New Sightline article by Kristin Eberhard

And there's this.

http://www.sightline.org/2017/05/09/glossary-of-executive-officer-voting-systems/#comment-205214

Unbelievable.

Hi all,

I hope you’ll read the full article for yourselves. It describes some problems with the way we elect executives such as presidents, governors and mayors, and how different voting systems might solve those particular problems. I’d be happy to hear comments, questions, and criticisms.

Clay is a passionate proponent of his views, but I think most of his statements mischaracterize my observations and conclusions or overreach the evidence. To pick one example, he says my description of what happens with two major-party candidates and a weak third-party candidate like Nader is “false.” But his reasoning is based on an example of three strong candidates, a situation the article specifically addresses in the very next paragraph. How would he conclude that “weak third-party candidate like Nader” means someone who got 32% of the vote? That’s a strong candidate, an order of magnitude stronger than Nader, who received less than 3% of the vote.

Sightline and I are supportive of both IRV and SRV, and particularly passionate about Proportional Representation. Mostly we want to see reform. If people in the reform community spend energy tearing down those who are fighting for reform, the defenders of the status quo will keep laughing all the way to the bank.

Clay is a passionate proponent of his views

Meaning..I am biased?

he says my description of what happens with two major-party candidates and a weak third-party candidate like Nader is “false.” But his reasoning is based on an example of three strong candidates

Not true. LePen would lose by a 2/3 supermajority to either rival in my example. If that's not weak enough, just add a few more candidates and it becomes possible for an arbitrarily weak candidate to win with IRV due to vote splitting. Also, I suspect Nader would have done about this well had there been a runoff. The ANES exit poll from 2000 said that of professed Nader supporters, 90% claimed they voted for someone else—suggesting he actually did have about as much support as the LePen in my example.

Here's even a hypothetical worst case scenario where the IRV winner is the favorite of only two voters, and would lose by a head-to-head majority to all by one rival. Now of course the more extreme these examples get, the less realistic. But you didn't make any reasonable caveats. You said, "it is always safe to rank a weak third-party candidate like Nader." Speaking of "overreach the evidence".

If people in the reform community spend energy tearing down those who are fighting for reform, the defenders of the status quo will keep laughing all the way to the bank.

I'm not trying to "tear down" anyone for the sake of being negative or spiteful. But your piece contained multiple blatant inaccuracies. The one where you mischaracterized Arrow's Theorem in two different ways. The one where you cited Wikipedia (which claimed that Score Voting strategy is Approval Voting) and then made the drastically different assertion that it's like (sincere) Plurality Voting. I was particularly bothered by your un-nuanced bald assertion that the Bayesian Regret figures were based on unrealistic assumptions, when your supporting statements revealed that you don't understand the methodology.

I take voting reform as seriously as (perhaps more seriously than) climate change. I sometimes see well-intentioned people making specious and overly strong claims on that subject, and I feel a similar obligation to point that out to the extent my knowledge allows. When I do that, I don't expect to be harangued by other activists asserting that I should just let it go lest the infighting stifle progress. I mean, that's a horrifying notion. I think misinformation does more harm than any appearance of infighting.

Also note that the real world "experience" with score or approval voting that Clay refers to is merely public opinion polls of voters about how the would score candidates, not actual voting behavior. In such non-binding surveys there is no voter concern about the crucial "Later-No-Harm" problem that indicating any score for a less preferred choice can hurt one's favorite choice, simply because none of these "votes" can actually help or hurt anyone --- they were not real votes. Also the voters were not subjected to campaign calls for bullet voting (as all competent campaign managers would have advocated). The only ACTUAL election data we have from the handful of nongovernmental organizations that have used Approval Voting is that the vast majority of voters do indeed bullet vote.

not actual voting behavior… crucial "Later-No-Harm" problem

You can't honestly have it both ways. Given the lack of robust real-world elections with score voting, the arguments that later-no-harm is a crucial problem are at least as weak as claims that score voting works out in great ways without problems.

real world "experience" with score or approval voting that Clay refers to is merely public opinion polls

Simply false. For instance, the 2015 Republican Liberty Caucus straw poll had significant political consequences. Ted Cruz and Rand Paul attended and bussed in supporters specifically to bullet vote for them. There have also been uses for real contentious elections within organizations.

The only ACTUAL election data we have from the handful of nongovernmental organizations that have used Approval Voting is that the vast majority of voters do indeed bullet vote.

Simply not true. From Dartmouth math professor emeritus Robert Z. Norman:

In 2007 there was a per voter average of voting for 1.81 candidates. Hence the proportion of bullet votes had to be fairly small (or else nearly everyone voted for one or all three candidates, but not two, which would seem crazy).
[Specifically, if all ballots approved either 1 or 2 candidates, there must have been 19% approve-1 and 81% approve-2 ballots. Norman in later email later hypothesized that actually there may have been a strategy of "either voting for the petition candidate or voting for all [3 opposing] nominated candidates." If that was the only thing going on then 60% of the votes would have been approve-1 and the remaining 40.5% approve-3s, but in this case approval voting was clearly showing its immense value by preventing an enormous "vote-split" among the 3. In any case the fraction of "approve≥2" ballots presumably had to be somewhere between 40.5% and 81%.]

And even the bullet voting we do see isn't necessarily a "problem". If your favorite candidate is a frontrunner, bullet voting is perfectly reasonable. Or if you prefer her so much over her rivals that it's a sincere expression—also not a problem.

The kind of scenario where bullet voting could be a problem is e.g. where a voter feels {Nader=5, Gore=4, Bush=1} but bullet votes for Nader, despite knowing the odds. It's neither strategic nor honest. I've seen no evidence this pattern happens in any substantial amount.

In such non-binding surveys there is no voter concern about the crucial "Later-No-Harm" problem that indicating any score for a less preferred choice can hurt one's favorite choice

I've seen no evidence that failure of Later-no-harm is a problem. My view is that satisfying it is a problem.
http://scorevoting.net/BulletBugaboo.html

Note that a ranked voting method cannot simultaneously satisfy Favorite Safe and Honest Second Safe. IRV satisfies neither.

Also the voters were not subjected to campaign calls for bullet voting (as all competent campaign managers would have advocated).

This is precisely the message voters get with Plurality Voting, and yet the dominant strategy is not "vote for your favorite candidate", but "vote for your favorite VIABLE candidate". If that's your third or fourth overall favorite, so be it. Score Voting allows voters to keep going up from there. It obviously don't change the strategy to bullet voting.

"If people in the reform community spend energy tearing down those who are fighting for reform..."

Kristin, this gets to the heart of the problem. Reading your article and response to Clay's concerns shows that you are taking the feedback about IRV and these concerns personally. Please don't. Even though you've put a lot into supporting IRV in your career I think it's important to look at these systems objectively and not write scientific articles like this one from a defensive or outcome oriented perspective. On the flip side, reading this article feels like you are intentionally trying to mischaracterize SRV. As you know there is a scientific system that can measure voting systems relative accuracy across all criteria. This is called Bayesian Regret and the inverse, Voter Satisfaction Efficiency (VSE). Omitting this key information makes no sense in this context. Why do it? Condorcet and VSE are the two best tools we have for establishing measurable accuracy. Our most important criteria for a new voting system.

Saying that in SRV people would "bullet vote" and only score their top 2 makes no sense for tons of reasons and scenarios I'm sure you are aware of.

"Later No Harm" is a criteria that I disagree with because it directly prevents a voting system from electing a strong compromise candidate if there is no strong majority winner candidate. That is a negative outcome for the electorate and I believe we should revisit this criteria. The ideal here should be that a voter doesn't hurt the best possible outcome for the electorate as a whole by honestly rating or ranking the candidates. If voters honesty show that they would be okay with a candidate other than their favorite and show how much they would support that candidate and under which circumstances, that is a good thing. This is a key tenant of consensus decision making and is one of the reasons that SRV (Star Voting) is able to out preform IRV in situations where there is not a simple majority winner with just the first choice votes counted.

Many of us think the inverse of "Later no harm" is a better criteria and that it should be called something like "Compromise Criteria". That a voting system should allow and encourage voters to express themselves with enough detail to help find the winner that can make as many voters as happy with the outcome as possible. This is what SRV (Star Voting) offers and this is one reason that it outperforms IRV in Voter Satisfaction Efficiency (VSE).

The other reason SRV (Star Voting) outperforms IRV in VSE is that in IRV if your favorite is eliminated in later rounds your other preferences may not be looked at. Ignoring some of the voters preferences on the ballots directly causes less accurate results.

Note to Sara: Kristen's summary definitely mentioned explicitly the concern that IRV counts some voters' later choices while ignoring that of others (although I like to add emphasis that the voters who alternate choices were ignored include those whose first choice also loses).

It would be better to point out concrete concerns about exact things Kristen wrote, such as specific sentences that are invalid or wordings that should be improved (perhaps with suggestions). That would do a better job at helping everyone understand one another (which is sorta what Clay did, but he wrote in a way that was liable to lead Kristen to respond defensively and personally).

Clay wrote a lot of things that violate the rules, assume bad faith from Kristen,

She cited a Wikipedia article which said the ideal Score Voting strategy was Approval Voting, then she wrote that it was bullet voting. To me this is pretty egregious, and it makes it hard for me to see how it could be an honest mistake. It's also stated in a broader context that Sara aptly characterized like so:

reading this article feels like you are intentionally trying to mischaracterize SRV.

I have to give Kristin some credit for responding here at all, but if she's acting in good faith, why isn't she at least apologizing for the particularly blatant and objective errors in her piece?

If there's one thing that gives me some pause in this assessment, it's this:

I have not spent my career promoting IRV. I’ve spent my career fighting climate change. As it became clear that climate action, and other popular and broadly beneficial actions, were blocked by broken institutions of governance, I started working on democracy reform.

My wife procures renewables at the utility company, and I've worked on renewable energy portfolio management software. I feel a similar concern for the connection between climate change and voting systems. There's something very human and relatable here.

I also understand that voting methods are famously counterintuitive. So when Kristin says:

when electing a president or governor or mayor, I don’t think a small group who feels intensely should count more than other voters.

Clearly she's falling victim to a logical trap. She's unaware that it's logically proven that a group may prefer X to Y even if a majority of its members prefer Y to X.
https://sites.google.com/a/electology.org/www/utilitarian-majoritarian

If you don't understand the implications of that, it's kind of understandable that you'd mistakenly think this is just some subjective thing and "of course" majority rule is preferable.

I don't really know what else to say. But several objectively false claims were made and I think it's fair to expect some acknowledgement of that and even an apology.

The simplest way to put it is that you can at least offer benefit-of-the-doubt with wording like Sara's (which describes her impression without sounding conclusive about Kristen's actual intentions) or just follow Rapoport's Rules strictly as a challenge to practice. Those Rules are totally effective against opponents who are acting in bad faith still.

By expressing your opponents view so that they say they couldn't say it better, it shows everyone that you get them. Then by expressing what you've learned or where you agree, it shows your thoughtfulness and good faith. Then criticizing their points comes across as devastating in that it's someone so understanding and respectful who nevertheless has serious critiques.