Score, Then Automatic Runoff
May 29, 2018 6:04 AM   Subscribe

"We propose a new voting system."

"Instead of limiting the voter to supporting just one candidate, STAR Voting allows voters to rate all the candidates similar to how we rate books on Amazon. The winner is determined in two simple steps. First, all the scores for all the candidates are added up. The second step is an automatic runoff between the two highest scoring candidates. In the runoff, your full vote is automatically assigned to whichever of the top two you rated higher."

Star.vote is an online web application that lets you run own STAR Voting polls, or participate in a range of public polls posted by others.
posted by seanmpuckett (66 comments total) 6 users marked this as a favorite
 
similar to how we rate books on Amazon.

This ... is not the greatest way to sell this idea.
posted by chavenet at 6:12 AM on May 29 [35 favorites]


I suspect nobody would only score things 5 or 0 in order to maximize their vote.

Poll.
posted by Artw at 6:31 AM on May 29 [5 favorites]


Agreed, it's a frankly terrible pitch even without the additional phrasing about restaurants on Yelp. And if they wanted to make a less appealing website to explain it, they'd probably have to use the blink tag.

The decision science behind the idea seems strong, and it's reasonably understandable -- not just in how to vote, but in how the winner's chosen; it's possible to do it manually without going nuts.
posted by seanmpuckett at 6:31 AM on May 29 [5 favorites]


similar to how we rate books on Amazon.

This ... is not the greatest way to sell this idea.


Why stop there?

"Electronic ballots will also include recommendations based on past voting patterns, e.g., 'People who voted for candidate X also voted for candidate Y.'"
posted by mandolin conspiracy at 6:35 AM on May 29 [18 favorites]


Voting min or max only -- doesn't capture the expressiveness one would want in a larger slate.

Like in our riding right now for the Ontario Provincial there's six candidates, two preferred, three viable, two jokers, and one hot garbage, and one of which I clearly prefer.

I can express all this by voting 0, 5, 5, 4, 1, 1 -- where the first 5 is my preferred but probably not viable, second 5 the viable and second preferred, the 4 the acceptable alternative, the jokers 1, and 0 the guy I really don't want.
posted by seanmpuckett at 6:37 AM on May 29 [2 favorites]


This is basically a dumber version of the system used in iirc Scotland, where you rank candidates in order in preference and if your prefered candidate drops out, your next prefered one is chosen?
posted by MartinWisse at 6:38 AM on May 29 [2 favorites]


I suspect nobody would only score things 5 or 0 in order to maximize their vote.

More and more online services like Netflix have abandoned star-rating systems as meaningless for just that reason, yeah. I don't know why we're thinking it's useful here. Ranked voting makes more sense to me.

Also "this voting system has never been used in an actual election, so we don't have the experience of potentially millions of people trying to game the system to demonstrate the stuff we didn't think to protect against" is not a great selling point. Especially given that changing that system after it's gamed will require legislative action by a majority party that may have gotten to power on said shenanigans.
posted by middleclasstool at 6:41 AM on May 29 [7 favorites]


Can we just use playoff brackets so we increase the drama, maximize television advertising and we get to witness Teresa May in a debate with Lord Buckethead?
posted by arcticseal at 6:45 AM on May 29 [8 favorites]


As usual, this proposed 'solution' is needlessly complicated. The 'best' method available is Approval Voting. It's no coincidence that most societies of mathematicians and statisticians use it!
posted by dbx at 6:56 AM on May 29 [14 favorites]


My favorite part of Amazon book ratings are people who get the star- rating the wrong way round, so you get reviews like:

★★★★★ awful book
The worst book I ever read

★☆☆☆☆ five stars!!
I loved this book!!!
posted by BungaDunga at 6:59 AM on May 29 [11 favorites]


The thing about voting systems is that there's enough good enough ones that you can have something that's a much better expression of the actual will of the voters over two party, first-past-the-post voting, and you're not going to get a perfect one anyway. And yet, many countries haven't moved to something that works better, and it's that which is the actual problem no-one seems to want to tackle.
posted by Merus at 7:06 AM on May 29 [6 favorites]


I think it is a good idea to have an automatic runoff, but it can be easiest achieved by allowing a single split-vote based on voter indecision, rather invite clones with a range voting technique. This also allows a manual counting method and, like any runoff method that features two separate votes to settle the election, this one collects it all at once. It also has a tie-breaker (whoever has the most un-split votes). This method supports both a party candidate a voter needs to support, and a preferred candidate, bringing reliable information to help keep a candidate alive to win another day (essentially narrowing the next election cycle). Single vote-splitting conforms to the majority rule at every stage, even in the voter's mind. In principle, voting should never be for minor preferences or whimsy, which misses the point.
posted by Brian B. at 7:08 AM on May 29


This is basically a dumber version of the system used in iirc Scotland, where you rank candidates in order in preference and if your prefered candidate drops out, your next prefered one is chosen?

No, that's Instant Runoff Voting. We use that in Australia as well. STAR is more like a hybrid of approval voting and IRV.

These simulations are interesting.
posted by flabdablet at 7:09 AM on May 29 [6 favorites]


More and more online services like Netflix have abandoned star-rating systems as meaningless for just that reason, yeah. I don't know why we're thinking it's useful here. Ranked voting makes more sense to me.
While there are many flaws to star-rating systems and I personally find them particularly unintuitive when it comes to art/media, I think it's important to distinguish the purpose of ratings on commerce websites (where the intended goal is to give an idea of the quality of a product) and the purpose of rankings in a voting system (where the intended goal is to produce A Final Winner).
posted by inconstant at 7:10 AM on May 29 [3 favorites]


And yet, many countries haven't moved to something that works better, and it's that which is the actual problem no-one seems to want to tackle.

For that matter, there are some places where laws have been passed by referendum requiring the implementation of Ranked Choice Voting, yet the implementation of those laws is being opposed and restricted at every turn by those with a vested interest in keeping FPTP intact. E.g.
posted by tobascodagama at 7:14 AM on May 29 [4 favorites]


They offer a comparison of STAR voting to a few other voting systems, including approval voting. The explanation for the advantage of STAR over approval or range voting is that the IRV step creates an incentive against strategic voting in the rank step, as ranking all candidates as either 5 or 1 means you essentially have no say in the IRV step.

Mathematically, I can see how this works. Practically, I think the added complexity is enough that most voters won't understand the implication of the IRV step, and will end up trying to do a strategic vote anyway. I think the creators of this system seriously underweight the importance of simplicity for a ballot.

Personally I've been leaning towards a three-level approval/range voting system, with added minimal requirements for victory. The ballot for each candidate may be marked "approve," "disapprove," or "no opinion," with blank ballots or entries counted as "no opinion." To be elected, a candidate must receive at least X% approval votes and no more than Y% disapproval votes, and the candidate with the most approvals meeting this criteria is elected. If no candidate meets the necessary criteria, the election is considered to have failed to produce an acceptable candidate and is reheld.

I believe the system has the advantage of being much simpler than STAR or IRV, and somewhat directly enshrines what a democratic election should accomplish: elect the most popular compromise candidate. Psychologically for the voter, it's dead simple: just say whether a candidate is acceptable to you. Practically, it has two advantages over unmodified approval voting. First, it gives an effective veto to minority blocs, allowing them to reject candidates who are completely unacceptable to them. Second, it reduces the spoiler effect of dark horse candidates, who have a minority bloc that strongly approves of them but a majority that doesn't know much about them.

Anyway, anything is better than first-past-the-post, so on that basis I guess I would approve of STAR.
posted by biogeo at 7:28 AM on May 29 [3 favorites]


I think it's important to distinguish the purpose of ratings on commerce websites (where the intended goal is to give an idea of the quality of a product) and the purpose of rankings in a voting system (where the intended goal is to produce A Final Winner)

Sure, but my understanding from that video is that it is very much a ratings system, not a ranking system. They specifically say at one point that if there's a runoff between two candidates and you've given them the same score (which is possible), your vote basically isn't counted. That isn't ranking. I can give everyone a five or a one if I want. That's a rating system. Different goal, sure, but still a rating system, and one that's not entirely clear. "Give me your top three choices in order" seems way more straightforward and difficult to misunderstand than "how would you rate Jill Stein on a scale of 1 to 5 stars"
posted by middleclasstool at 7:34 AM on May 29


It seems to be asking people to trade off between maximising their initial vote (5s and 0s) and fine tuning their run off vote (Giving the 5 you are slightly less into a 4, giving the 0 you are slightly more into a 1, but risk helping or hindering them in the first phase). I guess you could help this by allowing fractions so you can vote 5, 4.9, 0.1, 0 - but really just straight up ranking seems easier and i'm not seeing any disadvantages.
posted by Artw at 7:40 AM on May 29 [1 favorite]


I vote they get a more engaging spokesperson.
posted by dobbs at 7:45 AM on May 29 [2 favorites]


I hear Noah Ward is available.
posted by Artw at 7:47 AM on May 29


Practically, I think the added complexity is enough that most voters won't understand the implication of the IRV step

Which will make the vote suspect in the eyes of many voters. Any voting system which cannot be easily explained to low-information/low-education voters is illegitimate, whatever its theoretical advantages might be.
posted by justsomebodythatyouusedtoknow at 7:47 AM on May 29 [11 favorites]


The problem with any voting system is that you get a single discrete outcome: one person is the winner. That person will not, of course, be everyone's or really anyone's ideal candidate. So, until the Composite Superman runs for office, we'll be stuck with voting for someone who is merely human.
posted by SPrintF at 7:56 AM on May 29 [1 favorite]


This reminds me vaguely of Slashdot.
posted by ZeusHumms at 7:58 AM on May 29 [1 favorite]


This is my favorite voting system so far! I don't think that 5 discrete levels of ranking in particular is that important a feature of the system; the core of the method is (as the name says) score (of whatever kind), then automatic runoff. I'm not a marketing person; if someone can come up with a better number of score levels that people will like more, I'm all for it.

Reasons why I like it:
1. It shares the advantage of score systems in general that an honest ballot is never that much worse than a strategic ballot.
2. But it addresses a weakness of score methods in that in STAR, a fervent minority is less likely to defeat a less-enthusiastic majority. Once you get to the runoff stage, you need broad-based support, not just enthusiasm.

I'm a little sad that MeFi does not appear to cotton to this marvelous idea. Is this FairVote propaganda at work?!?

Incidentally, if people like the idea of ballots in which you rank candidates, note that the FairVote has been using the term "ranked choice voting" (which is a broad umbrella for many techniques) to refer to IRV (just one method). It's like their unintentional trap street or something. Anyway, instead of eliminating the candidate that has the fewest ballots in which they are ranked first, there are other possible elimination rules that make just as much (or more) sense. Consider instead the possibility of eliminating the candidate that has the worst average rank.

Standard disclaimer: every non-intentionally-shitty method is better than FPTP, so if the only option of replacing FPTP is IRV, I'll get behind it.
posted by Jpfed at 8:31 AM on May 29 [2 favorites]


@middleclasstool

Is this really a rating system, though? My understanding is that you rate one thing against a set of criteria, but you rank a group of things based on preference after comparing them (like when choosing one candidate among many). If you were to rate a single candidate, you might give them terrible marks across the board (rating = low), but still prefer them (rank = top) to all others in an election.

In the STAR system, you are still voting to ensure that the candidate you like best (or maybe second best or at least third best, you hope) gets into office. Thus you'd give them "ratings" that essentially rank them. Among five candidates, you'll likely have a clear favorite, a few somewhere in the middle, and one to avoid at all costs. If you get to be an informed voter, you'll find some minor difference between the middling candidates to assign them ratings of 2 through 4 and, if you're uninformed, you may still do so based on how they come across on TV. Even if you rate all the middling candidates the same, that's really just abstaining from voting in a runoff among them. Because there's no set of criteria to compare these candidates to in order to assign them a rating, you can only compare them to each other, which is a ranking.

One thing they sort of gloss over is that this five-star system, for any election in which there are more than five candidates (in America? Imagine!), forces you to rank at least two candidates as equal, even if you would rank one higher in a ranked-choice system. One answer, to have as many stars available for voting as their are candidates, would make a race with too many candidates pretty tough for voters.
posted by Grimp0teuthis at 8:40 AM on May 29


Practically, I think the added complexity is enough that most voters won't understand the implication of the IRV step, and will end up trying to do a strategic vote anyway.

The sort of people who'll strategically vote are trying to produce a specific effect. If they can't work out how to get that effect, they're not going to strategically vote.
posted by Merus at 8:42 AM on May 29


I like it better than first past the post, so there’s that.
posted by Artw at 9:45 AM on May 29 [1 favorite]


I am open to believing that popular pressure could improve the voting system, first locally and in civil society institutions until people demand it regionally and nationally and internationally. Those in power will stymie this at every opportunity and show their true colors. Ranked voting is my vote of choice but as a USAian newrly anything beats the clusterfkcu we got
posted by Anchorite_of_Palgrave at 9:55 AM on May 29


The sort of people who'll strategically vote are trying to produce a specific effect. If they can't work out how to get that effect, they're not going to strategically vote.

That should be something like "If they can't work out something that they think will get them that effect..." Lots of people in lots of circumstances have what they think are excellent strategies for something that are in fact crap.
posted by GCU Sweet and Full of Grace at 9:56 AM on May 29 [2 favorites]


Voting min or max only -- doesn't capture the expressiveness one would want in a larger slate.

Like in our riding right now for the Ontario Provincial there's six candidates, two preferred, three viable, two jokers, and one hot garbage, and one of which I clearly prefer.

I can express all this by voting 0, 5, 5, 4, 1, 1 -- where the first 5 is my preferred but probably not viable, second 5 the viable and second preferred, the 4 the acceptable alternative, the jokers 1, and 0 the guy I really don't want.


That's all great if you think a vote is a tool to express your personal beliefs and allegiances, but if you view a vote as a tool for affecting who actually gets elected, you've probably just weakened your vote, compared to voting all 5s and 0s. For example, if candidate four is running closely with candidates two and three to make the runoff, you've just helped them kick out one of your preferred candidates by giving them a 4. Conversely, if either of candidates five or six is contending for a spot in the runoff, you've just helped them knock off one of you preferred candidates by giving them a 1 instead of a 0. Sure, your individual vote is unlikely to swing the election, but that's true of any decent voting system. It's still true that a coalition voting honestly is at a distinct disadvantage to a coalition voting straight 5s and 0s.

Ultimately, figuring out how to use your vote to maximize the probability of preferred outcomes in this system requires fairly deep knowledge of both the system and the specific state of the electorate in a given election, and for me that alone is reason enough to dislike it. (For the record, I have similar issues with IRV and its non-monotonicity problems). That's why I generally prefer something dead simple like approval voting.
posted by firechicago at 9:58 AM on May 29 [5 favorites]


The problem with any voting system is that you get a single discrete outcome: one person is the winner.

Not always! Multi-member districts can be combined with most voting systems. Instead of handling each elected representative as a separate election, you combine two or more into a single election, and the top however many you need win. It works very well with instant-runoff systems, hampers gerrymandering, and improves proportionality of representation.
posted by vibratory manner of working at 10:08 AM on May 29 [6 favorites]


(What, no one yet? Fine.)

STAR Voting: 3 stars
First Past The Post: 1 star
Instant Runoff: 5 stars
Approval Voting: 5 stars
posted by Four Ds at 10:23 AM on May 29 [11 favorites]


STAR voting doesn't always elect the Condorcet winner (i.e., the candidate who would win in all pair-wise elections against the other candidates, if such a candidate exists), which is a dealbreaker for me (although it's still better than first past the post).
posted by Pyry at 10:50 AM on May 29 [1 favorite]


Welcome to the wide world of voting system theory, where there is no right answer, and everyone is an expert.

To begin with, STAR is a hybrid of a rating system (AKA range voting, score voting, cardinal ratings), plus a second stage ranking runoff between the two highest scoring candidates.

The pairwise runoff --- which is NOT instant runoff voting, as asserted above --- adds in a way to ensure that preference ranking is included in the outcome.

What I meant above by "no right answer" is that it isn't possible to have a voting method that satisfies all voting systems criteria. If it satisfies one set, it will fail another. It happens that FPTP fails many criteria, however, so some replacement is desirable.

Personally, I would be satisfied with Approval, but my current favorite is a variant of Majority Judgment, which I call EXACT (EXcluded Approval Comparison Test):

* Find the highest rating r, for which there is at least one candidate X who is rated at or above level r on more ballots than any candidate is approved on ballots which rate X below r.
* If there is more than one such candidate X, then if there is at least one candidate Y who is rated above r on more ballots than the highest approved candidate on ballots that rate Y below r, elect the candidate Y with the most ballots rating Y above r.
* Otherwise, elect the candidate X with the most ballots rating X at r or above.
* If no candidates satisfy the first criterion, for any approved rating r, elect the candidate with the highest approval over all ballots.

The EXACT grade for a candidate is tuple similar to Majority Judgment's "majority grade":

* EXACT grade for candidate X = (r, s, t)
where r is the rating at which X's votes at or above r are greater than those of the highest approved candidate on ballots excluding X at r or above;
* If the number of ballots with X at rating r+1 and above is greater than those of the highest approved candidate on ballots excluding X at ratings r and above, then
s = r+1, and
t = votes for X at r+1 and above.
* Otherwise,
s = r and
t = votes for X at r and above.

By sorting these tuples in descending order, one gets, as with Majority Judgment, an EXACT ranking for the candidates.

This method is immune from the irrelevant ballot problem -- if extra ballots are added that do not rate any contending candidates, they won't change the outcome. This avoids a problem with Majority Judgment and other median rating methods.

My preference is for a 3-star (4 rating level) rating system: 3-stars = Best; 2-stars = Good; 1-star = Acceptable; 0-stars = Reject/Disapprove. This is sufficient to allow expression without getting into other problems.
posted by Araucaria at 12:07 PM on May 29 [5 favorites]


Maine is doing Ranked Choice Voting in the upcoming primary.
posted by theora55 at 12:12 PM on May 29 [1 favorite]


"My favorite part of Amazon book ratings are people who get the star- rating the wrong way round, so you get reviews like:"

My favourite are when they rate a product based on some insanely irrelevant non-sense.

1 Star: Product worked great but I'm an idiot and couldn't do something it wasn't intended for.

1 Star: Thermometer says it works to check temperature of meat but would not work on my son

5 Star: Product arrived inoperable and on fire and burned my house down but customer service were aces about the whole ordeal.
posted by GoblinHoney at 12:26 PM on May 29 [4 favorites]


Don't the Oscars or one of those dumb award orgs use a system like this? In that context it seems to produce winners from mediocrity, the least offensive or controversial of a group. As such the things that win tend to not be very compelling or great and certain biases and tendencies seem to become more and more entrenched.
posted by GoblinHoney at 12:27 PM on May 29


2018 Oscars: Best Picture ballots will be counted (and recounted) under preferential voting system

This year, voters ranked the nine Best Picture nominees. If one nominee garners more than 50% of the first place votes, it will win Best Picture.

If, as is more likely, no nominee reaches this threshold, the film with the fewest first-place votes is eliminated, with its ballots being reapportioned to the second-place choice.

Should no film cross the required 50% + one ballot threshold, the film with the fewest first-place votes is again eliminated, with its ballots being apportioned to the next choice still in play (i.e., if the second-place choice is no longer in the running, then the ballot would be reapportioned to the third-place choice and so on.)

This process of elimination and reapportion continues until one film reaches at least 50% + one ballots.


I guess the 50% thing is supposed to shortcut the process and prevent them having to do the full process of elimination?
posted by Artw at 1:02 PM on May 29


The Hugos use straight up IRV for finalists, though thats after a nomination phase that uses some more complicated methods to avoid gaming by slates.
posted by Artw at 1:03 PM on May 29


Find the highest rating r, for which there is at least one candidate X who is rated at or above level r on more ballots than any candidate is approved on ballots which rate X below r.

So this sentence looks a little tricky because you mention both "rating" and "approved"; do you mean by this that when you are checking a possible r during this step, you consider a candidate "approved" by a particular ballot if that candidate is rated r or higher?

And when you say "more ballots than any candidate is approved on ballots which rate X below r.", do you mean more ballots than any one other candidate? I feel like this can't be what you mean because in almost all elections, r would be the highest possible score because someone will have the plurality of those highest scores. But I'm not sure how else to interpret it.
posted by Jpfed at 1:52 PM on May 29 [1 favorite]


Jpfed, by approved, I mean giving any score above zero, as in the 0 to 3 star example. So 1, 2 or 3 stars would all count as approved.

While I'm at it, I should give credit to Chris Benham, who came up with IBIFA (Irrelevant Ballot Immune, Fallback Approval), from which EXACT is derived

If irrelevant ballot immunity is not your cup of tea, and you prefer that your method satisfy Condorcet and Chicken Dilemma, Benham also developed a Condorcet completion method called MinLVSME.
posted by Araucaria at 3:36 PM on May 29 [1 favorite]


The Hugos use straight up IRV for finalists, though thats after a nomination phase that uses some more complicated methods to avoid gaming by slates.

Notably, the Hugos also always include a 'No Award' choice, so people can express whether the nominees, in their estimation, are 'worthy' of a Hugo by placing them below 'No Award'.
posted by Merus at 6:47 PM on May 29


I'm a little sad that MeFi does not appear to cotton to this marvelous idea. Is this FairVote propaganda at work?!?

I imagine there are basically three categories of people not excited by this: those who don't care about voting systems at all, those who have heard good things about IRV from FairVote or other places and like it, and those who are nerdy enough about the topic to have strong opinions about which voting system is best and therefore don't get excited about anything except their own pet system, but will unenthusiastically support any non-FPTP system as an improvement.

Anyway, it's a good FPP and interesting idea to discuss.
posted by biogeo at 10:35 PM on May 29 [2 favorites]


> Notably, the Hugos also always include a 'No Award' choice, so people can express whether the nominees, in their estimation, are 'worthy' of a Hugo by placing them below 'No Award'.

"No Award" would have placed very highly in every US Presidential election since 1912.
posted by Phssthpok at 12:54 AM on May 30 [1 favorite]


The problem with any voting system is that you get a single discrete outcome: one person is the winner.

Nope. The state and federal upper houses I vote for have a number of parties receiving a number of seats depending upon the vote (“proportional voting, in Australianese).

Incidentally, if people like the idea of ballots in which you rank candidates, note that the FairVote has been using the term "ranked choice voting" (which is a broad umbrella for many techniques) to refer to IRV (just one method)

We’ve always called IRV “preferential voting”.

I love the “no award” idea. I recall a Jello Biafra rant about the concept (“Vote 1 None of the Above!”). That said, I don’t know it would combine well with compulsory voting, and a population of smart-arses.
posted by pompomtom at 4:18 AM on May 30 [1 favorite]


It’s fine for the Hugo’s to skip a year... not sure it’s impact would be quite so good for things where you actually need someone to fill a role. I guess if you did get NONE OF THE BELOW as a winner because the candidates are all completely unacceptable then it would mean doing the entire process over?
posted by Artw at 4:58 AM on May 30


I recall a Jello Biafra rant about the concept (“Vote 1 None of the Above!”).
I don’t really know when he would have started with that. But we have this very persuasive ad for voting for non of the above, courtesy of Richard Pryor: Brewster’s Millions (1985): Vote ‘None of the Above’.
posted by Martijn at 5:02 AM on May 30 [1 favorite]




Well yes, because the Jungle Primary manages to be First Past the Post taken to a stupid extreme and whoever came up with it is an idiot.
posted by Artw at 7:17 AM on May 30 [1 favorite]


The Jello rant was on (IIRC) "I Blow Minds For A Living" (so 1991, I gather), and was couched as "the last Soviet elections were more democratic than the US elections, because if 'none of the above' won, they'd have a new ballot with new candidates". So, you know, context-free TRUTHBOMB, but if one really considers the electorate too clueless to understand something beyond FPTP, it would be a way to mitigate the demonstrated awfulness of that system.
posted by pompomtom at 7:23 AM on May 30 [1 favorite]


"Differently undemocratic" I could get behind...
posted by pompomtom at 7:25 AM on May 30


Ideally, then, considering how I want to express things, i'd like to see a combination of inclusion and ranking. There are people on the ballot I'd be okay with. There are people I absolutely do not want.

I would like to be able to have the first round be "any of these people are fine" and the second round be "if it's going to be one of those people, which one do I most want"

As was pointed out, having to rank candidates for "which one" on a granular "star" scale can cause trouble on the "scoring" round. With finer graduations, one could vote 0, 1000, 999, 998, 0 and relay the important information without your score causing trouble with the electability of candidates.

But then why not just Approval voting. It's really simple to explain, doesn't have a lot of weird strategizing, and you just count everything up and you get a winner right away.
posted by seanmpuckett at 8:25 AM on May 30 [2 favorites]


I guess because Approval voting would tend to drag everything towards a centrist "ho hum" candidate that no one hates, but also no one really likes.

This shit's hard.
posted by seanmpuckett at 8:26 AM on May 30


*gestures broadly at everything*

Ho hum sounds pretty ok, actually.
posted by tobascodagama at 8:49 AM on May 30 [2 favorites]


I guess because Approval voting would tend to drag everything towards a centrist "ho hum" candidate that no one hates, but also no one really likes.

As long as we are dreaming, note that ho hum candidates have some value, and truly representative bodies have value, so why not both? You could have one house of a legislature use approval (say, for the "lower house" of many smaller districts), and another use a PR method (for the "upper house" of fewer, larger districts).
posted by Jpfed at 9:24 AM on May 30 [2 favorites]


pompomtom: We’ve always called IRV “preferential voting”.

That's just as bad as ranked choice voting, in that it gloms onto the general term for a ranked ballot and presumes that the only way to tally such a ballot is with the single-winner version of Single Transferable Vote.

Preferential voting would be just as accurate to describe Condorcet methods.

Jpfed: You could have one house of a legislature use approval (say, for the "lower house" of many smaller districts), and another use a PR method (for the "upper house" of fewer, larger districts).

Genetic algorithms solve problems by simulating diversity, then applying selection pressure. A century ago, we had 800 different automobile makers, with many different designs, while now we have just a few, with mostly similar ergonomics and mechanics. In legislation, an analogue would be to have a diverse lower house to introduce legislation, with an aggregative upper house to consolidate bills and approve/disapprove.

So unlike your Australia-style proposal, I think it would be more optimal have a lower house elected with Proportional Representation, and an upper house with larger single-winner districts.
posted by Araucaria at 11:47 AM on May 30


If we're talking about electing an entire assembly of people, then the best method in my opinion is straightforward random selection (i.e., if you have 500 spaces then pick 500 ballots at random and whoever they voted for wins): strategic voting is impossible (the only strategy is to vote for who you actually want) and in expectation it's perfectly proportional.
posted by Pyry at 12:55 PM on May 30 [1 favorite]


@pyry Jacobin is arguing for a legislature by lot, leaving behind elections altogether.
posted by Grimp0teuthis at 1:30 PM on May 30 [2 favorites]


I'm all for skipping directly to sortition as well.
posted by Pyry at 2:58 PM on May 30 [2 favorites]


From that Jacobin piece:
the notion that regular people can make good, competent, political decisions if (and this is a big if) they are immersed in a well-designed deliberative context: a space with equal participation, access to pertinent information, skillful moderation, small group discussion, and the absence of all types of coercion and force except that which the preeminent philosopher Jürgen Habermas called the “unforced force of the better argument.”
I am not sure how this would practically work in the environment we have today. Having a bunch of regular old folks suddenly responsible for decisions affecting billions of dollars seem rather ripe for a whole lot of manipulative techniques. How are you going to ensure the "absence of coercion" of these people and their families/friends, etc. It seems ... implausible.

But as pipe dreams go, it's an interesting one.
posted by seanmpuckett at 3:04 PM on May 30 [1 favorite]


So unlike your Australia-style proposal, I think it would be more optimal have a lower house elected with Proportional Representation, and an upper house with larger single-winner districts.

I may be unduly influenced by living in Wisconsin, where a single party in control of both houses was not recently unable to reach a budget agreement for months. I purposefully chose Approval for the lower house and PR for the upper because I didn't want to make the houses exemplify crazy-lower-house and deliberative-upper-house. Rather than exaggerate their strengths and weaknesses I would want to temper them. But it's not too important a point to quibble over. The main idea for me is that by using different methods for different houses, you can kind of average over the different methods' strengths and weaknesses.
posted by Jpfed at 10:02 PM on May 30 [1 favorite]


Preferential voting would be just as accurate to describe Condorcet methods.

Fair point, but still it is the accepted usage where I live.

...and isn't descriptivism prescribed on Mefi?
posted by pompomtom at 4:54 AM on May 31


pompomtom: Fair point, but still it is the accepted usage where I live.

Fair point. However, "red" used to mean both yellow and orange as well, which is why we call people with orange-colored hair redheads. Usage can change over time as nuance becomes both important and useful. In Spanish, rubio (redhead) is applied to any hair color lighter than brown, from red to towhead blonde.

In the case of voting systems, describing a tallying method for a voting system by a ballot format that could be used by many other competing systems is a conflation that adds confusion, not to mention petitio principii (assuming the consequent) if there is a reasonable debate about the best method to use.

What's wrong with calling the method single-winner STV?
posted by Araucaria at 9:58 AM on May 31


Nothing at all, but you’d have to explain what you meant to 20M Australians, who actually use the system regularly, and get taught the less precise term in school.

Personally, I’m used to rephrasing to “instant runoff” online, which I thought was the Americanese term (and which I’d ordinarily have to explain, were I speaking to a compatriot). Is that similarly imprecise?
posted by pompomtom at 4:06 AM on June 1 [1 favorite]


I understand the Australian usage. It's probably because the Hare-Clark method was first used there, over a century ago. Before the discovery of sub-atomic particles, relativity, quantum theory, Goedel's incompleteness proof, or computers. I'm just saying that to illustrate that we've learned a few things in the meantime, and we've changed our language in ways that reflect that knowledge in other areas.

On the other hand, I have no problem with the term Instant Runoff (IRV). It's an accurate description and doesn't confuse the issue. That's a good compromise.

Incidentally, Condorcet methods have sometimes been called Instant Round Robin, to highlight that all candidates are compared pairwise.
posted by Araucaria at 11:52 AM on June 1


It's probably because the Hare-Clark method was first used there, over a century ago. Before the discovery of sub-atomic particles, relativity, quantum theory, Goedel's incompleteness proof, or computers. I'm just saying that to illustrate that we've learned a few things in the meantime, and we've changed our language in ways that reflect that knowledge in other areas.

So, are there other actually employed 'preferential voting' methods of which I should be aware? If I'm going to confuse someone by using the general term 'preferential voting', rather than the specific 'Hare/Clark Single Transferable Vote', what might the confusion be with? Most of my discussions, even online, have historically between the Australian STV system, and dumbarse FPTP.
posted by pompomtom at 7:04 AM on June 2 [1 favorite]


« Older Magical Voxels   |   Bless the Rivers down in Africa Newer »


This thread has been archived and is closed to new comments