failures as the failure to
eliminate a pairwise losing candidate (roughly "a" Condorcet loser,
although not "the" Condorcet loser of the contest) during the top-three
counting round. I find that most people more easily understand this
reasoning, especially when using the metaphor of a soccer team that
loses every match against every other remaining team.
Personally I think it's worth educating election-method reformers that
the mistakes in Burlington and Alaska are easy to remedy. That
awareness can yield improved legal wording that can be refined later,
just by adding two sentences about eliminating pairwise losing
candidates when they occur.
In contrast, the legal wording FairVote originally tried to push through
the Oregon legislature would have made this refinement very difficult.
Yes, these details are too subtle for voters to think about.
Yet if we don't get the legal wording right, it will be easy for
anti-RCV folks to exploit the same misrepresentations that the fans of
STAR voting are exploiting.
Richard Fobes
The VoteFair guy
On 12/18/2023 11:17 AM, Greg Dennis wrote:
> Richard, I don't work for or represent FairVote, but I'd like to correct
> some claims and seek citations for others:
>
> Overvotes cannot be counted, and it's not worth attempting to count them
>
>
> I have not seen any FairVote claim that overvotes cannot in theory be
> counted. To the extent FairVote made any related claim, it is likely a
> truthful claim about the practical reality: no certified tabulation
> software exists that is capable of counting them. For example, the only
> options in Dominion's Democracy Suite are to either truncate the ballot
> as soon as an overvote is encountered or to skip over the overvote to
> the next ranking.
>
> The winner in Burlington VT was the correct winner.
>
>
> There being no universally agreed upon meaning of "correct winner," that
> statement cannot be a "misrepresentation," only an opinion. If there
> were such a meaning, this list, whose purpose it is to understand and
> balance the various criteria, would have no need to exist.
>
> Electing the Condorcet winner is not important, and Condorcet
> methods are not worth considering
>
>
> That's just not true. As you know from the Zoom meeting with Deb Otis,
> FairVote does consider the Condorcet criterion important, which is why
> they collect and publish data
> <https://fairvote.org/resources/data-on-rcv/#condorcet-winners> on how
> often IRV elects the Condorcet winner. Furthermore, they
> have "considered" Condorcet methods, as evidenced by their various
> pieces on Condorcet and other alternative methods (example
> <https://fairvote.org/archives/alternatives-to-rcv/>). It would be fair
> to say that they do not place as high a level of importance on the
> criterion as you and others do, but I don't think it's fair to say that
> they consider the criterion unimportant or not even worth consideration.
>
> Ranked choice voting software must be backwards compatible with
> prior elections in Australia.
>
>
> I do not know the source of this claim and have never heard or read
> anyone from FairVote state this. If you have a source, I'd be glad to
> read it.
>
> Thanks,
> Greg
>
>
> On Mon, Dec 18, 2023 at 1:36 PM Richard, the VoteFair guy
> <electionmethods at votefair.org <mailto:electionmethods at votefair.org>> wrote:
>
> On 12/17/2023 6:30 PM, Michael Ossipoff wrote:
> > "Ballots that do not help voters’ top choices win count for
> their next
> > choice."
> >
> > That's the 2nd sentence about RCV at FairVote's website.
>
> I agree with Michael Garman that this is not a significant
> misrepresentation from FairVote.
>
> In particular, it's not the significant(!) kind of misrepresentation I
> had in mind when I started this thread.
>
> Examples of what I believe are significant misrepresentations from the
> FairVote organization are:
>
> * Overvotes cannot be counted, and it's not worth attempting to
> count them
>
> * The candidate with the fewest transferred votes is always the least
> popular candidate
>
> * The winner in Burlington VT was the correct winner
>
> * Electing the Condorcet winner is not important, and Condorcet methods
> are not worth considering
>
> * Ranked choice voting counting rules should use the wording
> supplied by
> the FairVote organization or the Ranked Choice Voting Resource Center
> (even though that wording is awful, and seemingly worded to make it
> difficult to change to anything better)
>
> * Ranked choice voting software must be backwards compatible with prior
> elections in Australia (even though their ballots require a voter to
> write numbers in a box, and their counting process is based on
> shortcuts
> that arose to minimize how many times each paper ballot had to be
> looked at)
>
> (There might be others, but the one Michael Ossipoff presents is not
> one
> of them.)
>
> For comparison, the biggest misrepresentations from the fans of STAR
> voting are (off the top of my head):
>
> * Ranked choice voting is vulnerable to vote splitting (this is a
> big lie!)
>
> * Ranked choice ballots cannot be counted in ways that are fair (which
> is implied by presenting IRV as if it's the only way to count ranked
> choice ballots)
>
> * Claiming they "officially" support a method for counting ranked
> choice
> ballots ("Ranked Robin") yet never mentioning that method as a possible
> alternative to IRV (when they villify ranked choice ballots)
>
> * STAR voting is resistant to tactical voting and strategic nomination
> (which ignores the case in which a large minority offers two similar
> candidates and tells their voters to top-rank both of those candidates
> and bottom-rank all other candidates)
>
> * Summability is still important (even though we now have very fast
> fiberoptic speeds instead of slow modem speeds)
>
> * Monotonicity failures should never occur, and are worse than other
> failures (including Condorcet failures)
>
> * Pairwise counting is not important (even though STAR voting's second
> step is to do pairwise counting between the top two)
>
> * STAR voting is a better kind of ranked choice voting
>
> Here are some misrepresentations from the Election Science Foundation:
>
> * Score voting would be a reasonable choice in elections
>
> * Approval voting would be suitable for general elections
>
> * The simplicity of Approval voting justifies not pursuing any method
> that uses ranked choice ballots
>
> * Ranked choice ballots cannot be counted in ways that are fair (by
> presenting IRV as if it were the only option)
>
> (I might be forgetting one or two more.)
>
> Finally here's a misrepresentation that applies to all three
> organizations:
>
> * Our organization taught the voters in such-and-such city (or state)
> about the evils of vote splitting so our organization's preferred
> method
> should be adopted in this city (or state) and no other method should be
> considered by the voters or elected officials
>
> I'm pleased to learn (from the discussion between the two Michaels)
> that
> Rob Richie is no longer the leader of the FairVote organization. He
> was
> the source of lots of misrepresentation.
>
> About a year or so ago I participated in a three-way Zoom meeting with
> Deb Otis who works at the FairVote organization. I was pleased that
> she
> did not make any misrepresentations about IRV or STV.
>
> However, I disagree with her claim that correctly counting overvotes is
> not worth the extra effort needed. Besides being an issue about
> spoiled
> ballots, counting "overvotes" also allows a voter to rank their
> most-disliked candidate lower than all other candidates when there are
> just 6 choice rankings and more than 6 candidates (which affects ballot
> real estate, which election officials regard as very important).
>
> Also, I disagree with her claim that the failure in Burlington was
> acceptable because it was just one election out of about 400 elections.
> (This was before Alaska's special election that also had a Condorcet
> failure.)
>
> I'll clarify that I've softened my opinion about Condorcet failures.
> I'll accept a few such failures if the election method gains
> significant
> other kinds of advantages.
>
> However, I strongly believe that when IRV eliminations reach the top
> three candidates, the presence of a pairwise losing candidate (who
> would
> lose both one-on-one contests against the other two candidates) should
> not cause the majority-supported candidate to lose.
>
> Typically I shorten this point by saying that pairwise losing
> candidates
> should be eliminated when they occur. Yet because Michael Ossipoff
> likes to jump on wording-based issues, I'll clarify that I really don't
> care about the exact order of elimination of the candidates who get
> eliminated prior to the top-three round.
>
> Getting back to my main point, my hope is that the people leading the
> three main election-method organizations (FairVote, STAR ..., ESF)
> recognize that making misrepresentations undermines our goal of ending
> "first past the post" (academically known as "plurality voting").
>
> And I'll repeat my specific request to please(!) avoid
> misrepresentations that might block the state-legislature-approved(!)
> referendum for ranked choice voting that will be on Oregon's 2024
> November election ballot.
>
> Richard Fobes
> The VoteFair guy
>
>
> On 12/17/2023 6:30 PM, Michael Ossipoff wrote:
> > "Ballots that do not help voters’ top choices win count for their
> next
> > choice."
> >
> > That's the 2nd sentence about RCV at FairVote's website.
> >
> > To reach that website, google "FairVote, Ranked-Choice Voting".
> >
> > On Sun, Dec 17, 2023 at 8:12 PM Michael Garman
> > <michael.garman at rankthevote.us
> <mailto:michael.garman at rankthevote.us>
> <mailto:michael.garman at rankthevote.us
> <mailto:michael.garman at rankthevote.us>>>
> > wrote:
> >
> > I’m quite familiar with it…which is why I am skeptical of
> your claim…
> > On Sun, Dec 17, 2023 at 8:11 PM Michael Ossipoff
> > <email9648742 at gmail.com <mailto:email9648742 at gmail.com>
> <mailto:email9648742 at gmail.com <mailto:email9648742 at gmail.com>>> wrote:
> >
> >
> >
> > On Sun, Dec 17, 2023 at 17:03 Michael Garman
> > <michael.garman at rankthevote.us
> <mailto:michael.garman at rankthevote.us>
> > <mailto:michael.garman at rankthevote.us
> <mailto:michael.garman at rankthevote.us>>> wrote:
> >
> > > " RCV, what FairVote is selling, is promoted with the
> > intentional lie your vote for Middle over Worst is
> > guaranteed to help Middle against Worst if Favorite
> doesn’t
> > win."
> >
> > Where does this claim appear from FairVote at all? Oops!
> > Michael Ossipoff hasn't produced any evidence.
> >
> >
> > Only throughout FarVote’s promotional material.
> >
> > “…hasn’t produced any evidence”?
> >
> > I hadn’t yet been asked for it. I thought that you’d have
> > already seen FairVote’s promotional material.
> >
> > But, since you evidently haven’t, then I’ll post an example
> > here. …one of many instances of FairVote’s repetition of
> that lie.
> >
> >
> > I'd appreciate it if you at least did me the courtesy of
> > spelling my surname correctly. I know it's hard to
> find --
> > not like it's in my email address, display name, or
> anything
> > of the sort.
> >
> > On Sun, Dec 17, 2023 at 8:01 PM Michael Ossipoff
> > <email9648742 at gmail.com
> <mailto:email9648742 at gmail.com> <mailto:email9648742 at gmail.com
> <mailto:email9648742 at gmail.com>>> wrote:
> >
> > You might want to specify what you’re talking about.
> >
> > Oops!!! Michael Garmin forget to say what my
> unsupported
> > claim was !
> >
> > On Sun, Dec 17, 2023 at 16:56 Michael Garman
> > <michael.garman at rankthevote.us
> <mailto:michael.garman at rankthevote.us>
> > <mailto:michael.garman at rankthevote.us
> <mailto:michael.garman at rankthevote.us>>> wrote:
> >
> > You might wish to consider substantiating your
> > claims instead of forwarding them to the list
> > without backing.
> > On Sun, Dec 17, 2023 at 7:54 PM Michael Ossipoff
> > <email9648742 at gmail.com
> <mailto:email9648742 at gmail.com>
> > <mailto:email9648742 at gmail.com
> <mailto:email9648742 at gmail.com>>> wrote:
> >
> >
> >
> > ---------- Forwarded message ---------
> > From: *Michael Ossipoff*
> <email9648742 at gmail.com <mailto:email9648742 at gmail.com>
> > <mailto:email9648742 at gmail.com
> <mailto:email9648742 at gmail.com>>>
> > Date: Sun, Dec 17, 2023 at 16:14
> > Subject: Re: [EM] Legacy IRV limitations
> > To: Michael Garman
> > <michael.garman at rankthevote.us
> <mailto:michael.garman at rankthevote.us>
> > <mailto:michael.garman at rankthevote.us
> <mailto:michael.garman at rankthevote.us>>>
> >
> >
> > The falsity of FairVote’s lie is well-known
> > among the electoral-reform community.
> >
> > The term “Know-It-All” is properly used
> to refer
> > to someone making incorrect statements.
> Oops!!!
> > You forgot to specify the incorrect
> statement.
> >
> > “The perfect is the enemy of the good”?
> >
> > You evidently think fraud is good.
> >
> > I wasn’t criticizing STE. I was
> criticizing fraud.
> >
> > …intentional lying to sell a product.
> >
> >
> >
> > On Sun, Dec 17, 2023 at 16:05 Michael Garman
> > <michael.garman at rankthevote.us
> <mailto:michael.garman at rankthevote.us>
> > <mailto:michael.garman at rankthevote.us
> <mailto:michael.garman at rankthevote.us>>> wrote:
> >
> > Sanctimonious know-it-alls like you
> who let
> > the perfect be the enemy of the good
> are the
> > greatest obstacle to any progress
> whatsoever.
> > On Sun, Dec 17, 2023 at 7:04 PM Michael
> > Ossipoff <email9648742 at gmail.com
> <mailto:email9648742 at gmail.com>
> > <mailto:email9648742 at gmail.com
> <mailto:email9648742 at gmail.com>>> wrote:
> >
> > I didn’t say that
> > Successive-Topcount-Elimination
> (STE) is
> > a fraud. I said that RCV is a fraud.
> >
> > RCV isn’t STE. RCV, what FairVote is
> > selling, is promoted with the
> > intentional lie your vote for Middle
> > over Worst is guaranteed to help
> Middle
> > against Worst if Favorite doesn’t
> win.
> >
> > i.e. FairVote is selling RCV as
> > Condorcet. RCV is a nonexistent
> > Condorcet-properties method being
> > fraudulently sold by FairVote.
> >
> > Thus, RCV is a fraud.
> >
> > Sorry, but I can’t abide dishonesty.
> > Fraud shouldn’t be supported.
> >
> > Don’t let a fraudulently-promoted
> > product be successfully sold to the
> > people of Oregon.
> >
> >
> > On Sun, Dec 17, 2023 at 15:39 Michael
> > Garman
> <michael.garman at rankthevote.us <mailto:michael.garman at rankthevote.us>
> >
> <mailto:michael.garman at rankthevote.us
> <mailto:michael.garman at rankthevote.us>>>
> > wrote:
> >
> > Oh come on Michael. You can’t
> claim
> > the system itself is “fraud”
> because
> > you dislike one of the many
> > organizations that advocate
> for it.
> >
> > On Sun, Dec 17, 2023 at 6:37 PM
> > Michael Ossipoff
> > <email9648742 at gmail.com
> <mailto:email9648742 at gmail.com>
> >
> <mailto:email9648742 at gmail.com <mailto:email9648742 at gmail.com>>>
> wrote:
> >
> > Yes, many RCV opponents were
> > formerly RCV advocates…until
> > they found out that
> they’d been
> > lied to by FairVote.
> >
> > As I often say, RCV’s worst
> > problem is FairVote.
> >
> > Lying to sell something is
> > called fraud.
> >
> > RCV is an intentional
> fraud, &
> > yes, people don’t like
> that when
> > they find out.
> >
> >
> >
> > On Sun, Dec 17, 2023 at 12:20
> > Richard, the VoteFair guy
> >
> <electionmethods at votefair.org <mailto:electionmethods at votefair.org>
> >
> <mailto:electionmethods at votefair.org
> <mailto:electionmethods at votefair.org>>> wrote:
> >
> ...
> ----
> Election-Methods mailing list - see https://electorama.com/em
> <https://electorama.com/em> for list info
>
>
>
> --
> *Greg Dennis, Ph.D. :: Policy Director*
> Voter Choice Massachusetts
>
> e :: greg.dennis at voterchoicema.org <mailto:greg.dennis at voterchoicema.org>
> p :: 617.835.9161 <tel:617.835.9161>
> w :: voterchoicema.org <https://www.voterchoicema.org/>
>
> :: Follow us on Facebook <https://www.facebook.com/yeson2rcv> and
> Twitter <https://twitter.com/yeson2rcv> ::
More information about the Election-Methods
mailing list