Well as noted already in a perhaps less appropriate place than here, I actually thought in terms of fusing the district “Electoral” and systemwide positive proportionality “Party” vote into one choice, and I gather that some people lament, some quite a lot (calling it a “fatal” flaw of MMP) a mode of strategic voting in MMP that uses the split of Electoral and Party vote into two. Not that I arrived at the fusion this way, I just started with the simplest extrapolation of familiar US style district voting I could think of that would lead to proportionality.
All the criticism I have gotten does not typically even mention the issues I figured people would object to most strongly–one of them being that ultimate total body membership is decoupled from exact district proportionality, that is if we say take the approach of having twice as many seats as districts, there is no guarantee each district gets two representatives; geographical demographic proportionality has to be satisfied with just half the body being representative in that respect, and the origins of the persons who wind up topping off party delegations are in fact going to be from wherever they are strongest–in fact there is a tendency for many representatives to come from districts that are most divided. So with a doubled up house, typically about 70 percent of all districts get 2 reps, about 1 in 5 are stuck with just one, their geographical district rep elected in the first round alone, and the remaining 10 percent or so can wind up with 3 or more! I’d argue this is fair because the body represents the people, not the territory, and the people chose these reps, and it is good that districts that are divided sharply get many reps and those whose voters overwhelmingly approve one district rep can be very happy with that one; the job of citizens getting representation is done.
Of course we can accomplish much with creative labeling too. We could refuse to say the top off people are “from” the district they actually ran in and that they really are at large–but I don’t think that fools anyone, nor do I wish to be deceptive. I think it is just fair that the free form scattering or concentration of seats arising from voter choices works out that way and is more representative than pretending one person represents everyone in the place they came from, when we darn well know that typically an American Representative was lucky to get more than 53 percent or so (of an already depressed, low turnout election, more often than not). About 40 percent of the electorate win no direct representation in FPTP.
Trying to engineer the top off process so that good party reps of a give party come from places where support for them is strong, while also guaranteeing something like two reps per district exactly, is something of a bizarro world nightmare–I have developed procedures to try optimizing it but it never seems to work out in a really satisfactory way. Especially because when we are properly, inclusively proportional systemwide, the small parties elected (such as with 871 House of Representatives in 2016, 11 Libertarians, 3 Greens, and ten other single seat winners including a bunch of independents could have won, with the same votes actually cast historically) taking the second seat typically must preempt one of the two big parties who came in second whose candidate would have taken that seat by merit of the size of their vote had not the little party candidate gotten in the way. If we rig things to guarantee as many people as possible in the immediate region are satisfied, we block out the little parties and independents, and I value those quite highly as agents challenging the status quo and putting it to the test. So just letting the chips fall where they may in terms of where the top off members come from, prioritizing the strongest support, seems entirely right to me.
So–first of all, consider the problems of decoupling the Electoral and Party Votes. In normal MMP systems on the German model, district seats modify the Party contingent from vanilla party PR as practiced in say Denmark only by subtracting the number of district seats that party won FPTP from their share of total seats, meaning that a semi-predictable at best outcome of those Electoral votes shortens the party list of a party, but all the top off members beyond those are in fact from party lists, traditionally chosen by party insiders without the public getting any direct input. Various ways of modifying party lists including I gather giving voters a third vote letting them see the list and register their highest approved candidate, or fudging, as I see it, the whole question by shifting over to multimember district approaches such as Ireland’s STV vote, in which large districts apportioned 3-5 members IIRC use Ranked Choice Voting (as a new member @kevinbowman, you should be notified most everyone here really hates RCV–I probably hate it less than most but I see it as flawed at best, and actually want to figure out how to make cardinal methods work for the very question you ask here to bypass it–though as noted, I actually think a single choice can serve remarkably well, certainly better than RCV!) in Single Transferable Vote form to elect those handfuls. People say “that’s proportional and behold, no pesky party lists! No need for parties at all” but it deviates considerably from properly systemwide inclusive PR as one gets with classic no-district Danish style PR, or in MMP using traditional party list systems. It tends to filter out the small parties, and the proportionality is lopsided to say the least. It can come close to a fairly PR outcome with lots of district reps, but then we have quite large districts, kind of getting away from the point of a district. And it still is off the mark versus voter true intentions as I think we can measure.
Which brings me to another point that has great bearing on introducing “Cardinal PR.”
Think about this. I don’t know what party systems you are familiar with, I am guessing you are probably a USAian or else actually NZ, and good chance Canadian or some other Commonwealth country like Australia. Well, I trust you have at least heard of US parties since I imagine they make themselves pretty infamous around the world! So let us say I am a lifelong Democrat, and thanks to recent voter reforms (in some better alternate universe, these are all ahead of us in this time line! Well except for the fairly good ones–STV as it happens–we did have in city governments back in the mid-20th century until McCarthyism smeared them as a Commie plot and scared everyone off) I like the direction the D party is heading now, and want to stay the course basically. But a Green candidate I like and an independent who is pretty nifty too just so happen to run against a lackluster local Dem, and unlike my slate of reforms I linked to, I can’t vote out-district. So my choices are what is on my local ballot–say a Republican, the Democrat, the Green, the progressive lefty indie I like, and also a Constitution party (pretty right wing, I think of them as basically Republicans but I suppose a Republican and Constitution party faction would get into hot arguments about why they are night and day different) and a right wing independent too. That’s a pretty fantastic array of candidates to choose from by US standards, many races are just two choices and an appalling number have just one (in Florida, Pennsylvania, Oklahoma and Louisiana, if someone registers to run and no one shows up to register to oppose them, they just leave their name and office off the ballot and they walk into office with zero votes recorded–even to US House of Reps from Florida and Oklahoma! The guy Trump appointed to run NASA was re"elected" to his House seat from OK twice that way in the mid 2010s, and Florida routinely has about 3 or 4 Democrats in Congress who win that way cycle after cycle).
OK, so now I decide I am going to use my newfangled approval vote to approve the Democrat, the Green and the Indie, and worst thing that happens, the Democrat wins. Well second worst, it could be the Republican (from my partisan point of view, not meaning to shut down any conversation with conservatives here. That might be another topic!) Anyway in a district vote for a single winner, it is hard to see how I could be worse off than I was with FPTP, and who knows maybe a lot more people than the press and polls are letting on like this independent, or this Green, I like and they win in an amazing upset! Whoever wins gets the most approval votes, end report. The rest is post-mortem.
But now someone says “hey we can integrate all the district votes into a systemwide proportional vote!” That sounds like great news–it means my vote for the Democrat, the Green and the Indie all give each of them a second chance if none of them won the district approval race. Or that is what it should mean.
But wait. What is “proportional” now? When I had a single choice, I had to decide–dare I withdraw support from Lesser Evil Lackluster Dem who is too timid about working class issues and taking on big money and so forth for my taste, and waffles on say gender issues, but certainly is not a conversion therapy touting enemy of gay folk and Young Earth Creationist with discernable white supremacist ties who wants to re-criminalize marijuana and Get Tough On Crime Again which means breaking the budget with more prisons for more misery (but big bucks for private prison firms which he has ties to). Oh, and he thinks we should nuke Tehran too. Also Paris, France. And San Francisco. I kid you not if you are not familiar with the USA, we get candidates like this, and lots of them win too. (Here in Nevada a few cycles back, 2014 which was a Republican wave year across the nation, we had an Assemblyman who said “I would vote to reinstate slavery if my constituents wanted it.” I presume he was just trying to say he is the humble servant of his community and will comply with the will of the people, but he surely could have picked a better example–because we do have people who really would want that, and it would not be at all implausible some of this very fellows supporters have gone on record saying they wish they could. I think it was a kind of Freudian slip like the Bushes used to have so many of–“I want to make sure every American who has a job, wants a job!” It would be an amusing little slip up if in fact that were not the actual effect of Bush Sr and Jr’s and Reagan’s policies–we all lost our jobs and got desperate for any other…)
So Lesser Evilism is a serious thing in FPTP, you can’t even consider daring to break duopoly lockstep in a tight duopoly race, and any third party votes you do ever dare cast because you judge one or the other has a lock on it anyway will always be symbolic and ignored. That’s why we want cardinal methods, right?
Now if a single choice were all I had but in a second round, my vote for some single candidate might or might not elect them in the district, but then is counted a second time as a vote for their party, I might think differently about that local race. Yeah, if I “desert” this Democrat, probably the totally insane and vicious Republican (I know they aren’t all like that, but some really are, especially in places I’ve lived in) almost certainly walks in–unless perhaps he too is deserted by people who always wanted to vote Constitution or for that other right wing independent who makes the Republican I fear look like Kermit the Frog. But if I cast my vote for either the Green or the indie, they surely (well probably) won’t win the district seat, but their party benefits–the indie does not have a party, but they get ranked as their own party and might win a suitably inclusive form of PR, winning a seat on far less than a quota as typically half a dozen or so do in Hamilton’s Rule for instance. The indie is a long shot still, so regretfully I focus on the Green, who surely has a national party with other candidates getting votes too, so they will gain for sure with my vote, and my vote for this one might promote them to being one of the delegation–anyway some other Green elsewhere in the nation will win, and help balance out the scary Republican and maybe prompt the Democrats to move seriously left, and if not the Greens might shake down into a solid and growing national party who get my vote every election after.
So my strategy balances both my reckoning of the local situation and my evaluation of systemwide prospects, and opens me up to think more about who I really want to back. I just said I made a strategic choice not to go for my very favorite, but I picked a new party that I like better than my old one, and let the chips fall where they may in my district; systemwide I made things better as I see it.
This means that my choice of a Green is very significant. It means I have decided the Green is my best shot and best representation, I am lining up with the Greens. My vote therefore weighs into how many total seats the Greens get and that satisfies me. I would be better satisfied if that indie had seemed to have better chances of winning and I had backed her to victory, but my calculation that she just did not have the numbers to win was sober. It might have been wrong, but I tried my best. Hopefully she joins with some wildcat party of allied independents and next time around, knowing that, I might switch past the Greens to her group and help put them on the map. My party choices tracked by my vote record where in my judgement the best representation I can reasonably expect to actually get is.
But now go back to having Approval instead of one choice. Should I be counted as a Democrat? A Green? For the indie? All three equally? My vote no longer records who is the best candidate in my judgement. It records who is within the range of those I will accept. If each approval were simply added up, I would have multiplied myself from one voter to three, and unfairly weighed the outcomes–the temptation of say conservative voters to counter my liberal-hippie-commiesymp spamming of the system by resolving to run not one but two right wing independents, not just vote for the Constitution party guy but get someone in from the newfangled American Hawk party they just made up and vote for all of them and the Republican too so they are all voting 5 votes and tip the legislature back right where they assume it belongs is pretty high. Soon we are all voting for 27 candidates and it is anyone’s guess who anyone actually likes.
Now we can try various tricks to arrive at some meaningful notion of what "proportional"means when everyone is approval voting. We can for instance simply count up all the approvals on one ballot and divide each one by that total. So my triple marked ballot counts in full toward winning the district race, or not, but my ballot adds to the precinct totals for “proportional” figuring as 1/3 Democratic, 1/3 Green, 1/3 for the beloved indie. Now actually that does not look so bad does it? I mean, I have traded off pushing hard for one to giving a piece to three. Remember I loved the indie and wrote her off regretfully in the single choice scenario, but now not only have I given her a chance to win in an upset in my district, assuming that long shot fails, she gets a 1/3 vote toward quota. If I didn’t think she’d get a large enough fraction of quota with all of my single choice behind her systemwide, odds are she’ll be even lower in the general pool of small also rans now…but actually she got zero from me in the other scenario, she is now ahead, and what if a lot more than three other people in my district had the same regrets and worries, and took the same lifeline…she could well get pushed up even with my lukewarm fraction.
Meanwhile my Green vote and Democratic vote, albeit small for each, go toward a sure thing. Surely those parties will gain something, and I helped. Worse case, I risk 1/3 of my total systemwide weight, but surely conserve the other 2/3, and I might win the lottery when the indie wins and being an independent, she has to come from my district. I look forward to the possibility she is sitting near the crazed Republican in the body, shaking her head and correcting him when he claims to speak for my whole district. “No, sir, you are mistaken to say all the good people of our home town support you on this. I am here to tell you, some of us think quite differently!” Heck maybe the Green candidate wins from my district too, and they team up on him.
And maybe it turns out he was just putting out all that crazed talk but is really having a crisis of conscience and rethinking things a lot, and goes back to his conservative supporter and explains how his views are changing, and actually they should rethink a lot of stuff. Like he didn’t realize it was an Iraqi missile and not an Iranian one that killed his good friend from high school serving on the USS Stark, and nuking Tehran is probably not going to help anyone. And his service in the legislature has introduced him to some really decent gay people, some of whom are fervent Christians and he has prayed with them. And actually the police department would be better off focusing on serious crimes and not trying to stop young hipsters from smoking dope, and treating meth addiction as a medical thing and not a criminal plot is liable to be much better for everyone. And so it goes, maybe. Maybe in a cycle or two, I split my approvals two ways, one for the Indie who is now a leader in my favorite party, and one for this guy who saw the light and is a leader in the new Lincoln Party. The really extremist right wingers who voted for him before all vote for some Constitution guy who keeps getting in on top off, but some 60 percent of my district now votes as I do.
But I hope this happy fairy tale does not distract from the plain fact that no one can tell from any of my votes, which party was my favorite that first year? I am split and spread between several.
The fear here is that if we were to define proportional that straightforward way, no one would do it, we’d all just bullet vote for who we thought was our favorite, and then it would be known, who everyone thinks their favorite party is. But these amazing synergy stories of cross-party cooperation and win win negotiations would never happen, whereas cardinal is supposed to mediate that. Well, I just told a fable in which it did, mostly because I was getting tired of turning that suited guy into a flesh eating monster when a whole lot of my relatives remain conservative Republicans and I darn well know some of them have principles that put some real limits on how far down some dangerous paths they will be led, and I can talk to them about stuff. It happens in real life after all.
But I don’t assume there is any great magic to be conjured by letting everyone rate everyone. It does not change voter intentions, though it might unfortunately obscure them, especially if everyone gets all strategic and deeply concerned about God Forbid some small scrap of their precious vote might vanish into the ether. It is amazing to me how they don’t seem so worried about that in the single winner races, where at the end of the day, one candidate wins, the others lose, but heaven forfend they risk a quarter share of a vote on some long shot whose party might not win.
A lot of those concerns are more sensible when one realizes that quite contrary to traditional MMP, they really don’t want to see a Party vote at all; they often feel that party influence is always corrupting and baneful and everyone should be voting for candidates as individuals with party shunned and not taken into account at all. If that is the case, then a systemwide “proportional” vote in which my many approvals are split into fractional votes for individuals instead of their parties becomes fraught with many of the strategic conundrums paraded before me in this thread which all referred to that scenario–voters might concentrate more support on one candidate than they need, thus scanting others they are allied to (or that the voter thinks is also good anyway) or vice versa spread their votes too thin so one support group by failure of coordinated strategy fails to win anything close to their proportional share. This kind of guesswork makes CV disfavored–but it relates to the specific demand that we not acknowledge partisan alliance. If one assumes as I do that voters shall expect and demand parties to exist to help guide their decisions, and created parties because they were useful, then it makes sense to recognize them and route support through them. But the contrary position drives us to other approaches.
I may over time come to understand these alternatives and perhaps recognize them as having advantages, but right now honestly I can’t even visualize what they would look like in practice. The basic paradigm for “proportional” in the generalized sense is an iterative process, conceived as analogous to the iterative form divisor methods like Jefferson’s method. Say we proceed by observing which party has the highest total score, that is won most approvals overall and say OK that is one of 150 seats for that party. Then identify some subset of all the different combinations of ballots that voted for that party, and using some criterion subtract some of them–say we start with the bullet voters, who approved only that party, well, a quota of them should be satisfied now, they won a seat, so purge them. That lower that party’s score and maybe now it is some other party that has the highest, so we turn to that party’s bullet voters and deduct them, which shifts their score down, and keep going round robin like that as the seats get distributed until eventually first one then another gets their bullet votes exhausted and it is time to turn to the choices where the voter voted for two options. But golly, now it gets weirder and weirder. Should the votes for the other party get deleted along with those for the one that just won the seat?
So others say no, don’t subtract, divide! Like Thomas Jefferson told us to! (Actually Jefferson developed his method on a table of all contenders together pretty much the way Iike to do Jefferson on a spreadsheet, someone else, maybe d"hondt , realized this was the same as successively dividing by a divisor dependent on the seats won thus far. So now instead of subtracting any each party gets all the approval vote combinations approving that party first cut in half, now look has the high score now, alternate as before; the biggest parties get the first ones but drop down fast, so there is a more inclusive round robin of all parties toward the beginning I guess, but as the seats for the big parties rise the divisors make less and less relative difference; the smaller scoring parties that showed up moderately early rarely surface again and it becomes an alternation between the largest two with some intermediate scoring party sometimes gaining one more seat. But again, when we discount scores for the party gaining a seat, we also discount all the choices by the same voters for alternates. Thus intuitively it seems we pretty much settle just about every seat on the largest scoring parties and the odder choices of the various voters who also voted for the big ones get effectively dropped out of consideration. So much for Indie Babe, eh? So much for minority views, and diversity. They might get a little, but will they get a fair shake?
What does proportional even mean then, when it is being processed allegedly on cardinal votes? I found the CV approach, and perhaps a weirder “vector” based approach I am trying to hash out, fairly gratifying and fair and transparent in concept, and actually pretty good for diverse long shot choices too, or so it seems.
What we need, to be properly scientific about this as the site more or less claims ambition of aspiring to, is some test samples–elections in fairly realistic scenarios with some verisimilitude, where we can reasonably believe voters will behave like this. That’s why I spend a lot of time digging up real election data and applying models to them. But of course part of the ambition here is to change people’s behavior by changing their options, and for that we have to speculate on what makes people tick, which in turn brings us to controversies about what human beings are and why we do anything, and what people are capable of. These same notions about what people will do and how they will react, which relate to notions of how people should be and ought to react, are also important for evaluating outcomes. Specific methods people propose, if they can define them concretely, can be tested, projecting how we think reasonably representative sample models of people will change their voting in response to different options and opportunities–then we test the process for evaluating their ballots, and see what sort of outcomes they make, and judge whether people are likely to accept those outcomes of what they collectively did as reasonable, or crazy and capricious if not downright rigged.
If we put computational resources into play, it should I think be about honing these models that generate simulated voter behavior, and if it takes long iterative runs of a computer to output the election results, consider carefully whether we can get the processing of the votes done a lot simpler in a way that is acceptably recognizable as reasonable.
That’s why I dislike any sort of iterative process, even if we can all come to clear consensus on which score combinations to discount first and whether to subtract them or divide them or what have you.
I wonder whether some people for instance are just unaware that divisor methods–Jefferson, Webster, with special restrictions Huntington-Hill–generally have a straightforward table processing mechanism where all the final values pop out of the slot as it were in one go. (with divisor methods, typically involving fiddling around with a multiplier). Some easily accessed sources fail to mention these alternatives. Iterative divisor methods seem to have a kind of authoritativeness to them kludging around with a multiplier until the numbers add up in a spreadsheet might seem to lack, but the methods are equivalent the answers you get are the same.
So I would consider it pretty important to a good Proportionality-Like processing of cardinal votes that we can just enter the votes in and the right answer just pops out of a spreadsheet. Indeed the Cumulative Voting framework seems to be just that; the question that still troubles me is, can these iterative methods provide outcomes from the same inputs that people will recognize as somehow more valid and proper than those the quick processing would.