June 10, 2019 at 9:29 pm #26095
I agree with pretty much all the recommendations in Réal’s paper.
My own criteria for MMP is basically that any implementation is acceptable if it ensures highly proportional outcomes and minimizes the risk of tactical voting. I’d be okay with a moderately proportional version of STV (e.g. 4-member districts everywhere). But for MMP, issues can arise if the results are not highly proportional.
A very simple MMP model that meets my criteria is to use closed regional lists with a single-vote ballot. But I recognize that having 2 votes and a bit more choice is an expectation for many. Réal’s proposal is a realistic way of implementing open-list MMP without risking unintended consequences (i.e. distorted results, tactical voting, the “Bevan-Baker problem”, etc.). It makes sense to me to mention more ambitious features like biproportionality and ranked ballots, but not to strongly recommend them unless perhaps there’s an indication they’d be seriously considered.July 7, 2019 at 4:58 pm #26290
Like Tony, I was invited by Réal to participate here. Well, at least he pointed me to this discussion board after years of me offering my opinions to him and Anita (and Kelly and Gary and Wayne and Larry…) I expect that means he thinks my participation might be useful. So… here goes.
I should start by saying that my observational perch lies a fair distance outside the MMP “bubble” and, to be honest, beyond the PR “bubble” to which FVC confines itself. That said, my electoral system ideal — the “bull’s eye” on my electoral reform target — is that every voter gets a rep of their choosing and every rep represents an equal number of voters. This is a definition that embodies the underlying (foundational) principle [sic] of PR (inclusion) without necessitating making the special dispensation for parties that PR demands. In other words it is more a first principle than is PR. As well it inherently covers the problems that Réal is struggling to fix in MMP (which revolve around the Charter right of equality) but which both PR and MMP in their simplest terms ignore. A system that delivers on that ideal… or in the real world approaches it most closely — solves all of Réal’s problems and should be the one ER advocates propose. And we might also choose to abide by William of Occam’s dictum as well.
And therein lies the essential problem. The premises underlying the choice of MMP and to a lesser extent, PR — and therefore much of FV advocacy — are actually antithetical to the obvious essential statement of principle. The decision to go with MMP… to hold it up as an exemplar — is the author of the problems, with the solution and with the process of reform. The troubles are self inflicted. And this is obvious to me from my seat.
I’ve suggested that FVC is operating in a bubble. In fact it’s much like a board game wherein the rules of play are internal and not necessarily reflective or respectful of the larger group (everyone else) whose larger perspective leads them to different choices. Thus, whilst in MMP Land dual candidacy seems to be an unquestioned given, out in the real world candidates running in more than one riding is naturally prohibited and I would say for very obvious reasons (cf the Parliament of Canada Act). I respectfully suggest that if you actually want to achieve electoral reform rather than trying to force the larger public to play your game using your counter-intuitive rules, you start listening harder to people who haven’t bought into playing the game and aren’t likely to.
Of course these are exactly the people you need on board.
The easiest way out of the design conundrum is, like “War Games”, not to play. The central conceit of MMP is the single-member local riding and it is that premise that is at once the central misstep and the core roadblock. But I won’t go there… yet. Instead, let’s look at Réal’s essay and look at the problems it raises.
“Design Considerations” tries to navigate us through the weeds to solutions for two problems: compromised proportionality and effective voter choice (and the related issue of accountability).
I’ll address the proportionality one first because it’s the most straightforward. Réal indicates that he is against overhang seats because “there are better ways to increase the proportionality of the [regional MPP] system” than employing overhang seats.
Well I, for one, would really like to know what these “better ways” might be.
The maths are clear and, I think, uncompromising: proportionality increases with effective district magnitude. That caps performance of the design unless you “fix” things. The greater the district magnitude the greater the proportionality can be. All things being equal, compared to a national district (at large) approach, regional will deliver lower inherent degrees of absolute proportionality. Add in a proportion of single-member seats and the potential for plummeting proportionality performance increases dramatically. As the real-world results in Wales demonstrate.
The problem is not just bad luck. The smaller the region the more homogeneous the electorate and the more likely it is that local single-member elections will produce homogeneous, single-party results. And high disproprotionality. The obvious cure is to tie the proportion of regional seats to the number of viable parties contesting seats in a way that deals with the potential sweep. Therefore, if there are three parties competing the local seats should make up just a third of the total seats in the region. If there are four “contentious” parties, local seats should be only a quarter of the total. That’s just good engineering.
Failing that — if the objectives of the seat split include maximizing the number of local seats — then the design should include provision for overhangs… all the while assuming that proportionality is the grail. But if Réal has a better idea…
I want to spend some time on the notion of lists and in particular Réal’s idea of “fully open lists”.
Here again, the insularity of limiting oneself to the universe of the game board means that proven best practices magically become an anathema to the “game’s” premises.
Offer a list up to have a choice made and long experience teaches that to avoid unintentional spurious results and so end up with the true choice according to the intent of the choosers, there is one critical obvious feature to employ: you randomize the list so that the order of presentation doesn’t affect the outcome unduly.
Réal argues that allowing — nay, encouraging — the parties to order the list somehow represents providing “useful information for the voters”. I’m all for parties disseminating (true) information but this is categorically not the way to do this.
If a voter only cares about voting for their party of choice — implying that they couldn’t care less which candidate becomes their representative (or if they even have “a” representative) then, just maybe, their vote could/should reflect that too. It would if ballots featured Robson Rotation. Their votes would get distributed randomly amongst the candidates on the list. Perfect.
Now here’s the thing. If overall and specific voter intention is better articulated and reflected by using randomized lists then how is a party-ordered list “fully” open? The truth is, it isn’t.
And unfortunately this deficiency is compounded by the other design choices made here including dual candidacy and conditional tabulation of votes which both serve to close down what might otherwise approach an open list. Yet Réal’s “fully open” design includes both features. The only deleterious feature from which we are spared is flexible lists.
In short these lists are open in name only and in all fairness represent close to the least “full” open list that might be implemented. This package, then, has been mislabelled. And this is sadly consonant with Fair Vote’s take on lists generally. After the OCA in 2007, where the only meaningful feedback obtained (other than the rejection of MMP) was a rejection of “appointed MPPs” Fair Vote merely suggested that it would no longer push for closed lists. However, the other equally complicit active ingredients of “appointment” — dual candidacy and party-ordering of the lists — remained and have still remained essentially unexamined and undiscussed.
The upshot is that what might be considered as a “fully open” list within the narrow confines of the MMP game world are nothing like it to ordinary folk who live, work and play in the real world. And you wonder why MMP doesn’t get the traction you think it deserves from the voter in the street?
On a more positive note Réal has seen some utility in ordinal ballots… otherwise known as preferential or ranked ballots. This is promising.
Preferential ballots might represent a partial remedy for dual candidacy. If a voter’s regional candidate of (first) choice wins her local contest then that voter’s second choice might keep their franchise alive. Indeed, an “open” mixed design including dual candidacy but without an associated ranked ballot on the regional list side might be considered akin to malpractice.
I want to discuss one last aspect of a sufficient electoral system, one that aspires to meaningful articulation of voter intent at the candidate level and approaches the idea I outlined earlier and that is equality.
MMP as commonly configured is terrible at addressing equality. Here’s one reason. Typically under FPTP only half of all voters get a rep of their choosing. 50%. This low rate of performance will almost certainly persist on the local constituency side of the divide. In a typical 60:40 design that means that 50% of the voters will — if they’re lucky — obtain representation of their choosing by only 40% of the seats in the assembly. Profound inequality is built into the system where maximizing local seats is a design criterion.
The fact is that you only approach true equality and inclusivity (two sides of the same coin) when the mandates of each rep are high (>>>50%) and equal. One path to this result is traced by the Droop Equation and via electoral formulae that deal with surplus transfers in addition to the usual bottom transfers of ranked ballot counting.
I think that’s enough for today.
Hope this helps.
You must be logged in to reply to this topic.