IRV-MMP

What do folks think of this idea, proposed by Mark Roth in the thread on open-list MMP?

I do not believe it is entirely necessary to have two votes; though I don’t oppose the idea. Essentially I would have IRV-MMP. An instant runoff determines which candidate wins the local seat in each district. First preferences determine who receives the at-large seats. If a voter wants the Greens, but knows that they won’t win locally, a vote 1 Greens 2 Labor has the effect of supporting a winnable local candidate and helping the Greens secure seats in general. I would allow transfers to second (or lower) ranked parties should the first choice(s) of parties not reach a threshold. I would also be inclined to allow a List Party that isn’t running a candidate to appear on the ballot anyway; probably marked to indicate that the List cannot win the local seat. The candidates who lose in their local race would be selected to fill the at-large seats based on their personal vote counts. List order would only be a tiebreaker.

Decoy lists would technically be possible, but they would stick out like a sore thumb, require voter coordination to ensure that the “right” candidate gets the vote in the district level races, and would still need to front candidates in local races to have enough warm bodies.

As I say at the other thread in a comment of my own, I like it much better than the “AV+” idea of having two votes (one ranked-choice for local candidates and the other for list).

35 thoughts on “IRV-MMP

  1. I have two concerns. First, this is relatively easy for election methods geeks to explain to each other, but it might turn out to be hard to explain to rank and file voters. Having a separate vote for a party is fairly transparent, at least once you get used to the idea of voting for a party rather than a person. But explaining how the same first preference can contribute to the election of both a district representative and a compensatory list representative might be a chore. Can you prevent skeptics from arguing that some voters get two bites at the apple while others get one? With respect to IRV, we are pretty skilled at rebutting that bogus argument. Is the rebuttal just as clear with respect to IRV-MPP?

    Second, I think this might be hard to sell to “independent” voters and candidates. Mark addresses this in his description but to me he sounds a little cavalier about it. The myth that government and politics can be “non-partisan” is widely believed in the United States, making it necessary for any electoral reform to presented in a way that treats independent candidates and voters fairly. I can think of ways to make list systems (including the list part of MMP) fair in this sense. I can’t figure out how IRV-MMP would do so.

    • When combined with list MMP, IRV really comes with 2 votes, for a local candidate and the candidate’s party, that may split in certain circumstances. Thus, it will be very hard to make the case that no voter gets two bites at the apple: decoy lists still are feasible and not much harder to manage than in standard 2-vote MMP (just visit an SNTV jurisdiction).

      When looking at linkage between vote(s) for district candidate and preferred party, the basic choice in MMP is “freedom enabling fraud or fetters enforcing fairness”.

      • I’m not saying you’re wrong. I think, for the U.S., decoy lists and “Independent Republicrats” would be a problem if MMP were ever adapted and solutions are needed. But my thought with IRV-MMP would be that a decoy list and its partner list would end up splitting the vote if they ran two candidates in every district. If they voted for the real party, so its guy would win the local seat, the vote would never transfer to the decoy list. If they put the decoy candidate first, either the decoy candidate would win, reducing a decoy to a two-party alliance proportionally represented, or another candidate would win, and both “parties” would end up needing compensatory seats.

        How does a single ranked ballot allow a party to gain unearned extra seats?

    • I understand the concerns of independent candidates, though I don’t care (and don’t really buy) the claims of candidates who will be “independent” if elected while still having an R or a D next to their name.

      I would allow actual independent candidates to run as one man lists. As a list, they would appear on every ballot in every district. They would only be an official candidate in one district, but if they don’t win that district but accumulate a minimum number of votes overall, they would be awarded a seat. My “threshold” would be 5% of first preferences, 5% of local seats, 3% of first preferences and 3% of local seats, or be a one candidate list that won 1/Nth of the vote, N being the number of local seats.

      For partisan candidates, the appeal should be advertised as the ability to vote for whoever you want. Maybe you’d only get your second or third choice elected locally, or neither, but you can help get an ally of your first or second choice into office. It would require a campaign to change the way people think. But many Americans seem glued to to the idea every seat must be contested separately and whoever gets more votes than the others wins. To the point that a failed candidate in Maine is suing because he claims the Constitution requires first pas the post. If we’re going to have to make big changes for any system, we can free ourselves from assumptions of “independentess.”

      Or we can just have STV-MMP and let the same ballots seat 5 people per district and few top-up candidates.

      • A decoy list could be employed in IRV-MMP whenever a party expects to win so many district seats that it won’t be eligible for any top-up seats. (Admittedly, this is more limited than the possibilities in standard 2-vote MMP.) Party A creates a dummy party X. Its supporters with January-July birthdays are told to rank candidates A, X, … and supporters with August-December birthdays to rank candidates X, A, … In each district candidate A gets more votes than X and so party A will win just as many district seats as it would have without the decoys. The August-December voters earn top-up seats for party X.

        The maximum payoff in IRV-MMP would be reduced by slightly more than in Bavarian MMP. Probably that would be enough to inhibit the practice in New Zealand (but maybe not in Italy or Lesotho?).

  2. It’s an interesting idea but I don’t think it is viable. All it means is that the major parties have a huge advantage winning the single member districts and the minor parties are restricted to the 50% of seats in the list election. It would not be proportional at all.

    I am unclear how list seats are to be allocated, but if they are to be decided on first preferences then you have a nightmare in electoral education, I have no idea how would explain to electors that they have a fully preferential vote in the district seats that is also a non-preferential vote in the proportional seats. STV has a quite well-developed algorithm for deciding which candidates are elected. Superimposing a non-STV list tier on an STV district tier leaves you with the disadvantages of STV without any of the advantages and a result that will not be proportional at all.

    • The same, ranked vote would be used twice. Within the district it would determine a winner with 50%+ support. Major parties would have an advantage whether the local races are FPTP or IRV, that’s just a fact of their larger support. All ballots would pool together to fill up the remaining seats proportionally.

      A ballot A,B,C,D may help candidate C get elected for a local seat, though C clearly is the voter’s third choice, but still preferred to D and Ed. If Party A doesn’t meet the threshold, the vote will transfer to Party B at the national/regional level and help that party get a seat. Any at-large seats would be awarded to create proportionality, it would not be parallel.

    • Surely major parties already have a huge advantage in winning single-member districts, under any form of MMP? I don’t see how this proposal changes that, and letting party votes be transferred away from parties below the threshold seems like a very elegant mechanism for combining the two votes.

      My broader objection to this proposal is to do with the idea of meddling with MMP to avoid the possibility of decoy lists. If this is a proposal for the US, PR already has the problem of the ban on House districts that cross state boundaries, and MMP makes this problem considerably worse. Taking away the two-vote mechanism, or making it way more confusing for the ordinary voter (as IRV-MMP would do) makes the single-member districts much less worthwhile, and if political norms aren’t strong enough to prevent decoy lists forming, just avoiding MMP and going for a simpler list or STV system seems like a better option.

      • There’s no mention of the list tier as compensatory and no mechanism is offered to make that tier compensatory.

      • Surely the mechanism would just be the ordinary mechanism for MMP-allocate every seat by PR, distributing second preferences of parties below the threshold, and then subtract the number of district seats won by each party to get their number of list seats?

      • The proposal does not state that the list tier is compensatory and offers no mechanism for making it compensatory. ‘Surely’ does not cut it. Using a limited subset of STV ideas without actually losing the whole of the STV algorithm puts us back with a vote that is fully preferential in districts and not fully preferential in the list tier. What you end up with is IRV/non-STV.

      • The term used for a mixed-member electoral system where there is no compensation in the list tier is mixed-member majoritarian, which Mark does not use. Instead, he describes the system as IRV-MMP. Under such circumstances, I don’t think it’s particularly weird to assume that such a system uses the compensatory mechanism used in Germany and New Zealand, two textbook examples of MMP, or at least *some sort of compensatory mechanism*. Ending up with IRV/non-STV is kind of the point of this proposal, and is also perfectly compatible with a compensatory mechanism being used to distribute the list seats.

      • I did directly call it IRV-MMP because the list seats would be compensatory. I just didn’t spell it out. Every party above threshold will be entitled to a number of seats, and would receive seats for their best losers if they did not win enough districts. Every vote for a below threshold party would transfer to a qualified party, if the ballot shows a preference for one.

  3. Alan, it is in fact preferential for the list tier: “I would allow transfers to second (or lower) ranked parties should the first choice(s) of parties not reach a threshold”. I’m also not sure why you say this wouldn’t be proportional, I’m assuming if we’re calling it “MMP” it means the list tier would be compensatory.

    I’ve always felt uncomfortable with thresholds in PR systems, since it preserves the need for strategic voting. This seems like a reasonable way around it—though you could also just get rid of the local vote entirely, and have “Preferential List PR”.

    • That is why I qualified my comment. The proposal both that list seats are decided by first preferences and then mentions a series of exceptions to that rule that are somewhat unclear. The exceptions do not amount to actual STV. That could just be a problem with the way the proposal is expressed.

      Some time ago I proposed a facially similar model where the list tier would be compensatory. You would transfer wasted votes, those that did not contribute tot he emotion of district candidate, into a common pool that would be counted according to the standard STV rules.

    • Then what is the compensatory mechanism in the proposal? You may care to address the fact that the proposal effectively hives a major party supporter 2 votes while a minor party supporter gets only 1 vote. Fred votes 1 Labor 2 Green in the district of Upper Middle Bogan. He gets a Labor member. If his vote then goes into the list tier it gets recounted a second time. Dag votes 1 Green 2 Labor so his district vote does not elect a member and gets counted only once, in the list tier.

      If the compensatory mechanism is that Fred’s vote does not get counted twice, then the proposal is identical with one that I made at this blog some years ago by which only wasted votes would go into the list tier. I have been unable to find the comment but I do remember that Ed and i discussed it at some length.

      • As I understand it, the system works like this. If we assume Labor has a first preference majority in the district of Upper Middle Bogan, then it’s true that Fred’s vote elects a member and Dag’s does not. This is how single-member districts work.

        However, under MMP, all the votes for parties are tallied up nationwide, and each party is allocated a total number of seats in the legislature based on its proportion of the vote. Dag’s vote may or may not transfer to a larger party for this tally, depending on whether there’s a threshold and whether the Greens pass it. The number of single-member district seats won by each party is then subtracted from their number of seats in this allocation. In the nation/region-wide list count that determines how many seats each party holds, both Fred and Dag’s votes count exactly the same. If you replace ‘votes for parties’ with ‘list votes’, that is how MMP works in New Zealand.

      • If I close my eyes and click my heels together 3 times, I seem to remember previous attempts to marry transferable vote with a list system. That attempt encountered the slight problem that electors were completely unable to relate the outcomes to the way they voted. In this system, if the proposal is as you say, electors have a vote that is fully preferential in Upper Middle Bogan and may or may not be even slightly preferential depending on a series of rules about thresholds that are unknown in transferable vote.

      • The rules appear so unclear because there are a number of options for the system. If it were to be implemented, a particular one of those options would be chosen. You could have first preference votes automatically counted as list votes. You could transfer votes of parties below the 5% threshold. Either way, I’m unconvinced the system is quite so complex: explaining to voters that “for your constituency, your preferences will be transferred until one candidate gets a majority, for the party-list seats, your vote will be transferred to your later preferences if the party you vote for first doesn’t get enough votes” doesn’t seem too hard to me.

    • I have used the term “Preferential List PR” to refer to the existing classes of list-PR systems in which the voter can cast one or more preferences votes. Perhaps I should not have used that term, but it is in print. So I want to nominate “Ranked-Choice List PR” as the name for a system like the one Vasi describes in which votes can transfer from one list to another.

  4. I prefer STV myself, but I put up a pet MMP proposal earlier on this site. I think Mark Roth has a good proposal and the approach is similar, though I think mine might be easier to explain to voters.

    Essentially the country would be divided into a tier of multi-member electoral districts and a second tier of single member districts, with each multi-member tier including several single member districts. Parties would run at large candidates for the first tier and candidates in the single member districts. Votes cast for candidates in the single member districts would also count for candidates from the same party/ list in the associated multimember district, but if a candidate in a single member district was elected, the votes needed to elect him or her (essentially the runner up’s total) would be deducted/ not counted towards the party vote in the multi-member tier. Its not really proportionate, but it means that minority parties would rarely be shut out of representation of a region by losing all the single member seats, unless they lost all of them by huge margins and had little overall support in the region.

    I had pluralities determining who won both the single member and multi-member electorates. I suppose IRV could be added but I’m not sure if its worth further complicating the system.

  5. A concern I have is the following. Would surplus votes for the district winners be transferred to each voter’s next preference, thus meaning that this would switch parties for the proportional allocation? How about district votes from eliminated candidates, if their second preferences went to a losing candidate in the following rounds, would the first preference party get those votes back for the proportional tier?

    • Derek, as Mark stated above, “The same, ranked vote would be used twice”, with first preferences aggregated nationally (or regionally) determining overall distribution of seats to parties, and all necessary preferences counted to determine single seat district winners.

  6. In agreement with much of what has been said above, I think this proposal is elegant, but probably too complex to be entertained outside this orchard (perhaps with the exception of Australia, where people are already so accustomed to everything to do with IRV in general, so saying something like “your first preference will be used to determine the parties’ share of seats” might be relatively easy to relate. Maybe.).

    But I also completely agree with Henry’s comments regarding the applicability of this to the US, specifically at the federal level. Certainly, it’s a creative (and, as I said, elegant) answer to the problems of decoy lists – as well as to the the drawbacks of regular, nonranked, single-vote MMP. But MMP would always have to be regional for the US House. So my practical conclusion is exactly as Henry put it: “if political norms aren’t strong enough to prevent decoy lists forming, just avoiding MMP and going for a simpler list or STV system seems like a better option.”

    • Most Australians would see this as a repeat of that other system where votes were transferable at some stages of the election and non-transferable at other stages of the election. Modified D’Hondt was used at 2 elections. After experiencing the system in 1988 a public consultation process for its replacement was already complete by the 1992 election. Modified D’Hondt and earned such universal opprobrium that it was not even on the ballot for the indicative referendum in 1992. Hare-Clark prevailed over single member districts by 65.30%. A referendum to entrench Hare-Clark was carried by a similar margin in 1995.

      A major factor in the rejection of Modified D’Hondt was that it required every ballot to be counted at least twice. This meant that the count took over 7 weeks at the first election.

      I would think Australians’ experience with a fully preferential vote makes us even less likely to adopt this system than Americans. Modified D’Hondt was described as an elegant system in its time. But that was before anyone had actually used it.

      • Alan, that’s just silly. I see virtually no similarity between the systems. Ballots would not even have to be counted twice. Is the ACT’s experience with D’Hondt-STV even that well-known?

      • jd

        I do not propose to comment further on this matter for reasons which must surely be obvious to you.

  7. What if a 2 Round System is married to a List system? It would be a 2 Round MMP system. The first round could just be only for single member districts, a candidate is elected outright with 50%+1, and the two round, single member districts just where the top two candidates go on from the previous round if they didn’t win a majority, and the party list vote. There could be variations and modications of this, one could have a two round MMP system where in the first round, the list is open, and in the second round the list is closed. There could be problems with such a system of decoy lists.

    • I’ve thought about that, but it contains potential issues, especially for areas not used to multi-round voting…

      If turnout differences affect participation (one round has more turnout than the other) the proportionality can be thrown off; essentially one electorate sets up the runoffs and a different one determines PR vote. This could be exacerbated in areas without a runoff in the second round; how long will it take people to learn to show up if they’re only voting for a list.

      There is a the problem where second round votes can be influenced by the first round. Party A wins enough seats in the first round? Vote for likely ally Party B. Party A faces a negative vote weight? Go to anyone else. Two allies face off in a runoff, vote for the guy who’s party is less likely to win big.

      The biggest issue I’ve foreseen is the potential for a pointless runoff. It may not happen all the time, but what effect will there be when everyone knows that both candidates in a second round are literally guaranteed to be seated. Will they turn out for an effective list only vote? Will it make people take the system less seriously.

      Doing it all on one day seems so much neater.

  8. There are several advantages in using FPTP for the SMD tier in 2-vote MMP: one-to-one relationship between an elector and her/his representative; small district size; the method of selecting a winner can be understood by most six year olds.

    The two big problems with MMP are overhang and false-flag candidates (decoy lists, fake independents). In replacing FPTP with IRV we lose its brute simplicity without solving the false-flag problem. A more fruitful approach is to re-assign the “party” component of split votes whenever that would reduce overhang.

    To Illustrate, consider a region having 10 single-member districts and 6 top—up seats. The local candidates of Party A receive 50,000 out of a total of 100,000 votes and win 7 of the SMD seats. Party A got 40,000 party votes, not the 43,750 for a 7/16 share of 100,000 votes. The prescription is to transfer party votes to Party A from parties that benefited from split votes. In this example, a 37.5% tax would be levied on those 10,000 split votes, bringing the Party A total up to 43,750 party votes.

    This would completely nullify decoy lists and fake independent candidates, but could leave residual overhang. An unfortunate feature is that in blocking “criminal” vote splitting, the rule would misconstrue the intent of many “innocent” vote splitters.

    • Really, the decoy list problem goes further than just actual manipulation, doesn’t it? If SNP and Greens got all their voters in Scotland to vote SNP for the constituency but Green for the list, you’d have effectively the same problem as if they ran decoy candidates/lists, wouldn’t you?

      • Quite true. MMP provides strong incentive for two parties like SNP/Greens to drift into such a symbiotic relationship. It may start innocently with individual voters maximizing their electoral clout, but turns into systematic abuse (if your concept of MMP is that an elector’s 2 votes should affect the seat total of only 1 party).

        The incentive for Green supporters would be inhibited by the threat of having split votes reassigned, as I proposed on 20/11/2018. The idea gets complicated when you stir in Independent candidates, thresholds and multiple overhang parties, though — technically manageable but tricky to explain to the public.

  9. I would use STV for the seats. Then rank the candidates by % of vote gotten to allocate the top up seats.

    STV means that if your party gets x.9999 of a quota then then excess votes still get to pick a candidate.

  10. Hi,

    If I understand correctly, this proposal would switch the ballot structure in a MMP from dividual to ordinal, using Gallagher’s & Mitchell’s terminology from their chapter in the OHOES.

    Regarding this idea, I see mostly discussion on how the change in the system would have an effect on the individual-level. Like maybe a voter’s idea on whether they were able to vote in accordance with their preference or, on the other hand, whether they get confused by it or not.

    I am a little out of the loop as I just found this blog and am only a student myself. But I am interested in these ballot structures, and have been wondering which varieties in ballot structures have which effects on both the individual and macro-level (if there are any).

    I would for instance say that, with an ordinal or dividual ballot, the vote you give at one election is either not, partially, or completely the same as the vote you gave in the previous election. With a categorical ballot, your vote is either the same or not. Wouldn’t this ability to have a vote that deviates only partially from the vote you gave before add up to aggregate vote volatility going down?

    For instance, a voter for party A has to switch to party B with a categorical ballot if this person’s preferences change. Dividual or ordinal ballots would allow a voter to abandon A ‘partially’ by giving e.g. a local vote to A but the list-vote to B in MMP (presuming both votes went to A previously).

    If voters switch votes gradually, should vote-volatility logically be more gradual too? To me it seems logical and I suspect I am not the first to have this line of thought, but I have trouble finding theories / findings on ballot-structure effects on system level stuff. Has such a link ever been documented?

    • Dividual or ordinal ballots would allow a voter to abandon A ‘partially’ by giving e.g. a local vote to A but the list-vote to B in MMP —> maybe the panachage system in Luxembourg is a better example

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.