I had the pleasure of sitting down with Robert Wiblin of 80,000 Hours to talk about The Center for Election Science and voting methods. 80,000 Hours is an effective altruism focused podcast, which we’ve written about before. We had a lot of fun in the interview, which you can find on the 80,000 Hours website. We want to give a big thank you to their team for providing this opportunity.
Like many 80,000 Hours interviews—which is a great podcast to listen to in general—this interview was long and detailed. Fortunately, there’s recently been more of a demand for these types of podcasts.
While Robert provided every opportunity to provide nuance, there were some opportunities I missed where further detail may have been beneficial. Also, this interview was recorded at an earlier date, so some of my responses are out of date. Below, you’ll find these miscellaneous issues addressed.
There was a section about conferences and events. You can find the events we have on our events page. We also have a number of events in June including socials and charity dinners that you can RSVP for through Eventbrite.
An additional reason why better voting methods are higher impact than voter turnout is that you can have 100% voter turnout and wind up with a terrible outcome. This is from the computer science principle “garbage in, garbage out”. When you bring in bad information, you’re going to get bad output. It doesn’t matter that you have everyone voting if the way that they’re voting is fundamentally flawed.
We’ve made our hires since the interview, including our local campaign coordinator in Fargo, ND. They’re great, and you can learn more about them on our Who We Are page.
Organizations Using Approval Voting and Score Voting
It’s tough to remember them all, but here’s a list (surely incomplete) of organizations that use approval voting and score voting as well as instances of early use.
Other International Organizations In the Voting Method Space
There’s another organization, FairVote Canada (no affiliation to the FairVote in the US), that does a good job promoting proportional representation in Canada. There are also some other similar organizations with various voting method platforms in Europe.
I offered some criticism of other organizations in this space who offer electoral support to governments to conduct and design their elections. My gripe was that they were offering inferior solutions to single-winner elections. Such organizations are International IDEA, International Foundation for Electoral Systems, and National Democracy Institute. This isn’t to say that they don’t provide a number of excellent resources or do many other aspects very well. I personally use them as a resource quite frequently. But I think they have a ways to go in their performance of implementing single-winner voting methods.
The Concept of Majority
We touched on this a bit. It’s a deceptively unintuitive concept—particularly because it seems like it should be easy. On our blog, you can find a thorough analysis of the majority concept as applied to cardinal methods like approval voting as well as runoffs and instant runoff voting.
Approval Voting Critiques
Robert really dug in to make sure that we expanded on approval voting criticism. If I missed anything, you can probably find it in a previous article I wrote on this topic.
Voting Method Research
Asking about voting method research, it would likely be good to see more research on multi-candidate races and see how the election would differ based on the voting method. One way research could, in general, be better would be if raw ballot information on non-plurality methods were easily made available to the public. Because so little information is provided in plurality ballots, all the information is inherently there. But for methods that ask for more information, having access to all the information is essential for research.
There’s also some research on voting methods and outcomes. I’m not personally aware of research that looks into identifying policy support as an outcome measure that has the voting method as a predictor variable. It’s not possible to use random assignment when looking at these outcomes, but statistical models can control for potential confounding variables, and this statistical control is commonplace.
Our Work In Fargo, ND
Awhile back we were contacted by a Fargo, ND election commission member. We’ve been offering advice and intellectual support to help that person, Jed Limke, set up Reform North Dakota, which is operating as Reform Fargo locally. We’re working alongside local support there and have also hired locally for the campaign coordinator. We hope that our support with locals in Fargo empowers their citizens to have the most meaningful vote in the country. Fargo would then serve as a replicable model for other cities.
We also have a recommended books page if it’s helpful.
We broached the Gibbard-Satterthwaite theorem but didn’t really go into Arrow’s Theorem. If that’s the kind of thing you’d like to learn more about, I wrote an article on it shortly after his death. I also interviewed Kenneth Arrow awhile back.
Proportional Approval Voting
I alluded to a proportional approach to approval voting. One approach in the literature (made publicly available here—it’s the sequential apportionment one) is to have a normal approval voting election and then keep going. But every time a candidate gets elected, every voter who approved of that candidate gets the weight of their ballot deweighted. This would repeat until all seats are filled. This deweighting upon getting someone elected is a common approach within proportional methods.
One deweighting method for ballots upon getting voters’ approved candidates elected is 1/3, 1/5, 1/7, … Of course, there are different deweighting schemes that have subtle pros and cons. There’s also research being done to have proportional approval methods deal with the surplus votes that go over the threshold for getting a candidate elected. Deciding where the election threshold stands is one of the factors to consider.
In any case, the simple approach outlined above provides a proportional outcome. It can work for at-large council elections. It can also be scaled up to have many multi-member districts so that an entire legislature can be filled. Using multi-member districts of at least five while using a proportional voting method creates a kind of bullet-proof vest for gerrymandering attempts. This is also much easier than independent commissions and will surely lead towards better outcomes.
Using an approval-voting-style ballot also provides an easy ballot. That’s essential when you have a multi-winner election which tends to draw many more candidates than single-winner elections. For comparison, Robert referenced the difficulties that STV had with this issue because the method insisted that the voter rank their choices. (Try approving within a list of 30 options versus ranking; it’s a huge difference in effort.)
Reiterating the Target
I stated this at least two times in the interview, but it’s worth honing in on. While we look at all kinds of voting methods as we figure out this problem domain, it’s unambiguous that plurality voting is the worst voting method there is, and it needs to go. I responded candidly when asked about the reform instant runoff voting/ ranked choice voting (IRV/RCV). We’ve critiqued IRV just as others have critiqued approval voting. But the place where we spend our overall energy is replacing plurality voting with approval voting wherever possible. We commend others for doing the same even if we disagree on the best solution. That said, we’re quite comfortable with our case for approval voting.