[EM] GOLD voting: the most practical PR proposal?

[Jameson Quinn]

2017-07-11 13:39 GMT-04:00 Toby Pereira <tdp201b at yahoo.co.uk>:

> I think GOLD looks like it's probably a good method, and it looks like an
> incremental modification of previous methods you've defined and discussed
> (such as PAL).
>

Yes. I regard GOLD as the culmination of the series of method proposals
that includes PAL and OL/D.


> A few things though:
>
> You mention that candidates must publicly declare a preference order of
> the other candidates for when votes are transferred. Although it's not
> explicitly mentioned, I think under PAL party candidates were compelled to
> list candidates of their own party above all others. They also had a strict
> number of layers, so, for example, all candidates from one party would be
> tied in their rankings rather than a candidate having the ability to give
> individual ranks to every candidate. Does any of that apply here?
>

I have been deliberately vague on that point in defining GOLD as a general
idea, because I think that different implementations of GOLD could
reasonably differ on this detail. But when I tried to redraft the Fair
Representation Act
<https://docs.google.com/document/d/1nfpFrGNXA8KJHw0Rd6YByyXyx2Gj_awi16apSUr9L0s/edit>
to institute GOLD instead of multimember STV
<https://docs.google.com/document/d/1AWhaCWxG2TkvsASkbYt6Gf4NihvzCupcJL5cM9Qu9BQ/edit>,
I mandated a maximum of 6 rating levels, a minimum of 2 candidates in any
level used, and a rule that no same-party candidate may be rated strictly
below a different-party candidate.


>
> I'm not sure I fully follow the vote tallying section, but I'm assuming
> it's close enough to STV to make it generally proportional while fitting
> other criteria you want, such as making sure each elected candidate is
> popular enough in their own area.
>

That's the idea, yes.


>
> But I do wonder if having precisely one candidate elected per area is
> going to look that good though. With mixed-member, you get a local
> candidate anyway without worrying about proportionality, and then the
> top-up MPs are there to make the result proportional. And with STV, the
> proportionality comes from having wider areas with several MPs elected.
>
> I presume the philosophy behind this is that it seems most fair to have
> one MP per constituency, and you don't end up with two different types of
> MP like in mixed-member. STV gives complex ballots and people won't feel
> quite so connected to MPs that cover a wider area - it might be that all
> the MPs in your area happen to come from the other side geographically. But
> I think some people might be annoyed that their MP is not the most popular
> candidate in their area, and they might perceive it as their MP being there
> just to make up the proportional balance. And they will look across at the
> next area and wonder why they were allowed their most popular candidate and
> not forced to give them up for some sort of "greater good" (proportional
> representation). I know you have implemented rules to make sure that the
> elected candidate is not a "nobody" in their own area, but I don't think it
> will appease everyone. I think that that is arguably one advantage of
> mixed-member, in that every area gets the MP most popular with the voters
> (well, according to the voting method being used).
>

These are reasonable concerns. I'd answer them with the following points:

   - As a relatively recent addition to my definition of GOLD, I've
   included a mandate¹ for winning parties to assign each district that they
   didn't win as "additional territory" to one of their winning
   representatives. (Search for "Once all the winners..." in the text below to
   see this requirement.) So, say there was a member of party X who happened
   to be among the plurality of their own district but still got a local
   winner from party Y due to transferred votes. Such a person would still be
   in the "additional territory" of a party-X candidate; in almost all cases,
   one whom their vote had helped elect. So they'd still have a sympathetic
   representative responsible for hearing their petitions.
   - This aspect of GOLD is an advantage in the US context because it makes
   the method compatible with current federal law.
   - It's also an advantage in terms of making the system more acceptable
   to incumbent politicians, because it means that the GOLD outcome will
   differ from the FPTP outcome only insofar as the latter is
   disproportionate, and/or insofar as GOLD motivates or allows different
   behavior from voters (including cross-district voting). In other words: if
   you are an incumbent from a party which has not benefitted from
   gerrymandering, and if you are not unusually unpopular with your
   constituents so that they'll be especially likely to vote out-of-district,
   and if you do not expect the number of seats that go to third parties under
   GOLD to be greater than the gerrymandering disadvantage your party suffered
   under FPTP, then your seat is as safe with GOLD as it was under FPTP. The
   first two criteria there are likely to hold for most incumbents in any
   place that has any chance of passing a PR reform; and the third criterion
   is likely to be close enough to holding that you can reasonably
   de-emphasize it when talking to incumbents.
   - Finally, the "one per district" rule of GOLD helps to justify the
   "pre-elimination" step, which helps discourage excessive party splintering
   (a la Israel). One could still add a pre-elimination or other quota-like
   step in OL/D (that is, GOLD without the one-per-district rule), but it
   would seem more arbitrary. I think that having a mechanism to discourage
   excessive splintering is an important feature of a practical PR method.
   (Note that the corresponding rule for MMP, such as Germany's 5% quota,
   leads inevitably to wasted votes and thus to voters strategically ignoring
   new parties, while in GOLD the votes can still transfer usefully and thus
   smaller parties can still play a productive role and exercise real power
   via their transfer orders even before they are large enough to win any
   seats. I think that's clearly an advantage to GOLD over MMP.)




>
> While I'm here, I might as well throw in another method, which I described
> here - https://groups.google.com/forum/#!msg/electionscience/
> aP7ybKMb1zs/giaYAh6wAwAJ It is a mixed-member method, but it uses
> approval or score ballots rather than plurality/First Past the Post. I also
> discuss it in this video https://www.youtube.com/watch?v=EjeBEBZjm9Y If
> you don't want to watch that, I think it's still worth going to 2:33 to see
> the score ballot and 12:10 to see the approval ballot, which I think are
> both nice and simple for the level of proportional representation you get
> from them.
>

I like this idea, but I don't want to get this thread off-topic.


> Toby
>
>
> ------------------------------
> *From:* Jameson Quinn <jameson.quinn at gmail.com>
> *To:* electionsciencefoundation <electionscience at googlegroups.com>; EM <
> election-methods at lists.electorama.com>
> *Sent:* Monday, 3 July 2017, 18:15
> *Subject:* [EM] GOLD voting: the most practical PR proposal?
>
> I sent essentially the same message as below, with a different subject,
> earlier today, to the same two lists. I'm re-sending it now because on
> second thought it should be in its own separate thread. The only changes
> between the below and the earlier message, besides the subject line, are:
>
>    - I've fixed the copy of the webpage at the end to reflect the latest
>    adjustments.
>    - I mention precinct-summability as an advantage of GOLD.
>
> -----
>
> I designed GOLD voting
> <http://wiki.electorama.com/wiki/Geographic_Open_List/Delegated_(GOLD)_voting>
> specifically for situations like BC, and I put the finishing touches on the
> system in response to my simulation results based on the BC outcome. So I
> think that GOLD is basically the ideal proposal for BC, as well as being
> uniquely well-suited as a reform in the US context.
>
> (Note: at the end of this email, I'll paste the current version of the
> electorama page)
>
> What's good about GOLD? I think the starting point for asking that
> question is first to establish what voters (and, insofar as they're not in
> conflict with voters, politicians) are looking for from a voting system.
>
> And that brings us to the mydemocracy.ca poll. That poll had some serious
> flaws: it tended to pose its questions in the form of "A or B" false
> dilemmas, rather than just asking "How important is A? How important is B?"
> In fact, the entire poll can be seen as a strategy by the Liberals to
> deep-six election reform. Nevertheless, the poll is, as far as I know,
> unequalled for its sample size in probing the electoral values of an
> English-speaking population accustomed to FPTP.
>
> So, what did it find? What do Canadians think their electoral system
> should be like?
>
>    - It should lead to a greater diversity of views in Parliament (65%
>    agree, 13% disagree; and 70% want power to be spread over more than just
>    one party).
>    - However, it should not lead to too many small parties (59% agree,
>    41% disagree).
>    - Ballots should remain relatively simple (59% agree, 41% disagree).
>    - MPs should be more accountable to their constituents than to their
>    party (77% agree, 23% disagree).
>    - The poll did not ask whether voters believe that the “constituents”
>    mentioned above must be defined by geographic area (riding). However, when
>    asked to pick the important values from a list of various possibilities,
>    “governments with strong representation from every region” was picked by
>    49%, and “MPs who focus primarily on the interests of their local
>    community” by 42%.
>
> These results were spun by Trudeau and the Liberals as “mixed”, and he
> reneged on his promise to support and implement PR. But there is nothing
> “mixed” about it. There’s a clear mandate for reforming the voting system,
> and it’s possible to design a voting system that fulfills all of the above
> values.
>
> GOLD is one possible outcome of such a design process. Furthermore, having
> gone through this process, I think that many of the decisions involved in
> GOLD's design are clearly the right ones. I'm certainly not saying that
> every detail of GOLD as it's currently written up on electorama is beyond
> debate, but I do think that the basic skeleton of the idea — in particular,
> optional delegation and localized ballots — are pretty well-justified.
>
> I'm posting this message to both electionscience at googlegroups. com
> <electionscience at googlegroups.com> and election-methods at lists.
> electorama.com <election-methods at lists.electorama.com>. On these lists,
> we like to talk about theoretical aspects of voting methods. Give us a good
> debate about Phragmen vs Ebert or about Sainte-Laguë vs D'Hondt, and we're
> happy. But I actually think that, when it comes to proportional methods,
> these debates are secondary. Between FPTP and PR, there's a yawning chasm;
> between Sainte-Laguë and D'Hondt, or between the honest outcomes of
> proportional range voting and of Bucklin transferable voting (to give
> arbitrary examples) the difference is in almost all cases a small handful
> of seats or fewer.
>
> So the important questions aren't the details of the proportionality
> formula, but rather practical matters like ballot design.
>
> As far as I can tell, the PR options generally considered for a case like
> BC are dual-vote MMP; multimember-district STV; or open list. All of these
> are decent systems. But none of them comes close to matching GOLD at
> fulfilling the values above.
>
>    - Delegation and localized ballots allow GOLD to give much simpler
>    ballots than STV.
>    - The fact that voters have the option to vote out-of-district and/or
>    not to delegate means that voters have much greater power and politicians
>    have much more accountability than MMP or open list.
>    - Pre-eliminating all but the "top two" (occasionally one or three) in
>    each district provides an incentive against party fragmentation. This is
>    better than STV, and transfers let it naturally avoid the wasted votes of
>    similar mechanisms that can be added to MMP or open list.
>    - One important advantage of GOLD is that it only changes the FPTP
>    outcome insofar as it has to to achieve proportionality. This makes it
>    much, much less of a threat to incumbent politicians, especially insofar as
>    those politicians won "fair and square" under FPTP. (That is, without
>    relying on natural or artificial gerrymandering, negative campaigning, or
>    other pathologies of FPTP.)
>    - In the US context, compatibility with current federal law is an
>    additional advantage of GOLD.
>    - GOLD is also precinct-summable, unlike STV or many other "advanced"
>    proposals.
>
> ...
>
> So I think GOLD is the gold standard. But is that just because I'm the one
> who designed it? The burden of proof is on me to show that it's not. In
> order to meet that burden, I'm interested in having as broad a dialogue as
> possible about practical PR reform proposals. Does anybody here think that
> MMP, STV, or open list is a better proposal (that is, given similar
> investments in activism, more likely to catch on and spread) than GOLD? Are
> there other new proposals that offer as many advantages as GOLD does?
>
> Jameson
>
> ps. Here's the current GOLD webpage from Electorama:
>
> Geographic Open List/Delegated voting (GOLD voting) is a proportional
> voting method for electing legislators to a multi-seat body. Its main
> advantages are: simple ballots, minimal wasted votes, and "do no harm"
> (that is, it doesn't change FPTP outcomes unless they're non-proportional).
> It assumes the voters have been divided up into one equal-population
> riding (aka district or constituency) per seat being elected and that each
> candidate has publicly declared their preference order for the other
> candidates ("if I don't win, then I want the votes I hold to go to her,
> then him, then him, etc."). Precisely one representative per area
> (district, riding, or constituency) will win.
> Here are the rules. Items in italics are mere explanations or
> justifications; the rules themselves are only the non-italic portions.
> Voters make two different choices in each race:
>
>    1. *Choose a candidate*.
>       - The ballot lists the candidates running locally, with their
>       parties and their first three transfer preferences (explained below).
>       - Voters may write in candidates from further away.
>    2. *Choose a transfer method* for when your first choice is no longer
>    in the running. There are 2 basic options:
>       - *Open list*: Trust the *voters* of your chosen candidate’s party.
>    - If your first choice is no longer in the running, your vote is
>       transferred to the remaining candidates from your chosen party, in
>       proportion to the number of direct votes they got.
>       - This is the default if you vote for a local, non-independent
>       candidate.
>       - *If every voter chose this option, this would be like an “open
>       list” voting method; that is, seats would be divided proportionally by
>       party, and go to the highest vote-getters within the party.*
>       - *If you choose this option, your vote will never be transferred
>       out of the party. Since independent candidates are considered to each be in
>       a party by themselves, voters for those candidates should only choose this
>       option if they do not want their vote to be transferred.*
>    - *Delegated*: Trust the *candidate* (that is, the pre-declared
>       preferences of your chosen candidate.)
>    - Each candidate must publicly pre-declare ordered preferences between
>       the other candidates. If the candidate is no longer in the running, these
>       votes will go to the highest remaining candidate on their pre-declared
>       preference list.
>       - *This is the default if you vote for a non-local and/or
>       independent candidate.*
>       - *If a voter mistakenly marks both transfer methods, the default
>       applies (as if they had chosen neither).*
>
> The basic vote-counting process has 5 steps (based on Single Transferrable
> Voting):
>
>    1. Tally votes
>       - Each ballot counts as 1 point for the chosen candidate.
>    2. Eliminate candidates without enough support in their riding
>       - The top candidate in each riding, counting local votes only, is
>       never eliminated.
>       - The second candidate in each riding, counting local votes only,
>       is eliminated only if their local votes are fewer than half those of the
>       top.
>       - Others are eliminated by default, surviving only if their local
>       votes are more than half those of the top AND their total direct votes
>       (including non-local write-ins) are more than those of the top local
>       candidate. (For this rule, "top" is counted by local votes only, but "those
>       of" includes non-local votes.)
>       - *This makes sure that no riding is badly mis-represented just
>       because a given party "deserves" more winners.*
>       - *It also helps discourage voters from splintering into small
>       single-issue parties. If a party can’t pass this threshold in even one
>       riding, it won’t get seats. But those votes can still be transferred, so
>       those voters can still be represented by a relatively sympathetic candidate
>       from a slightly larger party.*
>    3. Find winners and transfer leftovers.
>       - If V is the total number of valid (non-exhausted) votes, and S is
>       the number of unfilled seats, then a “quota” is defined as Q=V/(S+1). This
>       ensures that each full “quota” of voters will get a seat, with less than
>       one “quota” of vote left unrepresented even though they still have a valid
>       preference.
>       - Any candidate with a full quota of votes at any time is elected.
>       If their winning vote total is W>Q, then the leftover fraction (W-Q)/W of
>       all of their votes is transferred.
>       - Whenever a candidate wins, all other candidates from their riding
>       are eliminated.
>    4. Eliminate the candidate who's furthest behind in their riding and
>    transfer votes
>       - *If a candidate's current full tally is 1000 votes (including
>       local votes, direct write-ins, and transferred votes), and the top full
>       tally of any remaining candidate in their riding is 2000, then they are
>       1000 behind in their riding.*
>       - *This rule means that the last remaining candidate in a riding is
>       not eligible for elimination.*
>       - See above for the transfer methods a voter can choose.
>    5. If there are still seats to fill, repeat from step 3.
>
> Once all winners are chosen, each winning party is responsible for
> assigning each district they did not win to be "additional territory" of
> one of their winning representatives. Representatives are responsible to
> all citizens from their own district, and also to hear petitions from their
> "additional territory". That means that if you are in the minority in your
> district, you will still have a sympathetic representative to petition.
> Proportional or semiproportional?[edit
> <http://wiki.electorama.com/w/index.php?title=Geographic_Open_List/Delegated_(GOLD)_voting&action=edit&section=1>
> ]
> GOLD voting is proportional in a two-party context. If there are more than
> two parties, though, it is only semiproportional; smaller parties without a
> clear regional character may get less than their proportional share. But if
> that happens, their votes will not be ignored; they will have a say on
> which of the larger parties gets more seats, and even on which candidates
> from that allied larger party win. Thus, a smaller party will be able to
> promote their issues by favoring those candidates who prioritize those
> issues. Also, if there are two competing party coalitions, with all voters
> choosing one of the alliances and all candidates preferring same-coalition
> candidates over opposite-coalition ones, then GOLD will be fully
> proportional between the two coalitions.
> Note that other proportional voting methods sometimes are used with extra
> rules designed to stop fringe parties from winning seats. For instance, in
> the German mixed-member "proportional" method, a party that gets less than
> 5% or 2 direct seats does not get a proportional allotment of seats. Thus,
> technically speaking, even the German system is really only
> semiproportional, not truly proportional.
> Advantages[edit
> <http://wiki.electorama.com/w/index.php?title=Geographic_Open_List/Delegated_(GOLD)_voting&action=edit&section=2>
> ]
> The advantages of this method are as follows. First, the advantages common
> to all proportional representation methods:
>
>    - Equality: partisan gerrymandering is impossible, and each party gets
>    its fair share of seats.
>    - Representation: Almost all voters are truly represented; even if you
>    are a minority in your district, your vote helps elect a candidate of a
>    party you sympathize with.
>
> This method also keeps all the strong points of the current voting system.
> (The current system is horrible in general, but it still has its strong
> points.)
>
>    - Simplicity: you just choose one candidate, and the ballot is short.
>    - Accountability: voters, not parties, choose who is elected.
>    - Unity: discourages splinter parties, because candidates without a
>    strong local base of support are eliminated up-front.
>    - Geography: Everyone has a representative who lives relatively close
>    to them.
>
> Similar methods[edit
> <http://wiki.electorama.com/w/index.php?title=Geographic_Open_List/Delegated_(GOLD)_voting&action=edit&section=3>
> ]
> OL/D voting
> <http://wiki.electorama.com/wiki/Open_List/Delegated_(OL/D)_voting>:
> basically the same, but without the constraint of one candidate per riding,
> and with a slightly weaker elimination rule.
> Proportional 3RD voting
> <http://wiki.electorama.com/wiki/Proportional_3RD_voting>: a similar
> system, for a nonpartisan context without ridings.
> ----
> Election-Methods mailing list - see http://electorama.com/em for list info
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.electorama.com/pipermail/election-methods-electorama.com/attachments/20170711/f4502262/attachment-0001.htm>