A “Top Three” Version of California’s “Top Two” Elections
To combat polarization and increase the capacity of Congress to enact legislation addressing the nation’s needs, here’s a proposal worth considering.
Eric Maskin and I have written a Project Syndicate column describing a “top three” electoral system, which is a variation of California’s “top two” system. This system would implement the values I have been advocating here at Common Ground Democracy, producing winners who achieve the greatest degree of consensus among voters. It is, therefore, an antidote to the partisan polarization that is afflicting contemporary American politics.
As in California’s top-2 system, the first part of the top-3 electoral process would be a nonpartisan primary in which all voters would choose among all candidates who qualify for the ballot (by collecting enough signatures), without regard to the party affiliation of either voters or candidates. To keep the system simple, voters in this nonpartisan primary would select the one candidate on the ballot whom they most want to advance to the November general election. This part of the process would be exactly like California’s.
What would be different is that the three candidates who receive the most votes in the nonpartisan primary would advance to the general election, not just the top two as in California. Then in November, the top-3 system would present to the voters a ballot letting voters express their preferences between each pair of these three candidates: A versus B, A versus C, and B versus C. (In California, by contrast, the November ballot presents voters with only the single choice between A and B.)
In this top-3 system a candidate whom a majority of voters prefer to either of the two opponents according to these pairwise comparisons wins the election. For example, a majority might prefer C over A and also prefer C over B, in which case C wins the election (and this is true regardless of whether a majority prefer A to B or vice versa).
Theoretically, it could happen that none of the three candidates is preferred by a majority of voters over each of the other two, although in reality this situation is likely to be rare. Suppose, for example, that A prevails over B, 53% to 47%; B prevails over C, 51% to 49%; and C prevails over A, 55% to 45%. In this example, each candidate is preferred by a majority of voters over one other candidate but not the other. (A beats B but loses to C, B beats C but loses to A, and C beats A but loses to B.)
There needs to be a way to break this 1-1 tie among all three candidates. The tiebreaker that best serves the values of Common Ground Democracy would be to elect the candidate whose single pairwise defeat is by the smallest margin. Here, that would be C, whose single head-to-head loss to B is only a 2-point margin. By contrast, B’s loss to A is 6 points, while A’s loss to C is 10 points.
As these numbers show, of these candidates C is the one who has the most support when falling short of a majority against one opponent, and thus C is the candidate who comes closest of all three to having majority support against each of the other two. In this way, of all three candidates, C is the one capable of achieving the widest possible consensus among the voters in the electorate and, on this basis, is the fairest way to treat all voters as equals—and so most deserves to win.
As already indicated, the need to invoke this tiebreaker is very likely to be infrequent. In most elections, according to both political theory and available evidence, one candidate will be preferred by a majority of voters compared to each opponent. This candidate will be the one whose views most closely align with those of the electorate’s median voter. This candidate is also necessarily the one in the race achieves the greatest common ground, because this candidate is the only one for whom if elected a majority of voters would not have preferred a different outcome. Or, to phrase the same point another way, for every other candidate except this one there’s a majority of voters who would have preferred a different candidate to win.
This top-3 system is well-suited to current conditions of American politics. For the last few years, we have been watching the Republican party split into two different factions: the dominant MAGA wing of the party, led by Donald Trump, and the now-eclipsed traditional wing, represented by Mitt Romney, Lisa Murkowski, and Susan Collins, among others.
There really is a three-party system struggling to emerge, but existing electoral institutions are designed only to work with two parties. Americans increasingly say they want the option of a major third party. This top-3 voting procedure would make that possible. Moreover, it would do so in a way that does not require Ranked Choice Voting, which regrettably has become demonized in some political quarters.
The top-3 system is easy to explain to voters. It is easy for voters to cast their ballots, indicating their preference between each pair of contenders. Equally important, it is easy for election administrators to tabulate the results. Unlike Ranked Choice Voting, it doesn’t require any new software or other technology. It just requires calculating the winner of three head-to-head races between two candidates, instead of calculating a single head-to-head race between two candidates, as California currently does.
It's not hard to see the difference that using this top-3 system would make. One current race that especially illustrates this is Arizona’s U.S. Senate election, which I’ve mentioned previously on Common Ground Democracy. If the top-3 system were in place in Arizona, Kristen Sinema, the independent incumbent, could run alongside Ruben Gallego, the Democrat, and Kari Lake, the likely Republican nominee. All three assuredly would advance from the nonpartisan primary to the November general election, where Sinema positioned in between the two other candidates would have a good chance of beating both. Most Republican voters would prefer her over Gallego, and most Democrats would prefer her to Lake. Combined with whatever percentage of voters like her best of all, Sinema could put together two different coalitions, so that a majority of voters favor her over each of the other two.
By contrast, Sinema is shut out of the current electoral system. Previously a Democrat, she became too much of a centrist to win her own party’s nomination, which is why she is now an independent. But if she runs as an independent in the existing plurality-winner general election, she likely will finish third behind both Gallego and Lake according to polls. That’s the effect of partisan polarization: even though Sinema would receive majority support against each of the other two separately, being the compromise choice of each side against the other, each of the two major-party nominees would run ahead of her in a fractured three-way race with none having a majority.
Nor could Sinema survive in California’s top-2 system. Gallego and Lake would be the two candidates receiving the most votes in the nonpartisan primary, for the same reason that the two of them would get the most votes in a three-way plurality-winner general election. (Only if another Republican in the nonpartisan primary siphoned enough votes away from Lake would Sinema have a chance of being one of the top two.) Instead, the top-3 variation on the California model is necessary to counteract partisan polarization and enable Sinema to show that she is majority-preferred compared to each of the two major-party nominees.
Arizona’s current U.S. Senate election is hardly the only one where the electorate’s preferences would be better served by using this top-3 system. If we look back at the 2022 midterms, we can see a lot of races where the general election was a choice between a Democrat and a Trump-endorsed MAGA Republican, while there was a non-MAGA Republican who couldn’t survive the GOP primary but whom the voters in November likely would have preferred to either major-party nominee. The Project Syndicate column that Eric and I wrote used Ohio’s 2022 U.S. Senate election as an illustration. There, Trump-endorsed J.D. Vance beat Tim Ryan, the Democrat, in November. But if Ohio had used the top-3 system for that race, Rob Portman, the incumbent non-MAGA Republican, could have run in the nonpartisan primary and finished among the top three, rather than having to run in a MAGA-dominated Republican primary (which he declined to do). Then, in the top-3 system’s general election, Portman easily would have defeated both Vance and Ryan—picking up the support of Democrats against Vance and the support of Republicans against Ryan.
A similar story can be told about North Carolina’s 2022 U.S. Senate race. A Trump-endorsed election denialist, Ted Budd, defeated non-MAGA former governor Pat McCrory in the Republican primary. Budd went on to defeat the Democrat, Cheri Beasley, in the general election. If North Carolina had used the top-3 system, McCrory would have beaten both Budd and Beasley in November. McCrory was the compromise candidate whom Democrats would have supported against Budd and Republicans would have supported against Beasley, but the existing electoral system was incapable of showing that McCrory was the majority-preferred candidate of the whole North Carolina electorate against each of the two major-party nominees. This example succinctly shows how the existing system exacerbates partisan polarization, whereas the top-3 system would counteract it.
The top-3 system doesn’t inherently favor either Republicans or Democrats. Instead, much more that the current system, it encourages both parties to field candidates close to the electorate’s median voter in order to build a winning coalition. If New Hampshire and Pennsylvania had used the top-3 system for their U.S. Senate elections in 2022, Republicans rather than Democrats likely would have won both those seats, and thus Republicans would have become the majority party in the Senate.
In New Hampshire, Maggie Hassan, the incumbent Democrat, beat an election denialist Republican, Don Bolduc, in the general election. But in the GOP primary Bolduc barely beat a much more moderate Republican, Chuck Morse, who was president of the state’s senate. Comparing the results of the Hassan-Bolduc race for the Senate seat with the state’s gubernatorial election at the same time, where moderate Republican Chris Sununu easily beat the Democrat, it is highly likely that in a top-3 system Morse would have achieve majority victories against both Hassan and Bolduc.
Similarly, in Pennsylvania, the Democrat John Fetterman beat Trump-endorsed Mehmet Oz in November. In the GOP primary, Oz’s victory over the more traditional Republican, David McCormick, was so close it went to a recount. In a top-3 system, McCormick quite likely would have prevailed over both Fetterman and Oz.
Because of polarization, we are currently watching Mitch McConnell lose his grip over Senate Republicans, as the MAGA wing of the party becomes increasingly dominant. But if Senate elections nationwide were conducted with the top-three system, McConnell’s leadership position likely would be far stronger. Considering the four states we’ve just examined, instead of there being two MAGA Republicans (Vance and Budd) and two Democrats (Hassan and Fetterman) antagonistic to McConnell, he’d have four more allies to work with: Portman, McCrory, Morse, and McCormick. Consequently, he’d have had a much better chance to muscle through the compromise border bill that he deputized Senator Lankford to negotiate.
Trump also wouldn’t be able to exert the same kind of pressure against Republican incumbents to kill the border bill because to win reelection in the top-three system they wouldn’t need to worry about surviving a Trump-dominated GOP primary. As long as they received enough votes for one of the three tickets out of the nonpartisan primary and on to the general election ballot, they could build majority coalitions against each of their other two opponents to their right or left. All in all, it would be much easier for solution-oriented Republicans who genuinely want to govern on behalf of the public to achieve legislative successes in a top-3 electoral system than in the current one.
We can make the same point about the House of Representatives. If the top-3 system were used for House elections, the current Republican majority in the House would be much more conducive to finding bipartisan legislative compromises that could survive a Senate filibuster and be signed by President Biden. Speaker Mike Johnson would not have declared the bipartisan compromise border bill “dead on arrival” in the House. Indeed, it’s almost certain that “MAGA Mike” Johnson would not be Speaker, and instead Kevin McCarthy would have been able to retain the gavel.
Just one House election will suffice to illustrate that, even without eliminating gerrymandering or the way House seats are districted based on the geographical self-sorting of Democrats and Republicans into their urban and rural divisions, the top-3 electoral system would produce a less polarized House. Arizona’s congressional districts are drawn by an independent redistricting commission and thus not the product of partisan gerrymandering. The state’s second congressional district is Republican-leaning, and the 2022 election for its seat was won by a Trump-endorsed election denialist, Eli Crane. Replicating the Senate examples we’ve already considered, Crane beat a non-MAGA traditional Republican in the GOP primary.
Had the top-3 system been in place for this election, that non-MAGA traditional Republican—who was an institutionalist serving in the state’s legislature—almost certainly would have won, securing a majority of all the district’s voters against both Crane and the Democrat. But having vanquished the institutionalist in the primary, Crane defeated the Democrat in November and went to Congress to represent the district. Once he got to Congress, he immediately became one of the House’s most hardline Republicans and was one of the “crazy eight” who sabotaged McCarthy’s speakership. While Crane is just one member of the House, he exemplified the kind of candidate who can win under the current system but would lose to a more moderate Republican in a top-3 election.
In sum, if we want a Congress that is functional, that actually gets stuff done on behalf of the American people, we ought to consider adopting the top-3 system instead of what we currently have. In addition to accomplishing things, a Congress elected by means of top-3 elections would more accurately reflect the preferences of the voters who elected those members of Congress. Indeed, it is precisely because a Congress populated by means of the top-3 system would be more genuinely representative of the voters that the Congress would be much more inclined to achieve legislative results in the public interest than the current Congress is.
But what about presidential elections? Could a top-three system be used for them? That’s a topic that will need to await a future Common Ground Democracy column. Meanwhile, we should be thinking of ways we can implement the top-3 system in congressional elections, for both Senate and House, as a key step to improving American democracy.