I got a lot of pushback on the last thing I wrote about the “Bernie or Bust” movement, so let me try again to explain why, rationally, it makes no sense to vote third party within the current system. If you don’t want to read the whole thing, skip to the end, where I link to some ideas of how to make a true multi-party democracy flourish, and end the two-party duopoly.

The reason I’m on this again is because of the following Facebook post by Robert Reich, who has been a gigantic Bernie supporter:

The blurb doesn’t do a good job of highlighting the actual point of the post, which was, basically, “Fight like hell for Bernie as long as he has a chance, then fight like hell for Hillary if and when he doesn’t.” This is exactly what I was trying to say in my last post. Reich’s post attracted the usual comment ranters, who accused him of “vote shaming,” or said that Hillary “represents everything we’re fighting against.” And, quite frankly, that’s just bullshit. It’s spoiled, childish thinking, and I’m going to attempt to use a little game theory to explain why.

Voting Third-Party is Bad Strategy Under the Current System

A lot of the reason that voting for a non-Democratic, non-Republican candidate is useless, or worse, actively harmful to your interests as a voter, has to do with the structure of our electoral systems in general, and the way we elect the President in particular.

In the U.S., we elect candidates for office in what’s known as first-past-the-post (FPTP) voting. That means that the candidate with plurality support wins, period. In a two-way race, that’s fine, because a plurality will always be a majority in a two-way matchup. This is, in fact, why FPTP has hung on so long here; there is a vested interest by the two major parties to preserve it, because it marginalizes third-party and independent candidates quite effectively. If you want to see how FPTP breaks down in a multi-way race, just take a look at what happened in the Republican primaries. Despite not winning a majority in any single state until April, Donald Trump is now (effectively) the Republican nominee, despite the fact that approximately 60% of Republicans voted for someone else. At an abstract level, a candidate can, in theory, win a plurality in any n-way race with only 1 more than 1/n of the votes cast.

On the Presidential level, all of this is compounded by the unfairness of the Electoral College, and specifically by the way all states (except Maine and, for now, at least, Nebraska) allocate their electors on a winner-take-all basis. This means that a single “spoiler” vote in a close election could, in theory, tip a lot of electors from one side to the other. Remember that in Florida in 2000, George W. Bush won by only 537 votes out of 5.96 million votes cast. That vote margin, a mere 0.009%, gave Bush all 25 of Florida’s electors, thus winning him the presidency. A lot people blame the 97,488 people who voted for Ralph Nader on the Green Party ticket (I’ll admit to some bitterness myself), but if you look at the numbers, any of the third-party voting blocks could have covered that margin. The lowest ballot-line candidates, on the Socialist Workers’ Party ticket, got 562 votes.

Game Theory Applied to Presidential Elections

Let’s try a little exercise in game theory. We’ll posit an FPTP presidential election in which there are three candidates. You can assign each of your candidates a score from 0.0-1.0 representing your estimate of their quality. (There are a lot of dimensions to “quality” – affinity for their policy positions, belief in their ability to enact said policies, their likely cabinet and court appointments, their grasp of foreign policy, general trust in them to wield executive power responsibly, and so on. Those are not important for the game theory part of this.) They also have a score from 0.0-1.0 representing their chance to win. This latter score doesn’t have to add to 1.0 across all three candidates to make the example work; there is uncertainty in every estimate.

So, let’s say that you give the candidates the following scores:

Obviously, you don’t want Candidate C to win, as you judge Candidate A to be 5.1 times better than them. But there is a genuine choice to be made between Candidates A and B.

If you multiply the two numbers for each candidate, you get the following, representing their “composite score”:

This clearly shows that the rational thing to do is to vote for Candidate B, because the current rules of the game could hand the election to Candidate C if you don’t vote tactically. In fact, you would have to judge Candidate A to be 0.64 better than Candidate B before the math would balance out to be worth the risk.

This is a very clinical examination, to be sure, and if Candidate B really only had a quality score of 0.28 (which is what they’d have to have to rationally justify voting for Candidate A in this scenario), then maybe you would be justified in saying there’s not enough difference between them to abandon your strong preference.

“So what,” you’re thinking. But here’s the wrinkle. Let’s say that, if 2/3 of Candidate A’s supporters decide instead to vote for Candidate B, on the theory that he or she is “good enough,” then Candidate B’s probability of winning shifts from 0.39 all the way to 0.43, and Candidate C’s drops to 0.39. If that’s the case, then it’s even crazier to insist on voting for Candidate A, who only represents a 0.17 improvement over Candidate B, if the risk is that it will give the election to Candidate C, for whom you estimate a 0.71 difference in quality (almost an entire Candidate B), right?

Let’s put some real names and faces to this, shall we? Let’s say you took the I Side With quiz, and got a 92% match with Bernie, a 73% match with Hillary, an 88% match with Jill Stein of the Green Party, and a 32% match with Gary Johnson of the Libertarian Party. You have a 21% match with Donald Trump (don’t ask me how you get even that high a match if you’re a 92% Bernie supporter, but bear with me).

Let’s set up our model like this: we’ll assume that Hillary wins the Democratic nomination (a >98% probability at this point, unless the superdelegates override her expected majority among pledged delegates). Let’s further assume we live in a state that tracks the national average, allows write-in candidates, and has the Green and Libertarian parties on the ballot. We’ll use the latest Real Clear Politics head-to-head poll averages (as of 5/9/2016) for Clinton’s and Trump’s chances of winning, and we’ll make educated guesses for the other candidates. Given those conditions, we’ll assign the following scores:

I assign Bernie such a low percentage because, in this scenario, he is a completely independent, write-in only candidate, with no party infrastructure to support his general election campaign. (Incidentally, you can’t just write in someone’s name for President in most states. In 35 states, you can only be a valid write-in candidate if you have filed paperwork (and in some cases paid a filing fee). In North Carolina you have to circulate a petition and get enough signatures to be a valid write-in. In 6 states, write-ins aren’t allowed at all.) Jill Stein has a higher chance than Bernie, because the Green Party is an established party, and her name appears on the ballot in our example. Gary Johnson is slightly better positioned, as in addition to being on the ballot in our hypothetical state, he has a decent chance of picking up defections from Republicans who are disgusted by Trump.

Point is, the combined score for Hillary still dominates the field for this hypothetical voter. She’s also worth 73% of a Bernie, and 83% of a Stein, so is that really so bad? In addition, every Sanders write-in or Stein voter who decides to vote for Hillary instead increases her chances against Trump; conversely, every voter who casts a ballot for one of the spoilers on the left tends to strengthen Trump’s chances of winning a plurality.

When you extend this example to a 50-state scenario, where the relative strength/weakness of the major-party candidates fluctuate, and the Green and Libertarian parties might not even be on the ballot (as of this writing, the Greens are only on the ballot in 20 states; the Libertarians, 33), it becomes even more quixotic to vote independent or third party.

What’s Your Point?

The point of all this is that, if a major-party candidate is “good enough” to meet the majority of your desires as a voter, and the other major party candidate is deeply opposed to most of your preferred policy positions, you are taking a big gamble by “voting your conscience,” “sending a message,” “refusing to choose a lesser evil,” “punishing the party for a rigged process,” or however you want to formulate it, under the current system. Now, it is absolutely your right as a voter to do any of those things. But let’s be honest – did Gore’s loss in 2000 because of Florida send shockwaves through the party, causing a massive realignment around liberal/progressive values? Not directly. The aftermath of seeing GWB start an unnecessary war; appoint far-right federal judges; allow the EPA, the SEC, FEMA, and other government agencies to atrophy; and sign legislation and executive orders that rolled back years of progress, is what caused that realignment, by awakening an entire generation of voters to the dangers of a Republican president. But the cost of that was profound: thousands of American soldiers (and many more thousands of foreign civilians) dead in Iraq and Afghanistan; millions of homeowners and retirees wiped out by a preventable economic crash; hundreds dead in New Orleans, and an entire city that is still rebuilding 10 years later; civil rights protections for millions of disadvantaged voters being inexorably rolled back; a campaign system that is flooded with Super PAC donations, drowning out the voice of the regular voter. All of this was made possible by voters who thought there was no difference between the candidates, or who thought there were enough other Democratic voters to cover them as they cast their protest vote.

But We Can Change the Game for 2020 and Beyond

But the other part of Reich’s post is just as important. You have to start working immediately after this election if you want to change the game in 2020. Where he is mistaken is in leaving it at building a third party, as you see from the difficulty the Greens and the Libertarians, two long-established third parties, have in getting on the ballot. Another independent party is not enough. In fact, a left-wing third party that can’t reach a plurality of voters, but can clear the 5-10% threshold to be a real national party, just magnifies the problem. If we really want to pluralize American politics, we have to fundamentally change our voting processes. Here are some things that would make a big difference, that we could push through with ballot initiatives (in the states, like Colorado, that allow them), if our states’ legislatures can’t be lobbied to put them in place:

• Multi-party primaries: this is how things work in Louisiana for everything but the Presidential race. There are no party primaries at all. There is a first round of voting open to every candidate who qualifies for the ballot. If no candidate receives an outright majority, the top two candidates advance to a run-off election. There is still some opportunity for spoilerism here (for example, if there are three left-wing candidates who split 54% of the vote more or less evenly at ~18% each, but two right-wing candidates who split the other 46% 23-23, both right-wing candidates will advance to the runoff). This is probably more practical for Congressional elections than Presidential; can you imagine an 18-way open primary, which is what we would have had if every declared candidate from the two major parties, let alone third parties, were to appear on the ballot? With that said, we need to re-think how we elect Congress, too.
• Majority to win: dismantle the FPTP system, and require that a winning candidate for office must receive an outright majority of the ballots cast. This will force the two major parties to take the concerns of voters at the fringes seriously, and put together coalitions and platforms that represent the true majority will. It would also allow voters to freely vote for the candidate of their choice, knowing that they aren’t denying their second choice a plurality. However, this will cause a lot of run-off elections if we retain the single-candidate ballot, which is why we need:
• Ranked voting systems: there are a lot of options here (instant runoff / alternative vote is probably the easiest to implement on current voting machinery, and the easiest for voters to understand). The important takeaway is that a ranked voting system of any kind will allow a consensus majority winner to emerge in any election without a second round of voting.
• National popular vote: The NPVIC would establish a system where, once enough states have adopted it that there are at least 270 electoral votes among its member states, all the members of the compact will award their presidential electors to the winner of the national popular vote. This will, in effect, end the Electoral College, and make it so that no Florida 2000 situation can ever happen again. It also happens to have the advantage of being completely constitutional, not requiring a constitutional amendment, and not requiring any of the compact’s member states to unilaterally disarm.

If we work toward these reforms, we can actually create a real system where multiple parties can flourish and begin to attract voters, while still helping to ensure that a splintered coalition doesn’t allow a minority party to seize control of the government against the will of the voters. I will sign any petition put in front of me to adopt any of these reforms, even though I am a proud Democrat and don’t identify with any third party. We need to give voters more choices, and we need to end the two-party duopoly. I fully agree with the “Bernie or Bust” crowd on that point.