Kennedy and Condorcet

Robert Kennedy, Jr. would not be the "Condorcet winner" in a well-designed presidential election system.

Robert Kennedy, Jr., the independent presidential candidate, claims he would beat Biden or Trump if he were to run one-on-one against either of them. He bases this claim on recent polling. Echoing Kennedy’s claim and using technical terminology, some political scientists have recently asserted that Kennedy would be the “Condorcet winner” in a three-way election with Biden and Trump, and they use this assertion as a reason to reject the concept of a “Condorcet winner” as a measure of representational fairness in an electoral democracy.

To understand this assertion and its potential significance, we need to explain the idea of “Condorcet winner” and then apply the concept to this year’s presidential election. For some readers of Common Ground Democracy, “Condorcet winner” will be thoroughly familiar term. Other readers will welcome a bit of a refresher, and still more readers (hopefully) will be coming to this concept for the first time.

A “Condorcet winner” is a candidate who in an election involving more than two candidates, from which only one will prevail, is preferred by a majority of voters over each other candidate when compared one-on-one against each of these opponents. A simple sports analogy will help grasp the concept: think of a round-robin tournament, in which competitor faces each other competitor in a head-to-head match. If there is one competitor in the round-robin tournament who wins every match against each other competitor, that undefeated winner would be the “Condorcet winner” of the round-round tournament. Thus, in an election, a “Condorcet winner” would beat every other candidate if the voters voted separately between each pair of candidates as if the election were a round-robin tournament.

The concept of a “Condorcet winner” is named after an Enlightenment-era French philosopher, the Marquis de Condorcet, who studied the mathematical properties of elections and who argued that candidate who would prevail against each other candidate in the way now defined as a “Condorcet winner” is the candidate most deserving to win the election. Condorcet considered a “Condorcet winner” most deserving because, almost by definition, no other candidate could claim to being more preferred by a majority of voters. If any other candidate attempted to make that claim, the “Condorcet winner” could challenge that claim and, based on the same preferences of all the voters in the election (as revealed on ballots that would permit voters to express the full set of preferences among all candidates), the “Condorcet winner” would be preferred by more voters than this other claimant. Thus, if elections should be governed by the principle of majority rule, as Condorcet believed they should be, the “Condorcet winner” should be the candidate elected to the office at stake—because, again, more voters prefer that candidate over each other candidate in the race.

Although political scientists use the term “Condorcet winner” to describe a candidate whom a majority of voters prefer to each opponent, it turns out that Condorcet was not the first philosopher to define the concept and advocate for it as the standard by which any electoral method that yields a single winner should be evaluated. Rather, a thirteenth-century Majorcan monk named Ramon Llull came up with the same idea 500 years before Condorcet, but his writings on electoral methods were lost to history until rediscovered in the twentieth century. By then, Condorcet had gotten credit for the concept, as his writings had become very influential in the further development of political science.

Whatever we call the concept—and I will continue to use the term “Condorcet winner” at least in this particular essay, now that we’ve defined it and discussed its origins—it is true that Robert Kennedy is claiming that he would be the Condorcet winner in a three-way race with Biden and Trump. But we still need to consider the strength of this claim and, even more importantly, its relevance to deciding whether or not electoral systems should be designed so that they successfully elect Condorcet winners.

I’m not convinced that Kennedy would be the Condorcet winner based on just one poll this far out from the November election. As voters learn more about Kennedy besides his famous family name, it seems likely that his level of support will tend to diminish. If so, it’s likely that either Biden or Trump would beat Kennedy one-on-one.

But even if we assume that Kennedy would be the Condorcet winner in a three-way race with Biden and Trump, I would view that as more of an indictment against the two major-party nominees and the primary process that produced them, rather than a reason to think that the concept of a Condorcet winner is an inappropriate measure of an electoral system’s representational fairness.

Most importantly, however, it is fundamentally mistaken to think that in a well-designed “Top-3” presidential election system, Kennedy along with Biden and Trump would have emerged this year as the top three finalists for a Condorcet-based November election. Let’s imagine that, after the 2020 election but well before this presidential election cycle began, the electoral system had changed so that the presidential election would have been a two-stage process, in which the first stage would be a nationwide “preliminary” vote in September among all candidates who gathered electronic signatures from 5% of registered voters nationwide.

In this preliminary stage, each voter would cast a single vote for whichever of the qualifying candidates on the ballot they most preferred. More than one candidate from each political party could be on this preliminary ballot as long as they gathered the requisite number of electronic signatures. The political parties could still hold their nominating conventions to nominate their most preferred candidate from their own party, and states could continue to hold presidential primaries for the purpose of choosing which delegates to send from the states to each party’s nominating conventions, but this party nomination process would not preclude additional candidates from the same party from qualifying for the September preliminary ballot.

The three candidates who receive the most votes in the September preliminary round would advance to the November general election, at which voters would receive ballots that permit them to express their preference between each pair of these three candidates: A versus B, A versus C, and B versus C. If one candidate beat each other opponent in these head-to-head pairwise matchups, that candidate would be the Condorcet winner and, according to the rules of this system, would be elected president.

We cannot know for sure, of course, how the presidential election would have unfolded if this system were the one in place in advance of this year. (This system is the same as the one I described in a previous Common Ground Democracy column.) But we can make some educated guesses, and those educated guesses do not include Robert Kennedy being one of the three candidates advancing to the November general election ballot. He might well be one of those collecting enough electronic signatures to qualify for the September preliminary round. But that hardly means he would be one of the top-three vote-getters in that round.

Trump surely would be one of the three advancing to the November ballot. The steadfastness of his core support, amounting to roughly a third of the American electorate, would assure his finishing among the top three in the September ballot. As for Biden, it seems quite likely that he would have bowed out of the race, knowing that there would be room for another Democrat to secure a spot on the November ballot without directly taking a spot away from him. And if Biden didn’t bow out, he would certainly have drawn additional Democrats into the race, like Gavin Newsom, Gretchen Witmer, and other prominent members of the Democratic “bench” waiting for their chance to run. And given Biden’s relative unpopularity even among Democrats, he almost certainly would have come in behind some other Democratic contender in the September preliminary round. The prospect of this ignominious humiliation is one reason why I believe Biden would have declined to run in the first place if this were the existing electoral system. For sake of argument, let’s assume this is correct, and Gavin Newsom would have ended up being the second candidate to qualify for the November ballot.

Who would have been the third? There would have been a vigorous competition among several serious competitors for this spot. Joe Manchin almost certainly would have entered the presidential race if this system had been in place. Nikki Haley likely would have stayed in after losing the Republican nomination to Trump, because she could have secured a spot on the September preliminary ballot without being the Republican nominee, and towards the end of her campaign she was essentially running as a third-party alternative to Trump’s MAGA version of the Republican party on the right and the Democrats on the left.

I also think Larry Hogan would have run for president rather than seeking Maryland’s Senate seat if this kind of Condorcet-based system had been in place for the presidential election. Remember, a Condorcet-based system tends to favor candidates who are closer to the center of the overall electorate, because to be a Condorcet winner you must be able to secure more than 50% of the votes against every other candidate in the race, whether that opponent is on the right or the left. Thus, Hogan would have understood that he had a good shot at being the Condorcet winner in a three-way race against Trump and Newsom, being a popular centrist governor in a solid blue state. For the same reason Chris Sununu, who also toyed with a presidential bid this year but ultimately declined, might have been more tempted to enter the race if the electoral system were Condorcet-based. Similarly with Liz Cheney.

Even assuming that all of these additional candidates—Manchin, Haley, Hogan, Sununu, and Cheney—gathered enough electronic signatures to qualify for the September preliminary ballot, my guess is that most of them would have dropped out and endorsed one of these other candidates before the September voting began. The reason is that all of them would have known that they would be splitting the support of centrist voters, thus diluting the chance that any of them would finish third. Indeed, the very presence of Robert Kennedy, Jr., on the same September ballot, combined with their collective desire that he in particular not be the one to receive that a third spot as a result of them splitting their centrist support, would have caused them to consolidate around a single one of them—probably whoever was polling the highest heading into September (or whoever collected the most electronic signatures as a signal of their strength). It’s purely speculative as to which of these would-be alternatives to Trump and Newsom would have been most successful. My guess is that it would have ended up Haley or Hogan. Let’s say Hogan, only because he probably would have appealed to more centrist independents and moderate Democrats who, thinking Newsom might be too liberal for their tastes, also viewed Haley as a bit too conservative for their liking. In other words, I’m speculating that Hogan would have been most successful in straddling the middle of the electorate with Trump to his right and Newsom to his left.

The main point, though, is that a Top-3 Condorcet-based electoral system would have produced something like Trump, Newsom, and Hogan as the three November finalists, and it would not have included Kennedy.  Thus, in evaluating whether a Top-3 Condorcet-based system would be desirable for presidential elections, one should do so with something like a Trump-Newsom-Hogan race in mind, rather than the Trump-Biden-Kennedy contest produced by our existing electoral system.

So, what would we think if this year’s presidential race had been a Trump-Newsom-Hogan Condorcet-based election? Trump certainly would not be the winner, as he is unable to reach 50% approval among the nation’s voters. Hogan certainly would have beaten Trump head-to-head, with all of Newsom’s supporters voting for Hogan over Trump. But what about the head-to-head between Newsom and Hogan? That’s harder to say. If Newsom could have convinced over 50% of the country to support him and the Democratic ticket, then he would have been the Condorcet winner. Otherwise, Hogan would have beaten Newsom with a coalition of both MAGA and traditional Republicans, plus independents, supporting him against a Democratic minority—and then Hogan would have been the Condorcet winner.

Let’s say it would have been Hogan, not Newsom. Would that be a reason to condemn the concept of a Condorcet winner? I don’t think so. To be sure, Newsom would have had more enthusiastic supporters, wanting him more than either Trump or Hogan, than Hogan would have had (wanting him more than either Trump or Newsom). But that doesn’t mean that Newsom, rather than Hogan, should have been elected president. If Newsom couldn’t reach 50% support against Hogan among all the voters in the election, why should Newsom rather than Hogan be elected? Conversely, if more than 50% of the whole electorate would prefer Hogan to Newsom, isn’t that a good reason why Hogan is more deserving that Newsom to be elected?

Ultimately, the decision whether to use the criterion of a Condorcet winner as the basis for judging the fairness of a single-winner election is a philosophical one. It is true that a Condorcet winner can be something of a lukewarm compromise candidate, about whom relatively few voters are passionately enthusiastic.  But this kind of Condorcet winner can win a Condorcet-based election only because the other candidates, about whom voters are passionately enthusiastic, are also extremely divisive—despised by the opposing side, just as they are adored by their own supporters. A Condorcet winner can be most voters’ second choice and few voters’ first choice, but is it not better for the whole society to settle upon a widely admired compromise candidate rather put into office someone who can’t secure the support of a majority of voters against an opponent, and whom a substantial number of voters strongly oppose?

In future Common Ground Democracy essays, I plan to probe this philosophical issue more deeply. In the meantime, though, it’s enough to refute the spurious assertion that the concept of a Condorcet winner should be judged by considering the possibility that Robert Kennedy, Jr. might be a Condorcet winner against both Biden and Trump.