As Los Angeles County worked through more than 700,000 uncounted ballots after the June primary, the argument focuses on the speed of the count. But it is also worth wondering just who is taking so long to count our votes.
In California’s largest counties, the permanent employees who process and tabulate ballots are represented by some of the state’s most politically active unions. For example, election workers are represented by SEIU Local 721 in Los Angeles County and by SEIU Local 521 in Santa Clara County. These unions endorse candidates, fund ballot-measure campaigns, and contribute to the county supervisors who ratify their contracts and sit atop the election apparatus their members run.
That is a conflict of interest, and it should trouble people regardless of political party. The organizations whose members handle the ballots have a direct stake in who and what those ballots decide, and they often take positions on the very races being counted while helping elect the officials who approve their raises. Even assuming complete integrity from every employee, a system in which an organized political actor staffs the machinery of its own elections is hard to defend on principle. We do not let companies audit their own books or judges hear their own cases.
The concern is not merely about appearances. California’s public-sector unions can strike, and election workers have shown they will use that leverage when it bites hardest. In 2020, Santa Clara County went to court to block SEIU Local 521 from striking days before the March primary, arguing that roughly 1,700 members tied to the registrar’s office could cripple the election. Fortunately, the strike was averted.
Union contracts also make the organized workforce harder to bypass or supplement. Los Angeles, like other counties, must give notice and bargain before contracting out work its represented employees have historically done, so a registrar facing a slow count cannot simply bring in outside help.
This dependency deepened when California began mailing a ballot to every active registered voter during the COVID pandemic. A mailed ballot adds a labor-intensive chain not required for in-person ballots, including signature verification, envelope opening, extraction, and scanning, often performed by permanent unionized staff over weeks. The count has thus consolidated into a single central operation dependent on a unionized workforce.
None of this requires believing that any California election has been stolen; the evidence for that is, at best, inconclusive. Observation rules, chain-of-custody procedures, and the post-election manual tally exist to catch error and fraud. The point is narrower: sound institutions do not rest on the good faith of interested parties.
The best ways to reduce these conflicts are to go back to the way California successfully administered elections in the past.
For example, California should end the blanket mailing of ballots and restore Election Day in-person voting as the norm, while keeping a mailed ballot available to anyone who asks. A ballot cast in person skips the entire back-office chain and is handled by a temporary, non-union workforce spread across hundreds of sites. The more of the electorate that votes that way, the smaller the share of the count flowing through the one workforce with a political stake and the leverage to act on it. Access need not suffer, since early voting, drop boxes, and vote-by-mail on request can all remain.
The second reform closes the time window. California today counts any mail ballot postmarked by Election Day and received up to seven days later. If no legible postmark is available, election workers are supposed to accept the ballot if the voter dated (or backdated) it on or before Election Day. These rules guarantee a weeks-long count precisely when the workforce holds maximum leverage. Requiring every ballot to be returned by Election Day, as most states already do, would compress that window and end the discretion a postmark-and-grace-period process invites over which late ballots to honor.
The country is moving this way: several states ended their grace periods in 2025, and the U.S. Supreme Court is now weighing Watson v. Republican National Committee, which could require ballots nationwide to arrive by Election Day.
These two reforms are relatively easy, but the ultimate solution is politically more challenging: unions that donate to candidates and ballot measure campaigns should not be allowed to represent election workers.
Although this last change may be a bridge too far, hopefully the national embarrassment of California’s 2026 primary election will produce reforms that minimize conflicts of interest in the vote-counting process. The integrity of an election should not depend on trusting the good faith of those with the most to gain from its outcome.
Marc Joffe is a Visiting Fellow at the California Policy Center.