December 20, 2016

Ranked Choice Voting — should Greens embrace it or be concerned?

Ranked Choice Voting, also known by many as Instant Runoff Voting (please don't claim there's technical differences; I know that IRV is one of several ranked voting systems) will be how Maine conducts its state elections in 2018.  (Unfortunately, though I know of no U.S. Constitution bar to the idea, per that second link, it does NOT cover the presidential election. And, that's not because this is a trial for 2018 only; the ballot measure stipulates elections after Jan. 1, 2018, and not JUST in 2018.)

Green Party presidential candidate Jill Stein and the party both touted Maine voters approving this on Nov. 8. But, some Greens think it will actually hurt them. The fear is that this will co-opt Green voters into becoming accommodationist to Democrats. Given that the issue of how much Stein and others in one wing of Green leadership, like David Cobb, already may be doing this, this is a legitimate concern — if it's legitimately grounded. 

But, is it? 

I do appreciate the concern, but think it's overblown. 

First, a foot in the door is a foot in the door. Unless it's an openly accommodationist foot, it's a foot that shouldn't be looked in the mouth, to mix metaphors. 

And, I don't think it's an openly accommodationist foot, and I certainly don't think that's its primary intent. 

Second, it IS a foot in the door. Some people, especially in narrow races, who see “lesser evilism” and “greater evilism” in the two major parties, as in this presidential election, may be afraid of rewarding “greater evilism” in our current voting system. I voted Green; I didn't vote for Donald Trump by doing so. Nonetheless, I understand the concern, even the fear, and think it's a greater concern than some Greens have about RCV being accommodationist. 

Third, if that IS a concern, the accommodationist issue, there's a way to avoid that. 

RCV, of course, has voters rank candidates.  

A typical election with the four largest parties having candidates on the ballot might see a Green, or Green leaner, do:  
1 = Green 
2 = Democrat 
3 = Republican 
4 = Libertarian. 

Well, who says you have to do that?  

You can not only not give a 3 to the Republican or a 4 to the Libertarian, you can also choose not to give a 2 to the Democrat, at least hypothetically. (Note: I do not know if Maine's law requires a voter to rank every candidate. However, per the Ballotpedia link, the second from top above, this does not appear to be the case. Feedback is appreciated.) 

In that case, as soon as the Green candidate misses the cut, so does your vote. 

And?  

That's no different than the current system. 

So, RCV may not be perfect. And, it may not even be the best ranked voting system, though it's very arguably the easiest to understand.

And, it's better than the current, IMO. 

That said, I do agree with the Green Party Power statement that RCV, or any ranked voting or plural voting system, is not in and of itself a cure-all, or close to it. (And, in case it's not clear, that this needs to be across the country, not just Maine.)

And, having been in extended discussion in a Facebook group about Greens and RCV in Oregon (group is "open" so not violating confidences), while I'm not wedded to any one alternative system of voting, I am most certainly wedded to opposition of people who seem to think that their particular ranked choice system is the best even when they claim it's not. And, to the degree that approval voting does tend to elect more moderate candidates, I am wedded in opposition to it. Approval Voting on Wiki.

I need ConservaGreens no more than I need ConservaDems. 

That said, I did freely admit that it would surely be easier to recalibrate voting machines for approval voting than RCV. But, the idea that RCV is a possible dead end, possibly harmful to the whole idea of alternative voting or that only RCV is vulnerable to what happened in Burlington, Vermont (Bernie-ville!), rather than noting that NO alternative voting system to our current voting plurality system can guarantee a majority winner? I am wedded in opposition to that, too.  I stand by my observation, otherwise, in discussion here.

Update, Dec. 25. Range Voting, with its problems as well as goods, at Wiki.

13 comments:

Clay Shentrup said...

> I need ConservaGreens no more than I need ConservaDems.

You need "the highest possible expected utility". Approval Voting is markedly superior to Instant Runoff Voting as measured by Bayesian Regret.

http://ScoreVoting.net/BayRegsFig.html

Not to mention superior for third party visibility/viability and a host of logistical concerns.

https://www.youtube.com/watch?v=2q_eMUGCU5U

Gadfly said...

I've already sadi, under logistical concerns, that I agree on ease of voting machine configuration. Related issues, I would also agree.

I'll take a look at the Bayesian Regret.

That said, I question or reject your other claims.

First, approval voting has no guarantee of improving third party visibility vs. any other alternative voting system to FPTP. None. Indeed, that's not really even an empirical statement; it's opinion.

Second, as already noted, NO alternative to FPTP can guarantee a majority winner. That includes approval voting. I've seen approval voting enthusiasts give lip service tot that, then ignore it.

A corollary to that is the one I've already stated: If approval voting is adopted enough, it WILL have its own "Burlington moment."

Gadfly said...

Oh, and I'll see if I can find something on Bayesian Regret that's not on a site dedicated to approval voting and similar.

Gadfly said...

And, this page from a Google Group, in which I notice Craig is a contributor, does have people voice concerns about the utility of Bayesian Regret, etc. https://groups.google.com/forum/#!topic/electionscience/LUzDXlZxceM

To which, I'll add two other things.

One, relying on a 2000 computer simulation where the biggest simulation had just 200 voters isn't that statistically strong.

Second, per other observations I've made about some approval voting touters, the fact that most the first page of Google returns is devoted to range voters' and approval voters' touting of Bayesian regret for their desired voting system, rather than, say, a Wikipedia article, etc., is itself interesting.

Speaking of, via the Free Dictionary, here's what Wiki has to say about BR, and not just limited to alternative voting systems. IMO, it does give at least some credence to some of the alternative voices on that Google Group: http://encyclopedia.thefreedictionary.com/Bayesian+regret

Gadfly said...

And Fair Vote (yes, I know its oars row mainly in IRV) comments on problems with multiple other systems: http://www.fairvote.org/why-approval-voting-is-unworkable-in-contested-elections

Gadfly said...

Finally, speaking of the "Burlington problem," Fair Vote argues that Range Voting could have seen Goldwater defeat LBJ.

http://archive.fairvote.org/rangevoting.pdf

Look, touters of approval or range voting — it's been a while since I've done much swimming in these waters. But, I have done a few bits of such swimming before, and I'm not a pilgrim. Before I, or others, comment further, and as I consider a follow-up post, I just wanted to throw that out there now.

Clay Shentrup said...

> approval voting has no guarantee of improving third party visibility vs. any other alternative voting system to FPTP. None.

This is simply false, as I demonstrate in the video. Systems like Score Voting and Approval Voting, which produce a total sum for each candidate, allow a simple comparison of the relative strength of every candidate. Whereas Instant Runoff Voting for example provides no such metric.

Additionally, Score Voting and Approval Voting satisfy the Favorite Betrayal Criterion, so it's never strategic to NOT vote for your favorite candidate. The general strategy with ranked systems is to bury your favorite if she's unlikely to win, causing an exaggeration in the weakness of already weak candidates. Warren Smith has called this the "nursery effect".

http://scorevoting.net/NurserySumm.html

> that's not really even an empirical statement; it's opinion.

I just cited empirical evidence for it.

> http://www.fairvote.org/why-approval-voting-is-unworkable-in-contested-elections

Essentially every claim in that piece is rigorously refuted, e.g. here.

> Fair Vote argues that Range Voting could have seen Goldwater defeat LBJ.

FairVote is known to regularly make deceptive and false statements.

> this page from a Google Group, in which I notice Craig is a contributor, does have people voice concerns about the utility of Bayesian Regret, etc. https://groups.google.com/forum/#!topic/electionscience/LUzDXlZxceM

That thread strongly supports the robustness of Warren Smith's Bayesian Regret figures, even when you consider the mild "criticisms". Warren says there:

"as an abstract concept/framework, I think [Bayesian Regret] doesn't really have any weaknesses, it's just "the right approach." The difficulties arise when you try to convert the concept into practice; i.e. whenever you develop a computer program it has various ingredients like voter behavior models etc which can always be criticized. BR with the right voter behavior model (& etc "knobs") is (I think) clearly the right thing to do -- but that doesn't mean we know what that right model is. Fortunately in the single-winner case it seems not to matter much what these "knob settings" are. [I.e. Score Voting beats all other "non-exotic" systems regardless of any of the knob settings.] If conclusions had been very sensitive to that, then BR would have been a lot less useful."

Gadfly said...

I never said Fair Vote was perfect, and will look at your links. I did already say that I knew its oars were primarily in the water for IRV.

I've only mildly expressed my thoughts on Bayesian "anything." I consider Bayesian anythings to often be "stretched." I've already written extensively on this site about Bayesian probabilities and Jesus mythicism. And, when I do a second post, which I will, I'll be sure to make further reference to that issue.

And, approval voting? I've thought of one way in which it might be more confusing for voters than many of its touters think.

As for other possible refutations of Fair Vote, or other sites that present issues with score voting, or approval voting, I'll continue to look at neutral sites. I've already referenced Wikipedia's take on various alternative voting issues. And, its two basic issues:
1. That NO alternative system can guarantee a majority winner and
2. That ALL alternatives are subject to tactical voting
Will top that next blog post, too.

Clay Shentrup said...

> 1. That NO alternative system can guarantee a majority winner and

Or to be even more accurate, there's no such thing as a "majority winner".

And even when there's an unambiguous majority winner, that's not even necessarily the best option.
http://scorevoting.net/CondorcetCycles.html
http://scorevoting.net/XYvote.html

> 2. That ALL alternatives are subject to tactical voting

False. I know of three systems that are completely immune from tactical voting, though they all include randomness and are thus not well suited to political elections.

In any case, I have a lot to say about tactical voting.

Gadfly said...
This comment has been removed by the author.
Gadfly said...

Wikipedia was clearly talking about systems for political elections, re tactical voting. And, on the issue of majoritarian voting ... well, that's a new version of giving lip service to approval voting having the same faults as others, but then sweeping it under the rug.

I'll add that "utilitarian winners," specifically as for what's most utilitarian, are in the eye of the beholder. I already told Aaron that approval voting's tendency to elect moderate candidates is a specific fault in this beholder's eye.

You got one comment left, and then, for now at least, I'll be closing comments here.

Clay Shentrup said...

> I'll take Wikipedia's statements and observations over yours.

Wikipedia's information on voting methods is actually generally inferior to the state of the art as discovered by Warren Smith and some others of his ilk, such as Forest Simmons. Particularly its treatment of tactical voting makes the very fallacy I highlight in the graph on this page.

To give a few concrete examples, the Wikipedia entry on tactical Score Voting says:

>This presents an additional problem as compared to the approval system if some voters give honest "weak" votes with middle rankings and other voters give strategic approval votes.

Simply false. Score Voting outperforms Approval Voting with any mixture of strategic or honest voters. This should be obvious, since Approval Voting forces a distortion of sincere preferences.

> A different path to minimize this problem is to use median scores instead of total scores, as median scores are less amenable to exaggeration, as in majority judgment.

Utterly false. MJ is subject to the same strategy, which can produce more severe problems. It also produces significant information loss for sincere ballots.

> Wikipedia was clearly talking about systems for political elections, re tactical voting.

The Wikipedia entry I referenced is wrong, regardless of what whether we're talking about voting for politicians or names for pet rocks.

> I'll add that "utilitarian winners," specifically as for what's most utilitarian, are in the eye of the beholder.

Simply false. You can objectively measure the utility efficiency of a voting method. Some methods will make you statistically more satisfied with election results than others.

> I already told Aaron that approval voting's tendency to elect moderate candidates is a specific fault in this beholder's eye.

You shouldn't care about centrist or non-centrist. You should care about your satisfaction with election outcomes. Expected utility.

Gadfly said...

Wow,a mass of contradictions.

I say that I care about centrist vs. non-centrist. I'm told not to care about that, yet told to care about my satisfaction?

I otherwise disagree with you and Aaron.

And, I'll stand by Wiki, and others, rather than you.

I already noted I'm no pilgrim. I first wrote about FPTP alternatives 9 years ago. And, already back then, some of you approval voting folks came off as semi-cultist.

That too will be on my next post.

As I said was likely, comments on this are closed for now. (And just because I don't address other things doesn't mean I think you're right on them.)