The fight over 20,000 “likely noncitizen” registrations in California is not really about a single list; it is a stress test of how American elections reconcile two imperatives that can collide at scale—exclude ineligible ballots with speed and certainty, and avoid wrongfully sidelining eligible voters when evidence is probabilistic and timing is tight.
The Short Version
- Federal officials built a list of roughly 20,000 California registrants flagged as likely noncitizens using DHS databases plus third-party analytics; DOJ sought to segregate and pause associated ballots until verification.
- California counties countered that spot-checks revealed duplicates, inactive records, U.S.-born individuals, and minimal voting history—evidence the list was too noisy to drive emergency ballot segregation.
- The Ninth Circuit rejected DOJ’s emergency bid, reflecting judicial reluctance to reorder ballot processing midstream on contested evidence and uncertain authority, though underlying legal issues continue in the trial court.
- Beneath the litigation sits a perennial election-administration problem: large-scale record matching produces false positives and false negatives, so remedies must balance integrity with due process and feasibility.
What DOJ asked for and how it built its case
The Department of Justice sued California after DHS-affiliated teams concluded that more than 20,000 registrants were “likely” noncitizens, and asked courts to compel election officials to segregate and refrain from counting ballots linked to those registrations pending verification. The government’s filings describe a multilayer workflow: matching voter rolls against immigration-related records held by USCIS and Homeland Security Investigations, supplementing with third-party datasets, applying scoring logic to prioritize cases, and then conducting manual review—culminating in declarations that the government “reasonably believes” the set contains noncitizens registered to vote. The complaint and associated statements anchor the request for urgent relief, framing the harm as irreversible once ballots are processed and intermingled.
That framing matters technically. California law permits preprocessing and tabulation-ready handling of mail ballots weeks before Election Day; DOJ argued that waiting turns a solvable eligibility concern into an administrative fait accompli. In oral argument, government counsel emphasized the pace at which counties—Los Angeles foremost—were moving through large volumes, contending that only immediate segregation could preserve the ability to prevent any ineligible ballots from being tallied. The theory is straightforward: federal law bars noncitizen voting in federal contests, so if the government presents a vetted list of likely ineligible registrants, states must at least pause.
Why counties pushed back: noise, authority, and feasibility
California officials did not simply dismiss the allegation; they interrogated the dataset’s reliability at the individual-record level. Los Angeles County reported that in a preliminary 200-record review, nearly 40% of the entries with usable birthplace fields indicated U.S. birth—strong, though not absolute, evidence of citizenship. A majority had no voting history, and more than a hundred were already inactive or canceled, implying no ballot was sent in the first place. Santa Clara County flagged apparent duplicates—three on the first page—casting doubt on deduplication and linkage discipline. These are classic data-quality failures in cross-file matching, which become fatal when the remedy risks sidelining legitimate ballots.
Local counsel also challenged federal authority to direct county election workflows absent statutory citation or court order, warning that ad hoc segregation demands—particularly on a compressed calendar—could cause operational breakdowns. In parallel, state officials underscored that California’s system relies on sworn attestation of citizenship at registration and ongoing list maintenance under state and federal rules; they argued that a late-breaking, probabilistic federal list does not justify pausing ballot counting without individualized due process. Collectively, the counties’ rejoinders reframed the dispute from “ignore noncitizen ballots” to “don’t disenfranchise citizens on a noisy match.”
What the Ninth Circuit actually decided
The appellate panel turned away DOJ’s emergency request to set aside the ballots—an extraordinary remedy—largely on procedural and equitable grounds. The court’s ruling reflected skepticism about late-stage federal intervention into state election administration and the adequacy of the evidentiary proffer for such relief; it sent the case back toward the district court rather than greenlighting ballot segregation on the spot. That outcome is consistent with a long judicial pattern: courts hesitate to impose disruptive remedies close to elections when factual disputes are unresolved and when federal-state authority lines are contested. The decision did not bless every California practice or condemn federal concerns about eligibility; it said emergency relief on this record, at this time, was not warranted.
The jurisdictional posture also mattered. DOJ sought appellate action before the trial judge had developed a factual record. Absent findings on data accuracy, operational feasibility, and the balance of harms, the Ninth Circuit declined to override the status quo. Timing cuts both ways in election law—delay can entrench harm, but last-minute orders can destabilize systems measured in millions of ballots and thousands of poll workers.
The real problem under the headlines: classification at scale
Stripping out the partisan valence, the core is a high-dimensional matching problem. Voter files do not contain canonical citizenship markers; immigration databases can be incomplete, outdated, or mismatched to voter-file keys like name and date of birth. Add third-party data with opaque provenance, and you increase coverage and also error. At scale, small false-positive rates generate large absolute numbers of wrongly flagged individuals. Social scientists and election-administration research have documented these dynamics for years: record linkage yields both false positives (eligible voters ensnared) and false negatives (ineligible records missed), and small changes to matching thresholds can swing outcomes materially. The engineering reality presses for careful validation before punitive action.
Because ballots are secret and processing begins well before Election Day, the remedy space is narrow. You can preclear eligibility through rigorous registration vetting, but that risks new barriers for eligible citizens. You can create a parallel provisional-ballot lane for flagged registrants, but that imposes due-process burdens and must be disclosed and appealable. Or you can pursue post-election audits aimed at prosecution rather than preclusion, which preserves votes but may allow isolated ineligible ballots through. Each path reflects a policy choice about which error—wrongful exclusion or wrongful inclusion—is more tolerable, and at what frequency.
How to judge competing claims responsibly
The federal case presented a specific, numerically bounded claim backed by sworn declarations—a stronger starting point than generic suspicion. But specificity is not the same as accuracy. County spot-checks found indicators of both database noise and administrative irrelevance (inactive records, no ballot issuance), while the courts balked at emergency relief without robust fact-finding. On this record, the evidence weighs against segregating ballots based solely on a probabilistic list assembled late in the cycle. The more durable takeaway is methodological: without transparent matching criteria, auditable provenance, and individualized notice-and-cure procedures, large-scale flags are not a sufficient predicate for remedies that risk disenfranchising citizens.
None of this forecloses appropriately targeted enforcement. California can and should investigate credible, individualized evidence of noncitizen registration or voting and update rolls continuously. Federal authorities can bring criminal cases where statutes are violated and facts are clear. But the decisive venue for structural fixes is process design—clear statutory authority for intergovernmental data sharing, standardized matching protocols with published error metrics, and prebuilt remediation workflows that can be triggered early enough to be fair and effective.
#ELB: Judge Bumatay in Ninth Circuit Dissent in California Ballot Case Appears Incorrect That Congress Enacted Ban on Noncitizen Voting Under Its Election Clause Powers; Its Sponsor Said It Was Under Its Immigration Powers https://t.co/WqksfkBIWv
— Rick Hasen (@rickhasen) October 30, 2026
What better systems would look like
Three elements raise the floor. First, transparent, conservative matching: disclose fields, thresholds, and expected error rates; confirm matches with secondary data before flagging a voter. Second, early, individualized notice with cure: months before ballots go out, contact flagged registrants with simple, multilingual paths to verify citizenship—sworn statements plus documentary options—and guarantee that unresolved flags route to a provisional process rather than silent exclusion. Third, bounded emergency remedies: reserve ballot segregation for cases where evidence is individualized and strong (for example, a recent naturalization mismatch reconciled by authoritative records), and where administrators have pretested the operational plan. In short, build the remedy before you need it, and scale it only as far as the evidence can bear.
Sources:
cbsnews.com, justice.gov, npr.org, nbcbayarea.com, ashtalks.substack.com, thebayareabrief.com, wbhm.org, therothmanreport.com, yahoo.com, justfacts.com
