What the department says
The analysis “had nothing to do with” the group’s legal standing, so it broke the judge’s order, and the matter should go to the Justice Department for possible criminal contempt.
What the filings show
The sworn analysis names no one. Its central figure is that 8,880 of the list’s 20,930 names, or 42%, are not on California’s voter file. The department’s Sunday response on standing does not dispute it.
The thumb drive reached a lawyer in Los Angeles at 6:03 p.m. Eastern on Friday, Oct. 30.
Inside was the list the Justice Department says shows noncitizens registered to vote in California. A federal judge had ordered the government to hand it over, and the clock to Election Day was running.
Chief Judge James Boasberg had ordered the list produced that afternoon in a lawsuit by the Coalition for Humane Immigrant Rights, known as CHIRLA, and a naturalized citizen named Pedro Trujillo.
The file’s name ended in “reduced columns.”
UCLA data scientists Matt Barreto and Michael Rios say in a sworn declaration that the spreadsheet had eight columns. A Homeland Security analyst’s declaration in a separate California case, they note, describes a 30-column version of the same work, including voter ID numbers.
The file’s properties, they say, show it was created at 2:06 p.m. the same day, with the author listed as Ty Trejo, the Justice Department lawyer on the case. The version sent to Los Angeles County and the state had been produced on Oct. 13.
They filed their analysis the next day.
The declaration is dated Saturday, Oct. 31, and the numbers are blunt.
Of the 20,930 names, only 20.7% matched a record on California’s current voter file by first, middle and last name, birth date and exact address. Another 36.8% matched when the address was dropped.
That leaves 8,880 names, 42%, that the analysts could not find anywhere on the file. We checked the division: 8,880 over 20,930 is 42.4%.
Read that against what the list is for. The government is asking California counties to set aside ballots from the people on it, and a name that is not on the voter file has no California ballot to set aside.
Two cautions belong here. The analysts say they matched against the current file without the voter ID column, so some misses could be matching failures, which they argue is the department’s doing. And the declaration discloses that Barreto and Rios have held a contract with California’s Secretary of State since 2020.
This is the plaintiffs’ experts’ work, produced in a day. It is also the first sworn outside analysis of the full list that we found.
The analysts also searched CHIRLA’s own roster of naturalized citizens, members and clients. They report 22 entries that match the government’s list on name, birth date and exact address, and 66 more that are near-certain matches on name and address.
Those are the 22 people the coverage describes. The department’s own Sunday filing repeats both numbers.
Then comes a projection: if CHIRLA’s match rate held across California’s 3.8 million naturalized-citizen voters, the analysts say there would be 3,743 certain wrong matches.
That is a projection from a membership list, not a count of anyone. The sworn count is 22.
The department’s reply came Sunday, Nov. 1, in a 12-page filing we read in full.
It argues CHIRLA still has no standing. Of the analysis, it says: “The report’s political grandstanding about alleged inaccuracies that purportedly impact California voters generally, or comparisons between the government’s list and the full ‘California voter file,’ has no relevance to whether the particular plaintiffs in this case have standing.”
It does not say the 8,880 is wrong. It says the number does not matter to this question.
The filing also describes the judge’s Oct. 30 order as barring public dissemination of the “identities” of people on the list. The declaration reproduces none. Its tables of matches went to the judge in camera, and the department objects to that too.
The response mentions a motion to strike filed alongside it, and the contempt request appears to sit in that filing, which we could not retrieve. The department says the analysis violated the order’s limit to arguing standing, and it asks Boasberg to refer the matter to the Justice Department “for investigation” of “possible criminal contempt.”
We found no ruling on it by Monday morning.
One more passage in that Sunday filing deserves its own look.
Arguing that CHIRLA’s members cannot be harmed, the department tells the court that its letters to California officials “are not self-executing; they do not compel anyone to do anything.” States and counties, it says, “may disregard the government’s demands,” though “at their own legal peril.”
Hold that next to the warnings the department has issued, that officials who knowingly let noncitizens vote could face criminal charges. The same demands compel nothing in court and carry a threat of prosecution outside it.
Courts have so far declined to enforce the demands. A federal judge in California refused to pull 20,000 ballots, the Ninth Circuit turned down an emergency appeal on Oct. 30, and a San Francisco judge barred the department from threatening election workers. The department appealed that last ruling on Sunday.
The contempt theory needs the analysis to have nothing to do with standing. That is a legal claim, and the two filings let you test it.
The plaintiffs’ supplemental brief argues that a citizen “erroneously flagged as a non-citizen by the government” has standing, and cites an appeals court decision for it.
The department’s response lays out its own five-step chain for how a member could ever be harmed. The member has to be on the list, the listing has to be wrong, the county has to fail to confirm eligibility, the county has to set the ballot aside, and the ballot has to go uncounted.
Step two asks whether the list is wrong about real citizens. For 22 entries on CHIRLA’s roster, the only sworn answer so far is yes.
An analysis that speaks to step two is at least arguably about standing. That makes “nothing to do with” a contested claim, which is a poor foundation for a criminal referral.
We think the contempt request is the story.
The analysis can be wrong. It was done overnight by experts the plaintiffs hired, on a file missing its key column, and the department holds the full spreadsheet and the voter file, so it can run a better one. Challenging an analysis is what an answer looks like.
Asking a judge to send the lawyers to a criminal investigation is something else. Under the federal criminal rules, the default when a judge suspects criminal contempt is to ask the government to prosecute, so this request would hand the people who published the numbers to the department they are suing.
If the department believes the 42% is wrong, the remedy is a declaration saying so, with its own match. California counties facing its demands would want that before Tuesday.
Until it answers the arithmetic, a contempt referral reads as a substitute for an answer. The sworn count on the table says 42% of the names the government wants ballots set aside for are not on the voter file.