The city I live in, Eugene, OR, USA, is currently voting on ballot measure 20-349 which, if passed, will change the voting system for selecting the mayor and city council to use “STAR Voting”. This is an electoral reform proposal in the same vein as Ranked Choice Voting but which its proponents argue is a better system. I’ve been looking into it since I want to decide how to vote on the ballot measure, and since it’s potentially interesting and newsworthy I thought I would write about it.

What is STAR Voting?

They call it STAR Voting for two reasons: One: because you give the candidates running for office a “star rating” from zero to five stars, and Two: because it’s an acronym for Score Then Automatic Runoff.

The common method in the US is the have a two stage election process, first we have the primary election where you figure out who the candidates are (usually one candidate for each officially recognized political party, but it’s more complicated for “nonpartisan” offices), and then a general election between the candidates. Voters register one vote for their preferred candidate, and whichever candidate gets the most votes wins. Critics say that there are some bad incentives with this system, for example you might feel obligated to vote for the candidate you feel is most “electable” even if you’d prefer someone else, and you don’t get the option to choose any of the losers from the primary, even if they might be a better compromise choice than whoever the party faithful chose to be the standard bearer.

With STAR voting there’s no separate primary, instead there’s a single election where you rate every candidate who’s running from zero to five stars: you give your favorite five, people you don’t like zero, and others somewhere in the middle based on how appealing you find them. Then the election officials tabulate the votes: the two candidates that got the most stars are the two options, and then any ballot that rated A higher than B counts as a vote for A and any that rated B higher than A counts as a vote for B and whoever has the most of these votes wins. Since “zero” is the default vote, every ballot is involved in this Automatic Runoff part, but if you voted the same value for both candidates then you don’t have a preference, so it’s kind of like abstaining from the vote since you said you’ll be equally happy (or equally miserable) with either option.

(image from starvoting.org)

What is the benefit?

Because your ballot still has the opportunity to matter even if your most-preferred candidate doesn’t have a lot support, the advocates say that this system means there’s less need for “strategic voting” or settling for a “lesser evil” candidate, you can express your true preference without worrying that will be a wasted vote. It’s also simpler than ranked-choice systems where each voter gives a full rank-order of all the candidates. That can be tricky to specify on a ballot, and requires processing in a centralized way to implement the instant-runoff algorithm. Since STAR voting just involves adding up stars and reporting results for the various head-to-head matches it can be done in a more distributed way (which can be important, for example, since many election laws require the votes to be tabulated at the county level).

What are the drawbacks?

Any new system will have implementation costs, and any change in voting procedures is likely to cause confusion for some people as they have to learn a new system. Some people seem to be uncomfortable with the idea that your vote is set aside in the “no preference” category in the Automatic Runoff stage if you voted equal ratings for the two candidates that are in the running, that seems a lot like your vote not being counted. There are also some ways that counterintuitive results could manifest:

A weird example

Consider a town with 100 people, 33 of whom are members of the Enthusiastic Party and 64 who are members of the Grumpy Party. Two people from the Enthusiastic Party decide to run for mayor, Mr. Excitement and Mrs. Glamorous. There’s only one Grumpy candidate, Miss Taciturn. The Enthusiatic Party all love Mr. Excitement, and every one of them gives him 5 stars. They’re thrilled with Mrs. Glamorous, too, just not as much, so everybody in that party gives her 4 stars. And of course they’re not fans of Miss Taciturn so they give her zero stars. The Grumpies hate all the candidates, but they figure Miss Taciturn is the least bad, so all the members of that party begrudgingly give her 1 star and zeroes for everyone else. And now the results are tabulated: Mr. Excitement has 165 stars (5×33), Mrs. Glamorous has 132 (4×33), and Miss Taciturn has 64 (64×1). As the two candidates with the most stars, Mr. Excitement and Mrs. Glamorous go to the automatic runoff, and Mr. Excitement wins 33-to-0. Congratulations, Mr. Mayor! But wait, isn’t that a crazy result where the town now has a mayor that 64% of the people tried to vote against!? Yes. But also, it’s arguably the Grumpy people’s own fault for not giving their favorite candidate the full five stars like the ballot told them to, so it’s not clear whether it’s worth worrying about hypothetical situations that only come up if the voters vote with strange patterns.

A more plausible example

Imagine a town of 100 people that is split down the middle of the typical left/right spectrum, with 34 hardcore Democrats, 33 hardcore Republicans, and 33 independents. The candidates are Demi Kratz, on the left, Moe Derrit in the center, and Roy Publican on the right. The Democrats all give Demi 5 stars, and 0 stars for Roy, and figure that Moe at least isn’t as bad as Roy, so he gets 2 stars. The Republicans are the opposite: 5’s for Roy, 0’s for Demi, and 2’s for Moe. Moe doesn’t have a big party apparatus backing him up so he isn’t able to fire up the voters very much, and his centrist policy positions aren’t very exciting, but he does manage to convince every independent to give him 4 stars. The independents also see some things worth supporting in both Demi and Roy’s platforms, so they give those candidates 2 stars each. Now lets tally the results! Demi has 236 stars (34×5 + 33×2), Roy has 231 (33×5 + 33×2), and Moe has 266 (33×4 + 34×2 + 33×2). Demi and Moe go to the Automatic Runoff, where Moe was preferred over Demi by 66 voters (33 republicans + 33 independents) and Demi was preferred over Moe by 34 (34 democrats). Moe wins! Is that a good result? Well, he doesn’t have the intense support, but he’s the broadly-liked compromise choice. But Demi’s supporters might be angry: she was the top choice of 34 people, which is higher than the 33 people who said Moe was their top choice and 33 people who said Roy. So it might depend on your point of view about whether this system was better.

But those are simplified examples

In order to make the examples simple I had the various groups voting in lockstep, but that's probably not realistic. Different people in different groups feel differently about different candidates, and a system like this may encourage lots of candidates to run. How will that manifest in actual election results? That's a lot harder to say.

My thoughts

Personally I’m not sure how I feel about this system. It seems to me that elections with lots of candidates are harder to engage with, it’s usually easier to make a choice once things are narrowed down to two candidates, so I’m not sure I like that this system has room for lots of candidates right up to election day. However, with the US parties so nationalized most states and localities are very lopsided and the head-to-head matchup in the general election often ends up being noncompetitive, whoever wins the primary for the dominant party ends up in office, so I don’t think the current system works very well either. Since the current system isn’t working well it can be useful to explore alternatives, but I’m not sure the “problems” that this system “solves” are the key problems in current politics.

Discussion

What do you think of this system? If you were a resident of Eugene would you be voting for it or against it?