Canadian electoral reform ideas

Ideas for replacing FPTP with some form of PR have been floated in Canada many times in the past, but so far no serious reform process has gotten underway.

This month, the former leader of Canada’s Liberal Party, Stephane Dion, has advocated a new system. He calls it P3, for “proportional-preferential-personalized vote”.

As best I can tell, this would be an amalgam the likes of which we have never seen before. District magnitude would be 3-5, and voters would undertake two voting steps:

    1. They would rank parties in order of preference, and a process apparently akin to single transferable vote would be followed to determine how many seats each party would win in the district.

    2. The voter could cast a single candidate preference vote (non-transferable, it seems), and these would determine which candidates would win the seat(s) each party was entitled to after the completion of the phase of party-level allocation.

In other words, it is party-STV-open-list PR!

Meanwhile, Benjamin Forest, a geography professor at McGill, has advocated other solutions to get “effective representation for national minorities“, by which he means French speakers outside Quebec and aboriginals. He proposes either separate voter rolls or minority-majority districts. The first of these ideas is akin to what New Zealand practices for Maori voters: separate districts for the minority, with voters of the minority group eligible to vote in those separate districts. The second idea–which Forest appears to prefer, given “difficult legal issues” with the separate rolls–would be the affirmative gerrymander widely practiced in the US.

Both of these districting concepts strike me as highly retrograde. As for Dion’s proposal, whatever one might think of it, one has to give him credit for originality.

36 thoughts on “Canadian electoral reform ideas

  1. Aaaaaaargh! It sounds like the notorious ACT modified d’Hondt system, which was a disaster.

    I think personalised proportional preferential could work, but only if you used the principle of pooling wasted votes across the compensatory tier instead of using a second vote.

  2. Credit for originality, yes, but from the observation of how difficult it is to promote STV to electorates used to FPTP, for me this just reinforces the perception that Dion … lacks the common touch.

  3. Dion’s proposal seems to be a disguised version of STV. Why not just advocate STV?

    Actually there are ground to think that STV is particularly suitable to Canada; big regional differences, tradition of local MPs, robust multi-party system, and distinct minority groups. But the comments to the article are pretty depressing and indicate why Canada won’t get proportional representation any time soon.

    You can frame an argument against proportional representation for the federal parliament, however, on the grounds that the existing voting system produces most of the positive results that you would get from proportional representation. Five parties are represented in the legislature, and parliaments with no overall majority are fairly common.

    I agree that the racial gerrymandering as practiced in the US is pretty atrocious and should not be imitated. I actually don’t mind NZ style separate electoral rolls as much as I understand inclusion on the separate roll is voluntary on the part of the minority group members. Currently in Canada, electoral districts are provided for First Nations areas that have much smaller populations than normal, guaranteeing representation from this community but violating the principle of one man one vote. There is a strong argument for replacing this arrangement with a Maori roll type system. But I don’t think this has to be done for anglophone and francophone minorities.

  4. Reformers who want to replace FPTP with a form of proportional representation have made no progress partly because the discussion has been about the best form of PR, and how all our political problems can solved in one revolutionary change. Revolution is scary and breeds resistance.

    What is needed is a form of PR to replace FPTP. Reformers should think more about a form of PR that looks to be very similar to, and is as simple as, FPTP.

    I am advocating a form of PR incorporating a non preferential voting system that works with the existing system of single member districts/constituencies.

    MMP/AMS has simple voting, but still needs constituency boundaries to be redrawn, and alienates some because of the system of List and Constituency MPs.

    Direct Party and Representative Voting (DPR Voting) is a form of PR incorporating a non preferential voting system that works with the existing system of single member constituencies. Voting and counting is very simple. Introducing it would involve only the smallest change to the FPTP electoral system.

    It would achieve a form of proportional representation at minimum cost and with minimum disruption, while maintaining a degree of political continuity that cannot be achieved by other systems.

  5. In the Canadian context (or anywhere PR is already on the table), I agree with MSS and Ed that racial gerrymanders are a bad thing. But I feel differently about current practice in the U.S., precisely because PR is not in the table here. Opposing racially-conscious district boundaries will not hasten the arrival of PR. Until it arrives, the legacy of slavery and Jim Crow is still too recent and too pervasive to accept anything else.

  6. Its drifting away from the topic, but I disagree with Bob Richard.

    I view racial gerrymandering as the political counterpart behind affirmative action. Much of affirmative action seems to based on the idea that some members of racial minority groups can’t get jobs since nepotism and prejudice plays a greater roll in the job market than merit. So we will basically carve out an additional non-merit based job set-asides for racial minorities, on top of the existing nepotistic set-aside for connected members of the racial majorities. This at least advances some members of racial minorities and may be better than what we had before, but its kind of a crappy solution to the problem of people using every consideration but merit in hiring decisions. To the extent that working class white employees were replaced by middle class minorities, it may have worsened income inequality.

    So politicians crack and pack racial minorities in order to create districts that are tailored to certain incumbent politicians? Fine, we will mandate that they also gerrymander districts tailored to racial minority politicians as well. This may address one problem, the low number of elected minority politicians, while making another problem, gerrymandered districts that cut across communities of interest, even worse.

    And if members of racial minority groups actually manage to move into integrated areas, the system doesn’t know how to handle that. In the northeast and midwest the majority African American district lines have been getting more ridiculous as African Americans spread out into the suburbs. On the other hand, where housing remains segregated, its not difficult for an independent commission to create majority minority districts on purely community of interest considerations and that is where the need is greatest.

  7. Majority-minority districts do have one advantage over separate rolls or reseved seats for minorities, in that minorities can choose to not elect a member of the minority, if they so wish.

    Of course the problem being addressed by Min-maj districts is the historic reluctance of more than a minority of white voters to consider voting for minority candidates, regardless of their merit.

  8. The idea behind minority-majority seats in the US is not affirmative action in the candidates elected, but rather giving minority voters the ability to elect their ‘candidate of choice.’ Indeed, in three of Texas’ majority-Hispanic districts, Latinos overwhelmingly supported non-Hispanic white men in 2012.

    They are only created initially where minority populations have the geographic concentration to consist of a majority of the citizen voting-age population in a district that isn’t unnecessarily gerrrymandered to do so AND the must have consistently voted in a cohesive manner for specific candidates (regardless of their race) and seen those candidates blocked by an Anglo majority voting against them.

    Once established, they then cannot be reduced in number unless the minority also reduces in number, because in doing so, this would violate Section 2 of the Voting Rights Act, which prohibits diluting the voting power of a minority group which historically has had limited access to voting rights (those who faced literacy tests, poll taxes, property ownership requirements, and lack of availability of ballots in their language in particular).

    While I feel proportional elections (and in particular, STV elections) would overcome this without using specifically-drawn ‘opportunity’ districts, we are nowhere near a situation where the partisan legislatures draw maps which are fair to minorities (just look at the number of maps denied preclearance each year).

    I personally wouldn’t be opposed to voluntary minority rolls for separate seats, but there is no chance that would happen in the US. It’s also a matter of the fact that ethnicity and race on the Census are determined by the respondent, so the rolls would almost certainly have to be open to anyone who wanted to register for them and not just members of that minority group.

  9. The Maori electoral roll in New Zealand is only open electors of Maori descent. There is a definition of Maori descent in the electoral act. I do not think there is any history of non-Maori trying to enrol fraudulently. The number of Maori seats is not census-dependent in the sense that the electoral officer merely applies the same quota to both electoral rolls. There would obviously be complications in a federation unless you simply made the indigenous seats separate from the state delegations.

    It is almost tempting to disregard the population numbers and simply treat the indigenous roll as if it were a state or province and seat indigenous senators in the senate.

    Hopefully the indigenous delegation would be elected by PR rather than uninominally.

  10. In Canada, the current ‘indigenous’ seats are two whole territories, Northwest Territories and Nunavut.

    I think that an indigenous roll could work in the USA (with eligibility restricted to voters enrolled in a federally recognized tribe or to holders of a Certificate Degree of Indian Blood) , but there are very few states where the indigenous population is high enough to merit a seat (Alaska and the Dakotas have only a single seat each, New Mexico only 3, and Oklahoma only 4; Arizona is the only state where it could be a possibility).

    However, for the minorities that currently qualify for ‘opportunity districts,’ that is African-Americans and Latinos, there is no federal register of ancestry. Traditionally anyone with a ‘single drop’ of black blood was discriminated against as black, and indeed, at least one black Congressman by all appearances looks like he is ‘white.’ Latinos are not a race, but an ethnicity, and are all colors of the rainbow, from blond hair and blue eyes to pure-blood indigenous to African to Asian.

    The way it is determined whether these groups get seats is based on the number of people who indicated that race on the census, but there is no way to measure whether a person claiming to be black or Latino actually has that heritage; therefore any ‘minority roll’ would have to be open to any voter who claimed that heritage.

    One way for Democrats to avoid being gerrymandered would be to encourage white Democrats to indicate they are racial minorities on the Census, therefore guaranteeing them representation under the Voting Rights Act. Aside from being dishonest, immoral, and distorting the purpose of the districts, thought, I also feel that any coordinated campaign to get people to lie about their heritage on the census has to be a federal crime of some sort.

  11. The better alternative may be to let people establish their own separate rolls, subject to voting only once and a uniform quota. There would be no issue with definitions if you elected to enrol in the Mixed Irish/Zulu Flat Earth Roll because the ethnos would be self-defined. As an aside, I am acutely uncomfortable with the blood quantum rules that the US uses to define Native Americans.

  12. The blood quantum rules vary by tribe, though I believe one must have at least 1/8 blood quantum to qualify for federal affirmative action programs. The Cherokee Nation, the second largest tribe, simply requires one ancestor to have been listed on the Dawes Rolls of Cherokees, which were compiled almost 150 years ago. Therefore, some tribal members may be of less than 1% Cherokee ancestry at this point; the current Principal Chief is only 1/32 Cherokee.

    I wouldn’t at all mind the ‘form your own roll’ rule. The rolls could be finalized 30 days before the election, with seats assigned to the rolls proportionally, and the members of any roll with less than 1 seat’s share reassigned to the ‘general’ roll for that election. It would certainly be an unorthodox form of PR (and indeed, I wouldn’t be surprised to see partisan-based rolls form were this system adopted, particularly by minor parties), but not necessarily less valid than any other form of PR. Aside from NZ, it reminds me of the confessional rolls found in Lebanon.

  13. Alternately, a more realistic rule in Texas at least might be to have a ‘Spanish language ballot’ roll, with a number of candidates based on registration and candidates required to be able to speak Spanish to contest that roll’s seats. I expect many English speaking Latinos would elect to register on the Spanish roll for increased representation.

  14. I am not offering any opinion on #13, but I would love to sit back with some popcorn to watch the lawsuits and exploding heads when that idea reaches the public imagination.

  15. I’ve been reading about the modified d’Hondt system Alan mentioned @1, and it does indeed seem to be a disaster. I’ve never heard of an election taking 65 days to count that didn’t involve extremely close races and disputed ballots.

    Dion’s system differs from ACT modified d’Hondt in using OLPR to determine who gets a party’s seats, and in not using group voting tickets. Optical scan ballots (which I’m not sure whether they use in Canada or not) would mean that the order in which a party’s candidates would be elected would be known before the number of seats a party gets is known, so it’s simply a matter of transferring those below the quota by party. It would be slower in that it does not use block exclusion for all parties below the threshold, as the ACT did.

    My biggest gripe with Dion’s proposal is that a person must transfer his vote to a party without having any say in which candidate is elected. I would prefer in a party-preferential system that either a closed list be used, or else that a voter may choose one candidate per party, and his “personalized” vote (to use Dion’s phrasing) would be counted for whichever party his ballot eventually counted for.

    Of course, because Dion uses such small districts, I don’t see any reason not to use traditional STV. Were they electing province-wide, with 10+ seats, I think a party-preference system would be more effective.

    The NSWLC system has for all intents and purposes become a party preferential system. They probably could make the count go a lot faster if they just abolished below the line voting, but that would take a referendum, and I don’t know if such a referendum would pass. I don’t believe any ungrouped candidate has been elected to the Legislative Council under STV, and I’m pretty sure no candidates have been elected while someone ranked higher on their party’s ticket has been (though the exclusions are routinely out of order; at the last election, Elaine Nile of the CDP-Fred Nile Group was the last candidate from that party excluded despite being #5 in ballot order).

    In any other group voting ticket system, it would be possible for candidates to be elected out of order (for instance, if some of those parties who rank the Greens 6-5-4-3-2-1 in the Senate transfer to the Greens, it might be possible for the #3 to pass the #2 and end up elected).

  16. Dion’s unusual proposal is NOT STV, because it has no surplus transfer. It’s simply a preferential party vote, with pure open list, in five-seaters.

    Its interesting element is to open a debate about district magnitude, and what he calls “moderate proportionality.” The term used by Professors John Carey and Simon Hix (2009) is “low-magnitude Proportional Representation” with districts in the six to eight range.

    The debate continues.

  17. I think the 2nd BC-STV referendum showed us that Canadians aren’t likely to support the huge geographic ridings necessary for STV or the Dion pitch; you’d have to have all three territories as a massive 3-member riding for STV or P3 there. Of course, since at the federal level, a much larger percentage of people didn’t vote for their “local” MP, the may be much less attached to the concept than in BC.

    Honestly, I see absolutely no reason to use Dion’s model and not real STV, but it’s possible his “P3” could be more palatable to the three major parties–particularly if he ditched single-preference open list, which just functions as a less-predictable form of Australian-style group-ticket STV, and moved to closed list instead).

    • Of course, the first BC-STV referendum showed that voters (at least in BC) might approve of large multi-seat ridings by an overwhelming margin! (Almost 58% yes, but 60% required under BC law; NZ’s MMP was approved with “only” 54% in 1993, by comparison.)

      Overall political context likely matters far more than the details of the reform system being proposed.

  18. They each have less than a seat’s population, so you’d either have to use AV or combine them into what would have to be the largest,constituency on earth.

    Honestly, though, Canada’s 3.5 party system (outside Quebec) might be able to survive with 2-seat ridings. It wouldn’t be easy to pass those, but easier than hetting all of northern Alberta to be one 7-seat riding.

  19. I think in earlier incarnations of Westminster history, two seat constituencies were standard.

    One halfway to PR variant would be to use two seat constituencies. The top vote getter in each constituency would always take the first slot. The second slot would go to the top vote-getter in the constituency representing the party who has received the most votes in the province/ county, from among a pool of parties, whose number of seats won in the province/ county (at that point) is less than the share they would have gotten proportionate to their vote. You would need a mechanism to determine the order in which constituencies received their “additional member”.

    I still prefer STV.

  20. @MSS, speaking of context, one article/blog post I read (I cannot remember where) suggested that in the first referendum, based on the phrasing of the question, many people were expressing approval of the concept of the Citizens’ Assembly and of its work rather than the specific proposal.

    It also mentioned that no riding maps or divisions had been provided, and that once they had been provided for the 2nd referendum, many voters soured on the idea.

    I do agree that the 60% standard was ridiculously high. I feel that it’s somewhat ridiculous to ask that a majority approve of a voting system designed to protect minority representation for it to be implemented. Of course, I also feel that in certain cases (like California Prop 8), the fact that a narrow majority of those who bother to turn up can overturn a law granting a civil right is also quite ridiculous, so I’m conflicted. In general I think referendums tend to protect/promote special interest groups and not necessarily the citizens’ rights they are intended to, and they also somewhat defeat the purpose of having representative democracy in the first place.

    @31–in the UK, there have been several variations of two-seat or larger constituencies. Wikipedia ( has a good article on the history of multi-member constituencies.

    By 1918, all multi-member constituencies except for certain university constituencies were abolished. From 1918-1950, university constituencies with two or more members actually used STV, but in 1950, the university seats were abolished.

    Another potential variant of two-seat PR would be using the Chilean system in fairly apportioned constituencies (if one can overcome the Pinochet connection, though that might actually bring some Thatcherites over to the idea). Of course, STV would be fairer than that, even in two-seaters (the largest party needs 2/3 of the votes to win a second seat, not just double the votes of the second party).

    On a side note, because it elects only 2 members, the Italian Senate region of Molise essentially uses the Chilean system as well.

    I think the idea of tying the “additional member” to a specific constituency could make it quite a bit more palatable, though I don’t know if parties would like the idea of a full half of seats being top-up seats rather than the 1/3 used in Wales or even the 15-20% Jenkins proposed.

    I think one way to construct the mechanism would be to use d’Hondt to determine total seats, and then allocate extra seats in order of most votes received, with all of a party’s remaining candidates excluded once it has reached its seat quota. Of course, this means that if the Greens or UKIP get a seat, they’re stuck in the leftover seat and not necessarily the one where they performed the best.

    I too still prefer STV, but one major issue that I’ve been noticing is that STV districts have to have high magnitude to be proportional to first count votes (especially if people don’t tend to transfer between parties, as is the case in Malta). However, larger magnitude brings to big issues in convincing people to adopt STV: it makes the count much longer and more complex, and it makes the area of each district larger, which may make it more difficult to convince people to replace FPTP with STV in the first place.

  21. @Ed

    Until the Third Reform Act 70% of the House of Commons were elected from multimember districts.

  22. The only working model of STV that produces reliably proportional results satisfactory to most voters and most significant parties is Northern Ireland’s, with all six-seaters. They had watched STV at work in the Republic of Ireland next door for generations. In fact, the nature of life in Northern Ireland is to watch the Republic, with either admiration or repulsion, but they watch it more closely than anyone else. And after all that, they all decided five-seaters are too small; six-seaters were necessary to accomodate Northern Ireland’s political diversity.

    In Canadian federal elections, with about 100,000 residents per MP, that’s STV districts of 600,000 residents. Only nine metropolitan areas in Canada are that large, containing 51% of Canada’s population. The other 49% would never buy it.

    Canadian geography requires at the federal level, inescapably, a mixed compensatory system. That’s what the Law Commission of Canada concluded after a lengthy study by many experts and 16 public hearings across the country. The geography has not changed, nor has the conclusion. The details will require more consideration, but the model is clear.

  23. A lot of the opposition to STV districts being too large involves (a) forgetting that we are talking about 3 to 7 MPs servicing the area, not just one, or (b) remembering that but assuming that they will all come from the single most populous town or area within the district (because this is what often happens under MNTV).

    I think this assumption is mistaken (up there with “fewer legislators means smaller government and less regulation”) but acknowledge that it does have a lot of emotional traction.

    For me, having districts (on average) 3 to 7 times larger would be a fair trade-off if it means having 3 to 7 times as many representatives I can choose among (and a corresponding increase in candidates). I mean genuinely “choose among”, comparing like with like, rather than assuming that Candidate Apple who lost with 45,000 votes in one district has somehow been “preferred” over Candidate Orange who lost with 40,000 votes in a neighbouring district.

    The various versions of MMP seem to offer districts, on average, 1.5 to 2 times as large in area and population but still only one MP who is guaranteed to be electorally removable. Some districts might, if they’re lucky, get a second MP or even a third but this bears no necessary relationship to population or voter turnout.

    Between the southern Irish model with an average of around 4 seats per district, and the Northern Irish model with 6, is considerable room for other variations. Uniform 5-seaters would avoid most of the Dail’s disproportionalities (as in Tasmania since 1998, or even the Australian Senate 1948-80 where they compensated for vast differences in population size). The NZ Royal Commission recommended, as its second choice, a workable STV model where at least 80% of districts would be 5-seaters but up to 20% could have 3, 4, 6 or 7 seats to accommodate local geography.

    Canadians can choose whatever system of government seems right to them but I can’t say (sorry guys) that I see the Dominion as world’s best practice for models of representation.

    While I have no inside track on why BC voters reversed their large pro-STV majority within a few years, experience with Australian referenda suggests that the shell-game tactic might have worked again. First referendum: “Vote no – this system is novel and untried, and there hasn’t been enough discussion yet.” Second or third referendum: “Vote no – this system has already been rejected by the people, so why don’t the politicians take no for an answer?” (See the simultaneous Senate elections proposals in 1974, 1977, and 1984 – the BC precedent is especially applicable since in 1977, 62% voted yes nationwide but States divided 3-3, and in 1984 great play was made of the previous “failure” of this proposal).

  24. Tasmania first used STV in 1896 in the cities of Hobart and Launceston and statewide in 1909. That is a longer history than Ireland’s. If length of use is determinative Tasmania’s system of 5 representatives per district should be considerably more persuasive than Ireland’s.

    In fact length of use is not really a good guide and there is nothing sacred about magnitude 5 or magnitude 6. Tasmania has a long experience of tied assemblies when it used even numbers for district magnitudes and that motivated the shift to odd number magnitudes in 1959.

    The ACT entrenches the principle of odd numbered magnitudes and allows districts with different magnitudes so long as the principle of equal voting is preserved.

    There is no reason, unless the principle of geographical propinquity is to accepted, that Canada should not emulate and Tasmania rather than Northern Ireland. The principle of geographical propinquity may require Canada to follow the different magnitudes used in Ireland rather than the the identical magnitudes used in Northern Ireland, it would depend on how you measured distance and closeness of observation between Canada, the Irish republic, and Northern Ireland.

    It would seem to me that Canadians could fairly easily provide for varying district magnitudes with small magnitudes in non-urban areas.

  25. @Tom

    Canada, like Australia, has some electorates with huge areas and very scattered population. I think they have a very good case for single member districts under STV.

    I would think that geographical similarity may be even more important than geographical propinquity and it would be appropriate for Canadian electoral reformers to look at a country with a broadly similar constitution, population, area and demography. The Australian electoral experience may be a tad more relevant than unitary countries with very small areas and distances.

    It is also possible to ensure local representation by a variant on the circuits idea that was proposed for by-elections to the BC Citizens Assembly.

  26. STV proposal:

    2/3 of the district members are allocated district-wide by STV

    1/3 of the district members are allocated proportionally by national vote.

    3-seater: 2 STV, 1 Top-up
    5-seater: 3 STV, 2 Top-Up
    7-seater: 5 STV, 2 Top-up

  27. If Canada is going to use MMP, then add a nation wide tier of at least 10%, and 90% of the seats are allocated to the providences. Divide the seats in half, as an example; Prince Edward Island elects 4 representatives; one will be from SMD, and the other three are at regional list. The other providences which are more populous can determine the ratio of SMD, and List Seats. It would be a two vote system. The first vote for the local riding, and second vote for both the regional and national list. It may even be wise for some of the large providences to be subdivided into multiple regions. This system looks to be extremely complex compared to the present FPTP system.

    This system could be be adapted towards an Two Vote STV system using small multi-member districts with another vote for national proportionality. Rural Areas could elected one to two members, and urban areas 3, 4, 5, 6, and 7, and it would still be proportionate.

    Maybe it would be better for Canada to use Proportional Representation for it’s lower house, and a SMD system for the Upper House (The Tasmanian Model) to ensure that rural areas are properly represented.

  28. A nationwide tier means amending the constitution, a process that is even more difficult in Canada than in the US. Using MMP with balance seats would also need a constitutional amendment because it would alter the distribution of members of the House of Commons among the several provinces and territories. The heavily localised distribution of support among Canadian parties would result in many overhang seats.

    ListPR or STV could be enacted by ordinary legislation, although it would be better to proceed by referendum, hopefully without the supermajority that was required in British Columbia.

  29. OK, I need to correct something here. I propose the very same thing many Maltese proposed for STV, although some district seats would be used to correct any disproportional results on a nationwide basis.

    If we go to Malta for a second, they have 65 seats, divided into 13 5-seater districts. Some proposed allocating 4 seats with STV and the remaining seat being a “Top-up” to correct the results (I believe in a national basis).

  30. Not sure if anyone noticed the United States Supreme Court heard the Alabama Legislative Black Caucus v. Alabama case involving affirmative/racial/partisan gerrymander.

  31. Back to Dion for a moment: he has clarified that most districts would be five-seaters. I discuss the result here. This has taken on new relevance since his Liberal Party has changed their position and proposes “electoral reforms including, without limitation, a preferential ballot and/or a form of proportional representation.” Dion’s model meets his objectives, and even lets the Green Party win at least six seats if they double their vote, very possible since half their support shown by polls always falls prey to strategic voting. So it now must be taken seriously. Still, about 164 of the 332 MPs to be elected from his multi-MP regions – 49% of them — are lumped into regions much larger than their communities.

    Ed, above, has an interesting idea: “two seat constituencies. The top vote getter in each constituency would always take the first slot. The second slot would go to the top vote-getter in the constituency representing the party who has received the most votes in the province/ county, from among a pool of parties, whose number of seats won in the province/ county (at that point) is less than the share they would have gotten proportionate to their vote. You would need a mechanism to determine the order in which constituencies received their “additional member.” Doesn’t that amount to the Baden-Wurttemberg model, with one additional rule: no constituency can have more than two MPs? Now, Ed, just give me your “mechanism to determine the order in which constituencies receive their additional member” and we might have a serious model here. Unless your mechanism is either too opaque, or produces too weird results, or both.

    • I’ve thought about Wilf’s question, and I think the problem here is determining which constituencies wind up represented by politicians running for fringe or minor parties who have taken low shares of the national vote, and low shares of the constituency vote. This might be called the “Rhinoceros Party problem”.

      Take the example of a party that gets 6% nationwide, and gets that 6% spread out evenly, it never runs more than 8% of the vote or less than 6% of the vote in any one constituency. That means about 6% of the constituencies would wind up with one member of parliament who took the most votes in the constituency, and one who took about 6% of the vote and came in third or fourth in the voting. Now the combined total vote of the minor party and the dominant party in the constituency might be quite high, since there are some constituencies where parties in multiparty systems where parties get percentages in the 70s. Bob Uecker roomed with Eddie Matthews one season in which Uecker hit one home run and Matthews hit 399. Ueker would say “between me and my roommate, we hit 400 home runs.” The optics of some two member seats having representatives who 90% + of the voters rejected would be bad.

      The way around the Rhinoceros Party problem, to modify my proposal, is to take the pool of two member constituencies (nationwide, provincial, regional, county), and once the votes are counted, rank all the candidates in the pool by the number of votes they received. Each party would be proportionally, by whatever mathematical system of translating votes to seats the country uses, a quota of seats tied to their aggregated candidate’s proportion of the total vote. The top vote getter of all the candidates in the electoral region would be declared elected for his two member constituency, then the second top voter for his two member constituency, and so on. However, once both seats in a constituency are filled, the remaining candidates in that seat would be disqualified and removed from the order. Once a party has elected a number of candidates equal to its “quota”, its remaining candidates in the electoral region are likewise disqualified. The larger the number of seats in the electoral region, the friendlier the system is to minor parties, keep the pool at the provincial or county level and Rhinoceros Party candidates don’t get elected, in a nationwide electoral region or pool some would.

      To take a practical example, take a 300 member parliament in a system where there are just four parties. The members are elected from 150 two seat constituencies, and the one electoral region is a nationwide one. Party A wins 40% of the vote and comes first in 85 constituencies. It would also elect 35 “supplemental members” from its also rans. Party B, which wins 30% of the vote and comes in first in 50 constituencies, would elect an additional 40 members from its also rans. Party C, which wins 20% of the vote and comes in first in 15 constituencies, would elect an additional 45 members from its also rans. Party D, which wins 10% of the vote and comes in first nowhere, would wind up with the second member from 30 constituencies. There would probably be about 50 constituencies where someone other than the top two vote getters gets elected. Mathematically, it would be possible for the top vote getter in an unusually small or low turnout constituency to not be elected, because its party has already exhausted its quota, but hopefully this situation would be rare!

      Generally, within a constituency, the two two candidates would be elected, and within the number of candidates run by a party, its top vote getters would be elected before the others. In seats where a party gets huge vote totals, such as the Alberta seats for the Conservatives in Canada, it would make sense for the dominant candidate to run two candidates.

      The minor party candidates would wind up as the second representative from seats where the major parties get relatively few votes, because the constituency is overrepresented (low member to population ratio, such as the Artic and PEI seats in Canada), or turnout is low, or maybe a major party dominates but can’t or won’t run two credible candidates. In the Canadian situation, the Territories would probably retain their three single member districts, and the provinces would be the electoral regions. There would be a few three member constituencies where the province has an odd number of MPs.

      Dion’s proposal, which I like better, is basically “STV lite”. This idea is basically “additional member lite”, its the additional member system, and retains the proportionality of the additional member system, but reduces the party managers’ role in selecting the additional members.

      • That sounds similar to the system proposed in Quebec in 2004. It would be really interesting to do simulations, to figure out how likely strange results would be.

Leave a Reply

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

You are commenting using your 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 )

Google+ photo

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

Connecting to %s