On September 4th, Massachusetts had its primary elections. The biggest headline grabber was the 7th US congressional district (mine), where Ayanna Pressley managed to unseat incumbent Michael Capuano to be the Democratic nominee and thus likely the next member of the US House of Representatives. But also dramatic was the 3rd district, where Lori Trahan won the Democratic nomination with 21.6% of the votes in a multi-way open race; just 145 votes ahead of her nearest rival.
Isn’t democracy supposed to mean “majority rules”? How can under 22% be enough to win? This is the kind of perverse outcome you can expect in a choose-one voting system when there’s a multi-way race without two clear frontrunners. And better voting methods would fix this, improving democracy in many ways.
Right now, the better voting method that’s best known is called Ranked Choice Voting, RCV. Voter Choice Massachusetts, a reform nonprofit, is building a powerful movement to pass RCV in MA, probably in 2020. That would almost certainly have improved matters in the 3rd district, ensuring that the winner had to show more than 22% support.
I definitely endorse Voter Choice MA’s work, and think RCV would be a huge improvement over the status quo. But RCV is not the be-all and end-all of reform. It solves some, but not all, of the practical problems with choose-one voting. Other reforms can do better. I’m going to discuss
I’ll explain RCV by showing how it solves the problem of minor spoiler candidates, like Nader in Florida in 2000. In RCV, voters rank candidates in order of preference, and then if no candidate has over 50%, the weakest candidate is eliminated and their votes are transferred to those voters’ next highest preferences. In 2000, that means that Nader would have been eliminated. So instead of being wasted, the votes which preferred Nader first would have helped decide the outcome, probably for Gore. Nader would not have been a spoiler.
This would probably also work in a crowded primary like MA-03. As the weaker candidates were eliminated, votes would transfer to the stronger ones, so that it would take far more than 22% to win. A few of those voters would not have bothered to rank every candidate, so some ballots would be “exhausted” and the winner might not have over 50%, but they’d certainly be a lot closer. It’s a more democratic way to vote.
But it doesn’t solve every election problem. An example where it would probably have failed is the Egyptian presidential election of 2012 — the first (and last) truly democratic election that country has ever had. This was run with a two-round runoff system, in which all but the top two candidates are eliminated simultaneously, and ballots are cast again for the second round. Though this is not the same as RCV, it’s similar, and results in this case would probably have been the same.
We can explain this by focusing on the top 3 candidates. There was Morsi, an Islamist candidate supported by the Muslim Brotherhood; Shafik, an old-guard military candidate associated with deposed president Mubarak; and Sabahi, a dissident and reformer. Pre-election polls suggested that a reformer candidate would win a one-on-one race, getting support from most old-guard voters against an Islamist candidate and from most Islamist voters against an old-guard candidate. But because Sabahi was in 3rd place, he was eliminated, and in the second round his voters split between Morsi and Shafik. The result was that Morsi won the second round with 51.7%—but many of those votes were more “against Shafik” than truly “for Morsi”. As president, Morsi tried to consolidate Islamist power, but most voters were unhappy with that, and he was deposed in a coup a little more than a year later, ending Egypt’s short period of democracy.
Theorists have warned that this kind of “center squeeze” problem could happen with RCV as well. In fact, in Burlington, VT, 2009, a similar thing happened—a Democrat was squeezed between a Republican and a Progressive Party candidate, prematurely eliminated, and unhappy voters repealed RCV soon after.
Such center-squeeze problems do not happen in every election. So RCV is still better than choose-one voting, where many voters are forced to dishonestly pick a “lesser evil” every time if they want their vote to matter. In particular, RCV works well when there are only 1 or 2 ideological groups, and besides that the main factor is candidate quality; as in most partisan primaries, and the Bay Area local elections where it’s been tried.
There are voting methods which fix this. First, and most simply, there’s approval voting. In this voting method, voters approve as many candidates as they want to, and the candidate with the most approvals wins. Center squeeze is less likely, because voters on each wing (especially the wing that’s likely to lose) can also approve the centrist candidate. This would certainly have helped in MA-03, because some voters would approve more than one, so the winning total would be higher; but some voters would still approve just one, so it’s likely the winning total would still be under 50%.
Approval voting is being considered by voters in Fargo, ND this November. Check out electionscience.org for the latest updates on this campaign.
An method that’s even more expressive and strategy-resistant than approval is STAR voting. This stands for “score, then automatic runoff”. Voters give each candidate a score from 0 to 5 stars; the two candidates with the highest total score are finalists; and the winner is the finalist rated higher on more ballots. This would do even better than RCV, and much better than approval, at ensuring the winner truly had more than 50%; the only way it could fail to do so is if significant numbers of voters gave both finalists exactly the same score, suggesting that the two are so similar as to be nearly indistinguishable. It’s also more resistant to center squeeze than either of the other two methods; the centrist will almost surely get enough points from either side to be a finalist, and then is guaranteed to win. (Assuming that they are truly a centrist — the favorite of the median voter, and preferred by either wing over the other wing).
STAR voting is being considered by voters in Lane County, OR this November. Check out equal.vote/lane for updates on this campaign.