Garneau endorses a ranked ballot -

Garneau endorses a ranked ballot


The Liberal leadership contender writes to party members with his thoughts on democratic reform.

If elected, my proposal would be to reform Canada’s electoral system by changing our voting process to a preferential ballot, or a ranked ballot. Used by many other nations, as well as the leadership races for the Liberal Party of Canada, the federal NDP and the Conservative Party of Canada, a preferential ballot better reflects the will of the people.

Using a ranked ballot, Canadians would no longer tick only one box indicating their first and only choice. Rather, they would rank their choices and tick not only their first choice, but their second, third, fourth, etc. choices. If no candidate wins more than 50 per cent of the votes when the first choice votes are tallied, the bottom candidate is dropped and his or her second choice votes are allocated to those who remain. The process continues until one candidate has achieved at least 50 per cent plus one of the support from that riding. The preferential ballot fundamentally addresses the challenge of vote splitting. Parliament will better reflect the real preferences of its people.

I have lately taken a liking to the idea of a ranked ballot: far less complicated than the various proportional representation scenarios that have been proposed in the past (and thus, I suspect, easier to convince the general public to support), but likely to produce a more representative result.

Mr. Garneau also says he would, as leader, only appoint candidates in exceptional circumstances. Fair enough, but why stop there? Why not go all the way and propose that the requirement that a candidate have the endorsement of his or her leader should be eliminated? If empowering local riding associations is a worthy goal, commit to it fully.


Garneau endorses a ranked ballot

  1. I wonder, in a ranked ballot, does it often or rarely happen that the candidate ranked as the first choice doesn’t win in the end? I can see that such a ballot would (somewhat) silence those who attack the legitimacy of MPs elected with fewer than 50% of the votes, but I am unclear if it would really change much in the results.

    • Often. That’s how we got Ed Stelmach here in Alberta, for one.

      Of course, as demonstrated by Ed Stelmach, it doesn’t mean you’ll get a person who’s any good, just the one who isn’t “the other guy”

      • Regardless of how great the elected person is, you are absolutely guaranteed to elect someone who has majority support.

        • No. Just minority opposition. Which, I’ll admit, is still a step forward.

  2. The leader should retain both the option of appointing a candidate, and the right to refuse to sign a candidate’s nomination.

    This is to prevent one-issue candidates, or people there only to cause trouble….or a racist candidate like Forbes in Quebec surfacing during the last election.

    • Why can’t the electorate be entrusted to dispose of such lame candidates at the ballot box?

      • Because of the damage it does to the party. Not only the image, but the number elected.

    • But those people are nominated and elected under the current system. If the leader wasn’t the one appointing the candidates, then perhaps the leader would then be in a better position to disown nominees who are morons.

      • Leaders rarely appoint a candidate….but sometimes it’s necessary

      • “If the leader wasn’t the one appointing the candidates, then perhaps the
        leader would then be in a better position to disown nominees who are

        True. There are absolutely no examples of the local crazies running for congressional districts in the States saying something stupid and stealing the headlines from the much more important presidential campaign. None whatsoever.

    • More importantly, IMO, is that we live in an era if professional politicians, who have been schnoozing with associations for years. But is that all we need in parliament and in cabinet ? There are situations where special talents are required, Marcel Massé being one example. I would rather have a political leader keep the right to parachute a candidate, on occasion and in exceptional cases, say just a handful of ridings, than to have a bunch of non-elected ministers in cabinet. Don’t forget that in theory we can have a complete cabinet of persons who are not MPs.

      • We seem to prefer no-talents to special talents unfortunately.

        I would also prefer to have unelected ministers than some of the elected dodo birds we’ve got in there now

        I think it was MacDonald who said something to the effect of ‘if you want me to build a better cabinet, send me better planks’….but of course people rarely do.

        Some of them are THICK as a plank though.

        • Unelected ministers = senate appointments. You’ve got to pay these people somehow.

          • Elected or unelected,they have to be paid.

  3. Two simple changes to our electoral laws that can help transform our Democracy.

    The Canadian public has become progressively more disengaged from the political process in Canada over the last number of years. This disengagement is reflected in the
    decline in voter turnout and general participation rates.

    In one sense, the lack of concern, interest and attention is a sign of some positive history that led to this point. Canadians have, for over fifty years, enjoyed stable, relatively transparent, middle of the road, government. No matter who was in power, no matter the
    specific details of a platform, the Canadian government has trended towards stability and moderation. In the 20th century Canadian politics has largely secularized, especially in contrast with the United States. We have also seen the steady erosion of blatant cronyism and patronage that built the original Canadian political structure. The civil service is in large part non-partisan and professional in its development of talent and promotion. No longer the day after a change of government do huge swathes of public sector jobs and contracts change hands. Certainly there is still pork in the barrel, but the scale is diminishing. The point is that any of the four national parties who run candidates from coast to coast, could form a government tomorrow, and the lives of most Canadians would not change over-much, and certainly not overnight. The major parties now offer differences of degree, emphasis or of strategy, rather than fundamental restructuring of Canadian society. Canada will remain a pluralistic, mostly secular, Liberal Democracy with a free market economy with
    universal social services. The details will change, but the majority of Canadians will not tolerate a change to the fundamental structure or understanding of our country. So as long as no one is threatening radical change, and the policy differences are nuanced at a level that non-specialists (most of us) find difficult to distinguish, more and more of us will disengage. With no large patronage stake in which ‘team’ wins, or major ideological or moral fights on the horizon, many Canadians conclude that it doesn’t matter who wins, therefore don’t engage and thus don’t vote. This is the ‘good’ news.

    There is also discontent with the developments in the House of Commons, the loss of individual MP’s voice, the ongoing centralization of power in the PMO and the constant campaigning that never ends. The increasingly partisan nature of the attacks, advertisements and rhetoric is also making an ever growing number of Canadians turn away. Combine this with the perception that in a ‘First past the post’ system, many voters are wasting their votes, or feel forced to vote strategically, if at all, it is little wonder voter turnout continues to decline. The current system creates a paradox wherein it is possible, by increasing partisanship and increasing factionalism, to fracture opposition, solidify your base, and govern with a majority of seats in the house with less than forty percent of the electorate supporting your party.

    The two legislated changes I am proposing are intended to be as simple, practical, and as minimalist as possible. Proportional representation might be nice, but it would require constitutional changes, and with the recent provincial forays in this direction serving as guides, it isn’t going to happen. I don’t want the “Ideal Solution” to get in the way of a
    “Possible solution” therefore, the changes are:

    1) Every Member of Parliament should win his or her seat with 50%+1 of the votes in her riding. Introduce a preferential ballot or runoff election to make this happen. Preferential would be easier to administer with the current infrastructure at Elections Canada (don’t need a second election day), but the countries that use the runoff seem to like it and have had success with it, so it is an option.

    2) Candidates should not require the signature of the leader of their party be on the ballot, they should be nominated by the local riding association. he will of the local riding and community is to be paramount, and the candidates must be responsible to their local association and community over the leadership of the Party. Every Member of Parliament is intended to represent their constituency in Parliament, not the party to the constituency. *It is up to the parties how they constitute these associations and the method used to choose their own candidates, but it may be desirable to institute a formal primary type system,
    with registered supporters of the parties, administered by Elections Canada to prevent abuses of the system at a local level. Some discussion around this point will be important.

    The two simple changes are the goal. They lead to a constellation of smaller changes and altered dynamics of power throughout the system. MPs have to win 50%+1 to take a seat in the House, and constituents, not national leaders, put the names on the ballot. That is it.

    The difficulty is in implementing those two changes. On the surface, it would appear to not be in the interest of the major parties in the House of Commons. For the current incarnation of the Conservative party, that would certainly be the case.

    How to get there:

    The solution to implementation is one-time co-operation. The national parties that agree to the need for the proposed changes, need to sit down and negotiate a one election only
    deal to elect a majority sufficient to pass the changes. This is not merger, coalition building, or any form of permanent co-operation. The national political parties exist because they express, or should express, differences of policy and values. Many voters do not find that one political party or another agrees with them on every issue, nor does every candidate, but some align better than others. That should continue. However, for one election, parties that support electoral reform need to agree on a division of ridings amongst themselves, or joint nominations, or another option, in order to achieve the changes desired. Once implemented, there is no further need for formal co-operation or merger. In the future coalitions or other arrangements in the house will develop to help form Governments, but they need never be absolute nor permanent.

    That’s it. I only want to add as a biographical detail that I am not a rabid partisan of any of national parties. I am a Liberal party member at the moment thanks largely to my local Member of Parliament. However, in the not too distant past, for over fifteen years I have been an active Progressive Conservative, and I sought local nomination when Joe Clark was leader, I have been an active Green Party member and candidate in 2008, and for the record, I have voted NDP provincially as well.


    Michael Munday

  4. Unfortunately, this method (known as Instant Runoff Voting) is just about as bad as straight plurality voting and far more complex to process. It is non-monotonic meaning increasing your support can actually make you do worse. Plus you can’t combine results. You need to wait until all of the ballots are in before you start the simulated elections.

    Much better methods include Range/Score voting (rate each candidate on a scale), Approval Voting (a yes/no for each candidate), or Condorcet voting (ranked ballots as well but does pairwise preferences rather than sequential sub-elections in IRV). Score Voting is simple and has the least flaws of all known voting methods.

    It’s unfortunate that IRV (also known as Single Transferable Vote / STV) is the leading contender for change.

    • Just as bad? Non-monotonic results are only even remotely likely to occur in a three candidate election, and even then the highest legitimate estimate of that probability I’ve ever seen is 5% per election. (In an election with more than three candidates, which is 99% of elections in Canada, the probability drops to less than 0.1%) This is a tiny likelyhood of error for a system that involves one tiny change and is vastly more fair.

      So I urge you to think about it this way. Under first past the post, the more candidates there are, the more likely it is that someone will win their mandate with less than 20% of their constituents’ support. Under IRV (i.e. ranked ballots), the more candidates there are, the less likely it is that the voting system will have any hiccups, and the more likely it is that a greater number of voters will feel engaged by the election.

      • I actually said just about as bad, not as bad. I’m well aware of the first-past-the-post problems as well as IRV. The problem is you’ve created a false dichotomy between the two. There are much better ones like I listed above.

        As for problems with IRV, let’s take a realistic example. Suppose in a riding the IRV first round votes are 41% Conservative, 39% NDP, 9% Wild Rose, 8% Green, and 3% yet to vote (or 3% Liberal who get knocked out). Suppose all Wild Rose have CPC as 2nd choice and all Greens have NDP as 2nd choice. If those 3% go Green (now 11%), WRP get dumped and CPC gain those 9% and win it. The Conservatives win because Green gained support.

        Had that 3% instead voted NDP (now 42%), Greens get bumped out and that 8% goes to NDP who win it. Hence going from NDP to Green makes the election go from NDP to Conservative.

        It gets more complicated if those 3% are honestly Conservative supporters as voting Green first might be more in your interest to move those Wild Rose over to Conservatives, but that’s getting to complicated for here.

        This is part of the problem. Not only can your votes get your worst case elected, but strategic voting gets far more complicated, and you might sometimes be better off voting first for you worst enemy.

        Then there is the problem with counting votes. You can’t tally until all the ballots are in because you won’t know who even gets knocked out in the first round until then. IRV is sensitive to who is last place, not just first. Changing who loses in the first round can change the whole outcome. This means you can’t just add them up at polling stations. You need all of the preferences of all ballots, and all ballots accounted for before you can move to the 2nd round.

        The source of these problems is that large blocks of voters can be moved with small movements across relative thresholds. If WR or Greens get kicked out it makes a huge difference to the leaders as one or the other gets a big block of voters suddenly next round. It makes sense to even make sure your most similar party gets knocked out early.

        On the other hand, Score voting degrades gracefully. There is no value in voting strategically. The scores for each candidate (e.g., 1 to 10) can be tallied at polling stations and tallies centralized. New ballots can be added in as they come; there is no need to wait for all of them. There are no big jumps due to crossing thresholds. Voter preferences and preference orders are fully accounted for. It is monotonic.

        Score voting is so much better in just about every category of voting metric and ease of logistics for both voting and counting.

        • Thanks for the reply.

          I’m not very familiar with some of the other systems you listed above, so I can’t really have that debate. But what I do know is that ranked ballots/instant runoffs are politically feasible (since they are already used across the US and by Canadian parties and since they involve only one simple change to the ballot format) and are a huge leap forward over the status quo; if you’re an incrementalist like me and you’re also big on electoral reform, that makes them the go-to policy… for municipalities. For federal and provincial ridings I’m not so sure. One of the great features about IRV at the municipal level, where there are typically no parties, is that it prevents people from being pushed out of the running due to perceived vote splitting. This leads to greater and much more diverse participation (more young people, more women, more minorities). But in a party system there is a natural limit on the number of additional candidates who can feasibly campaign for office, so you’re unlikely to gain those benefits from switching to IRV.

          I understand what you’re trying to say with your example, but I think you are misjudging the system. First off, you say that switching your vote from NDP to Green makes the election go from NDP to Conservative, but this exact same thing could be said about an equivalent scenario in FPTP. It’s just that in FPTP it is clearly vote-splitting as opposed to a feature of the multi-round system. But then what are the chances of whole blocks of voters all going in the same direction anyway? Your scenario suggests that the electorate would be very cleanly fractured along partisan-party lines and that the vote would add up perfectly through multiple rounds to a ncie clean, left-right partisan result. But I’m fairly certain that the Green support would split left and right, and last time I checked, the Wild Rose party was basically an anything-but-the-PCs party at the provincial level, so in this campaign scenario, isn’t it likely that the Wild Rose candidate would seek out some common ground with their centre-left opponent in, say, the Liberal party or the Greens in order to garner some secondary support for later rounds of voting?

          This is the dynamic that I think you are missing. You are looking at the campaign through the lens of FPTP, and then you are guessing at how it would play out if you just superimposed an IRV system onto the voting results. What you’re missing is that the campaign dynamics would conform to the system, meaning that parties would organize themselves along less harshly partisan lines in order to cobble together a broader coalition of support, which means that whether the NDP or Conservatives won in your scenario, there would be nothing particularly unjust about it. Think of what happens in the current system when a Green party supporter votes with their heart: They know from the outset that their vote is basically wasted and that it will also likely feed into a vote split in many ridings. That, and no Green candidates ever win despite broad national support for their policies (although IRV is not actually designed to fix this).

          Now I understand that you are proposing a range system as a better alternative, but it only took me a few minutes of research just now to reveal that range voting has its own potential glitches. Here’s a couple of paragraphs from Wikipedia:

          “However, there are examples in which voting maximum and minimum scores for all candidates is not optimal.[9] Exit poll experiments have shown that voters tend to vote more sincerely for candidates they perceive have no chance of winning.[10] Thus range voting may yield higher support for third party and independent candidates, unless those candidates become viable, than other common voting methods, creating what has been called the “nursery effect”.[11]

          Range voting advocates argue that range voting systems (including approval voting) give no reason to ever dishonestly rank a less-preferred candidate over a more-preferred one in 3-candidate elections.[12] However, detractors respond that it provides motivation to rank a less-preferred and more-preferred candidate equally or near-equally (i.e., both 0-1 or both 98-99). This could lead to undemocratic results if different segments of the population used strategy at significantly different rates. (Note that traditional first-past-the-post voting forces all candidates except one to be ranked equally, so that all voters are compressing their preferences equally.)”

          It appears to me that tactical voting and irrational voting are both very real possibilities in a system where you have a fractional currency to distribute on the ballot. Personally, I think that if we’re going reform our electoral system, one of our primary goals should be to eliminate the phenomenon of strategic voting so that voters are expressing their preferences directly. I believe that IRV is the best system for this, even if it produces a slightly wonky outcome less than 0.1% of the time – which, I’ll reiterate, is the actual likelihood of a non-monotonic result in an election with more than three candidates.

          I’ll also reiterate that I don’t believe that IRV is necessarily the best option for a party system. I would actually prefer a proportional system using STV at the federal and provincial levels. But IRV would a nice change too.

          • Yes, FPTP is non-monotonic as well. I agree, which is why I’m not suggesting FPTP as a good method. But replacing a bad method with an almost as bad method is not a good plan and makes replacing it with a good method less likely as nobody will want to change it again shortly after implementing IRV. It’s better to wait to get support for a good method and then make the change.

            There also cannot be a perfect voting method meeting all desirable election properties following from Arrow’s Impossibility Theorem. The question is which criteria are you going to violate and how badly. Here are some typical criteria for judging:

            Favorite Betrayal Criterion: “No voter should ever have strategic incentive to vote a less-liked candidate over his favorite”. FPTP and STV/IRV fail this. Range Voting, Approval Voting, and Condorcet Voting pass.

            Defensive Strategy Criterion: “If a majority prefers one particular candidate to another, then they should have a way of voting that will ensure that the other cannot win, without any member of that majority reversing a preference for one candidate over another.” FPTP and STV/IRV fail this. Range Voting, Approval Voting, and Condorcet Voting pass.

            Monotonicity Criterion: “With the relative order or rating of the other candidates unchanged, votinga candidate higher should never cause
            the supported candidate to lose, nor should voting an unsupported candidate ever cause the candidate to win.”

            Participation Criterion: “Casting an honest vote should
            never cause the election result to get worse (in that voter’s view) than if she hadn’t voted
            at all.” FPTP, STV/IRV, and Condorcet fail this; Approval and Range voting pass.

            Clone-safe Criterion: “If a clone of a candidate enters or leaves the race, that should not affect the winner
            (aside from possible replacement by a clone).” FPTP fails this; STV/IRV, Range, Approval, and Condorcet pass.

            Precinct-countable: “If each precinct can publish a succinct summary of the vote (sub)total in that precinct,
            and the overall country-wide winner can be determined from those precinct subtotals,
            then the voting system is “precinct-countable.” Precinct-countable
            systems allow greater election transparency to reduce risk of undetected fraud.” STV/IRV fails this; the others pass. This one is the logistic problem I mentioned before. You need all ballots accounted for before you can determine who to knock out in the first round and which ones to transfer in the second round. You can’t tally IRV/STV independently.

            There are other criteria occasionally used, and other voting methods tested (Borda count, Shulze, etc.) but I don’t want to launch a whole thesis in a comment.

            In this context, FPTP (aka, Pluraity) fails many criteria and so does IRV/STV. You can see tables of methods and pass/fail in various election method studies such as here: (which doesn’t contain Range Voting but does have Approval as the best), or here:

            So the question isn’t whether Range/Score Voting is perfect or not, it is which method is the best over all of the criteria. So the question is whether the “nursery effect” is worse than the cumulative failures of STV/IRV.

            Further to that, I would argue the “nursery effect” is essentially a false problem. While it is true in principle that people can vote an undesired party higher than they honestly support, there is no incentive to do so. What does a voter gain by artificially raising a party above what they would honestly like? All your quote does is assert this without demonstrating how the voter gets value from doing this. In fact, they risk having their preferred candidate lose to their lesser preferred one by ranking the lesser preferred one higher.

            More importantly, the “unlikely to win” assertion is circular. There is no independent means to evaluate who is preferred (likely to win) outside of the results themselves. Allowing voters to vote for a small party they like, but is unlikely to win, without “wasting” their vote for is exactly what you want in a voting method. You want to aggregate the voters’ preferences which includes the opinions of parties with currently little support.

            As far as democratic, that is what these voting criteria measure. Democratic means respecting the voters’ preferences, and Range Voting does it much better than the rest, with Approval Voting as a close second. (Approval Voting is just a state within Range Voting whereby everybody just votes their maximum or minimum score for each candidate/party.)

            As far as strategic voting, I’m shocked at your suggestion that IRV/STV is a best means to eliminate it. IRV/STV is the only method that makes strategic voting even worse than FPTP as I demonstrated last time and the election criteria (and tables) show. With FPTP your strategic vote is best to abandon your favorite candidate in favour of your favorite of the top 2 likely to win. With IRV/STV your strategic vote is to vote highest for the low-polling party you hate the most, then your favorite party second, and rank your second favorite party lowest. You want to raise up the opposition’s fringe clones to keep their votes split and you want to knock out your own clones immediately so their votes move to your favorite party in round 2.

            In other words, if you are a Conservative supporter you are best to rank Green highest so that Green stays in round 2 and NDP doesn’t get their votes, and you want Wild Rose kicked out in the first round so that their votes go to the Conservatives in round 2. That keeps the “left” (NDP/Green) split while aggregating the “right” (Conservative/Wild Rose). Of course there is no value in this if the number of people doing it exceeds those brought over from Wild Rose, so it’d have to be coordinated. But that’s the problem. STV/IRV encourages collaborative strategic voting to game the system because it is sensitive to the lower polling parties relative positions.

            STV/IRV is not more democratic or more likely to reduce strategic voting. It arguably makes them worse or only minimally better. Range or Approval Voting are far more representative of honest preferences of the voters, with greater democracy and less/no incentive for strategic voting.

    • Is ranking really that different than scoring each candidate? The I in IRV stands for instant. There is only one round, the candidates are ranked by the voters. This is not different enough from your suggested Range/Score voting to seem so much worse than your ideal method.

      • That’s not how IRV/STV work. It is “instant runoff” not “instant”. There are multiple rounds. You tally all of the first place votes first. If a candidate has more than 50% they win. If not, you remove the lowest scoring candidate and move their votes to the second highest candidates on each of those ballots. If no candidate has 50% still, you repeat and move to round 3, and so on.

        It sounds good in principle because it always results in a winner getting a majority vote (>50%). The problem is the path in which it gets there is highly sensitive to the lowest ranked fringe parties, so small changes at the low end can cause huge changes to which candidate ends up with that 50%.

        It has horribly undemocratic properties and logistical problems, such as increasing support for a candidate can cause them to lose and the fact you can’t move to round 2 until all ballots are accounted for. Even finding a few missing ballots can mean re-running it all from the beginning because it affects who gets knocked out in round 1.

        I just responded above with a long message explaining most of the major criteria for judging voting systems so rather than repeat it all, I suggest checking it out. IRV is quite a horrible system so I can’t support it even if it is slightly better than FPTP. There will be no support for changing again any time soon so I think it is better to wait until support behind a good system builds up and only change it once.

  5. A busload of Nazis, or Tamil Tiger supporters, or anti-abortion activists, show up at a nomination meeting for, say, the NDP and get their candidate nominated, a candidate whose agenda is 100% at odds with the party he wishes to represent. Under the current system the leader can simply not sign the nomination papers; removing this power leads to some serious unintended consequences.

    • It is true that a lot of people don’t realize how easy it can be for a relatively small number of insurgent wingnuts to take over a riding association (or even a number of them). The whole sorry and weird history of David Orchard in the old PC party was a classic case of just that.

      • How about requiring two thirds of the other nominated candidates for the party agree to reject the candidate? Having that power vested solely in the leader gives them far too much power. The executive branch has effectively subsumed the legislative branch in this country, and that is a serious problem that requires addressing.

        • That would work. As I understand it, though, most of these rules are contained in the constitutions/constating documents of the political parties,so they’re essentially internal rules of private clubs (sort of like the constitution of the Elks Club or the Royal Order of Water Buffaloes). I’m happy to be corrected here, but I believe that such rules have nothing to do with our legislatures.

          • Legislatures could require recognized political parties to adopt these rules, or lose their recognized political party status and all associated perks.

    • Potentially yes. However, it has more to do with the rules in place for the selection of candidates and nomination process and eligibility of local voters. There are ways around the ‘busload of instant voters’ problem. A more formal primary system is but one way of addressing it., voting rules and procedures will also help.
      The problem is that the system our current evolved from had party leaders elected by the party caucus in the house. This meant that the leader was accountable to caucus, now with party members voting for leaders, the accountability flows the other way, with Members of Parliament becoming voting ‘puppets’ of their respective leaders, dependant on the leader for their promotion prospects and their nominations. I would argue that is undesirable.

  6. I’m happy to see this development. The political participation rate in Canada is brutal and a move to ranked balloting has the potential to greatly increase the appeal of politics generally. In a ranked system, debate becomes more civil and strategic voting that disenfranchises “smaller candidates’ becomes moot. A dialogue that is not toxically partisan and encourages multiple perspectives is more appealing and representative of democratic values.

  7. Preferential voting systems carry higher bayesian regret with strategic voting than does plurality, and are acutely more vulnerable to strategic voting.

    One can find numerous instances in canadian party elections where the preferential vote elected a third or fourth place candidate over the top two contenders, thanks to strategic voting. It’s a popular fantasy that it is superior to first-past-the-post — but like many fantasies, the reality comes up a little short.

    • True: Stelmach, Joe Clark, Stephane Dion. It’s the “anybody but ______” phenomenon, and it often leads to severe buyer remorse.

  8. I guess anything is better than we have now,but its not enough,the “real” issue is Federal recall,petition them out if they dont keep there promises,i will settle for nothing less!

  9. Canadians,, Dogs for punishment,no matter what you do to them,they always come crawling back!