Thursday, November 12, 2026

The 2026 Election Was Not Stolen. It Was Broken.

  From Staff

Six thousand ballots went missing. Hundreds of thousands were discarded on technicalities. The records that would settle the argument are sitting in county offices, unread.

For three weeks the loudest argument in American politics has been about a heist. The case was elaborate. A party winning twenty nine of thirty coin flip districts, phantom millions swelling one side of the popular vote, tabulators quietly weighting ballots at decimal values, Texas turned blue by forces that could only be artificial.

That case does not survive contact with the returns. Republicans won thirteen of the twenty closest races in the country. The single tightest contest, North Carolina's Fifth District, went Democratic by ninety nine votes. The national House vote moved from a Republican advantage of 2.7 points to a Democratic advantage of 13.6, which is what a wave is and a wave is precisely what makes thirty separate districts break the same way without anyone touching a machine. Texas did not fall by a whisker. It fell by twelve points, against a candidate who had spent two years under investigation, in a year when the national environment was thirteen points against his party. The "fractional voting" story requires a tampering that leaves no trace in any hand count and no signature failure in any transfer log. It is a story that cannot be tested, which is why it cannot be won.

And that collapse has been used, very conveniently, to close the file. The file should not be closed. Because the same election produced something far more concrete than a conspiracy and far more damaging to public trust than any of the theories now circulating on the internet. It produced a paper record of institutional failure, written by the institutions themselves, in press releases and county statements and secretary of state bulletins. Almost nobody has read it.

Start with the ballots that simply vanished from the count.

Latah County, Idaho, left 2,343 ballots out of its election night totals. The machines counted them correctly. The reporting system did not receive them. A full recount of all 18,104 ballots followed, conducted in public and no outcome changed. The closest measure on the ballot was a school levy that passed with 60.34 percent when it needed sixty. Three tenths of a point separated a functioning audit from a genuine crisis and the only reason anyone knows that is that a clerk's office went looking before the canvass.

Johnson County, Iowa, left 2,017 absentee ballots out of its election night totals because the boxes were set on a shelf instead of a cart. The county recounted more than 26,000 absentee ballots. The secretary of state issued a directive. No outcome changed.

Fulton County, Georgia, failed to scan more than a thousand early ballots. One scanner reported a paper jam that did not exist. Another had its power cord rolled over by a voter in a wheelchair. Seminole County, Georgia, reported roughly 388 more voter check ins than ballots, because a memory card never made it into the county's election management system.

Bannock County, Idaho, is the strangest and most instructive. A sealed ballot box from one precinct slid beneath the cab of a poll worker's pickup truck and sat there for a week. It held 333 ballots that had never touched a counting machine.

That is more than six thousand ballots, across four states, either never counted or never reported on the night citizens went to bed believing they knew the result. Not one of those jurisdictions was hacked. Every one of them was sloppy.

Now the number that should end the argument about which party is being cheated. North Carolina's mail ballot rejection rate reached 5.85 percent this cycle, with 3,672 ballots at risk and rural counties running between sixteen and thirty percent. Anson County threw out a third of the mail ballots it received. In the same state, more than thirty six thousand provisional ballots were cast, nearly ten thousand more than in 2022, and counties began rejecting them at rates that would embarrass a developing country. Hoke County counted 59 and rejected 105. Watauga County rejected 23 percent. Democrats cast 43.8 percent of those provisional ballots.

Mississippi is worse. More than 230,000 voters were listed as inactive by Election Day, many of them flagged by address lists the secretary of state bought from a commercial data broker for fifty thousand dollars. One man who had lived in Itawamba County his entire life was placed 160 miles away in a Delta town he had never visited. A student was moved to a county she had never lived in. In Hinds County, officials rejected 1,609 of 2,426 affidavit ballots, roughly two thirds, against a statewide acceptance rate near seventy seven percent. The state had quietly deleted its mail ballot curing regulation in August, and Mississippi has no recount law for close races at all.

California rejected 148,241 mail ballots in its June primary, about 1.7 percent of the total. Ninety three thousand arrived late. Fifty two thousand had signature problems. That is a primary, in a state that has spent twenty years telling itself it has the most accessible system in America.

Add it up and the picture inverts. The number of ballots plausibly stolen by a hacker in 2026 is zero, as far as any public record shows. The number of ballots thrown away by procedure, in the open, under color of law, is in the hundreds of thousands. One of those numbers is a theory. The other is a policy.

Here is what the people who spent November shouting about algorithms will not tell you, because it complicates the story. The system caught almost all of this and it caught it using paper.

Bannock County recovered its missing box because the seals were intact and matched the record and because the electronic poll book for that precinct recorded exactly 333 check ins. The count came to exactly 333. That is not luck. That is a chain of custody with a paper spine. Where that spine exists, errors are findable. Where it does not, they are not. The absence is the vulnerability.

Idaho's secretary of state called the Latah discrepancy significant and ordered a public recount with candidates, press and citizens in the room. Johnson County's auditor stood in front of the supervisors and explained, in detail, how her office misplaced two thousand ballots. These are not the reflexes of an institution covering something up. They are the reflexes of an institution that got caught being careless and knew it had no choice but to say so.

And the audits are real now, in places they were not five years ago. Texas ran its first ever statewide risk limiting audit this cycle, alongside a statutory hand count that requires the early voting ballot board, not elections staff, to count early, mail. Election Day ballots from a sample of precincts and publish the results. Georgia ran a risk limiting audit at a five percent risk limit and then a full ballot image audit of all 21,222 ballots in a congressional special election. And reported one hundred percent agreement. Orange County, California, streamed its manual tally live and generated its random sample by rolling dice on camera. Leavenworth County, Kansas, found nothing.

That is the test that settles these arguments and it exists. If a multiplier were shifting tallies, a hand count of randomly selected paper would find it. The audits keep drawing ballots until they either confirm the result or force a full count. Demand the reports. In Texas they are public documents for the first time in the state's history.

The good news stops where the map stops. Fifty states, more than three thousand counties and no common standard for any of it. Texas now audits. Mississippi has no recount law. North Carolina changed its envelope rule weeks before the election and watched rejection rates in its largest county hit fifty three percent on that rule alone. Mississippi deleted its curing regulation in August. Both changes were legal. Both changed how many ballots counted. Neither gave voters time to learn the new rules. In both cases the burden fell hardest on the people with the least broadband, the least spare time and the least reason to trust the process.

Then there is the vendor problem. Four companies sell the machines and the software and the poll books and they answer to almost no one. El Paso County's poll pads failed at sixty three percent of Election Day sites in March. The county commission handed the vendor a no cost support extension and called it the best option available. Fort Bend County's entire network of vote centers went dark for two and a half hours because an employee uploaded the wrong file from a May city council election. Broward County's ballot issuance software misfired on split precincts. Westchester County's scanners refused paper ballots countywide and officials put them in orange bags under police escort.

And the voter roll mess cuts in both directions at once, which is why neither party wants to describe it honestly. Mississippi used commercial address data to inactivate hundreds of thousands of people, many of them legitimately registered. Texas ran 2,724 names through a federal citizenship database and later confirmed that at least 578 of them, more than one in five, were citizens, most of them naturalized, and had to tell counties to put them back. Sixty seven million registrations went through that system nationally. In Forsyth County, Georgia, a private tool flagged four voters as dead. All four were alive. Of 69 challenged voters, thirty eight had simply registered using an initial.

The rolls are not clean. The cleaning is not competent. Both statements are true and the people who pay for the contradiction are ordinary citizens who show up. Are told they are not on the list, fill out a provisional ballot and never learn whether it counted.

Forget the multiplier. Demand the ratio.

Every precinct in America should publish one number after every election. Ballots counted divided by voters checked in. Latah's would have read 88.4 percent before the recount. Bannock's would have read 99.0 percent. Those figures would have screamed from the page and every citizen could have read them without a statistician.

Publish that ratio. Publish the hash of the certified software against the hash of the deployed software, before the polls open. Publish the audit report, with the random seed and the ballot manifest. Use hand marked paper, or at minimum a paper record the voter can verify. Keep two parties in the room for every transfer, with timestamps. Ban wireless components from tabulation equipment. Stop changing election rules within ninety days of an election. Give every state a recount trigger set by margin. And stop buying voter roll data from commercial brokers and stop bulk matching citizens against databases that were never designed to answer the question.

None of that requires anyone to prove a conspiracy. All of it would have caught every failure on the 2026 ledger, in public, on the record, before anyone had a chance to invent a story about it.

The election was not stolen. It was broken in a dozen ordinary ways by people who were tired, underfunded, running old equipment and following rules that changed under their feet. The institutions that ran it then did the one thing that could restore confidence, which is to admit the errors and open the recount to the public. The institutions that write about it did the opposite, spending three weeks mocking the people asking questions instead of answering them with the reconciliation data they already had.

That is the real damage. Not a hacked tabulator, but a system that loses six thousand ballots in a week and treats the resulting suspicion as a public relations problem rather than a summons to prove itself. Trust is not rebuilt with ridicule. It is rebuilt with a number anyone can check. Publish the number.


Sources:

The collapse of the steal case

Texas and HD 93

The lost ballot ledger

  • Latah County, 2,343 ballots missing from reported totals, full public recount, no outcomes changed, school levy at 60.34 percent: KIVI · Boise Gazette

  • Bannock County, 333 ballots in a pickup truck, intact seals, exactly 333 poll book check ins, District 29 margin narrowing to 2,024: Idaho State Journal

  • Both Idaho counties plus Seminole County, Georgia and the 388 missing ballots: Newsweek

  • Johnson County, Iowa, 2,017 absentee ballots on a shelf, 26,000 recounted: KCRG

Equipment and poll book failures

The rejection ledger

  • North Carolina mail ballots at 5.85 percent, 3,672 at risk, Anson County at 33.3 percent, Wake County inner envelope at 53 percent: Democracy Docket

  • North Carolina provisional ballots, 36,000 plus, Hoke County 59 counted and 105 rejected, Watauga at 23 percent, Democratic share at 43.8 percent: Democracy Docket

  • Mississippi, 230,000 inactive voters, the fifty thousand dollar Experian contract, the man placed 160 miles away: News from the States · Sun Herald

  • Mississippi curing rule deleted in August and the absence of a recount law: Democracy Docket

  • Hinds County rejecting 1,609 of 2,426 affidavit ballots; Harrison County at 746 accepted and 704 rejected: Meridian News · Newsweek

  • California June primary, 148,241 mail ballots rejected, 1.7 percent: IVN

  • California signature mismatch rejections at about 44,000 and 0.51 percent: Joe Hoft

  • National provisional benchmarks, 1,736,209 cast in 2024 with 73.8 percent counted and 25.1 percent rejected: The World Data (aggregating Election Assistance Commission survey data)

  • Cuyahoga County provisional rejection detail, 12.97 percent and 5.89 percent excluding unregistered and duplicate voters: Cuyahoga County Board of Elections

The voter roll ledger

  • Texas SAVE results, 2,724 flagged and at least 578 confirmed citizens: Democracy Docket

  • 67 million registrations run through DHS verification: AP

  • SAVE system modifications, SSA data limitations, Missouri misidentifications: Democracy Forward

  • Forsyth County, four dead voters all alive, 69 challenges, 38 initial registrations, ELLY tool: ProPublica

  • DOJ at zero for twenty eight in district courts on unredacted rolls: Democracy Docket litigation summary

The audits

  • Texas first statewide risk limiting audit, Section 127.201 hand count, 22 month retention, public reports: Texas Association of County Election Officials

  • Georgia risk limiting audit at a five percent risk limit and the full ballot image audit of 21,222 ballots: WNEG

  • Orange County manual tally and risk limiting audit with dice roll seed and live streaming: OC Vote

  • Leavenworth County, Kansas finding no discrepancies: Leavenworth Times

  • How audits work state by state, and which states do them after certification: Votebeat

  • Virginia risk limiting audit history and methodology: Virginia Department of Elections

The litigation and press angle

at

No comments:

Post a Comment

Subscribe to: Post Comments (Atom)

Nvidia: The Story, Public in May, Priced in November

 From Staff A short seller published the Nvidia story five months before it cost shareholders a third of the company. He was mocked, warned ...