In two months, Massachusetts and, likely in 2028, California will consider ballot propositions to adopt (MA) or discontinue (CA) top-two open primary elections. First enacted in Washington (2004) and later in California, the consensus on top-two among online politikers spans between “madness” (Liberal Currents), “a wreck” (Stephen Wolf), and a mild “Did it flop?” (CalMatters). That consensus is wrong. Liberals should be celebrating top-two as an accidental and underappreciated success.
California passed top-two by proposition in 2010 and began using it for all statewide, congressional, and state legislative elections in 2012. Most California localities followed suit. The record, contrary to the assertions above, is almost uniformly neutral to positive. Since 2012, there have been 1,275 of these ‘state’ races. Of those, 187 have resulted in a same-party contest in November. Which is to say, in those lopsided districts the real election took place when we said it would, not 3-6 months prior. More about this later.
The origins and official pitch for top-two probably explain most of the disregard. Top-two was put forward by moderate Republicans who wanted a check on the leftward movement of the state and by others alarmed by polarization. They saw around the country that non-competitive seats were becoming captive to motivated party-primary voters, with increasingly fringe candidates playing to narrower electorates. They believed that top-two would protect mainstream Republican candidates in conservative districts and centrist Democrats in progressive districts. It makes sense, but it hasn't turned out that way. To its credit, top-two has doubled the frequency of same-party challenges to incumbents, for the simple reason that a second-place finish is enough to advance. Those challengers, progressives mostly, often advance from the first round to win in November.
Why have the results defied predictions? Those expectations stem from a one-dimensional, left-right, mental model in which the more centrist candidate has access to more votes. But campaigns are contested along many dimensions and in a two-person race, particularly when party ID has been mooted, the fight is over which dimensions matter.
The practical objection to top-two is that it risks shut-outs of the district’s plurality party. If, in the first round, Republicans are disciplined and have two strong candidates while Democrats have three or more strong candidates, the possibility exists for two Republicans to advance. Or vis-versa. It's a legitimate concern and my response may not be entirely satisfying: the math is very difficult. Voters on each side have almost never divided their support equally between two candidates and three candidates, each.
There was anxiety when this year's CA gubernatorial primary was polling near that result. I saw handmade signs at No Kings distraught about this! In the end, if Hilton (R 24.6%) and Bianco (R 10.2%) had polled evenly at 17.4%... and Becerra (D 28.0%), Steyer (D 22.8%), and Porter (D 4.36%) had polled evenly at 18.4%... well, Becerra and Steyer would have advanced. For Hilton and Bianco to advance would have additionally required that Mahan (D 3.5%) pull another 3% evenly from the leading three Democrats. Not every year will be so lopsided, but the anxiety this year was misplaced.
The concern about shut-outs is legitimate but doesn’t call for complete overhaul. There have been 3 shut-outs of a registration-plurality party out of 1,275 ‘state’ elections since 2012. Those are clearly negative, but flukey. It's a related and solvable issue that the threshold to appear on the ballot in California is low enough to invite a comical number of vanity runs: the ballot this year had 63 candidates for governor.
Another principal objection is that the open primary dilutes the role of political parties. Any candidate can run under any affiliation and a candidate's voters can be of any registration. That's true but not strictly unique to top-two.
The modern party primary only dates to 1972 and was a further step in devolving power away from party elites. So, how effectively does it help voters embody their interests? Do parties in the 14 (to 23, depending) states with registration-based party primaries behave more like membership organizations? Are political outcomes improved? In practice, the difference between open and closed primaries for party organizational strength has been negligible.
That's not to say we don't need parties, we plainly do. Only that the mechanisms of candidate selection are not what they were and can't be returned to. Parties are caught between early models in which party leaders "discipline the base" and the modern reverse. Further, contemporary party leaders are in an awkward double-bind when it comes to coordination: “fail to winnow the field and you’re ‘in disarray.’ Clear the field and you’re an elite conspiracy. This also leads to a kind of paradox of (non)transparency: efforts by party leaders to shape nominations are largely viewed as illegitimate, so the processes are informal and potentially the most important conversations are happening out of public view.” (Julia Azari) Placing these problems at the feet of top-two is a mistake.
So, if top-two has not been a disaster in outcome or in party building, why affirmatively support top-two? Most of all, it gives lopsided districts a much better chance at meaningful contests in November. That’s most districts! Typically, only about 10% of congressional districts are considered competitive.
That might, in itself, be bad but explicable. What we can’t accept is how illiberal and corrosive it is to settle November elections in spring and summer – while telling the public otherwise and proceeding to conduct campaigns that can't but talk past one another. (Yes, readers here understand when the real electoral action happens in party-primary states. Shame on us for accepting that it comes so much sooner than we say it does and well before most voters are paying attention.) Campaigns and contests are important to democratic formation, to political education. When that's settled in June by voters from one party alone, we seed alienation from the democratic process.
The sequentiality of top-two also deserves appreciation. In the first round, we establish what sort of November election we will have. Who Are We, district? Then we all focus our attention on the two formations that emerge. It's simple, effective, and inclusive.
(Alaska’s top-four, to get that out of the way, guarantees against lockouts but sacrifices the focused contest. It matters that top-two is top-TWO.)
This year's California primary is a great example. For governor, top-two degraded gracefully into a very standard party primary with a very ordinary outcome: an establishment Democrat and a Republican, as befits the 37% share of the two-party vote that Republicans combined received. For Los Angeles mayor, meanwhile, the result was a top-two showcase. The leading conservative, Spencer Pratt, received 25.5% of the vote, not enough to beat the insurgent Democrat, Nithya Raman who received 29%, for second-place. LA now has a campaign and contest over its future that is meaningful and relevant to all Angelenos. It has pushed the otherwise disengaged Karen Bass to govern and compete for votes. It has spared us the dispiriting irrelevance of a Bass-Pratt contest; not simply that the outcome would be known but: what LA reality do they even share to fight over? That there's competition on real, local, intelligible terrain – D vs D in a blue city – is a democratic success.
The only statewide race this year to lead to a same party matchup is for California Insurance Commissioner (think: global warming) with two Democrats, Ben Allen and Jane Kim. Downballot party primaries like this generally draw little attention in spring-summer and, absent top-two are decided in November by party affiliation. With the party question settled, we can hope that by November, with a live contest and more voters paying attention, we’ll see more public engagement. Finally, positive examples of the promised moderation do exist, mostly on the right. (Why on the right? I'd speculate that this reflects a gap between elite rhetoric and the reality of what ordinary voters find moderate.) So, the California Federation of Labor Unions has regularly endorsed and campaigned for their preference among two Republicans in conservative district November elections.
Top-two hasn’t yet led to any 3rd party victories in state races, but it has advanced many to the run-off; mostly Libertarian Party candidates in conservative districts.
There's one more objection to address, that top-two is not as good as IRV, RCV, PR... take your pick. Maybe so. Each has (no hot take here) relative strengths and weaknesses. Simply, top-two is much better than the US standard and we should say so!
My advocacy, I think, comes down to how I weight the tradeoffs. I place a lot of value on contested November elections. Many alternative schemes don't pass this test. I don't place as much value on the 'expressiveness' of voting or the harm of strategic voting to avoid spoilers. (Most voters aren’t particularly strategic and, in any case, what voting isn’t?) I do value simplicity. ("Remember to fully rank all 5 ballot slots but don't rank the main opponent, mind you, expressively not strategically, of course.") Proportional representation can address most of this, agreed, and has real strengths – particularly to moot the problem of gerrymandering – but again, isn't on the ballot and is in no way forestalled by top-two.
Top-two belongs in the same conversation with innovative election models that are more typically championed on the democratic left. Top-two is a clear improvement over standard US practice and has a lot to recommend it. Importantly, it’s on the ballot this year and in 2028, perhaps owing to its intelligibility/legibility. While working toward proportional representation, we should be championing top-two for the improvement and accidental liberal success that it is.