The 2020 election: How competitive elections undermine democracy

 I teased this post with one of my "quick takes" last week, and I think I may actually have something to say here.  As in, I actually published on this stuff.  For real.

So.  This thing is over.  Finally.  The country faced a test, and we passed that test.  We failed that test in 2016, but we got ourselves a do-over, and after careful reconsideration, having to live with the consequences of our collective error, we passed the test in 2020.  But it was a close thing.  And that's rather my point.  It shouldn't have been close.

What are the characteristics of a well-run, small-d democratic election?  To a goo-goo*, first and foremost, a democratic election should be "competitive."  OK, what does that mean?  That's actually messier than you think.  I spent a whole chapter in my first book, Hiring & Firing Public Officials: Rethinking the Purpose of Elections (Oxford University Press 2011), going through all of the many and often conflicting definitions of a "competitive" election.  Most commonly, though, it means an election should have one of two related characteristics:  it should be marginal, meaning that the two candidates should be separated by relatively few votes in the final tally, and it should be uncertain, which... well, does that mean it should actually be a random process, or merely that we should be ignorant, even if the process is deterministic?  OK, we don't need to get into that.  Let's go with random.

So these two characteristics...  Goo-goos will tell you how much we need elections to have them.  If we don't have marginal, and uncertain elections, then everything sucks, and we might as well live in a dictatorship, boo-hoo.

I built my career, from a series of papers, through that first book, by calling bullshit on this.  Let's walk through the why's, with some application to what just happened.

In 2007, I published a paper in Public Choice called "The Social Sub-optimality of Competitive Elections."**  There's a bunch 'o math in it, but the essential argument was rather simple.  By various criteria, there can be an electoral outcome that is mathematically better.  If those criteria are met, then the ideal probability to ascribe to the worse candidate's victory is zero.  If we cannot objectively say that one candidate's victory would be better, then ascribing probabilities of .5 to each candidate's victory-- a competitive election by the uncertainty definition-- would be harmless, but also useless.  A competitive election never accomplishes anything.  A simple observation.

Let's apply.  As a general rule, I preach Kenneth Arrow's impossibility theorem.  That is, there is no way to combine the preferences of everyone in the electorate into one collective preference ranking that meets all of Kenny's quite sensible conditions for democracy.  Whatever you're doin', you're doin' it wrong.  However, there is more to an election than the aggregation of preferences.  To treat an election as such is to treat all opinions as valid, and to treat nothing as a matter of objectivity.

So here, we bring in the distinction between positional issues and valence issues, with my standard reference to Donald Stokes.  A positional issue is an issue about which we disagree on the goal.  Abortion.  Is the goal the protection of the fetus/embryo's life, or is the goal the woman's right?  Different goals, different policies.  A valence issue is an issue about which we agree on the goal, but disagree on how to achieve it, or who can achieve it.  Like, a strong economy.  The concept of the valence issue has morphed over time into valence characteristics, which are traits that we want leaders to have.  Positional issues are the realm of preference aggregation, subject to Arrow and the impossibility theorem.

Valence.  Valence brings us into the realm of objectivity.  When valence overshadows everything, positional distinctions must cease to matter.

When we, game theorists, incorporate the concept of a valence dimension into our models, we conceptualize the valence dimension as follows.  A candidate's score on that valence dimension represents traits such as, "competence," and, "honesty."  Straight-up, those are the two most commonly named traits.  I wrote a little paper, some years ago, on the many conceptualizations of "valence," but never really did anything with it, 'cuz you can't publish a paper about game theory that isn't a game theory model with an equilibrium.***  Nevertheless, we are mostly talking about "competence," and, "honesty."

So... Donald J. Trump.  I'm not going to belabor the point, but he is objectively the most unfit major candidate for public office in the history of the United States of America.  Without hyperbole.  Can you find some fringe candidate out there, somewhere, who was even worse?  Sure.  Can you find a major candidate?  Anywhere, anytime in American history, who was more unfit for public office?  No.  And before you give me any speeches from segregationists during their era, riddle-me-this:  what would Donald Trump have said or done, at that time, in that place?

Valence.  Donald J. Trump.  He is, beyond dispute, a sociopath.  Not in colloquial terms, but in clinical terms.  He lies more than any human in the history of the species.  His lies are so flagrant, so constant, and so egregious, that the only points of comparison are the propaganda machines from dictators past and present.  "Honesty?"  Yeah.  And as for competence, Donald J. Trump is the most inept, grotesquely ignorant person in the history of presidential politics.  He is a fool.

He is a fraud, a con artist, and despite claiming to understand business, he is a... mercantilist.

And all of this has been obvious, all along.  It was obvious in 2016, despite which he won.  And it is more obvious now, when we can observe valence outcomes, resulting from his dishonesty and his incompetence.

Donald Trump is, arguably, the worst president in American history.  There is a bottom tier, belonging to James Buchanan, Herbert Hoover, Warren Harding, and we can debate who else goes in there, but that's basically your bottom three.  Or at least, it was, until Donnie came along.  We can observe, quite directly, not just the economic costs, but the cost in human lives of his lies and incompetence.  We can look around the world, and see what happens when you have honest, good-faith, semi-competent leadership, and what happens when your president is the lying-est liar who ever lied a lie, who refuses to do anything else to manage a pandemic but lie, while also making the Harding administration look like our idealized image of Elliot Ness.

And to top it off, he tells us in advance that he wants to be a dictator, and that he's going to try to steal the election.

Objectively, Donald Trump is terrible.

Notice that I didn't say anything at all about abortion.  Tax policy.  Any left-right issue.  Why?  Because when you mismanage every valence-based issue that badly, no positional issue matters anymore.  That's the point.  Donald Trump is not defensible.  Period.  He is objectively not just bad, but the worst.  He is objectively an idiotic, narcissistic, lying sociopath, who has let a pandemic run amok, thinking stupidly that he could lie his way out of it and go back to giving his Nuremberg rallies where throngs of cultists chant his name as he governs by owning-the-libs instead of actually doing the fucking job.

Objectively, Donald Trump was the worst president in the history of this country.  At least James Buchanan was trying.  He sucked, but he was trying.

Among academics, you will find very close to zero Donald Trump supporters.  Even most Republicans recoiled from Trump.  You can find a small group, such as those who are motivated primarily by their opposition to political correctness/wokeness.  True, Donald Trump is politically incorrect, and by this terminology, Joe ain't the sleepy one.  Yet, the reason Trump is the enemy of wokeness is not a principled opposition to the excesses of call-out culture.  Rather, he actually is an unreconstructed racist and misogynist.  And to support him on these grounds is basically to say that the most important thing is to own-the-libs.  That, amid COVID, seems to me to miss the point.

I am reminded of a professor I had in grad school.  Ray Wolfinger.  Very smart guy.  He would often make a very sharp critique of an argument in a student's paper.  That critique was almost invariably right.  Yet, Ray would rarely have a sense of magnitude.  Instead, he would latch onto that observation like a dog with a bone.  He would run the risk of missing the bigger picture, and in a sense, being not exactly wrong, but losing the plot because he would focus on what turned out to be a minor and unimportant detail, even if his observation of that detail was generally right.  I have encountered arguments from people who support Trump out of an opposition to political correctness, and as an opponent of political correctness myself, I see the basic motivation, but that seems to me to miss the point.  The job of the president is not to own-the-libs, as much as I recoil from the modern left.  And when the cost in lives has grown so high, voting for an actual racist who has let a pandemic run amok just because he happens to own-the-libs through his actual racism... um... no.  I'm going to say that these people are objectively wrong.  As much as I find call-out culture and the excesses of wokeness noxious... no.  This is Ray Wolfinger, latching onto one small point, and losing the plot.  At best.

Objectively, the better outcome for the country was a Biden victory.  This is not the kind of statement I would make in most elections.  In 2012, for example, one's preference between Barack Obama and Mitt Romney would have been determined essentially by one's preference on positional issues.  Given that, I could not say that one outcome was objectively better for the country.  Neither candidate was objectively stupid, crazy, a possible threat to national security, nor anything like that.  Tell me you voted for one or the other, and that means nothing to me.  Yeah, Paul Ryan's taste sucked-- I'm not a fan of either Ayn Rand or Rage Against the Machine, and I'll spare you my commentary on that combination for today-- but who cares?  In a time of relative stability with neurotypical candidates adopting mainstream positions, I am not willing to say that one outcome is objectively right.  We're in Kenneth Arrow territory there.

Donald Trump was different.  He made Hillary Clinton, and then Joe Biden objectively the correct choices.  So let's apply my basic principles.  What probability should be ascribed to the suboptimal candidate's victory?

Zero.  Any other probability is socially suboptimal.  Donald Trump's probability of victory should not have been .5, merely because goo-goos like competitive elections.  Instead, those of us who wanted the autocratic dipshit out of office were sweatin' bullets because the autocratic dipshit's probability of victory needed to be zero.  Why?  Autocratic dipshit. Which part of that was unclear?  Why is that .5 thing important?  It isn't.  Goo-goos just like the idea of .5.  'Cuz.  It never creates any uniquely good outcomes.  And when you've got a Donnie Trump problem, you need his probability of victory to be precisely zero.  It should have been zero in 2016.  What went wrong in 2016?  What has been the source of our problems for the last four years?  The fact that it wasn't zero.

Instead, it was a close election.  Marginal.  That creates two problems.  First, there's the recount-lawsuit mess.  Here, we are somewhat saved by the grotesque incompetence of Donald Trump's "legal" team, and I think that one does deserve the sarcasm quotes.  Yet it did not have to be so.  As I have suggested in several posts, had they prepared in advance, gotten their act together, and not made fools of themselves in the courts with silly pseudo-arguments, we could have been in real trouble.  So here's the problem with a marginal election, and this brings me to an argument I made in an article called "The Statistical Properties of Competitive Districts," from PS: Political Science & Politics (also 2007).  Elections are basically subject to central limit theorem problems.  The closer the election is in terms of raw votes, the more susceptible they are to a statistical error in outcome.

An error can be induced.  Trump's plan was to induce an error through bad faith lawsuits.  Also known as "stealing an election."  He expected to lose, planned for it rhetorically, if not legally, and decided that his way to stay in power was to count on the courts to overturn his loss.  His lawyers didn't have a specific plan, but with a close election in marginal terms, that requires throwing out fewer ballots than a blowout election.  If you read through the opinions of the judges who keep eviscerating Trump's lawyers, what they are doing is as follows.  They are looking at the final tallies, looking at the specific complaints that Trump's lawyers are raising, and making a judgment about the feasibility of flipping the outcome.  This is about simple arithmetic.  In a close election, throwing out a few ballots can flip the outcome.  The wider the margin, the more ballots that one would need to toss to make that happen.

This was why Democrats kept saying that Biden needed a big victory to prevent Trump from stealing the election.  It appears that his victory was big enough to stop Trump from stealing the election, but that actually makes my mathematical point for me.  I was skeptical, prior to the election, that a big Biden victory would stop Trump from stealing the election because I didn't foresee how inept Trump's legal team would be, but given that ineptitude, they haven't brought any cases to flip anywhere near enough votes.  And that's where the math comes in.  If an election comes down to a handful of votes, tossing a few ballots-- maliciously or otherwise-- flips the outcome.

Do you seriously want that?  Marginality creates the lawsuit problem.  We were only saved by Trump's ineptitude.  Next time... we may not be so lucky.

But Trump's voters, recall, are mostly a bunch of idiots and dupes.  They believe every lie Trump tells them.  Why?  Idiots and dupes.  Trump has told them that the election was stolen.  From... him.  Did he present any evidence?  Of course not.  That's why he was laughed out of court.  Courts.  Plural.  Many plural.  He's just the lying-est liar who ever lied a lie.  He's doing his thing.  That thing being... lying.

However, his followers are cultists.  They'll believe anything he says.  So, most of them think the election was stolen.  From Trump.  And, they're pissed about it.

Of course, the losers are not generally happy about how an election turns out.  This, too, is something about which I have written in many contexts, mostly legislative, because it relates to district lines, but right now, we're talking about a presidential election.  So, what're they gonna do?  Secede?

I've written enough about the importance of the losers' consent in a democracy, but as a general rule, it is worth noting that if you care about maximizing contentment with outcomes, you maximize dissatisfaction with a close election.

And when those dissatisfied voters are told-- by the lying-est liar who ever lied a lie-- that the election was stolen, we're in some real trouble.

Yeah, we got it right.  This time.  But it shouldn't have been close, and the fact that it was both points to a problem, and creates its own set of problems.

I don't like competitive elections.

*Goo-goo:  An infantilizing name for "good-government" advocates.

**Got an award for it.  Started an argument in the journal.  Some voids shout back.

***Reviewers actually wrote things like:  it was really interesting, and would be cited a lot, but it shouldn't be published because where's-the-equilibrium?  Game theorists...  Useless people.