A New Jersey software error registered about 6,600 noncitizens who had identified themselves as noncitizens during motor-vehicle transactions, Governor Mikie Sherrill said. Fewer than 400 of them reportedly voted. [1] The first number establishes an administration failure. It does not convert every registration into a fraudulent ballot.
New Jersey has used automatic voter registration at motor-vehicle offices since 2018. Between 2023 and 2024, the affected people indicated that they were not citizens but were registered anyway. [1] That fact cuts against the simplest intent story: the system recorded information that should have stopped the registration and then failed to act on it.
It does not make the error harmless. Registration can produce election mail, administrative burden and loss of confidence. Ballots reportedly cast require review of citizenship at the relevant time, election, notice and intent. The fetched account does not provide a completed audit of those fewer than 400 cases, prosecutions or an outcome analysis.
The scale also needs its denominator. New Jersey has more than 6.6 million registered voters. An election expert quoted by the Guardian estimated that the potential ballots from incorrect registrations would equal about 0.009% of ballots in the 2024 general election, which Kamala Harris won by more than 250,000 votes. [1] That comparison limits an outcome claim; it does not excuse the system defect.
Donald Trump and allies inflated the disclosure into wider allegations. Trump cited a separate figure of 35,152 alleged noncitizens on New Jersey rolls, while the Department of Homeland Security had offered little methodology for its matches. Sherrill said the state had received no supporting identities from the department. [1] The 6,600 state error and the federal ceiling are not interchangeable.
Database matching is especially vulnerable to false positives when names and dates lack unique identifiers. Experts told the Guardian that public voter files ordinarily omit the immigration or Social Security identifiers needed for high-quality matching. [1] A headline total must therefore survive person-level review before it becomes an eligibility finding.
The opposite partisan shortcut is equally weak. A rare error can still expose a control failure worth repairing. The state should publish the software cause, affected dates, correction process, notices and audit result. It should explain which elections fall within the smaller ballot count and how investigators distinguish a person following state instructions from intentional unlawful voting.
Correction itself can create harm if handled carelessly. Notices must tell affected people what the state did, whether any ballot is under review and how to contest a mistaken citizenship determination. The audit should preserve privacy while publishing enough aggregate detail to test completion. Quietly deleting records would cure one database field while leaving voters and public confidence in the dark.
The failed X search adds no verified voter account or platform frame. The divergence comes from documented public arguments: mass-fraud rhetoric treats registrations as ballots and intent; minimization risks treating a small share as no failure at all. Both skip the work between record, eligibility, vote, correction and result.
New Jersey's disclosure is serious because the state registered people after they supplied information that should have prevented it. [1] Its political use is misleading when 6,600 becomes 6,600 fraudulent votes. The final judgment belongs to a corrected file and completed case review, neither of which was in the cutoff record.
-- SAMUEL CRANE, Washington