MMP for Quebec in 2022

Anybody have any connections in Quebec?

Quebec will hold another referendum in 2022 - on electoral reform.

The relevant details of the actual proposal to be voted on:

The number of seats in the National Assembly would remain 125. The new electoral law would divide Quebec into 80 larger electoral divisions that will mainly be the same as federal ridings, and 17 wider regions:

  • 80 MNAs would be elected according to the current system
  • 45 MNAs on a list of candidates would be divided among the parties, according to the percentage of votes obtained in the 17 regions

The advantage according to Legault, is that instead of a winner-take-all system, certain regions that “had quite a few votes but not enough to get an MNA, will have one or some of the 45 that will come from the list, so you’ll have a mix between the actual way of electing people and the and in proportion of the vote for each party.”

Why 17 superdistricts for 45 seats? 3 times 15 is 5, so why screw around with that creating irregularities? Why not therefore 15 superdistricts, or 5 to elect 9 in each? With 17, some superdistricts would be only electing 2!

Why subdivide the province at all? If they are going to do MMP with separate party list election, why not hold one of the latter for the whole province?

Now stipulating they want to hold the Assembly to a fixed membership of 125, it follows that reducing the districts electing members FPTP will dilute the constituencies, raising the size of the individual districts by 50 percent. Therefore it would do little more harm to define 62 districts, and elect the larger balance of the Assembly by proportional make up.

There is no need to empower parties to make up lists to be elected, with or without some kind of mandated public input, prior to the election in some kind of primary or by complicating the ballot further with some sort of candidate preference process, though the latter would not really be terribly burdensome. If in fact 80 districts are running FPTP votes, a large pool of suitable make up candidates are already being vetted by voters; if we just ran the 80 district contest and stipulated that votes for a candidate are in fact also votes for their party, we can readily compute PR for 125 members, and simply subtract the FPTP wins from each party’s total, then look at the candidates for each party that did not win the FPTP races and take the top vote winners to make up the lists. This takes power over the complete delegation out of the hands of party leadership and puts it in the hands of the voters. It results in the make up members coming from the districts where they performed pretty well, in small parties elevated to having any members solely by the make up process being represented from their strongest districts, in closely competitive districts having multiple representation of both leading factions.

It also results in uneven geographical distribution of the total weight of all candidates, but that is also a feature of the party list approach; I suppose chopping up the province into 17 superdistricts is more or less intended to offset this, but it is purely a game of definitional musical chairs to pretend it is a total cure.

In fact, if we took the approach I recommend, those districts that are left with just one member in the Assembly would be, assuming fairly even turnout, those closest to unanimous support for the FPTP winner, and these substantial majorities who tend to prevail locally across the board in their local affairs should be well satisfied with their single district representative. But meanwhile, the minorities who are not so well satisfied have a substantial say, in proportion to their small weight as a segment of a province-wide larger bloc of interests, in the overall makeup of the Assembly–their representative is probably not local to them, but the various parties are going to pay attention to where their support comes from and devote some constituency support to their far flung supporters from farther away from their power centers, if any.

For a third party to be very successful in a FPTP context, they need a regional or class stronghold corresponding to dominating districts that might be evenly scattered, but in certain kinds of neighborhoods. But going from a positive representation point of view I observe there are other persistent contenders who lack such a stronghold, typically, but nevertheless capture a substantial share of votes across the entire system. Our electoral systems should not be prejudiced in favor of voters who happen to be able to dominate some region; sectors of interest that never will prevail in any particular bailiwick remain legitimate. A specific example would be ethnic minorities suffering systematic discrimination; even without invidious manipulation of districts to prevent them from gaining an electoral base on a district basis, some groups suffering such discrimination are just inherently small and unlikely to have a stronghold naturally. But it remains important for them to be able to assert themselves as a group and present their claims without being supplicants to some larger interest group. To have majorities in the body rule in their favor, of course they must gain allies and make a solid case. But I should also point out, the role of an elected representative is not strictly limited to their ability to ram legislation through. They have protected and special standing in government to speak and be heard; they are accountable to publics that give them that standing and therefore more credible and their assertions are more weighty and less easy for responsible officials such as judges to dismiss out of hand.

Believing as I do that positive representation is another path, and frankly I think a surer and clearer one than murky and dubious claims that voters casting multiple weighted choices will somehow persuade politicians who I judge are likely, especially if we stick to single member per separate district elections or divide a whole Canadian province into even as few as 17 districts, to wind up coming from the same relatively narrow social circles as dominate FPTP victors, to actual functional consensus, whereby persons of diverse interest and standpoints come to realize their mutual dependence and negotiate the terms of it on a basis of fair balance of power, I share the goal I trust people here generally do that that society should indeed have a broad base of consensus. I believe that can only be based on each person, each sector of categories of persons, large or small, having a guaranteed fair shot on an equal basis per person of representation they have confidence in, which brings all sectors to the table on a level basis of political power. Then the majorities that prevail will be actively required to consider, not their subjective assumptions about what other persons ought to accept, prejudiced by what is most convenient for them, but what these others actually say. Meanwhile, if even the losers of particular decisions retain standing and presence, they will be able to keep reminding the body of their general shortchanged status, and if they are basing these claims on concrete reality, I would not set at naught the general sense of justice I think people generally share to move members of the majority to eventually take their side in something or other. And assuming for a stipulation the general sociopathy of all persons, which a neoliberal world view tends to reinforce, sooner or later some pragmatic issue will come up where large numbers, but a minority, of the body representing a large but minority set of various publics, wants something, and the hitherto neglected minorities can supply the balance of power needed to create a working majority in favor, and are willing to do so–for fairer consideration of their own priority issues.

So, if in fact cardinal methods can further improve and streamline this general balance of power by further facilitating an inclusive and considerate attitude, I am for that. But it seems essential there must be a guarantee of access to power in proportion to numbers, and not permit any systematic biases excluding some group or other organized on some crucial interest as they perceive it.

Now then, the MMP proposal for Quebec seems like a partial improvement to me, a step toward positive representation, but suffering unnecessarily in the following ways:

  1. unnecessary subdivision of the combined vote of the entire province in the Party List portion. It would be less bad with fewer divisions electing more members each; between 2 and 3 per superdistrict as I infer would be the outcome of dividing the province into 17 pieces for 45 members is not much of a step away from single member FPTP. I should point out, too, that FPTP as a term is really not bad as a description of the process, and that it describes just as well the election of multiple members from a district by successive pluralities. The practice of multimember FPTP was more widespread in the USA before the Civil Rights era court decisions and Voting Rights Act banned it for US House of Representatives races and cast a deep shadow over it in state elections, but I happen to know by happenstance that in West Virginia at any rate, it still continues. Now party list PR in a superdistrict electing 2 or 3 members is less bad than First, Second and maybe Third Past the Post, somewhat. But it makes far more sense to just subject the entire 45 seats to the entire vote of the entire province, especially with 80 seats guaranteed to be elected on a strict district basis only.

It is unclear to me by the way whether what is proposed is proper MMP as practiced in Germany or New Zealand, in which the party list race is taken as the yardstick for the entire body of 125 members, and make up seats (taken from party list in vanilla MMP and in the Quebec proposal, which is not I think the best way to do it) are chosen on that basis, versus a more restrictive mixed form in which the 80 seats and 45 are totally separate with no cross connection whatsoever.

  1. as noted, there is no need to hand parties the control of who has standing for the 45 make up seats by controlling closed party lists, nor introduce the heartbreak and controversy of running closed primaries versus open ones, nor complicating the election process further with additional voting on proposed party list members. The parties in fact supply a “list” in the form of nominated candidates in the regular single member districts, and voters are already voting on them. Why not use that single vote three ways, then, to determine district wins, to determine overall province wide proportionality, and to determine which candidates for a party get priority to fill the make up seats they might win in the middle stage? Each vote is a vote of confidence in a particular person to serve as a suitable representative for that voter, why subdivide the process?

I do think that it would be better to either make the membership variable in total size by using top off procedures meant to cancel overhangs by raising all parties to the same proportional level, or alternatively have a fixed multiplier of district seats on the scale of a factor of 2 or slightly greater; holding at fixed 125 members then implies fewer, 62 in fact, single member districts. But vice versa, at that scale, about 70 percent of districts will get typically 2 members; some will hold at 1 (about 1/5, but these will be those districts that in some combination either have very strong majorities for the prevailing candidate, or have very low turnout) and the remaining 10 percent, with some combination of very high turnout or very sharp competition between two or more parties, will have 3 or more.

With a multiplier little more than 50 percent, the representation is more uneven geographically; now only about 1/3 to 40 percent would have two members, about 1/20 or less more, and the majority only one, even many where the competition was rather strong. This seems unfortunate and so such 2/3-1/3 proposals strike me as falling between stools and not so satisfactory.

There may be other details, assumed as but of course, such as arbitrary percentage hurdles parties must clear in the party-line election, that are pernicious if present. There’s no valid point in doing that, save to gratify the large parties that are jealous of losing control and being faced with effective voter discipline by voters who have somewhere to go in knowing their votes will be conserved.

To discuss how much, we should turn to concrete evidence; fortunately a concrete proposal like this offers the opportunity to compare real world historical practices and outcomes with straight application of the proposal as given, or with modifications, to the actual recorded votes, or to behind-the-curtain modified ones based on a priori extrapolations from the data of how voters who did vote in certain categories might reasonably modify their choices to take advantage of the new system. The latter are always speculative and controversial, and dependent on one’s a priori assumptions about what is reasonable for people to desire, but grounding these extrapolations in real data I think is more valid that cooking up a hypothetical from no basis but pure ideology at all. We can do that too of course, but if we pretend to “science” we should acknowledge the limits of our assumptions and allow for them at least.

1 Like

I thought you were anti-MMP.
Anyway, if the party votes and riding votes are completely separate (which we know is a bad idea), there’s no excuse for using plurality within ridings.

Is there any hope of them changing to a smarter ruleset, or are the rules set in stone now?

This is an attempt to try and make it seem like the List seats are not all super partisans. It wont work.

I am. This is a net loss. Does not get rid of the biggest issue, vote splitting, and increases PR by increasing partisanship.

I am looking for people who want to try to get involved. The person in charge of this decision is Justice Minister Sonia Lebel. If we can at least get Approval for the Local districts that would make it better than the existing system. Even better would be to teach here about Cardinal PR systems. I think everybody should reach out and tell her she is using an old and flawed system. [email protected]

@MarkHFoxwell you seem to never be short on words. Want to write to her?

I notice a lot of PR reformers are rather comfortable handwaving the proportionality algorithm. The article didn’t mention how the party vote is converted into seats.

They are not concerned with representation. They want party votes. What we see as a side effect of gaining PR many see PR as an excuse to have party votes. Even fewer are concerned with representation of individuals like we are working on.

Is this ulterior purpose something you can demonstrate by quoting the advocates in candid remarks or other objective demonstrations of intent, or something you infer purely on the presumption all systemwide options must only result in “super partisanship,” or something in between or to the side like a reasoned analysis of what the logic of the system clearly is versus claims you say are therefore disingenuous?

I think there is a great deal of actual disingenuousness in much advocacy of many positions, and there is also quite commonly a lot of sincerity that has simply not considered other points of view and integrated their thinking on the subject. It can often be quite difficult to tell them apart, so rather than attributing dark intentions.

In this case I think that the specific reason they do not simply integrate the second party vote across all Quebec is simply that it is not usual to do that with PR or MMP. Most countries are indirect (and therefore inaccurate) in PR by doing something like this–mind I think it is more common for the multimember districts and the number of seats to be won to be closer to each other in magnitude, at least until the latter gets up above 5 or so–that is, if the entire Assembly were only the 45 seats mentioned as PR add-ons, and were exactly say 49, then seven districts with seven seats might be more typical, or considering seven members per district to be a bit high, make the number 50 and have 5 seats each in ten, or nine districts (to return to 45 seats).

Having 17 districts drops the average number of seats down to between 2 and 3, and makes the districts irregular in population. But again, it is difficult to construct a coherent evil plan from that; it seems likely at least to consider the thinking simply half baked and unscientific–something like a muddled compromise between a mentality that says that the geographic, local principle should be paramount and making the districts any larger than they absolutely must be is a disaster versus the logic of systemwide consolidation.

The straightforward way to balance systemwide integration with enabling strong localist priorities of some voters for various reasons which can be quite weighty is to take both to their logical extremes–systemwide integration of PR, combined with single winner in districts for either a fixed and large share of seats, best in a one to one relationship with the others IMHO, or make the systemwide integrated PR seats float as needed to achieve systemwide proportionality. The latter minimizes the size of the body thus allowing maximum district representation, at the cost of it being uncertain how large the body winds up being, the former provides a clear and transparently fair way to resolve the dilemma of different constituencies legitimately needing different strategies to effectively appear in the body in fair proportion.

We should neither be prejudicially excluding people who want regional or individualistic focus for choosing their most appropriate representatives, nor those who seek ideological, programatic unity and either are not concerned with regional stuff as a priority or are simply in too small numbers to show up proportionally if there are only 2,3, or even 5 members to be elected in their multimember district.

I think it is clearly wrong to say it is worse than useless. It is an improvement, just a pointlessly restricted improvement.

Also, there are details not spelled out that would need to be to judge it fairly. What for instance is the exact function of the 17 PR districts? Are they to elect 2 or 3 at most members out of 45 completely decoupled from the reduced but still predominant number of 80 plurality single member districts? Probably not, that approach has some other name than MMP. MMP as I understand it takes the party vote, uses that to set the overall membership targets for each party, then subtracts the single member district race seats won for each and awards the remainder to each party’s party list. And has evolved variations permitting voters to somehow influence those party lists, at the cost of a further element of voting not in any evidence.

Reformers often play fast and loose with terminology, either because there is no generally agreed on terms for various concepts, or perhaps sometimes manipulatively.

For instance, a couple decades ago it was pretty common to see IRV touted under that label, which is a fair label because it describes what the system attempts to do–seek a solid 50 percent plus one Droop Quota majority for one candidate, and hold that out as superior to simply accepting the largest single vote bloc in single choice–which I think is very aptly and fairly described, evocatively, as First Past the Post. There is in fact a “post”–the concept is, whoever has the lead when all ballots are counted deserves the seat more than anyone else does, having the largest base of support. The same format applies if the race is multimember; first, second, third, etc in total vote ranking wins. I see this applied not just in West Virginia today but in some US territory autonomous territorial legislatures, these latter on a much smaller scale electorate and house size than typical in US states of course. FPTP as a description is pejorative only if one questions the legitimacy of all people in a region being represented by whoever happened to command the largest single faction.

So, IRV attempts to guarantee that whoever does win does so with over half the total votes–“attempts” because of course in practice, lots of ballots get exhausted before someone is elected so the Droop Quota drops below half the number of ballots initially cast. The more weighty issues I have with it are 1) even 60 or 70 percent majorities are tossing the remainder of voters under the bus; if their votes can be conserved, they should be, and 2) it is highly dubious to claim that a person’s second, third, or lower rank choice winning with one’s low ranked support is exactly equivalent to that voter’s first choice winning; we should be discounting the weight of lower ranked choices–but that does more harm than good unless we can resurrect the electoral prospects of the eliminated alternatives in a fair and coherent way.

Anyway, Instant Runoff Voting does graphically, plainly and accurately describe what is happening in a single winner RCV option. It is no more satisfactory (and not less either) than two round runoffs, and has the advantage of one voting event for one hopes, maximum turnout, versus the terrible turnout bottlenecks of two round systems–generally the runoff is an optional second election with the low turnout, but California and Washington’s “Top Two” aka Jungle Primaries have extremely low turnout for that first election, that constrains the choices for all voters in the second “real” general election, and this tends to depress turnout for that too as well as other pernicious effects.

Either way, the premise of a runoff system is that “majority representation”=“fair representation,” and for someone who sees a flaw in that reasoning, it actually serves mainly to manufacture consent by persuading the losers they don’t deserve any direct representation whatsoever. Whether accomplished in IRV or in two rounds, a substantial number of voters must disregard their real prime choices completely and simply choose which of two options they already rejected are less unacceptable to them, so counting them as part of the “majority” inseparable from those who actually wanted the winner as their first choice seems awfully mendacious to me.

But of late, all of a sudden we see advocates who turn out to in fact be presenting IRV plans presenting them as “RCV.” This is technically true, IRV is as noted, Ranked Choice Voting to elect a single winner. But while say 2-member RCV, that is to say Single Transferable Vote in 2-member districts, is a pretty crude and lopsided and perverse way to take a step toward proportional representation, it certainly does take a single inadequate step in that direction. It is worthwhile, if one is being honest, to maintain a sharp distinction between IRV and even 2-member STV, although both look pretty much the same to the voter casting a ranked choice ballot–the only difference being, how many choices there are to rank.

So the recent downplaying of the term IRV when in fact concrete proposals being put forward are settling on IRV strikes me as quite possibly a cynical move, Machiavellian if not Orwellian, to manipulate various decisionmakers and publics with creative marketing. Some people out there in the masses have some vague idea that countries like say Ireland “have proportional representation” and use RCV, so labeling IRV as RCV both avoids association with negative experiences with IRV in the past, creating a false impression IRV is a step toward PR, which it decidedly is not. except in the vague sense of getting voters used to RCV, but if so why not push directly for STV immediately? Unlike IRV it was in fact employed in numerous US city governments for over a generation between the early 20th century and mid-1950s.

Always best to go to the source. It wasn’t as easy to obtain as I think it should have been–it being apparently customary to refer to a current bill in debate by its bill number, searching the Assembly site form for “tabled” documents (the press articles do mention it being “tabled” by Sonia LeBel) should have pulled it up, as should a general internet search, but no, most hits are for a previous session’s Bill 39, and the Assembly search comes up empty with that term. Using the bill author of record’s name, Sonia LeBel, works to pull it up, linky to PDF here.

Sometimes my browser pops open a PDF in a window, other times it downloads it and it opens in my document viewer (Preview, on my Mac). This was one of the latter.

Best I terminate this long reply and clean sheet a new commentary on the bill as “tabled,” whatever exactly that means in Canada, currently.

“Tabled” means it is being brought up for debate, i.e. the opposite of what it means in the US.

“379.1. The Chief Electoral Officer shall allocate the first regional seat
to the independent candidate for a regional seat or the authorized party who or
that obtains the highest quotient when the number referred to in subparagraph 1
is divided by the number referred to in subparagraph 2:
(1) in the case of an independent candidate, the total number of votes in the
candidate’s favour or, in the case of a party, the total number of votes in favour
of the party’s regional list; and
(2) in the case of an independent candidate, the number 1 or, in the case of
a party, the result obtained by adding the number 1 to half the number of
candidates of the party elected as division Members for that region; however,
if that half includes a decimal, the decimal is rounded off to the higher digit.
The calculation is repeated for the allocation of each subsequent seat.
However, each time a seat is allocated to a party, the number 1 is added to the
divisor mentioned in subparagraph 2 of the first paragraph, for the purposes of
the subsquent calculation applicable to that party. The Chief Electoral Officer
shall allocate each subsequent seat, up to the total number of regional seats to
be allocated for the region, to the independent candidate or to the party who
or that obtains, in a given allocation, the highest quotient.

So if I’m reading this correctly, the divisor is 1+0.5(riding seats won)+1*(previous district seats won). This would seem to be more favorable to larger parties than D’Hondt. Especially since large parties are known to be overrepresented in single-winner district results (in comparison to their vote proportions), and it is these seats only that are made cheaper than winning a seat in normal D’Hondt.

I’m not convinced this is proportional.

“379.2. To participate in the allocation of regional seats, an authorized
party must have obtained, across Québec, at least 10% of the valid votes cast
in favour of all its regional lists of candidates.

Wow, that’s a high threshold. I’m not sure why they’d need a threshold when their divisors are more favorable to large parties than D’Hondt, but

Yes but I would not out anybody. I have seen this in a number of communications but more so in private communications. The basis of it is maybe a little difficult to understand from a enlightenment/Individualist standpoint. It is instead based more on Hegelian roots and forms what we would call identity politics. Essentially they believe there are good an bad groups. The goal would be to design a system which forces the people who are neither “good” or “evil” to align with the good people against the evil. I can go into further detail but the point is that there is a basis for their goals of increasing polarization. This is why I think it is very important to be clear about your ideological underpinnings because they will influence the system you desire.

I disagree. The worst system is Party List and the second worse system is Single Member Plurality. Mixing them will not make a system which is better than either. SMP is a crappy system, adding partisan voting to it is not going to make it better. It will exacerbate existing issues. But thats just me and my liberal position.

OK, let’s test this method on some real world data. Since the ridings under this proposal are about as big as in federal elections, I’ll use the projections as data. There should be 4-5 ridings per region and 2-3 regional seats. So I’ll use the 5 ridings in/ just around Quebec City for my test case. Specifically:

The rounded vote totals (I don’t need that much precision to make my point here) in each riding are:
Charlesbourg–Haute-Saint-Charles PC 45 Lib 22 Bloc 13 Green 8 NDP 7 PPC 6
Louis-Saint-Laurent PC 47 Lib 21 Bloc 12 Green 7 PPC 6 NDP 6
Beauport–Limoilou PC 31 Lib 29 Bloc 19 Green 9 NDP 6 PPC 5
Québec Lib 32 Bloc 24 PC 23 NDP 10 Green 9 PPC 3
Louis-HĂ©bert Lib 38 PC 24 Bloc 21 Green 8 NDP 5 PPC 4

Assuming the ridings are roughly the same size (again, precision is beside the point), summing these percentages gets an estimate of what the party vote proportions might look like for this region.
So the aggregate is PC 170 Lib 142 Bloc 89 Green 41 NDP 34 PPC 24.

If we used D’Hondt party list for all the seats (i.e. no district seats), we’d get PC 3, Lib 3, Bloc 1 or 2, depending on whether there are 7 or 8 seats in the region.

Applying the method proposed in the bill: since only the PCs, Libs, and the Bloc are projected to beat 10% throughout the province (the NDP is close, but it doesn’t matter, they wouldn’t win a seat anyway), only those parties are considered for winning the regional seats.

Since the PCs won 3 ridings and the Libs won 2, the PCs start at a divisor of 3 (rounded up from 2.5, and the Libs start at a divisor of 2. We can expect that there will be either 2 or 3 regional seats (probably 2).
So the first regional seat count is
PC 170/3=56.6 Lib 142/2=71 Bloc 89
So the Bloc gets a seat. Their divisor increases to 2.
The second regional seat count is
PC 170/3=56.6 Lib 142/2=71 Bloc 89/2=45.5
So the Liberals get the second seat. If there’s a third seat, it clearly goes to the PCs.
So the final seat count is PC 3 or 4, Lib 3, Bloc 1
In the 5&3 case, this would appear to violate Droop Proportionality, since the Bloc had 2 Droop quotas (once the parties below the threshold have been ignored). While several other serious party-list PR methods don’t pass Droop proportionality either (Sainte-Laguë, Hare largest remainder), when they fail Droop proportionality, it is because they give a smaller party a seat that would need to be given to a larger party to maintain Droop proportionality. So I think that it is a stretch to call this proportional.

Note: Bloc Québécois doesn’t run in provincial elections, which this method would be used in. However, Parti Québécois does, and the two parties are closely tied.

1 Like

More MMP support federally. Also, says he will try to ram it through without referendum or tabling a bill. Scary stuff.

Doesn’t the NDP have no chance of forming a government, since it splits votes with Trudeau’s party?

They will make it a condition of coalition. He says as much in the video

The whole idea of a coalition formed by legislators, rather than voters seems backwards to me; it just creates conditions like this where a minority party can play kingmaker or otherwise destabilize things. The increase in responsiveness only occurs for the voters who are willing to play around with government for their ideals, which ties back to what you’re saying PR advocates really want for everyone. Is there any way Canada can try out Sequentially Spent Score or other cardinal PR systems at a local level? It’s crucial that these new systems can be placed into current discussions around voting reform in a way that people can fight for.

One interesting thought is that if a system like SSS was used on a federal or provincial level in Canada, the government premiers would generally tend to be the candidates who had been elected with the most points in their districts i.e. the consensus Score winners.

1 Like

Government formation also needs reform. I think i have a thread about that here

They would need the province to support it, probably. I think Canadian municipalities are less powerful than American ones.

You’re way ahead of me. Indeed that so-called method for the 17 regions (note, the language of the bill calls the 80 FPTP districts “divisions” and the 17 superdistricts “regions”) strikes me as a pretty appalling ad hoc hash. It seems to vaguely approximate Jefferson’s method.

Now having no data on a current “division” basis, let alone knowing anything about the purported 80 divisions and 17 regions (beyond what we can infer about the specially treated peripheral regions, and noting that the basic apportionment algorithm for number of divisions per region and number of surplus region seats are both by Jefferson’s method, that is successive divisors 1,2,3 etc (that’s how the bill defines it) I salute you for taking one sample example.

Looking at it more globally, I can show what would happen if Quebec were to adopt a system more like mine. Not exactly what I would do; I’d either make the number of seats float above 125 for greater representation on a top-off basis–and some 50+ seats would be needed to compensate for CAQ’s really massive lead over their proportional share however we figure it–or create 62 divisions for the FPTP single member races, then figure proportionality (using Hamilton-Hare of course, though one might live with Jefferson by throwing small parties and independents under the bus, and being less inclusive–or do that hybrid thing where we determine which parties are quota, which are subquota, find the share of total seats for each portion, do Jefferson on the quota parties and either give the subquota seats one each to the greatest remainder parties below quota, or open up asset voting, but that’s several layers of novel wackiness).

OK, to compromise a lot with conventionality, and not the special peculiarities of the bill all of which seem egregious:

  1. one point not to compromise is the Goddamned hurdle! 10 percent is not just a little high, it is outrageous. Only Turkey to my knowledge has such a monstrously high bar set. No reason for hurdles whatsoever, especially if we use Jefferson with no reservation of seats for the subquota contenders.

  2. given the nature of the data I have, I cannot guess how Quebec voters would respond exactly to greater freedom of effective choice, and going by the 2018 data for the province as a whole (I followed the links to the official sources, but they provide no data in useful form for these purposes than Wikipedia’s detailed list by party) in that year anyway, votes for independent candidates were negligible. They are included in my tables as though they were all votes for one candidate, but that imaginary candidate (actually there were 21) comes nowhere close to winning anything. As I have shown elsewhere, this is not a general feature of approaching positive representation by regarding votes for an individual candidate as votes for their party, and topping off–if we use Hamilton’s method, quite a few independents would be elected even with current US and British voting patterns, and I expect far more once people are no longer afraid of “throwing their votes away.” It would be a long shot under Jefferson, unless an independent does so well they simply win the plurality in a single member district (here, “division”) race. This is Quebec voters choosing to vote for candidates who happen to all be clearly associated with a party. Therefore, given such voting habits, it is no great difference to accept MMP and a second, separate party vote decoupled from voting for the division candidate. Mind, I anticipate trouble aplenty from this to me pointless separation, but certainly it is not problematic with 2018 data!

  3. I prefer to see 50/50 balance (or with odd numbers as in Quebec’s assembly, the extra seat going in the make up portion) or conditional top off to counter overhang, but let’s just roll with the decision to reserve 45 seats for top off and 80 for division seats. That works well with the 2018 voting pattern, if perhaps not generally or in the future with voters attempting to use the system to their legitimate advantage as the bill preamble chirps about.

  4. All the government has to do is apportion the 80 divisions of Quebec. No purpose is served by the 17 regions. Now, they are supposed to somehow reinforce the identity of the regions or some such, but I think that purpose would be filled quite well by 80 divisions; if the regions have informal significance, culturally, economically, politically or otherwise, patterns of voting in the divisions that would be associated with such regions suffice to give the regions pragmatic significance as something for electoral strategists to reckon with, and asserting their power. For once I am largely in agreement @Keith_Edmonds; the 17 regions appear to be arbitrarily there to handicap small parties and leave the big ones captive of central party insiders with their constituencies having practically nowhere to go. As you correctly estimated, 125/17 gives us a little over 7 and 1/3, and 45/17 is about 2 and 2/3, so if 11 regions have 3 regional seats, then that leaves 6 with 2, and if the regions were close to equal in population we’d expect about 4 or 5 divisions in each–actually I think the regions will vary in population quite a lot, being more like US states in their relative role as subdivisions of the whole. Anyway if the goal is to use MMP approach to proportionality, they are pointless. With 6-8 seats total, the lion’s share of which are determined already by division FPTP, there is little scope for useful proportionality in any region, and if in fact we wind up with a great variance in division numbers, we probably have the kind of conflicts we have trying to figure out proportional plans for the US House of Representatives–large state (or region) people wind up underrepresented and yet having more scope for approaching PR, small region people are pushed into nearly FPTP but have disproportional power in the body.

Versus all that, I say just apportion the 80 divisions across all Quebec on a straight equal population basis–and note that due to the PR leveling process, we can be pretty sloppy about that. So itty bitty regions that for geographic or traditional reasons or reasons of court order must have a division of their own, never mind they are way below Hare quota of the whole provincial population divided by 80, do little harm, nor does underrepresenting the people of say Montreal–it comes out in the wash. Underrepresented districts tend to win more make up seats.

So having defined 80 divisions, assuming for exercise purposes that the overall pattern of FPTP victories in those 80 regions maps proportionally to the historic pattern of wins that way with 125 divisions, we get

Historic wins Out of 80 Jefferson out of 125 Hamilton 125 Hamitlon 127
Coalition Avenir Québec 74 48 48 47 48
Liberal 31 20 32 31 32
Parti Québécois 10 6 22 21 21
Québec solidaire 10 6 20 20 20
Green 0 0 2 2 2
Conservative 0 0 1 2 2
New Democratic 0 0 1 1
Citoyens au pouvoir du Québec 0 0 1 1

I put in that last column because assuming CAQ would win 48 seats FPTP out of 80, they would still have one in excess on the basis of 125 seats; it is common in MMP nations to do some kind of overhang resolution–here we just note they had one extra seat, so that means add two (thus automatically preserving the evenness or oddness of the house size) and refigure proportionality on the basis of here 127 seats–then CAQ’s PR share rises to match its division victory share, leaving other parties to get the make up seats.

Because Jefferson’s method already gives CAQ an extra seat and leaves the 2 smallest winning Hamilton seats out in the cold, there is no need for overhang resolution with that rule, with these election results.

The four largest parties each ran 125 contenders and would therefore presumably have run 80 in the alternate system, and there is no need for any separate lists of “regional” nominees. CAQ does not get any make up seats; the top 11 (or 12, using Hamilton for 127 seats) out of 60 Liberal contenders who lost in their divisions would win on top off basis; as would the top 15 (or 16, under Jefferson) candidates for PQ, and 14 under all systems of QS. Even under Jefferson, the Greens of Quebec would see their top two vote winners, none of whom won any seats FPTP, win seats under make-up while the Conservatives would gain one under Jefferson, and 2 under the Hamilton variants. Under Hamilton, either version, the New Democrats and CapdQ would each see their top winner in whatever division that was for each win a seat.

If we were to instead retain the current 125 districts but add proportional expansion seats in the same ratio as the tabled proposal, that is 45 per 80, that would come to just under 70 and 1/3 seats, rounded down for a slightly lesser expansion ratio would raise the Assembly to 195 members. Obviously that saves the province the trouble of reapportionment, at the cost of an expansion of the house–which strikes me as a perfectly good thing to do, but obviously involves a one time round of expenses expanding Assembly infrastructure, and ongoing costs maintaining nearly 200 members–which, given the empirical realities of modern industrial republicanism in a global high tech interconnected world involving what I think should plainly be ongoing and indeed expanded state expenses of a social-democratic as well as typical state duty functions nature, and the need for more and not less democratic-state regulation and intervention in general social affairs, would all be quite sensible investments indeed. More democracy, not less, is what is needed generally and thus for Quebec in particular.

If I made that assumption and had the district by district candidate performance data, I could state precisely who these total 195 delegates to the Assembly would be, by name as well as party affiliation–where the handful of representative MNA’s for the bottom 4 parties would come from and just who they would be.

Lacking data on candidate totals district by district, I cannot do this, still less try to estimate how well they would correspond to the candidacies in only 80 districts expanded to 25/16 the population for each to make up 125 total. Presumably consolidating the existing divisions from 125 to 80 would tend to eliminate some structural biases in the existing divisions by semi-randomizing the makeup of the consolidated ones. If someone were to define 5 regions of Quebec made up by aggregating 25 of the current divisions, we could assume each of these are subdivided into 15, and anything applying to each block of 25 contiguous districts applies to the 16 formed from them taken together.

Which suggests yet another compromise–instead of defining 17 regions. Quebec could define 5 by equal population in each, and declare 16 divisions in each, or say subdivide them into 4 multimember large divisions each electing 4, and electing 9 “regional” members in the regions. Note that yet another approach then would be to elect not 9 but 8 regional members, do away with the regions and elect not 4 but 5 members by some proportional approach in the 20 larger now 5 seat divisions, and hold in reserve 5 seats to achieve province-wide balance as adjustor top off seats.

I think 5 reserve province wide seats would prove inadequate and demonstrate the fallacy of achieving “proportional” representation by summing up separated districts on any subdivided scale.

How much consequence is there in any of these reforms–as Tabled by MAP LeBel of CAQ, as modified by me in either Jefferson or Hamilton forms? I am going to drop the 125 seat Hamilton version because CAQ having 48 seats when proportionally entitled to only 47 is simply not acceptable, depriving another party (specifically the runner up rival Liberal Party) of a seat thus double dipping the imbalance as it were, and topping off by small expansions of the body is quite part of the standard worldwide MMP package purportedly being imported here.

To begin with, look at the total percentage of the electorate gaining any representation whatsoever respectively under current FPTP, Jefferson MMP or Hamilton with 127 MNAs. Historically under the current system, just 4 parties win any seats whatsoever. This is 95.4 percent of the whole electorate. This does not seem so bad, but recall that of course with FPTP “don’t throw your vote away” strategic logic at work, many voters are constrained in their practical freedom–that is inherent in the data, and we can only take note, but in truth the representativeness of the Assembly is actually less than the numbers make it look by any metric for this broad reason–nor can we draw absolute conclusions about the ratios among the smaller parties some people did vote for. We just don’t know for sure how people would have voted under any reform proposal.

Using Jefferson under my urged systemwide integration of all votes taken as party preference votes, the resulting inclusion of the next two largest parties, Green and Conservative, raises the inclusion of all voters by 3.15 percent. Note that both these parties, getting a bit over 60,000 votes provincewide, have nearly two quotas each with 125 total seats and would be clearly entitled–unless of course we indulge in the idiocy and mendacity of a hurdle of any large magnitude, and the Tabled 10 percent hurdle is of rather monstrous magnitude indeed! With the total vote being over 4 million, the provincewide hurdle magnitude would be about 403,000 and this would have the effect of limiting all distribution to only the same top 4 parties that won seats FPTP. With between 3 to 6 seats denied respectively 2 and 4 parties under either Jefferson or Hamilton with no hurdle, those seats would of course go to the top 4 parties.

Meanwhile, without data on the exact nature of the 17 regions and 80 divisions proposed, we cannot work out definitively the exact difference between these subdivided pseudo-MMP assignments and even the most restrictive Jefferson version, but there is every reason to expect it to bias the outcome toward the larger and away from the smallest parties included even further–though in fact even the least included party, QS, can count on retaining some division and gaining some regional seats.

Moving on to Hamilton, in terms of raw inclusion as a metric, versus Jefferson two more parties, each falling well below Hare quota based on 125 seats and even put together well below 2 quotas, would win 1 each, thus raising the total included percentage to over 99.45 percent.

To review then, if we assume the remaining just under 0.35 percent not included, coming to about 22,000 voters and thus as a whole under a Hare Quota (about 2/3) cannot reasonably be included by any decently balanced system (we could go over to asset voting for a small reserve of seats, of course, to stretch inclusiveness of the body a bit more) then versus this minimal exclusion, standard North American FPTP unnecessarily excludes 163,566 beyond that who amount to over 5 Hare quotas, and Jefferson being taken for the PR method excludes nearly 127,000, the excess being nearly 4 Hare quotas.

Thus we see that the difference in total inclusion between plain old fashioned winner take all FPTP and a Jefferson proportional system is substantially less than the difference between Jefferson and Hamilton PR approaches.

And that, my friends, is why scientifically speaking, from a positive representation values point of view, I so strongly disapprove of Jefferson as the allegedly best, most reasonable PR approach and affirm Hamilton as best from an inclusiveness approach–it is not just a little bit better, it is vastly superior in its scrupulous conservation of all votes cast, and the deviations from Jefferson outcomes are a small relative cost to the large parties. At the same time, applied to data based on FPTP elections where most voters are forced to choose between two leading duopoly parties, or just a few large parties where regional factors change the identity of the two parties contending in their district, it is the case that the difference between Jefferson and Hamilton will have consequences–for a party that under Jefferson commands the majority of seats, will fall short under Hamilton. But that’s good–it reflects the reality of the voters, while Jefferson’s shift reflects what amounts to an elitist bias, shifting power undemocratically into concentration of large parties that manipulative, coopting elites can concentrate their gatekeeping attentions on and that voters have disincentives to defy.

Against this, the objections raised of fissiparious tendencies seem minor at best. Furthermore, some mental analysis demonstrates to me that all other PR algorithms but Jefferson all have the same “hazard” to some degree, under any of them-I mean here specifically Webster-Sainte Legue and Huntington-Hill, others that are more esoteric might not have this feature–splitting a given bloc of voters can result in the two pieces gaining a seat. More rigorous math is needed to analyze whether the different methods vary significantly in the degree to which they pose this “hazard,” but I honestly think this tendency is nothing to worry about even at worst degree, and the degree is pretty low even with Hamilton, which might not even be the worst. It seems that valuing “thou shalt not gain by fission!” is of a piece logically with valuing “thou shalt have few large parties!” And if instead there is truly a consensus I defy here against countenancing alliances of candidates being recognized in any form whatsoever, I think that favoring anti-fission methods, which is to say Jefferson alone, is either highly irrelevant or will in fact permit de factor parties, in the form of distinct self-conscious factions among elected members, who can be targeted by elite interests for cooptation and thus censor legitimate minority interests, to form in less observable and therefore more difficult to counter politically by democratic activism.

Meanwhile, as all systems discussed with any specificity here all continue to be based on a single voter casting a single choice vote (or technically, two choices, one for a division candidate as a person, another for a party, but I assume for comparison everyone does in fact select the same party as that the candidate they vote for is running under), it is meaningful to compute Gallagher coefficients for the various Assembly forms. GC takes into account not only which voters are locked out of any representation whatsoever, but also the magnitude of the differences between outcomes and actual vote shares, on a partisan basis of course.

The Gallagher Coefficient is a sum of squares based measure, measuring the deviation between the number of seats a party wins and the share of its actual vote in the body, and summing them all for a number that is a share of the whole electorate, which can be represented as percent of the electorate, absolute numbers of voters, number of seats, or fraction of the body, that is out of alignment with the actual recorded vote. US and British Commons votes tend to have GC around 7 percent–again bear in mind, there is no objective way to measure definitively how much the apparent near satisfaction masks discontent due to voters having to compromise on lesser evil. Also, 7 percent can easily be the difference between a party controlling or not controlling the body unilaterally, and often is.

The current Assembly elected under standard rules exhibits a GC corresponding to 17.7 percent, which is awful even as FPTP bodies go! This amounts to over 714,000 votes, enough for over 22 seats, being in effect disregarded in favor of other factions benefiting. Using Jefferson for overall proportionality hones that down to just under 1.2 percent, a mere 48,000 votes deviation equivalent roughly to shifting 1.5 seats off track out of 125. And using Hamilton would instead cut these numbers down to 3/5 of a percent, or 24,000–about 3/4 of a seat. Note that this includes the total failure to include some small fraction of the voters noted above.

My consideration of including Asset voting for final cleanup was based on noting that even Hamilton leaves a substantial number of voters dissatisfied, although obviously in far lower than the huge percentages of voters typically thrown under the bus by FPTP methods. At some point one throws in the towel of total representation in any republican system, but I think that point should be pushed toward as much inclusiveness as feasible on principle, and allowing an Asset vote to settle the final seats gives even the smallest vote winners some aspect of influence, and thus can win them and their constituencies long term if only partial consideration throughout the term. The straightforward way to determine which seats are up for Asset Voting grabs seems to me to determine the share of all parties (including of course independents, regarded as parties of one candidate each) not amounting to a Hare Quota, taken together, to determine using what amounts to Hamilton’s method their share. Note the fissiparious phobia here does not apply at all; no meaningful faction seeks to be the “block of parties getting less than quota votes” and the only split possible is between those getting more and those getting less, so the issue just does not arise, and the simplicity of Greatest Remainder division and its plain transparency should prevail, as well as its inclusiveness.

So given these data, as noted, the same 4 parties that won FPTP seats, plus the next two (Green and Conservative) are the Hare quota parties, and so the same percentage as those left out completely by Jefferson approach, a bit under 3.15 percent, would be the share of 125 to reserve for the Asset round. That amounts to 4 seats. If we were to then bow to the pressure to use Jefferson instead of Hamilton for the quota seat victories, we would be distributing 121 seats to that bloc.

However, that would bump CAQ’s share down to 46 seats, two fewer than they are estimated to win FPTP in the 80 districts, so we need to refigure based on 127 seats in both steps as in Hamilton for the whole. This does not raise the Asset share above 4, and so distributing 123 seats to the 6 quota parties via Jefferson’s method gives the same quota seats to CAQ as Hamilton’s method with 127 total does, 48 thus ending the need for further anti-overhang topping off. Versus Hamilton for all with no Asset phase, the second and third parties come up one short, and the Conservative party is down to one seat from two, leaving the last two parties included under uniform Hamilton to try and win one (or more) each of the Asset seats to restore their wins under that system.

An Asset phase I think should include all parties, including those who have been awarded quota seats in any number. We’d subtract the number of seats won times Hare quota for 127 seats from each party’s total, and whoever manages to persuade others to give them more than a Hare quota when combined with their own remainder will sequester another seat, and any remaining seats will go to the top remainders then after transfers. The upshot of this subtraction is that all parties except CAQ have positive remainders, with the Conservatives having the largest.

Parties ranked In Hare quotas
by largest remainder
Conservative 0.86
New Democratic 0.72
Parti Québécois 0.66
Liberal 0.52
Québec solidaire 0.45
Citoyens au pouvoir du Québec 0.43
Independent 0.20
Bloc Pot 0.15
Green 0.14
Parti nul 0.12
Marxist–Leninist 0.05
Parti libre 0.05
Équipe autonomiste 0.04
Parti 51 0.04
Changement intégrité pour notre Québec 0.02
Alliance provinciale 0.02
Voie du peuple 0.01
Parti culinaire 0.01
Coalition Avenir Québec -0.47
Total (except CAQ)
Total of top 4 2.76
Positive remainder in play 1.71

From this we see that if no one takes action to change the outcomes by transferring votes, the same 4 parties that fall short of Hamilton with 127 seats will each make up their shortfall and the final distribution is identical to Hamilton. All parties except CAQ have the opportunity to offer vote shares to others in negotiation, or to try to persuade others to beef them up to gaining a seat, or theoretically all 4–indeed even CAQ could gain those 4 if every other party agreed to transfer all their surplus votes to that party. Needless to say this seems an unlikely outcome to me!

If you have read this far, you can see that NDP is dead last in the most inclusive, Hamilton with 127 seats, approach, getting the very last single seat. It could be that with voters knowing their votes will be pretty well conserved, they might have attracted more under these rules, but considering how very stubbornly the voters refused to be pushed into the 2-party box, the vast majority of them making it a 4 party race, and enough voters for nearly 6 seats by Hare Quota still refused to vote in that larger box either, it seems fair to say NDP is pretty unpopular in the Province!

I suspect this has something to do with voters who broadly agree with NDP’s general program and values being more attracted to a Quebec party that also includes some major emphasis on Quebec autonomy and opposition to federalism–CAQ’s own leader has ruled out seeking actual secession, but then remarks that having voiced this limit, many people supporting CAQ seem not to have listened to him.

Looking at outcomes then, the upshot is that in any properly proportional system, CAQ would not in fact have won “majority government” status in 2018 despite the spectacular rise in their vote–26, or over a third, of the seats they won come under what I define as “spurious” majority. Needing 63 seats to rule and being entitled to 48 at the most, an amount equal to what they would be most likely to have won in the divisions alone if they ran in only 80 divisions with the same FPTP pattern as historical prevailing, they would need to find 15 partners of other parties to gain control, which then would not be unilateral of course. They would have 54 seats to try to obtain this support from between the Liberals and Party Quebecois. From reading about the parties at Wikipedia, my impression is that the major split between CAQ and Liberals, the runner up party with more than twice the number of seats a CAQ led coalition would need, relates to Quebec separatism and presumably Francophone dominance; indeed one reason the CAQ leader gave for taking secession off the table is to attract Anglophone voters from the Liberals. So if the Francophone majority could accept that they are pretty securely in power and can get along with Liberals on issues of relevance for current priority, trusting the Liberals to be reasonable in the social matters of balancing French and English and associated cultural stuff, the outcome is to steer Quebec with a solid consensus majority toward conservative (by Canadian standards anyway, probably would not seem sufficiently reactionary for USAian right wingers!) autonomist but federalist membership in Canada, and the joined forces would be focused, on Federal matters, toward moving Canada as a whole toward those same moderate-conservative values.

Whereas, the alternative of seeking alliance with Parti Quebecois, which with 20 seats has adequate numbers for a lesser but secure majority assuming no defections or rifts, the emphasis is to become more agnostic and hence presumably in practice compromising on left versus right general issues, but taking a more firmly autonomous and separatist stance re Canada as a whole, and indeed have coalition members flirting with outright secession–those would be PQ people mostly though it might give license for CAQ people to break ranks with the party founder on this point if not others. PQ has a long left-wing legacy, having applied for but been blocked from membership in the socialist Second International, though that was quite a long time ago when I was a wee little lad, back in the early '70s, and IIRC any general social-progressivism is somewhat compromised, as I judge things, by indulgence by some in a general disapproval of displays of Muslim identity–the Wikipedia articles discussing a sentiment of supporting existing provincial bans on Islamic headgear and more people in the parties discussed wanting harder crackdowns on Islamic identity rather than relenting.

It is the nature of seeking positive representation generally that when spurious majorities are foiled, that if no single party or obvious coalition has a majority in the body outright, that in theory another party could form a rival majority. However, it is often also the case that realistically, taking into account the party values and programs, that such coalitions are often impossible despite being feasible on paper. If the current British House of Commons had been elected proportionally for instance, it would be possible for either the Tories or Labour to lead a coalition–but either would have to coalition with the Liberal Democrats, and then that would not quite be adequate; the Conservatives would have fewer seats to make up beyond that but would have far fewer, I judge to choose from as coalition partners; with LibDem alliance, Labour would be shorter of the goal but would have adequate numbers of likely parties (most of the nationalist parties outside Northern Ireland for instance caucus with Labour in real life–that is Scottish National and Plaid Cymru; Northern Ireland has its own Social Democratic party that is in practice actually a branch of Labour)

Here in Quebec based on 2018, however, it is still a CAQ game. The Liberals cannot expect PQ to join with them, being diametrically opposite on what appear to be the most salient issues, nor would that coalition close the gap by any means, being about a dozen short, but below PQ there are only 25 seats in play total even under Hamilton, which is simply not enough. That’s before we even get into noting how the Liberals would be just as opposite the views of most of those 25. PQ by itself is 41 seats short, and without any members from the top two parties could not possibly lead. Practically speaking, then CAQ would still lead based on honest proportional election, but would absolutely have to scrape up support from outside itself. Can they bypass both Liberals, PQ and QS too? No, there are only 5 seats to be found below the top 4 parties. If CAQ could somehow come to agreement with QS (the large party I judge I would be most likely to vote with were I settled as a voting Quebec resident, though their lean to separatism somewhat troubles me on broadest principles–given specifics of the Francophone-Anglophone rifts in Canada I am not familiar with, I cannot reasonably judge the particular case from where I currently sit–it is like my personal view Puerto Ricans would probably be better served with US statehood than independence, I defer that judgement except as a tentative if strong personal opinion to actual Puerto Rican-Americans) they would have less trouble and more profit in doing so with PQ, and making their coalition giant tent and pointedly leaving the Liberals out of it would have the effect of dragging the dominant partner farther to the left than they evidently want to go (or they would just stay in PQ!)

1 Like

It isn’t Hare proportional either, even if the threshold and potential for overhang is ignored.
Consider a region with 4 ridings and 2 region seats.

Division A B C D
Seat 1 37 32 17 14
Seat 2 37 32 17 14
Seat 3 32 37 17 14
Seat 4 32 37 17 14
Region 138 138 68 56

Party C has a Hare quota, since there are 6 seats total, but Party A and Party B start at divisors of 2, so they win the 2 region seats. (They each have 69 votes once the divisor is applied).

Democratic primaries have a 15% threshold.