Commentary & Analysis

How I Came to Care About Voting Systems

Frustration Raises Its Head

I’m concerned about the issues. I’ve never been politically apathetic. I care about healthcare. I care about education. I care about taxes. I care about all that.

Since my days as a public health graduate student, I’ve cared about US health insurance policies. So did my classmates. They cared about health insurance so much that they made a group about it. And I joined them.

We took time out of our busy schedules as graduate students. We had meetings. We did events. We were concerned students.

We did this for a while. Time passed, and the US Presidential primaries came around. We listened. We noted how those candidates thought about US health insurance policies. Many of those candidates disagreed with us. But some shared our view.

I thought, naively, that my classmates would vote a certain way. Really, I knew they would. I knew they would vote for the candidates sharing our health insurance stance. I was entirely wrong. My classmates did not vote that way. Instead, my classmates voted for candidates that opposed their stance.

Why would my classmates do this? These were busy graduates students. They had busy schedules. And they chose to use their precious spare time to devote to a group on health insurance—an issue they deeply cared about. This wasn’t even a fringe issue. Polls revealed that about half of Americans agreed with them on this. But my classmates wouldn’t even vote for the candidates that shared their view.

My classmates were worried about throwing away their vote. They thought voting for the candidates that agreed with them was a lost cause. But if even my dedicated classmates couldn’t bring themselves to vote for candidates that shared their cause, then who would?

The answer, unfortunately, was hardly anyone. Hardly anyone was willing to honestly vote their favorite. Our group had a speaker that did events for us. And he was just as devoted to our cause as we were. He even compared candidates on the issue during his presentation. But his presentation was odd. He left out the candidates that agreed on our view. He pretended they didn’t exist. He was just as disappointing as my classmates.

My classmates befuddled me. I argued with them. I made them uncomfortable. They were relentless. They were determined to vote against their interests.

So I got angry. This whole situation was wrong. Everything was clearly wrong.

Questing for a Solution

I stepped back for a bit. These were good people. And they were smart. Maybe it wasn’t their fault.

Something was holding them back. It was our voting system. Our voting system limited them to choosing just one candidate. This limitation caused vote splitting. Plurality Voting corrupted my classmates and made them betray their own values. There had to be a better way.

I read what I could find. As I explored, I became more aware of how the voting system affected those around me. Everyone I knew was turning their back, astoundingly, on themselves. It wasn’t just with healthcare issues. It was with everything. People I knew apologized for candidates that fought against them, and they ridiculed the candidates that adopted their views. Others showed a bitter indifference. This was not a democracy to be proud of.

I began to care about voting issues at the same level I cared about all my other issues—only combined. And I cared about voting theory because I cared about those other issues. The voting system was even more important than other voting-related issues. An election environment with 100% voter turnout, no voter or election fraud, and no corruption would still leave people voting against their values. And even this utopian election environment would continue to marginalize good candidates while encouraging and electing bad candidates.

The voting system itself was definitely the most important factor.

First, Second, Terrible

I learned about systems that let you rank candidates. I looked into Instant Runoff Voting (IRV). It was bad. I learned that more than a sixth of the time it would behave in this odd-ball way. IRV’s odd-ball behavior made it possible for you to hurt candidates by ranking them better and help candidates by ranking them worse. This was the Alice-in-Wonderland-like effect of voting systems. This walk through the looking glass could even happen in traditional top-two runoff systems. In fact, it did happen in Louisiana’s infamous 1991 gubernatorial runoff between Edwin Edwards and David Duke. Imagine being a candidate and knowing that more votes from certain voters would have caused you to lose in a runoff. Edwin Edwards was a real candidate that got that feeling.

Also, the way IRV calculated the winner could be complicated for impatient voters. It was time consuming, and even the results tables could take effort for people to absorb. Worse still, I learned that IRV didn’t let you vote your honest favorite—not without paying the price.

Admittedly, it buffered against the spoiler effect when those spoiler candidates had few votes. But the spoiler effect raised its ugly head again once those candidates became more popular.

I read about a 2009 IRV election in Burlington, VT. There, conservative voters learned the hard way that they would have been better off voting against their favorite. And people betraying their favorite was exactly what I was trying to correct in the first place. So IRV was a clear loser.

There were a number of other ranking systems. One called the Borda Count assigned values to the rankings. The Borda Count also encouraged voters to betray their favorite. The inventor of this system, Borda himself, had an infamous quote: “My scheme is intended for only honest men.” Unfortunately for Borda and the rest of us, we do not live in a world of honest voters. Most real-world voters are strategic.

I continued and learned about another ranking method, Condorcet. The Condorcet method simulated pairwise comparisons between candidates like a round-robin competition. Condorcet’s winner was the candidate that went undefeated in these pairwise comparisons. But then it got weird. This undefeated winner didn’t always exist. It was like a rock-paper-scissors match where one option could always lose to another. And the way to figure the winner in these situations could get EXTREMELY complicated. The Condorcet tabulation tables could be difficult for many people to understand as well. And, like the other systems, Condorcet didn’t let you vote your favorite all the time either. Strike Condorcet.

Some Voting Systems Are More Equal than Others

During this journey I also learned about Kenneth Arrow. Arrow is an economist that explores voting theory. He specifically focused on single-winner ranking systems in his most famous work. In this work he created a theorem that showed no ranking system could pass certain criteria. And he saw these criteria as fundamental.

But Arrow only intended his criteria to apply to ranking systems. Some academics even thought that certain nonranking systems either escaped his criteria or that his criteria simply weren’t appropriate for those systems. So, perhaps the takeaway from his work is to just not use Plurality or ranking systems?

I later came across a textbook where Arrow wrote a short essay on the public interpretation of his theorem. Even if we wanted to apply his theorem to all voting systems, his essay was still noteworthy. He stressed that the search for better voting systems was not futile.

He analogized voting systems with motor engines. He pointed out that the laws of physics keep us from inventing a perfectly efficient car engine. Of course, this lack of perfection doesn’t mean that some engines can’t be better than others. The fact that a Pinto’s engine obviously doesn’t run the same as a BMW’s engine is self-evident. Arrow’s point was easy to see.

He emphasized that the same principle was also true for voting systems. Some voting systems can still be better than others, even if none could be perfect. And after reading Arrow’s essay, I was seeing our current Plurality voting system as an antiquated, Model-T Ford from the turn of the last century. Frankly, I was ready to finally trade in that piece of junk for something better.

I Approve This System

I extrapolated from Arrow’s theorem to instead look at nonranking systems. Score Voting (a.k.a. Range Voting) was such a nonranking system. Here you rated candidates on a scale in the same way that you’d evaluate your professor at the end of a term or fill out a survey. The candidate with the highest total score (equivalent to highest average) won. This system was looking really good. It was expressive and overall very simple. It even let you rate your favorite candidate as best—always and without negative consequence.

But there was a system that was simpler still. A system called Approval Voting kept the highlights of Score Voting while still allowing you to always express your honest favorite. Approval Voting was so simple that it looked exactly like a regular vote-for-one Plurality ballot. The difference was that you could choose as many candidates as you wanted. This option to vote for multiple candidates removed possibilities of either vote splitting or a spoiler effect. It was exactly what I was looking for. Plus, results tables were just approval percentages. And that was easy for people to understand. Approval Voting was both good and simple.

Everyone Has Their Critics

Since settling on this method, I’ve had some people question how Approval Voting affects “one-person-one vote.” The question in itself suggests a misunderstanding of what that principle means. It was best defined in the US Supreme Court case Reynolds vs. Sims. In that case, the state of Alabama had set up its districts so that they varied wildly in population. Alabama’s worst extreme made it so that voters in one district had up to 41 times more voting power than voters in another. The Court rightfully saw this for the atrocity that it was. It said that each vote must be weighted the same. And it was the weight of the votes that the Court referred to, not how those votes were expressed.

But with Approval Voting, you don’t get an unfair advantage when you choose multiple candidates on your ballot. Let’s say I choose two people on my ballot, and you vote opposite of me by choosing the remaining four. You’ve voted opposite of me, and you’ve chosen twice as many candidates. Yet, your candidates are not winning over mine. All the candidates still only have one vote. Approval Voting, therefore, passes the “one-person-one vote” principle.

There are also some critics that think Approval Voting would just lead voters to choose only one candidate all the time. But does that really make sense? Imagine a strategic Nader supporter in 2000 that voted for Gore instead of her favored consumer advocate. If the Nader supporter used Approval Voting in that election, then do you really think that supporter would have voted for only Gore or only Nader? Of course not. Under Approval Voting, the strategic voter would have voted for both Nader and Gore. The strategic voter in Approval Voting doesn’t want to risk Bush winning, and that’s why she includes Gore. And she also wants to show support for Nader, so that’s why she chooses Nader as well. To choose only one in that situation would be ridiculous.

It’s not to say that it’s never appropriate to choose only one candidate with Approval Voting. Sometimes it is. But you get the option to choose multiple candidates when you need it—and that’s really what matters. Moreover, in every large-scale study on Approval Voting, most voters did choose more than one candidate.

The option to choose multiple candidates may be more important than you first realize. This option gives candidates challenging the major-party candidates an opportunity to receive legitimate support that they wouldn’t otherwise get under Plurality Voting. And challenger candidates getting legitimate support helps keep them from being marginalized by media and debate forums. This fair measure of support would also likely provide the motivation for people to fight back against restrictive ballot access laws. Our current ballot access laws frequently give obscene favoritism towards entrenching the major parties. With Approval Voting’s fair support for challenger candidates, voters would actually have an incentive to want their favorite candidates on the ballot.

I’d long since graduated from my public health graduate program by this point. Had I been able to ask my classmates though, I think they would have appreciated that past election much more with Approval Voting. Then the candidates supporting their favored health insurance policy would have stayed on the political radar. And my classmates would have appreciated those candidates in the limelight. That’s because under Approval Voting they’d have actually been able to vote for those candidates without fear of vote splitting or wasting their vote.

Probably one of the most ridiculous remarks on Approval Voting is repeated by an organization that, oddly enough, is dedicated to election reform. This organization imagines an election where more than half the voters have a clear favorite. And they say Approval Voting seriously risks not electing that clear favorite. But this organization’s contrived scenario requires that almost no voters show any preference whatsoever between the frontrunners—even though these particular voters do have an actual preference. With Approval Voting, it makes sense to distinguish between frontrunners and then vote for everyone else you like. And if (astonishingly) no one bothered to distinguish between the frontrunners, then it obviously didn’t matter much to the voters which frontrunner won anyway. This organization’s entire remark is either thoughtless or completely disingenuous.

You Say Majority, I Say Make Believe

This organization’s awry criticism frequently gets confused with the idea of a majority winner. But that term “majority” is weird in the election world. Typically, people think “more than 50%” when they hear “majority.” But the truth is that it is perfectly possible for this “majority” winner to simply not exist when there are more than two candidates.

As we’ve seen, runoffs and systems that transfer votes don’t necessarily create a majority winner because they too can eliminate candidates when they shouldn’t It’s also normal for those systems to have significantly fewer voters in later rounds than when they started. Others think that majority winner means the candidate that can beat everyone one-on-one. But then we already know from looking at Condorcet that this too does not always exist. Even Approval Voting doesn’t always guarantee a winner with greater than 50% approval.

People that seek a guaranteed “majority winner” with their voting system are simply chasing the end of a rainbow. Such a system does not exist because guaranteeing a “majority winner” is impossible.

A Mathematician Breaks Out His Yardstick

I became more curious about the different ways to evaluate all these voting systems. And I learned about a mathematician that used a computer program to simulate countless elections. It turns out that his process is common in the sciences. It’s used when it’s just not practical to take large samples over many conditions. His particular program had “voters” elect a winner under various scenarios. It then determined, on average, how close each voter was to electing their own favorite candidate. The closer the actual winner, on average, to each voter’s favorite candidate, the better the voting system scored.

This mathematician recognized that we did not live in Borda’s fantasy world where everyone voted honestly. So he astutely had his program factor in strategic voters. And the difference between voting systems given strategic voters was astounding. Score Voting and Approval Voting were among the top performers here with all the other systems bringing up the rear. But not even Approval Voting or Score Voting was perfect. Even they weren’t able to pick the best winner every time.

Naturally, “best winner” and “random winner” appeared on either side of the scale. I focused on strategic voters to be realistic. What was interesting was the difference between “random winner” and Plurality Voting. You could call this difference the gain from the invention of voting itself. This gain was approximately the same difference between Plurality Voting and Approval Voting. Profoundly, the simulation was saying that by switching from Plurality Voting to Approval Voting we would get roughly the same amount of gain as from the invention of voting itself.

To stress a point, I’m not saying Approval Voting is perfect. It’s not. It won’t read everyone’s minds and select the magic, ideal candidate. But it’s a whole lot better than what’s out there. It’s certainly much, much better than our vote-for-one Plurality Voting. You don’t lose anything by giving people the option to choose more than one candidate here. The transition is also really easy for voters. And it certainly would have fixed the personal struggles my classmates faced.

Cutting the Cake with Proportional Representation

I’m also not saying that Approval Voting is appropriate for every circumstance. Approval Voting is great for executive offices and other single-winner elections. But for elected bodies like councils and legislatures, proportional voting systems can make a lot of sense. Proportional systems give people their fair share of seats according to how many other voters share their views. If an ideology has 20% of the votes, for example, it gets represented with 20% of the seats. Proportional systems also work fine with independent candidates. These proportional systems are also virtually immune to gerrymandering.

Today, our elected bodies are either (1) elected as a disproportional at-large bloc or (2) gerrymandered up into single-member districts via winner-take-all elections. City councils frequently use a bloc system that completely denies competitive voices so that minority viewpoints get no representation whatsoever. Absent controversial “corrective” gerrymandering, minority voices get no representation when councils use single-member districts either. Proportional systems, on the other hand, give minority voices their due without these clumsy practices.

Beyond our winner-take-all system’s follies with gerrymandering, it makes other terrible mistakes. Political groups elected under winner-take-all frequently get more than half the seats while only getting less than half the votes. Canada’s national 2011 election, for example, gave their conservative party more than half the seats. Oddly, the party won those seats with less than 40% of the nation’s vote. And this odd result occurred despite Canada using independent redistricting commissions since 1964. This goes to show that independent redistricting commissions are not the saving grace.

This gaffe under winner-take-all elections happens all the time in the US as well—not that many Americans would know given how scarcely it’s reported. In the 2000 elections, for example, Democrats got less votes than Republicans for their US House seats in Texas, Wisconsin, and Arkansas. But that didn’t keep Democrats from taking the majority of seats in all those states anyway. And the table frequently spins the other way in favor of Republicans. Proportional systems, on the other hand, rarely exhibit this obnoxious behavior.

Now, legal barriers currently forbid proportional systems federally in the US. But proportional representation is still perfectly possible at the state and local level. We can do much better than our winner-take-all and bloc systems.

Traveling from 1885

Finally, remember Kenneth Arrow’s essay referring to car engines. Just because we can’t achieve perfection doesn’t mean some systems aren’t better than others. And do you really want to keep driving that clunker Model-T Ford we call Plurality Voting? Now, Approval Voting may not be as awesome as a real-life, time-travelling, 1981 Delorian. But it’s easily an upgrade that puts our democracy’s vehicle into the right millennium. And proportional representation should probably come along for the ride.