On 03/12/2019 07.00, robert bristow-johnson wrote:

Yes, it does work. I've just been busy with non-voting things (as well
as finding out how to design voting methods with desired properties).

What are the news from Burlington?

-km

okay, Burlington is pretty liberal.  While there is one Republican in City Council (who also happens to preside over the council and even more coincidently was the Republican candidate for mayor in the 2009 IRV election, gathering the most 1st-choice votes but losing to the Prog candidate in IRV) the Progs in City Council have reintroduced a Charter Change resolution to return to IRV, now they are calling it "RCV" but it is the same Single Transferable Vote procedure used in RCV elections everywhere.  The language of the Charter Change is posted below.

But there was a little pushback from other councilors (and myself) for Burlington to not repeat mistakes from the past, namely that of electing a candidate other than the clearly shown Condorcet Winner.

This is all in a hurry.  In two weeks Council will make the decision whether or not to include the question to the ballot for Town Meeting Day in March 2020.  On Monday Dec 9th, the Charter Change Committee will consider the language of the bill and vote on what shall be brought to the whole Council to include on the ballot.

I want to figure out some good language to change this from regular-old IRV to Bottom-Two Runoff, Single Transferable Vote, BTR-STV .  If someone is good at writing legal language or if this BTR-STV has had legislative language written for it somewhere else, I would like to see it.

The language that needs to be changed, to make this RCV Condorcet-compliant is:

"... The candidate with the fewest votes after each round shall be eliminated ..."

How can we clearly and concisely change that to Bottom-Two Runoff?

--

r b-j                  rbj@audioimagination.com

"Imagination is more important than knowledge."

All elections of mayor, city councilors and school commissioners shall be by ballot, using a system of ranked choice voting without a separate runoff election. The chief administrative officer shall implement a ranked choice voting protocol according to these guidelines:
(1) The ballot shall give voters the option of ranking candidates in order of preference.
(2) If a candidate receives a majority (over 50 percent) of first preferences, that candidate is elected.
(3) If no candidate receives a majority of first preferences, an instant runoff re-tabulation shall be performed by the presiding election officer. The instant runoff re-tabulation shall be conducted in rounds. In each round, each voter’s ballot shall count as a single vote for whichever continuing candidate the voter has ranked highest. The candidate with the fewest votes after each round shall be eliminated until only two candidates remain, with the candidate then receiving the greatest number of votes being elected.
(4) The city council may adopt additional regulations consistent with this subsection to implement these standards.

On 03/12/2019 16.54, robert bristow-johnson wrote:

My first stab would be: "After each round, of the two candidates with
the fewest votes, the candidate ranked below the other by the most
voters shall be eliminated".

Alternatively "on the most ballots". Or "among the two candidates". Or
"of the two candidates with the fewest votes in that round".

Presumably there has to be some tie-breaking language for the case where
more than one candidate is last or next-to-last. But I suppose there is
some similar language for IRV as is.

If there is no equal-rank or truncation, you can also say "the candidate
ranked below the other by a majority of the voters".

The strategy to turn IRV into BTR-IRV is probably the best one. Woodall
is better than BTR-IRV, and you could turn the method into Woodall by
checking for a CW before a round starts, but then you'd have to define
what a CW means in a way that doesn't confuse people who are unfamiliar
with Condorcet.

Looking again at the language, point (3) is already rather convoluted.
Perhaps it would be better to split up the definition. Something like:

(3) If no candidate receives a majority of first preferences, an instant
runoff re-tabulation shall be performed by the presiding election
officer. The instant runoff re-tabulation shall be conducted in rounds.
In each round, each voter’s ballot shall count as a single vote for
whichever continuing candidate the voter has ranked highest, and
subsidiarily to determine what candidate to eliminate. After each round,
the two candidates with the fewest votes shall be considered for
elimination. Of these two candidates, the candidate ranked below the
other by the most voters shall be eliminated in that round. The counting
in rounds shall continue until only two candidates remain, with the
candidate then receiving the greatest number of votes being elected.

(Maybe also call it something else than "instant runoff re-tabulation"
because it's no longer IRV, but I can't think of a better name at the
moment.)

I'm much less certain about Woodall. See below for something I cooked up.

(1) The ballot shall give voters the option of ranking candidates in
order of preference.
(2) If a candidate receives a majority (over 50 percent) of first
preferences, that candidate is elected.
(3) If no candidate receives a majority of first preferences, an instant
runoff re-tabulation shall be performed by the presiding election
officer. The instant runoff re-tabulation shall be conducted in rounds,
for which the counting shall be conducted according to the following
three points.
(4) If there exists a continuing candidate so that for every other
continuing candidate, the former candidate is ranked ahead of the latter
by a majority of the voters, the former candidate is elected.
(5) If there is no such candidate, each voter’s ballot shall count as a
single vote for whichever continuing candidate the voter has ranked
highest. The candidate with the fewest votes shall be eliminated, which
marks the end of the current round and the beginning of the next.
(6) The counting in rounds shall continue until a candidate is elected
or only one candidate remains, with the remaining candidate then being
elected.
(7) The city council may adopt additional regulations consistent with
this subsection to implement these standards.

(Strictly speaking, the latter half of point 6 is redundant because when
only two candidates remain, the winner beats the loser pairwise and
would be elected by point 4. But a potentially infinite loop looks more
dangerous than a finite one.)

I haven't written legislative language before, but maybe it can serve as
a starting point for others who have.

-km

My simple thoughts on this:

• You ask: Do you agree that in problematic situations, like what happened in 2009, the result should be different?
• They answer: Yes.
• You say: This is a problem that is fortunately quite easily to fix. The most straight forward way to enhance the method is to use the Bottom-Two Runoff. The losing candidate one of the two candidates with least votes will be eliminated.
• They say: Ok, let's see if that would be our ideal solution.
• You say: Ok, let's start from there.

BR, Juho

Juho, this is exactly my strategy.  But there is a little bit of an issue about "let's start from there".  Here is the latest news regarding yesterday's Council meeting and RCV:

https://vtdigger.org/2019/12/03/burlington-considers-instant-runoff-voting-for-most-city-races/

What has happened is that the Progs have "succeeded" in fending off an amendment to take more time and investigate alternatives to the IRV method of RCV.  They want, really badly, to get this on the ballot for this coming Town Meeting Day in March 2020.  The latest that the Council can decide to include this on the ballot is in 13 days.  And in 6 days is when the Charter Change Committee will consider this proposed ballot item, fix any language, and recommend it to the Council on Dec 16.

I am afraid that putting this on the ballot "half-baked", with effectively no change from what we had in 2009 and acknowledging no error from the 2009 election, will result in March in rejection by the majority of voters and will, again, set back voting reform for another decade.  So I want to make a good effort at changing the language from regular-old IRV to BTR-STV.  And hopefully get that change adopted by the Charter Change Committee and sent up to the Council to consider for inclusion on the ballot.  Then we can tell voters that this IRV is different from the 2009 IRV and would have corrected the failure of IRV we had in 2009.

I believe that Schulze is technically the best RCV, but since Schulze and Ranked-Pairs will elect the same candidate when the Smith set is 3 candidates or fewer, my favorite would be RP using margins for a governmental election because its method is easier to understand and encode into legal language that laypersons can read and understand.  Also, we would be able to say to the IRV haters that this RP RCV is not IRV at all.

However, we can't say that about BTR-STV because, after all, it is a form of Instant Runoff Voting, but and IRV method that will elect the Pairwise Champion (the term that I will use to denote the Condorcet Winner).  But, given the circumstances (that an IRV proposal is in the works right now), I think this is the best action I can hope to take.  I would have to admit that this Condorcet-compliant IRV is still IRV.  The Progs won't mind and other naive RCV supporters won't mind, but the IRV haters, particularly those who hated the IRV winner in 2009 will always hate IRV or any RCV, but I would still rather be promoting an RCV that is not IRV (no runoff rounds).

I am taking Kristofer's language suggestions and possibly modifying.  I would like to see language suggestions from others.  Either way, I will post what language I finally will submit to the Charter Change Committee.  They will probably reject my submission, but I will tell them plainly that if they are offering RCV that is no different from the IRV that resulted in the 2009 failure and was repealed by voters the following year, that their proposed charter change will be rejected by the majority of voters on Town Meeting Day.  And this will likely set back voting reform by another decade.

and also to you, Juho.  i am appreciative of any help or language or analysis suggestions from you or anyone.

--

r b-j                  rbj@audioimagination.com

"Imagination is more important than knowledge."

On 12/3/2019 2:56 PM, robert bristow-johnson wrote:

I'm jumping into the middle of this without having time to read about
BTR-STV, so pardon my misunderstandings about what you are attempting.

I suggest considering terminology such as:

"If there is a candidate who is less popular than every one of the other
candidates based on comparing each and every pair of candidates one pair
at a time, eliminate that pairwise-losing candidate."

This approach eliminates the need to mention the concept of a Condorcet
winner.

Also, this approach is not Condorcet compliant because it does not
ALWAYS elect the Condorcet winner, so it cannot be criticized as "a
Condorcet method."

Of course the key word is "if" because sometimes there will not be a
pairwise loser.  Whenever there is not a pairwise loser, your existing
words specify what to do.

Here's a description of a method that repeats this pairwise elimination
process and uses an upside-down version of IRV to resolve
rock-paper-scissors-like cycles:

https://democracychronicles.org/instant-pairwise-elimination/

I've called it Instant Pairwise Elimination, but the name is not important.

It's the concept of "pairwise counting" that I think is important, and
should be used in election counting methods.

Richard Fobes

I think that making sure that the voters can strictly rank from the top
however many candidates they wish is more important
than making sure the method elects the voted Condorcet winner.

I think that IRV/RCV  (with one-at-a-time elimination) that meets that
requirement  has the most merit relative to its traction and
comprehensibility.

But if we do want to insist on compliance with the Condorcet criterion,
necessarily at the expense of making the making the method vulnerable to
Burying
strategy and losing compliance with the (popular with IRV supporters)
Later-no-Harm criterion, then I don't like Bottom-Two Runoff.

It would be much better to check before each normal IRV-style
elimination that the candidate you are considering eliminating doesn't
pairwise-beat all the
remaining candidates. If it does the process should stop and that
candidate should win.

That would be a quite good Condorcet method.

Chris Benham

On 4/12/2019 9:26 am, robert bristow-johnson wrote:

How does this look to you guys?  This is the original language on the existing resolution to implement RCV that is not Condorcet compliant:

All elections of mayor, city councilors and school commissioners shall be by ballot, using a system of ranked choice voting without a separate runoff election. The chief administrative officer shall implement a ranked choice voting protocol according to these guidelines:
(1) The ballot shall give voters the option of ranking candidates in order of preference.
(2) If a candidate receives a majority (over 50 percent) of first preferences, that candidate is elected.
(3) If no candidate receives a majority of first preferences, an instant runoff re-tabulation shall be performed by the presiding election officer. The instant runoff re-tabulation shall be conducted in rounds. In each round, each voter’s ballot shall count as a single vote for whichever continuing candidate the voter has ranked highest. The candidate with the fewest votes after each round shall be eliminated until only two candidates remain, with the candidate then receiving the greatest number of votes being elected.
(4) The city council may adopt additional regulations consistent with this subsection to implement these standards.

and here is the original languate modified to implement BTR-STV that is Condorcet compliant:

All elections of mayor, city councilors and school commissioners shall be by ballot, using a system of ranked choice voting without a separate runoff election. The chief administrative officer shall implement a ranked choice voting protocol according to these guidelines:
(1) The ballot shall give voters the option of ranking candidates in order of preference.
(2) If a candidate receives a majority (over 50 percent) of first preferences, that candidate is elected.
(3) If no candidate receives a majority of first preferences, an instant runoff re-tabulation shall be performed by the presiding election officer. The instant runoff re-tabulation shall be conducted in sequential rounds. A "remaining candidate" is defined as a candidate that has not been eliminated in any previous round. Initially, no candidate is eliminated and all candidates begin as remaining candidates.
(4) In each round, each voter’s ballot shall count as a single vote for whichever remaining candidate the voter has ranked highest. The two candidates with the fewest votes in a round, herein designated as "A" and "B", shall be compared in a runoff in which the candidate, A or B, with lesser voter support is eliminated in the same round. If the number of ballots having A ranked higher than B exceeds the number of ballots having B ranked higher than A, then B has lesser voter support, B is eliminated, and A remains for the following round. Likewise, if the number of ballots having B ranked higher than A exceeds the number of ballots having A ranked higher than B, then A has lesser voter support, A is eliminated, and B remains for the following round. In the case that the aforementioned measure of voter support between A and B is tied, then the candidate with fewest votes is eliminated in the same round.
(5) This runoff re-tabulation, eliminating one candidate each round, is repeated until only two candidates remain, with the candidate then receiving the greatest number of votes being elected.
(6) The city council may adopt additional regulations consistent with this subsection to implement these standards.

Is this good language for BTR-STV?  Can any of you think of a hole in this description?  Can any of you make it better or more concise?

Thank you.

r b-j

...

--

r b-j                  rbj@audioimagination.com

"Imagination is more important than knowledge."

The reason for all the controversy is the limitations of IRV. And
political expedience over-looks scientific principle. From the latter
point of view, the limitations are that single vacancies struggle to
represent even half the voters. That's what you are all arguing about:
how to do even that much. The other IRV limitation is that eliminations
discard preferential information germane to the result.
The obvious answer is STV, which elects the first preferences, nearly
all of them with enough seats per district. That is the acid test,
despite the residual short-comings of traditional STV or Meek method. (I
have worked out how to further empower the transferable voting method:
my oft-mentioned FAB STV.)
A trouble with the US "choice voting" reform campaign is that IRV is not
choice voting, it is, as the French say "faux pas de mieux" (for want of
something better).

Richard L

On 04/12/2019 08:18, robert bristow-johnson wrote: