The top-two primary was a good idea.
Vote-splitting is spoiling it.

The answer is approval voting—let voters choose as many candidates as they like.

Current: Vote for ONE
Candidate A20% of vote
Candidate B19% of vote
Candidate C16% of vote
Candidate D24% of vote
ADVANCES
Candidate E21% of vote
ADVANCES
Proposed: Mark all you support
Candidate Aapproved by 55%
ADVANCES
Candidate Bapproved by 49%
ADVANCES
Candidate Capproved by 38%
Candidate Dapproved by 45%
Candidate Eapproved by 42%

Illustrative example. This pattern is playing out across California primaries right now.

scroll
ME

"The right way to fix California's primaries is to adopt approval voting. It's not by going back to the highly problematic partisan primary system."

Micah Erfan — Political Communicator & Consultant · 1B+ views · UVA Law '28

Choose-one creates division
even where agreement exists.

California's top-two primary was designed to give all voters a voice. But when voters who largely agree are forced to mark just one candidate, their support gets divided by the ballot structure itself—not by genuine disagreement. Some want to repeal the whole system. We think that's the wrong answer.

The repeal initiative goes backward

A new initiative wants to scrap the top-two entirely and return to closed partisan primaries—where party bases, not all Californians, decide who advances. That means more polarization, not less.

Approval voting goes forward

Instead of marking one candidate and hoping for the best, voters mark every candidate they genuinely support. The two with the most approval advance. No wasted votes. No vote-splitting. No strategic guessing required.

One sentence. Profound difference.

Change one sentence in the ballot instructions: instead of "vote for one," voters mark every candidate they approve of. The two candidates with the widest genuine voter support advance to the general election.

01

Mark every candidate you support

No limit. No ranking. Just vote your conscience—no need to think about how everyone else will vote. If you support three candidates, mark three.

02

Every mark counts equally

Supporting multiple candidates will never help the candidates you don't want to advance. Your full support goes to everyone you approve—without any tradeoffs.

03

The two most supported advance

Same top-two structure California already uses. Just better math—so the candidates with the widest genuine voter support actually make it through.

This isn't theory.
It's already working.

Approval voting has been adopted by real American cities—with documented outcomes that show exactly what we'd expect: candidates with the widest genuine support win.

63.5%

Fargo, ND — 2018

First U.S. city to adopt approval voting in a public election. Passed with a landslide majority.

68%

St. Louis, MO — 2020

Passed with 68% of the vote. In the first approval voting election, the machine candidate—later found corrupt—lost to two consensus alternatives who would have split the vote and lost under choose-one.

California's voting machines already support approval voting. No new hardware. No new software. No implementation cost. The only change is one sentence in the ballot instructions.

Simpler. Fairer. Better math.

Compared to every other proposed reform, approval voting is the most straightforward fix with the strongest track record.

$0

Implementation cost

California's existing voting machines handle approval voting natively. One sentence changes on ballot instructions. Nothing else.

VSE

Best-in-class outcomes

Voter Satisfaction Efficiency simulations consistently show approval voting produces outcomes closer to what voters actually want than choose-one or IRV.

Won with landslides

Both cities that adopted approval voting did so with over 60% of the vote. Voters understand it immediately and embrace it.

Why not RCV, Alaska model, or repeal?

Good questions deserve honest answers.

Why not ranked choice voting (IRV)?

IRV fails at its core job: it can eliminate the most broadly popular candidate. Burlington, Vermont adopted it, watched it knock out the most widely liked candidate, and repealed it. The problem is structural, not incidental.

Beyond that, IRV only applies to the general election—leaving the primary completely untouched. You're adding complexity where it's least needed while ignoring where vote-splitting actually happens.

Go deeper

IRV violates the monotonicity criterion—you can literally hurt a candidate by ranking them higher. It scores lower than approval voting on Voter Satisfaction Efficiency (VSE), a simulation-based measure of how close election outcomes come to maximizing voter welfare. Approval voting achieves higher VSE with a dramatically simpler ballot. Full analysis →

Why not the Alaska model or Final Five voting?

The Alaska model is radically complex: multiple candidates advance, voters must rank all of them, and results go through multiple elimination rounds. This isn't just inconvenient—it's a real barrier to voter confidence and participation.

Worse, it still uses choose-one voting in the primary. Vote-splitting remains intact at the stage where it matters most. All the complexity of reform with none of the actual fix.

Go deeper

In an analysis of 1,362 IRV elections, the winner came from 3rd place or lower exactly 4 times—less than 0.3% of the time. Advancing five candidates instead of two adds enormous complexity while changing almost nothing about outcomes. A two-candidate general is also the only election that is completely strategy-free. Approval voting achieves lower Bayesian regret scores with a single ballot instruction change. Full analysis →

Why not just repeal the top-two entirely?

Repeal means returning to closed partisan primaries—where only party members vote, and party bases decide who advances. That's the structure that produced the polarization Californians voted to escape in 2010.

Vote-splitting in multi-candidate primaries is already handing seats to candidates who don't reflect the majority preference of the district's voters. Approval voting fixes this directly. Repeal makes it permanent.

Go deeper

Closed partisan primaries systematically produce nominees further from the median voter, because the primary electorate is ideologically unrepresentative of the general population. This is one of the most well-documented structural causes of legislative polarization. The top-two was an attempt to correct this—approval voting completes that correction.

Why approval voting specifically?

It's the simplest possible fix—one sentence changes in the ballot instructions. Mark every candidate you support. The two with the most support advance. No new hardware. No new software. California's existing machines already handle it.

It's proven. Fargo adopted it in 2018. St. Louis adopted it in 2020 with 68% of the vote—and in the very first approval voting election, the machine candidate lost to two consensus alternatives who would have split the vote and lost under choose-one.

Go deeper

Approval voting achieves among the highest Voter Satisfaction Efficiency scores of any simple ballot method. Supporting an additional candidate can never hurt your primary preference—no tradeoff, no strategy, no wasted vote. Deep dive: wonk.blog/duopoly and on Later-No-Harm →

Help patch the primary.

We're building a coalition of voters, organizers, and donors committed to making California's primary work the way it was intended. Whether you have five minutes or five thousand dollars, there's a role for you.

Join the conversation

The organizing is happening on Discord right now.

Join our Discord

Get in touch

Volunteer, donate, or just learn more.

Contact Us →