Mixed Member Proportional Discussion

Home Forums Fair Vote Discussion Forum Mixed Member Proportional Discussion

Viewing 15 posts - 1 through 15 (of 17 total)
  • Author
  • #25913

    Lets talk about MMP and the 7000 apparent variations on it! One of the reasons the referendum failed in BC in my opinion was because the question of MMP was left too open ended and a lot of people were scared about all the different options. This even caused John Horgan to come out mid-way through the voting period to affirm that if MMP were the winner it would only be open lists. Do people here think that the BC referendum might have passed if the particulars for MMP had been more nailed down? How many different version of MMP are there theoretically?

    Robert Jarman

    I like the panachage system Switzerland has for open lists. Basically, imagine that a canton has an open list of say 9 members of parliament who must be chosen. This would in an MMP system be in addition to 9 or 10 or something like that other members of parliament elected from local constituencies.

    Each elector in Switzerland has a ballot with 9 slots on it. They can put the same name twice (and in theory, this would work up to 9 times). They can put 7 names once and one name twice. It does not matter which party they are from. Most parties have a list of candidates that you can use by default, but there is no obligation to do so. You can also take a party list and amend it the way you like, strike out names from it, substitute additional names onto it, put some names more than once, as long as the total is 9 names. It is mathematically giving the same power to each person for a person to be allowed to write 9 names each, just so long as all voters can do that.

    From there, the votes would be counted such that for each mention of a name who is a member of a political party, that party gets that many votes, and this is added for all the members of their party who are mentioned on any ballot in total. That part determines how many seats in parliament they get as a whole. In an MMP system, those seats would first be filled by those who won a local seat. If that is not enough, then all the members of the party or nominated candidates of a party who are mentioned on any ballot are added up along with the number of times they are mentioned and ranked in descending order, and from them, any MPs the party is entitled to based on the electoral result is allocated in descending order of most popular to least.

    This allows for candidates to be personally popular on the lists, it allows you to create your own list if you want, or use any list proposed by any group of electors or political parties that get enough signatures and/or pay the deposit to be on the ballot, you can amend any list, combine lists from two or more lists, strike anyone from any list, add anyone to a list, mention people multiple times, etc. In theory, any citizen group could propose a list too, or an independent. Independents get the added benefit of being able to have their name mentioned, so that a citizen who votes for a party list of 9 people has the same weight as someone who voted for an independent and put their name nine times on the ballot, and hybrid support for many independents or a combination of independents and partisans is possible. It comes close to single transferable vote in terms of how much choice you have. Doing so strictly without party lists would be cumulative voting, and earning a droop quota in cumulative voting will also ensure your election in STV as well.

    It is also called Free List proportional representation.

    jim in oakville

    Has fairvote not decided what MMP system it is proposing?
    I have seen proposals of regional votes, open or closed, which made me question if that could mean we would have multiple people from a party canvassing for votes in an election. One local FPTP, and several for regional. This I feel would be too confusing. Closed lists might seemed rigged.

    Also I have seen proposals advocating 40% regional reps. My analysis of election results shows this to be double or more what is needed on average.

    So it makes me wonder if the simpler, most democratic method would just be to allocate top up seats regionally to the party reps that got the most votes in any region.

    I think we must come up with a simple clear proposal that we can unify around, accepting no method may be perfect.


    There is a system called something like “the biggest loser” in which the top up seats are filled by those that did not win in the local election but won the highest percentage of votes of all other party members who ran in other local elections.

    I also do not think that FVC has decided on a version of MMP. There are so many versions I don’t even know them all.

    jim in oakville

    George. I think first thing we need to do is change the reference term to “Proportional Winners”. They definitely wouldn’t be losers.

    Robert Jarman

    Just call it the Baden Württemberg method.

    David Nash

    I am a supporter of PR and prefer MMP to other systems on offer for two primary reasons: First, it has the capacity to operate in a manner not dissimilar to our present system from the point of view of the act of voting. Second, it it has the capacity retain a strongly regional element in the choice of candidates and in therefore in the candidates that are elected.

    Having started a topic in which I suggested we try to define the parameters of a PR system that would be most acceptable to the non-PR-geek segment of the population, which is most us. I now find that it has become progressively bogged down in the kind of technical debates that my objective was to avoid, I decided to look around for a discussion that maybe had a narrower focus.

    In my original paragraph I mentioned two characteristics of MMP, both of are qualified by the statement that it “has the capacity” to do. What I would like for Fair Vote Canada to devise a simple statement which defines the nature of a system that actually does these things; my belief is that such a system would unite the clear majority of the population who, in theory, support us. We are an argumentative bunch and by behaving argumentatively this produce all sorts of doubts in the general public. I support the objective of Fair Vote Canada, but I do not support its policies, that we are non-partisan and that we are unwilling to plump for a single system.

    We all know the possibilities; what I, at least, want to see is a single clearly articulated proposal for the introduction of proportional representation based upon its public acceptability, not upon its imagined sophistication and niceties.


    I like that idea of calling them “proportional winners” over the biggest loser – it’s a glass half-full kind of thinking.

    As for being non-partisan, I think there is some value in that direction – I’m not sure it would be ethical for us to implement proportional representation on minority support and most parties do not get a majority of the popular vote (otherwise we might no be having pr discussion at all). An all party committee or citizens assembly without a referendum is a way that we can have a collaborative and inclusive process in a way where people feel included and consulted while coming to a unanimous but negotiated decision. We need to be able to better convince those in conservative viewpoints that a proportional system strengthens their voice and values while decreasing any one persons or parties ability to have absolute power.

    As for coming up for a single clear proposal, maybe it would be a good start to list some of the advantages of different systems and just starting talking about different options here. I really liked RUP in the BC election because it was relatively well defined and it took advantages of both STV and MMP together. If I were to propose our adoption of one system, it would be a custom made hybrid like that because it brings in both MMP supporters and STV supporters and if marketed right, can be sold to the increasing stark political divide between Urban and Rural voters without the perception that an only STV system would take advantage of rural voters or that an only MMP system would take advantage of urban voters.

    Wilf Day

    Has fairvote not decided what MMP system it is proposing? Yes. We told the Electoral Reform committee about MMP, here:

    Appendix 10: Made-in-Canada MMP

    Réal Lavergne

    Hi everyone,

    I think there are reasons that discussion of MMP tends to get into configuration issues for those who are deeply involved in our movement. MMP is, in fact, a complicated system to design properly because it combines two ways of voting. Furthermore, and this is significant, what FVC has put forward is a type of MMP that exists nowhere in the world: an open list, regional PR system, using dual candidacy, two Xs to mark one’s preferences, and the allocation of list seats according to the regional X vote.

    Other regional MMP systems do exist, notably in Scotland, Wales and Bavaria, but none combines all of those features. Baden Württemberg uses what we like to call a “best-runners up” system rather than “best losers.” Which is fine. But best-runners up is a one-vote system rather than a two-vote system.

    Bavaria uses open list, but it uses what Rhys Goldstein and others have called “Bavarian metrics.” This system allocates the regional seats by combining the local votes and the regional votes. It is the closest that exists to what FVC has been calling for, so in the Google Doc discussion piece I have been working on (linked below), I have included Bavarian metrics.

    Open lists are at once a core feature of FVC’s proposal, and the Achille’s heel of that model, since open list combined with dual candidacy can theoretically give you elected regional candidates based on very tiny fractions of the total vote. To make up an example based on PEI, let’s say in a provincial election that virtually everyone voting Green voted for the Green Party leader Peter Bevan-Baker as their prefered regional Green candidate except a handful of people (yes, this assumes the absence of strategic voting). Assume also that Peter Bevan-Baker won his own local seat and that the Greens deserved two top-up seats out of the nine. Those two top-up seats would now be allocated on the basis of only a handful of seats. This is a reductio ad absurdum argument, but you get the point. It’s nuts! Meanwhile, FVC has never looked closely at this issue.

    There are solutions, but I can think of only two good ones.

    Bavarian metrics helps in four ways:
    a) it does have a track record, and there are no other examples of open-list MMP in operation that I know of (the use of open lists in list-PR systems is irrelevant: list PR does not suffer from the Bevan-Baker problem mentioned above);
    b) it ensures that everybody has two votes and that each of those votes counts towards electing an individual candidate or not (each candidate is personally accountable to the voter);
    c) it ensures that winners have more than a handful of votes one way or another in two ways: in Bevan-Baker’s riding Green party supporters would not have the option of voting for Bevan-Baker both locally and regionally (only locally), so they would have to vote for other Green candidates and in other ridings in which Greens did not win, the local votes for those candidates will now count as regional votes for those candidates; and
    d) it discourages tactical voting as explained in the reference cited in my paper.

    So Bavarian metrics is an easy fix that goes a long way.

    Even better, I think, would be the use of ranked ballots. I keep suggesting MMP with ranked ballots as a solution, and so far I have not heard any convincing technical counter-arguments. The main counter-arguments I have heard are “I don’t want ranked ballots” and “no-one else seems to be suggesting this” which I don’t find terribly compelling as arguments. The counter-argument that I do buy is that this might be too much change to introduce all at once, and after all, the whole point of MMP is to introduce a change that voters and politicians can more easily accept. With this in mind, in my draft paper, I have proposed leading with Bavarian metrics and considering ranked ballots as a potentially better alternative down the line.

    I’m not religious about one solution over another, but I don’t think the challenges posed about open lists is something that we can afford to ignore.

    Here’s a link to my paper, for those who are interested in a discussion of MMP configuration issues.


    Haven’t read your paper yet, Réal, but I mostly like what you’re saying here! I think I’m in the camp of not ranked ballots at first implementation, even though I personally do like ranked ballots, but I think let people get used to the general two vote system first, then implement another tweak–or not depending on how well it works without that feature. Or to say another way: If it is still not giving people exactly what they are after (i.e., if their vote is still not counting) that could help make them feel at least a bit better.

    Antony Hodgson

    I’m responding at Réal’s invitation, and just want to offer a couple of brief observations:

    1. Now that there is no active electoral reform process on the table right now, it does seem to me to be a decent opportunity to see if there is substantial consensus (including beyond the FVC community) on what the best MMP system might be for Canada. I like Réal’s document because it lays out those issues on which FVC thinks it has reasonable consensus (and so can make recommendations) and those that FVC still considers open for discussion. I would recommend that FVC present these more as proposals than as definitive recommendations and seek input from other allied groups to see if a broader consensus could be developed (eg, all the various participants in the various referendum campaigns and the 2016 ERRE process would be a good place to start).

    2. Re: Bavarian metrics: I agree with Réal’s points – the standard approach to open-list MMP (ie, counting only the votes on the regional ballot) runs the risk that the most popular candidate (eg, Bevan-Baker) will win the majority of the votes and subsequent seats will be determined by very few votes. Using the Bavarian approach of adding both the local and regional votes together would tend to make it so that the most popular local candidates would win, whereas that might be a less certain outcome under standard OL-MMP. Of course, using STV for the open list ballots would be even more robust, but at the risk Réal mentions that this may be seen as too much change (and that it introduces another layer of complexity that might be difficult to explain). We should also remember that there is an additional element to the Bavarian system – namely, that both the first and second votes count towards the party vote total; this has the effect of reducing the incentive to cast your second vote for a smaller allied party if your first preference is likely to sweep the local seats.

    I would note that I added a comment in Réal’s document that we might also consider a slightly different one-part ballot organized in rows (one per riding) and columns (one per party, plus additional ones for independents). If voters could rank freely across the ballot (much like STV), then we could treat the voter’s rankings in the local riding as equivalent to a ranked ballot there, and could then use their global rankings for the regional runoff. The regional runoff could be done using some form of Multimember Instant Runoff Voting to keep it simpler than conventional STV.

    3. Re: names – for public consumption, I recommend that we avoid using terms such as ‘Bavarian metrics’, ‘Baden Württemberg’ or ‘biproportional’ – no-one outside our circle of geeks would understand these terms, and would mainly be turned off.


    I disagree that ranked ballots are too complicated at first implementation. STV and RUP both had strong showings in BC which show that there is a preference by a significant number of voters for a ranked system.

    You mention in the paper that “when combining open lists with dual candidacy rules is that voters will be called upon to vote for regional candidates without knowing which local candidates are going to be elected. It is likely that a very large share of regional votes will be for candidates who end up being elected locally, and are no longer eligible as list-seat candidates by the time the ballots are counted. Such votes still count as votes for one’s party of choice, but they do not help to elect a particular candidate. When this happens the regional candidates who end up being elected may lack the legitimacy they need because they are elected with so few votes.”

    You could solve this with an STV like regional ranked ballot with the candidates who are elected locally having their votes passed on to other regional candidates. The Bavarian model could still have everyone within a region vote for one candidate regionally when there are say 8 regional candidates, even with the local riding which the candidate runs not being able to vote for him regionally. Once the votes from the elected local candidates are transferred, you can go forward with MMP seat allocations.


    I think that the problem with the systems proposed in BC or the version of MMP proposed for PEI is that in both cases one was having to balance what was best for province/voters/democracy with what could be easily explained in a Referendum – those two criteria are often at cross purposes. I think having a ranked ballot for MMP (and I think down the road would be better than upon implementation) would be a great idea. Was glad I didn’t have to ‘sell’ it in a referendum, two Xs was much easier to explain and cut out the ‘confusing and complicated’ argument.

    Even the Open List will lose some supporters because we know it won’t likely have as good a result in electing women and other underrepresented groups. If we legislated that there had to be gender parity on Open Lists that would be better. Closed List could be viable in a non-referendum scenario if it was legislated that parties had to choose candidates (whether local or list) in a secret ballot process by party members and that it was illegal for party leadership to try to alter or reorder the list like in Germany!

    It is almost like the rigours of a referendum are dictating the specifics of a PR system.

    Tony, the simplified model of STV that you are talking about is that Local PR or variation of Local PR?

    Robert Jarman

    One idea would be to call a citizen’s assembly on it, hold a couple elections, and then 8 years later, we hold a referendum on whether to keep or ditch it. The referendum after a certain period of time is something that NZ did in 2011. NZ also had a referendum asking people what kind of voting system they would want if MMP was to be ditched, between FPTP, instant runoff, parallel voting, and STV. I don’t think they had the latter part ranked, which is a bit odd. Incidentally, FPTP did get a plurality of the latter question.

    I prefer citizens assemblies and elected constituent assemblies for this reason, studying the things is much more possible without accusations of partisanship. We didn’t do either when Canada was amending it’s constitution back in 1982, and we didn’t use them either when we were debating reforms in the 90s with Meech Lake and Charlottetown.

    Referendums are supposed to be used when politicians flat out refuse to make a decision or some power is so entrenched that there is little hope of getting something done without a referendum, or when confirming or denying specific options, like in Ireland how a constitutional amendment requires a specific amendment must be proposed by the Dail or Seanad, gets passed by both, and the population votes on that proposal and not a vague statement. Part of why I disliked the Brexit referendum, as there was no statement as to what that even meant. Constitutional amendments usually also require amendments to be adopted by referendum if they are possible, preventing politicians from flip flopping on the issue by debating whether or not to put it to a vote.

Viewing 15 posts - 1 through 15 (of 17 total)
  • You must be logged in to reply to this topic.