The recent presidential election has stirred up concerns with the Electoral College system, as well as the influence of third-party candidates on the overall result. Here’s how it went down:
- Of the 231 million eligible voters in the United States, only about 132 million (56.9%) actually voted.
- A handful of states (at least four) had the difference between the two major-party candidates less than the amount of voters that voted for a third-party candidate (meaning, if those third-party voters had voted for a major-party candidate, it could have tipped the balance the other way).
So, a huge number of people didn’t vote at all, and could have changed the result, and even those that did vote could have resulted in a different outcome, if their vote counted for a major-party candidate.
The non-voters likely didn’t do so out of protest for this particular election; Pew Research data shows that most of those non-voters aren’t even registered to vote (84.3% of registered voters did vote, though their data is based on the 2012 Presidential Election).
So, those who care to vote are voting for the most part. But then what about splitting the vote with a third (or fourth, or more) candidate? There’s many different voting systems to aggregate in more than one choice, if voters are given more than one choice when voting, and they can rank their preferences.
The current US system, where the candidate with the most votes (plurality of votes) wins, is more formally termed the First Past The Post system, and it tends to evolve into a two-party system, making it really hard for a third-party to gain enough momentum to matter, and stay there.
In order to change that, some change needs to happen to the ballot and the changes can generally be grouped into two categories: ranked choice (number all the options from favorite to least favorite), or approval choice (for each option, give your opinion on it).
If the ballot is modified so that each voter gives their first-, second-, third-, and subsequent-choice candidates for their vote, there’s several ways to utilize the second-and-subsequent votes for each person.
Some countries and states have adopted Instant Runoff Voting (IRV) as a means to allow votes for third options to not be “wasted” if that candidate doesn’t get plurality (Most recently, Maine adopted IRV for general elections of governor, and Senate and House state legislature). The way it works is:
If in the first round of voting, no candidate got a majority of the votes (note, majority, not plurality), the candidate who got the least is eliminated, and their votes redistributed by what the voters put as their next choice. That’s repeated until someone does have the majority.
That sounds good; people are able to vote for an underdog without fear of their least-favorite candidate getting their vote. Though it has a strange quirk where in some situations, it actually can harm a candidate to get too many votes (termed the Favorite Betrayal criterion, or Monotonicity criterion). And it’s not just theory; it has happened in some IRV-based elections in the US, and caused some counties/states to repeal IRV after trying it for a while.
In order to avoid that problem, and some other unattractive criteria, the Condorcet voting method looks like a very favorable system. In that voting method:
Each candidate is compared to each other candidate individually, and see if they would win if only those two were competing. If any one candidate beats all the other candidates, they are the Condorcet winner.
This seems to be a very fair outcome, however, there is the possibility of a loop being created (A beats B, and B beats C, but C beats A, so none are better than both the other two), and there being no Condorcet winner.
So, if Instant Runoff voting has problems with Monotonicity, and Condorcet voting sometimes doesn’t produce a winner, then what? Two other voting systems that Carneades.org presented in their recent video series on voting systems are the Borda Count and Bucklin Voting systems. Both these systems account for voters’ second, third, fourth and so on picks by giving them some additional weight in decreasing value. The Bucklin system has the least amount of math (you only need to tally up as many as needed to find a winner, not tally up all picks from all voters), but can lead to an unexpected winner, since someone who gets a lot of second-priority votes but none of the first-priority votes could be the winner (since the voters who voted for other candidates as first-priority cannot pick them again the second round). The Borda Count needs to tally all the priority rankings of all the voters (you give small fractions of a vote to even your last-place pick), but the benefit is then that every voter’s entire preference is counted.
So, how about this:
If there is a Condorcet winner, that candidate should win. If there is not a Condorcet winner, the Borda Count system should determine the winner.
In a ranked-choice vote, it seems this like a choice that eliminates most of the bad aspects, and emphasizes the good.
Instead of numbering candidates from best to worst, a ballot for an approval vote asks you to judge each candidate in isolation and give them a score. The simplest way a “score” can be given is a checkbox (an unchecked box is zero points, and a checked box is one point). This would be the least amount of change from a First Past the Post ballot: the only thing that would need to change is the instructions that currently say “pick one option” to say “pick one or more options”. Voters then vote for all candidates they wish to see in office.
Determining a winner for a vote done this way is:
Add up the total number of points each candidate got, and the one who got the plurality of points wins.
Because all voters can vote for more than one candidate, it’s possible to have a race where 95% of people voted for candidate A, and 93% of people voted for candidate B (the percentages don’t add up to 100%). The United States has long reported on an approval rating for candidates while they’re in office. This voting style moves that type of rating to the election process.
But then if you’re voting for two candidates, there’s no way to indicate you prefer one over the other. With a checkbox-style ballot, where you can only give one point per candidate maximum, all your approved choices are ranked the same. To alleviate this, an approval vote can be done more like a product review, where you rank each candidate on a scale (0–3, 0–5, 0–10, etc.), and they receive that many points for your vote. So, you can give your favorite candidate five points on the five-point scale, the third-party candidate you’d be okay with if they got into office two points, and the candidate you really don’t want to see in office you leave blank to give them zero points.
So which type of ballot/voting style makes sure voters can express their true intent with voting, while also not being too complicated to manage (for voters and for those calculating the results)?
One key thing I think needs to be able to be expressed by voters very clearly is the divide between candidates they would like to see in office vs the ones they would not. Here’s how you’d express that view in each ballot style:
- First Past the Post: The candidate you vote for is the one you want to see in office. All others you don’t want to see in office. This is actually not truthful; you may be okay with one or several of the other options being in office, but there’s no way to indicate that.
- Ranked Choice: In an election between three candidates, if there’s one candidate you do want in office (Candidate A), and one you don’t want in office (Candidate Z), and one you don’t care either way about (Candidate M), don’t just mark down a “1” for Candidate A (as you would have in a First Past the Post ballot). If you do that, and Candidate A doesn’t get enough votes and is eliminated, your ballot is then thrown out as if you hadn’t voted at all. Not numbering the rest of the candidates is like saying “I don’t care past this point; all the rest of them are equal in my mind”. So, in a ranked choice ballot you should fill out the entire ranking in order to properly show who’s your last-place choice. So, in this example, put “1” for Candidate A, “2” for Candidate M, and “3” for Candidate Z. And to make it even more clear, insert some write-in names into the ranking too, between Candidate M and Candidate Z, pushing Candidate Z even further down in your listing.
- Approval Voting: If the ballot is only checkboxes, this is the same as First Past the Post. If the ballot allows for a scale, it allows for the most expression of your views as a voter. Give your first-pick candidate the maximum number of points, and if you have second- or third-place picks, give them some number of points depending on your comparison of them to your first-place pick. Leave the ballot blank for any candidate you don’t approve of, to give them zero points.
Of those options, I think Approval voting most clearly indicates where a voter’s divide on each candidate lies, without needing to fill in the entire ballot for all candidates. Not needing to fill in the whole ballot (including all third-party candidates and needing to know leading write-in candidate names) in order to express my whole perspective puts less of an onus on the voters, which is good.
So, in my mind, after I’ve done all this research, I think having an Approval Voting style, with a ballot that gives a 0–5 scale gives a good balance between giving voters flexibility and at the same time not making it too complex for voters, and making it a more fair in overall result as it lets voters truly express their opinion, rather than worrying about spoilers and other unexpected results from voting truly.