This is a follow up to my 2016-02-28 blog entry about the 2016 Presidential election process and how the voting model used by the United States—plurality voting—is causing damage to the Republican party. I strongly urge you to read that entry first unless you are already familiar with plurality voting, score voting, and approval voting models.
On Super Tuesday (March 1, 2016), it became quite certain that the Democratic and Republican nominees for the 2016 US Presidential general election will be Hillary Clinton and Donald Trump, respectively.
This particular pair of major candidates is possibly the least-popular match-up in our lifetime. On one side you have a almost-certain felon being afforded sluggish prosecution by a affable Department of Justice and on the other hand you have a would-be autocrat with no clear conviction beyond narcissism. Has there ever been a more fitting application of the aphorism selecting the lesser of two evils?
How did we get here?
Why has our democracy thus devolved?
Put aside the matter of why these two candidates have any supporters at all because all candidates will have some support and these two demonstrate a charismatic indelicate control of power to a segment of the population. While it is potentially worthwhile to determine the reasons any given person would support either Hillary Clinton or Donald Trump, it's also important to realize that it's a fool's errand. A Democrat and Republican cannot understand the motivations of the opposite; why would we think we can understand the supporters of these particular specimens?
What I would like to think about here is how our voting system—the most fundamental instrument of democracy—is failing us, defeating us, in myriad ways.
In the United States the predominant voting model is called "plurality voting." Each voter casts one vote of approval and one vote only. Plurality voting requires that a voter not only support the candidate they select but support that candidate at the expense of
supporting all other candidates. The voting system dictates that you may support only one candidate. Support of multiple candidates is not permitted.
It is said that plurality voting enforces a principle of "one person, one vote
." But interpreting something meant to protect equal representation
as meaning each voter can only approve
of a single candidate is perverse. Clearly the intent of "one person, one vote" is to ensure that each citizen can submit one valid ballot and that ballot is counted in equal measure to all other citizens' ballots.
By disallowing approval of more than one candidate, plurality voting is pernicious and corrosive.
Origins of plurality voting
First, I want to think about why we use plurality voting. In my previous blog entry
, I argued that approval voting is not only superior but also more natural—it is the system I believe human beings instinctively use when forming a group consensus. To briefly reiterate, when forming a consensus, the natural process is to identify the one selection that is approved of by the most people. Axiomatic to that process is the assumption that each person may and very likely will
approve of multiple options.
For lunch, I could go for a burger, pizza, or tacos. You might want tacos, pasta, or a salad. Boom—decision made, and off we go to get some tacos. Another approval voting success story. They happen every day in our lives without even realizing the process has a name.
It is unnatural to constrain each voter to a single vote of approval as plurality voting does. So why is plurality voting widely used? I wasn't able to find an answer in spot searching. So here's my theory:
Approval voting is natural for small groups. However, processing votes from even a few dozen voters can make tallying of approval cumbersome, error-prone, and time-consuming. While it's easy to keep track of 5 peoples' approval votes for a movie or restaurant, a ledger might be needed to keep track of approvals from even a nominally larger group, say 10 people.
Requiring a single vote of approval from each voter also allows for a simple sanity check at the end of tallying: count the votes. Do you have a count of votes equal to the number of people in the group? If so, you probably have a valid tally. Meanwhile, with approval voting, the number of approval votes at the end of tallying votes from N people about M choices could range from 0 to (N x M). A simple count isn't enough to sanity check your tally.
I imagine at the scale of local, state, and federal elections, the complexity burden of manually tallying approval votes would have be too time consuming and too error-prone.
The genesis of plurality voting is therefore a matter of convenience for the people responsible for manually tallying votes (and presumably those who have to double-check or recount the votes).
Of course, computers have been with us for decades now. The persistence of plurality voting in the computer era has to be questioned. Plurality voting is an anachronism from an era of manual vote tallying.
The perils of plurality voting
Plurality voting may be a relic of a manual era, but if it's a good model, should we care that it's outdated?Yes, because plurality voting is not a good model.
During the primary process, each political party is charged with selecting a single candidate to represent the party. As in all contexts, plurality voting mandates that each member of the party support only one candidate.
This predictably engenders an emotionally-charged environment where members of the same party can see one another as adversaries divided by the (presumably) subtle variations between candidates. A supporter of Ted Cruz may strive to be cordial with a supporter of Marco Rubio, but until the primary season concludes, they are adversaries. If two candidates are even further apart ideologically, such as between Cruz and Trump, their supporters may find even cordial relations are strained.
Worse, with only a single approval vote to cast, the stakes of that single approval vote are extremely high for each voter. If a supporter of Cruz wants to convince a supporter of Rubio to join the Cruz bandwagon, it's necessary not only to make a convincing argument in favor of Cruz, but also to convince the Rubio supporter to stop supporting Rubio!
Plurality voting mandates support of only one candidate. Unsurprisingly, emotional attachment to that single candidate runs deep. In defense of their one candidate, party members are conditioned by the voting system
to poke holes in the armor of all the other candidates in the very same party. Republicans do the job of Democrats, and vice-versa.
This is the pernicious sectarianism of primary plurality voting.
Approval voting or score voting properly allow party members to approve of any or all candidates, at their discretion. In approval voting, you can be in favor of Rubio and Cruz simultaneously. In score voting, you can be in favor of Rubio at 100 points and Cruz at 80 points (or whatever).
Forced field consolidation
We've seen the bitter and damaging effects of forced consolidation in this year's Republican primary process. As the front-runner (Trump) continues to hold steady support from non-establishment voters, the "establishment" GOP votes have been split among a wide but slowly narrowing field. This is a consolidation process and it burns everyone it touches.
At the time of this writing, the GOP worries that the consolidation of establishment candidates did not happen quickly enough, leaving the very probable outcome of a Trump nomination.
Consolidation is forced by plurality voting since like-minded candidates realize they are splitting the vote of their mutual constituency. In this year's race we see that establishment candidates are realizing that without consolidation, no single one of them can harness enough support to defeat Trump.
As each candidate drops out, their supporters are left damaged and orphaned—forced to once again make a single selection among the remaining field. This year in particular, it is plausible that some Republicans have been through the consolidation wringer several times already.
Approval voting or score voting would not require this party-damaging consolidation. Candidates can still decide to drop out for financial or personal reasons, but by definition, it is impossible for a candidate to be a spoiler
to any other candidate. It is impossible to be tying up votes that would otherwise land in the hands of an ideologically compatible peer.
Furthermore, when drop-outs do occur due to financial or other reasons, voters are less burned because they (likely) approved or more than one candidate anyway. If a voter originally supported Rubio, Cruz, Paul, Carson, and Fiorina, it would be nearly a non-issue if one of those five dropped out.
Beyond the primary season, the most often cited criticism of plurality voting, which unsurprisingly comes from supporters of third parties, is what's known as Duvurger's law
. Duvurger's law says that a plurality voting system encourages long-term dominance by only two political parties.
This is due to the same consolidating force that occurs during a party's primary season. Third parties that are ideologically close to Democrats by nature siphon some voters from the Democrats; and the same goes for the Republican side. Historically, and in general, the Green Party takes some would-be Democratic votes while the Libertarian Party takes some would-be Republican votes.
If a third party is able to mount a meaningful campaign sufficient to pull in a sizeable number of votes, they will inevitably hurt the major party to which they are ideologically closest. In so doing, they will all but guarantee the success of the mutual adversary. This is called being a "spoiler."
Alternative voting models like approval and score voting are traditionally the academic playthings of third parties (though, as I mentioned in my previous blog entry, some third parties don't speak too strongly about voting models, despite replacing plurality being of crucial importance to the viability of third parties).
However, even the Republicans and Democrats should get on board with approval or score voting as insurance against such a "spoiler" effect in the future.
Because plurality voting inevitably creates a two-party system, and thanks to human tribal nature, in virtually all political matters, the Republicans and Democrats must
be opposed to one another. The only areas where we tend to find consensus is in aggregating more power in the state—power that both Republican and Democratic politicians wield when it's their turn at the throttle.
Plurality voting creates unnatural polarization due to its incentive to “stake an ideological claim” that doesn't yet overlap with someone else so that you're not at risk of splitting votes. This phenomenon is more visible in the primaries, where candidates who are (for the most part) ideologically similar will make seemingly outlandish policy claims in order to win votes.
We should replace plurality voting for the long-term durability of our country.
Primaries are not strictly necessary
Primaries are necessary to select a single candidate
representing each party. Strictly speaking, that need disappears with approval or score voting. In approval voting or score voting, each of a party's candidates will not act as spoilers to their peers.
It seems reasonable to allow any candidates that meet the basic qualifications (no trivial hurdle) in the general election.
Yes, we'll need computers
I suppose some precincts still manually count votes. Really.
But here's the thing. It's 2016. It's time to use computers. Manual counting should only be done if we have reason to not trust our computers.
As a registered Libertarian, I've long been suspicious of plurality voting. Decades ago, I built a web site that used approval voting to help groups make decisions that were agreeable to the widest majority of people. At the time, I didn't realize that asking voters to select all options they like was named approval voting.
More recently, I have been in favor of score voting since becoming aware of it thanks to the people at The Center for Election Science
My present rants about plurality voting are inspired by what is happening to the Republican Party. Even as an outsider, the disaster within the GOP worries me. The GOP is ideologically closer to and more accepting of libertarianism than the Democratic Party, so I feel somewhat attached to the GOP even as a non-member. And perhaps more importantly, one of the candidates selected by the two major parties is certain to be the next US President. A Trump versus Hillary general election is alarmingly awful.
It's impossible to know with certainty whether the situation would be different with approval or score voting. However, I think an honest observer will acknowledge that for the past two months the GOP primary situation has been Trump versus Anybody-But-Trump (ABT). This is a classic situation where approval or score voting can clarify the actual desire of voters. A majority wanted ABT, but plurality voting prohibited them from actually voting ABT. Instead, the majority had to split their votes among the candidates that constituted ABT. Trump, and indeed most analysts, incorrectly interpreted this as widespread support of him over ABT. But obviously he is just the beneficiary of a split vote thanks to plurality voting.
Data are hard to come by. Because our elections use plurality voting, few polling organizations ask voters to approve or disapprove of each candidate. Even those that do ask about approval probably do not coach the voter that for the purposes of the poll, it's acceptable to approve of more than one candidate.
Voters in the US are conditioned to expect to only approve of a single candidate because that is how we have to vote. So even if a pollster did ask for a voter to approve or disapprove of each candidate, voters are in the thick of a emotionally charged plurality vote primary; their opinions of the other candidates have already been poisoned.
Yet even with all of these caveats, where available, it was generally reported that Trump had low approval rating.
If the reader doubts the objective superiority (in terms of increased voter satisfaction) of the outcomes from approval or score versus plurality, I suggest further reading of the more academic (and less ranty) material at CES
Can I be sure the situation would be better with approval or score voting? Of course I cannot be sure. But odds say it would likely be better. And if odds don't matter, then we might as well decide elections via dice rolls.