Consider a couple of basic facts about the presidential election.
First, Donald Trump won the primaries with around 45 percent of the vote—and that includes votes taken after he’d effectively secured the nomination. Through the primaries, Trump’s opponents destroyed each other by splitting similar constituencies, while Trump skated through with minority support.
Second, minor parties won more votes than Trump’s margin of victory in some key states, including Florida, as Eli Watkins reviews for CNN. Watkins thinks third parties may have cost Clinton the election, but he may be wrong about what voters otherwise would have done. (Update: Sasha Volokh thinks Gary Johnson helped Clinton in the end.)
If it’s true either that Trump would have lost the primaries against a single amalgam of his opponents, or that Hillary Clinton would have won various key states and thus the electoral college in a two-way race, then Trump owes his victory to America’s system of winner-take-all plurality voting.
The way we vote is not a law of nature. It can be changed. I think there are good reasons to move to approval voting, meaning that voters can select multiple candidates. The main case for approval voting is that voters can better express their preferences. So, for example, in the primaries, a voter could have selected both Ted Cruz and Marco Rubio (or even everyone but Trump), and in the general election a voter could have selected both Stein and Clinton.
Absent approval voting, all we can do is make educated guesses about what might have happened in hypothetical two-way match-ups. I think it’s a good guess that Trump would have lost the primaries under approval voting and that Clinton would have won the general election under it, but we can’t be sure.
One reason I like approval voting is that we’d be more sure about people’s preferences. If, under approval voting, someone still voted for Stein only, that would mean the person really didn’t want to help Clinton win. I can even imagine some people voting both for Stein and Trump. Some people who voted for Gary Johnson would have also voted for Clinton, others would have also voted for Trump, and still others would have voted only for Johnson. People who voted for Evan McMullin—who did well in Utah—could have opted to vote also for Trump or Clinton. All of this is useful information that our voting system now misses.
Of course approval voting is only relevant with more than two candidates on the ballot, and it works especially well with two major candidates plus minor ones. In such cases, approval voting eliminates the “wasted vote” problem.
No doubt there would be losers under approval voting relative to the status quo. Trump probably would have lost with it. I have heard the argument that the current system is superior precisely because it allows candidates with minority support to sometimes win. But I think that’s an extremely dangerous game to play. In a three-way race, a dangerous demagogue would potentially need only a third of the votes (plus one) to win, and the percent goes down with more competitors.
Other things equal, a candidate with broader popular support is likely to be a better leader than a candidate with minority support. True, I other things aren’t always are equal. For example, I support the electoral college because it requires support across more states, even though it sometimes allows a candidate to lose with a higher popular vote, as Clinton did. (Generally if this happens the popular vote will be close.) But, in the case of approval voting, I think it assures broader popular support without bringing any important negative consequences.
All voter systems have their quirks, and this is true of approval voting in certain situations, such as a tight three-way race. Then voters would have to navigate tricky voting strategies. Let’s say you prefer Alice to Betty and Betty to Carl. Do you vote for Alice and Betty to stop Carl, or do you vote only for Alice in the hopes of blocking Betty but in the fear that you might help Carl win? But strategy can be at least as tricky for voters under the status quo.
The main alternative to approval voting (so far as I’m aware), besides the status quo, is ranked voting, in which a voter expresses a first choice, second choice, and so on. I prefer approval voting simply because it’s much easier to implement and understand. I grant that ranked voting can have advantages in certain cases, as in tight three-way races. Practically speaking, though, our main problem is minor parties “spoiling” elections, and approval voting solves that problem perfectly.
The main technical objection to approval voting that I’ve heard is that ballots might be prone to manipulation. It might be easy for someone handling a ballot to “sneeze” and put an additional mark on the ballot. If that’s really a potential problem, there’s an easy fix for it: Let voters mark yes or no for each candidate.
Even under approval voting, a candidate might win with less than majority support, say in a tight three-way race in which most people vote only for a single candidate. I don’t see any inherent problem with that, and I don’t think it’s a likely scenario anyway. But there is an interesting possible add-on to approval voting: Require that the winner receive a majority, else a new election is called with different candidates. Such a rule would encourage voters to err on the side of voting for multiple candidates when considering strategy. Again, I just don’t think this is much of a practical concern; I think simple approval voting generally would lead to winners with majority support.
At stake are not only the presidential elections, in which people have worried about Ross Perot and Ralph Nader “stealing” votes, but elections all the way down the ballot. For example, this year a Libertarian probably cost Colorado Republicans a state senate seat. The Republican incumbent in my area, Laura Woods, narrowly lost to Democratic challenger Rachel Zenzinger by 35,310 to 36,616 votes. The Libertarian got 4,573 votes—3.5 times the margin of victory. It’s a good bet that, under approval voting, Woods would have won because a lot of Libertarian voters would have picked her, too. Incidentally, Colorado Republicans maintained a single-seat majority in the state senate despite Woods’s loss. I discussed some other examples in 2014.
A note about primaries versus general elections: I think political parties, as private organizations, properly run their own affairs. I think parties would be wise to institute approval voting (or else use a caucus system with delegates), but government shouldn’t force them to. On the other hand, governments properly are responsible for general elections, so governments would need to institute approval voting in those elections.
I’m a little surprised that alternative systems of voting have not gained more attention given the obvious problems of “spoilers” every two years. Perhaps Trump’s surprising ascendency to the White House will encourage more people to take a look at approval voting, which probably would have led to a different outcome.
Update: Someone on Facebook reminded me that Maine voters considered ranked voting this year, and I see that the measure in question passed. I reiterate the point that ranked voting is complicated. In Colorado we had twenty-two candidates for president. Do we really expect people to keep their ballots straight with the complexity created by ranking? By contrast, it would be trivially easy to go through and “approve” or not each candidate individually. Regardless, hopefully the Maine measure will generate more attention for alternate voting systems.
Latest Updates | Image: Shannon
The Trouble with Ranked Voting
Ranked voting violates Arrow’s impossibility theorem, just as our current system does. (Our current system actually is ranked voting, with the proviso that we rank one candidate first and don’t bother with the others.) All fair systems (range voting, approval voting) discard the requirement that each voter make exactly one first-ranked choice. See http://skep.us/4281.
—Brian Dunning
Approval Voting Is Simple
I’m impressed that you got a comment from Skeptoid‘s Brian Dunning. I agree that approval voting seems to be the simplest system. And it would do a better job of expressing the voter’s preferences. It would be easy to implement. When I worked as an election judge our most common error was people “over voting.” If the check for too many votes was removed from the tally machines, we could do approval voting now.
—Mike Spalding