California keeps counting ballots until the Democrat wins, and the 2026 Los Angeles mayoral count was manipulated to deny a Republican the runoff
By The Conspiratory EditorsJuly 31, 2026
Where the evidence lands: ContradictedThat California's extended vote count is not an administrative process but a deliberate mechanism for producing Democratic victories, that officials keep counting and adding ballots until the preferred candidate is ahead, and that the 2026 Los Angeles mayoral primary was a live example, with the count manipulated to deny Republican candidate Spencer Pratt a place in the November runoff.
Contradicted. Keep the two layers apart. What happened is not in dispute: in the 2 June 2026 Los Angeles mayoral primary, Spencer Pratt led the early returns for the second runoff slot with about 30.4 percent to City Councilwoman Nithya Raman's 22.3 percent, and over the following days Raman passed him, with the Associated Press calling the second slot for her on 8 June. The claim this file rates is that the shift was engineered, that California deliberately extends counting until a Democrat pulls ahead. It is debunked. The count runs long because state law requires it to: mail ballots postmarked by election day are accepted for seven more days, signature problems trigger a mandated cure process, and counties are given up to 30 days to finish. Late-counted ballots have leaned differently from election-day ballots in California for years, which is a known pattern rather than a hidden one, and nobody produced evidence of manipulation. One documented detail is worth carrying: some of the loudest fraud posts were sponsored content, labelled paid partnerships with the prediction markets Kalshi and Polymarket, and were taken down once reporters asked about them.
Believed by: Large right-leaning accounts on X and their audiences. Reporting identified a group of influencers with a combined reach around 13 million promoting the claim, some of them in posts carrying paid-partnership labels from prediction-market companies. There is no polling on how widely the specific Los Angeles version was believed.
The full story
A week of counting, and a lead that moved
The facts of the Los Angeles mayoral primary are not contested by anyone, including the people who called it rigged. On 2 June 2026, Mayor Karen Bass took first place with roughly 34.8 percent of the vote, short of the majority that would have ended the race outright. The interesting fight was for second, and the first returns put Spencer Pratt, a former reality television star with MAGA-aligned backing, there on about 30.4 percent, with City Councilwoman Nithya Raman trailing on roughly 22.3 percent.
Then the count kept going, because in California it always does. Mail ballots postmarked by election day continued arriving. Signature mismatches went through a cure process. Provisional ballots were checked one at a time. Across successive updates over the following week, Raman closed the gap and passed him. On 8 June, with more than 90 percent of ballots counted, the Associated Press called the second runoff slot for Raman.
Somewhere in that week, the explanation arrived that California simply counts until a Democrat is ahead. This file rates that explanation. It does not rate the election, the candidates, or anyone's disappointment, all of which are outside its scope.
Why the suspicion lands so easily
Before the rebuttal, the strongest version of the case, because it is built on two things that are entirely true.
The first is that California really is slow. Results that other states settle on election night can take a week here and are not official for up to a month. If you are used to a race being over when the anchor says it is over, a count that keeps producing new numbers for days does not feel like diligence. It feels like a process still being worked on.
The second is that the late ballots really do break one way. This is not a perception. Across recent California cycles, ballots counted after election night have leaned more Democratic than ballots counted on it, consistently enough that political scientists have a name for the phenomenon. Someone who has watched several California races reverse in the same direction is not imagining a pattern. They have correctly identified one.
Put those together and the inference is short: a slow count, plus a reliable directional drift, plus a specific race where a Republican led and then did not. If nobody has ever explained why the drift happens, then a hidden hand is an economical answer. The rest of this file is about why it is the wrong one, but the reasoning that gets people there is not stupid.
The delay is written into the statute
California's count is slow by design, and the design is published. The state accepts mail ballots postmarked on or before election day and received up to seven days afterwards. Where a signature is missing or does not match the one on file, officials are required to contact the voter so the ballot can be cured, with that window running until shortly before certification. Counties are given up to 30 days to complete the official canvass. Provisional ballots are investigated individually.
Every one of those rules exists to count a vote that would otherwise be thrown away: the ballot posted on time that arrived late, the ballot whose signature had drifted with age or injury, the voter who turned up at the wrong precinct. The state has made an explicit trade, accepting a slower result in exchange for discarding fewer valid votes. You can think that trade is wrong. It is a different thing entirely to treat its consequences as evidence of a scheme.
The directional drift has an equally ordinary explanation. Ballots counted late are not a random sample of the electorate; they are disproportionately the ones posted at the last minute, cured after contact, or cast provisionally, and the voters who do those things skew younger and more Democratic in California. A partial count reported first is a partial count, not a smaller version of the final one. A lead built on it was never a lead across the whole electorate.
The part that was sponsored
Most claims of this kind are spread by people who believe them. This one has a documented wrinkle that separates it from the usual pattern, and it is the reason this episode has its own file.
A central piece of the argument was that betting odds on Kalshi and Polymarket had moved before the results did, which was presented as a sign that insiders knew what was coming. As NPR reported on 7 June 2026, several of the high-reach posts making that case carried paid-partnership labels from those same two companies. The markets whose odds were being cited as proof of a rigged count were, in some instances, paying for the posts that cited them.
The odds argument does not work on its own terms either. Prediction markets aggregate what bettors expect, and bettors were reading the same public returns as everyone else. In a race where the gap was visibly closing, the odds should move well before the call, because that is what the odds are for. They were tracking the count, not front-running a plot.
What happened next is worth recording, since it is unusually clean. After NPR put questions to Kalshi, the company said it had told the influencers to take the posts down, and several came down. Popular Information reported the same pattern and later reported that Polymarket ended the sponsorship arrangement. A claim that had looked like independent consensus across accounts reaching millions turned out to be, in part, a marketing spend, and it was withdrawn within about a day of a reporter asking about it.
Why a purchased chorus is so convincing
People do not usually evaluate a claim like this on its merits. They evaluate how many trusted-seeming accounts are saying it, and how quickly. Simultaneity reads as corroboration: if a dozen unconnected people all noticed the same thing within an hour, the thing is probably there. That heuristic is reasonable in ordinary life and useless when the simultaneity can be bought.
A paid-partnership tag is a few small words attached to a post that may travel through screenshots, quote-posts, and reposts that carry no label at all. By the second hop, the disclosure is gone and only the claim survives. Whatever the tag accomplishes, it does not follow the content through the network it was bought to reach.
There is a further trap in the specific argument used here. A moving number, updating live with real money behind it, feels like privileged information rather than opinion. Very few readers stop to work out that a prediction market is downstream of the same partial returns they are already watching. The odds looked like a second, independent witness. They were the same witness in different clothes.
None of which means the people who shared this were dishonest. The great majority were not paid and had no way to know that some of what they were amplifying was. That is the mechanism worth understanding: sponsorship does not need to reach the audience to work, only the loudest few accounts the audience already follows.
A claim that was withdrawn by its own sponsors
Two layers, kept apart. The documented one: a real primary, a real lead change, a slow count with causes written into California statute, and an AP call for Raman on 8 June that nobody has successfully challenged. The rated one: that the delay is machinery for producing Democratic wins, and that this race was steered. Nothing supports it. No filing, no official, no audit, and no evidence beyond a lead that moved in the direction California leads have moved for years.
What makes this case unusual is how it ended. Most conspiracy claims are never retracted, because there is no one with standing to retract them. This one had sponsors, and sponsors have reputations and legal departments. One reporter's questions were enough to get posts pulled and, before long, to end the sponsorship programme altogether. It is a reminder that when a belief is partly a commercial product, it has a weak point that a sincere delusion does not.
The frustration underneath deserves better than the theory built on it. Wanting to know who won before the following week is an entirely reasonable thing to want, and California's rules impose a genuine cost in public confidence that is fair to debate. That argument is available to anyone who wants to make it. It just has nothing to do with whether the count was honest.
What's still unexplained
- Slow counting is a legitimate policy argument even though it is not evidence of fraud. Whether a seven-day mail-ballot window and a 30-day canvass are the right trade-off against public confidence in fast results is a real debate, and people who want faster counts are not making a conspiracy claim by saying so.
- The disclosure question is unresolved. Paid-partnership labels are small, easy to miss, and were apparently sufficient to satisfy platform rules while the posts reached millions. What adequate disclosure would look like for political content sponsored by financial platforms has not been settled by anyone.
- Prediction markets now sit in an odd position relative to elections they profit from. Both companies acted when journalists asked, which suggests the norm was not deliberate policy, but no rule currently prevents a market from paying for commentary about contests it is running books on.
- How far the claim travelled beyond the accounts that seeded it is unmeasured. There is no polling on belief in the Los Angeles version specifically, so its reach is documented in impressions rather than in convictions.
Point by point
The claim: A candidate who was clearly ahead ended up behind. Counts do not just reverse on their own.
What the record shows: They do, predictably, when different kinds of ballots are counted at different times. California accepts mail ballots postmarked by election day for a further seven days, and those late arrivals have leaned more Democratic than election-day ballots for several cycles, a pattern documented well before this race. A lead built on the first tranche is not a lead across the whole electorate; it is a partial sample that happens to be reported first. The reversal was in the direction the known pattern predicts, which is the opposite of an anomaly requiring explanation.
The claim: The count dragged on for days. A jurisdiction with nothing to hide would finish on election night.
What the record shows: The duration is set by statute, not discretion. California law gives counties up to 30 days to complete the official canvass, requires that mail ballots postmarked by election day and received within seven days be processed, and obliges officials to contact voters whose signatures are missing or mismatched so those ballots can be cured, with the cure deadline falling two days before certification. Each of those steps exists to count valid votes that would otherwise be discarded. Speed and enfranchisement are in tension here, and California has explicitly chosen the second.
The claim: The betting markets moved before the results did, which shows insiders knew the fix was in.
What the record shows: Prediction-market odds are a measure of what bettors think will happen, not a source of information about ballots. They move because participants read the same partial returns everyone else can see and update accordingly, and in a slow count with a closing gap the odds should move well ahead of the final call. Treating that as foreknowledge inverts the relationship: the market was following the count, not anticipating a scheme. The circularity is sharper than it first appears, since some of the accounts citing the odds as evidence were being paid by the companies that ran those markets.
The claim: Independent voices across social media all reached the same conclusion at once, which is telling.
What the record shows: Some of them were not independent on this subject. Reporting by NPR and Popular Information found that several high-reach posts advancing the fraud claim carried paid-partnership labels from Kalshi and Polymarket. That does not make everyone who shared the claim insincere, and most people repeating it were not paid. But the appearance of spontaneous consensus was partly purchased, and it collapsed quickly: Kalshi said it instructed the influencers to remove the posts after NPR asked, several came down, and Polymarket subsequently ended the sponsorship.
The claim: If this were legitimate, officials would explain how the process works instead of stonewalling.
What the record shows: They publish it. The California Secretary of State maintains a public page on the official canvass and vote-counting process, has issued releases explaining that counting can lawfully take up to 30 days and why, and county offices post regular result updates through the canvass. The information was available throughout the week the claim spread. Being unpersuaded by an explanation is different from not having been given one.
Other readings
Angles that don't fit neatly into the claim or its rebuttal, laid out and weighed, not endorsed.
The 'fix the process, not the motive' read
A reading worth taking seriously holds that the fraud claim is wrong but the frustration behind it is pointing at something real: an election system that cannot tell voters who won for a week invites suspicion regardless of how scrupulous it is, and confidence is part of what an election is for. On this view California's rules are defensible in isolation yet costly in aggregate, and reforms such as earlier processing of mail ballots would remove the vacuum that speculation fills without disenfranchising anyone. This is a policy argument rather than a factual dispute, and it does not support the rated claim: nothing about wishing the count were faster implies the count is dishonest. It is included because the two get conflated constantly, usually to the advantage of the conspiracy version.
Timeline
- 2018-2024California builds a reputation for slow counts. As the state moves to universal vote-by-mail and long acceptance windows, results routinely take weeks, and late-counted ballots lean more Democratic than election-day ballots. Election scholars call the pattern a 'blue shift' and attribute it to who votes by mail and when those ballots arrive. Each cycle, the same shift is offered online as evidence of fraud.
- 2026-06-02California holds its statewide primary. In the Los Angeles mayoral race, incumbent Karen Bass takes first place with about 34.8 percent, short of the majority that would have avoided a runoff. Fourteen candidates were on the ballot.
- 2026-06-03Early returns put Spencer Pratt, a former reality television star backed by MAGA-aligned figures, in second place with roughly 30.4 percent, ahead of City Councilwoman Nithya Raman on about 22.3 percent. On those numbers Pratt would face Bass in November, and coverage reports him as holding the second slot.
- 2026-06-03-2026-06-07The count continues, as state law requires. Mail ballots postmarked by election day keep arriving, signature-cure cases are worked through, and provisional ballots are checked one by one. Raman gains steadily on Pratt across successive updates.
- 2026-06-05-2026-06-07Unsubstantiated claims spread rapidly on X that a plot is underway to deny Pratt the runoff slot. Posts point to moving odds on the prediction markets Kalshi and Polymarket as though a betting line were evidence about ballots. A former Infowars host with a large following posts that 'California just keeps counting until the Democrat wins.'
- 2026-06-06The detail that sets this episode apart surfaces: several of the posts pushing the fraud claim carry 'paid partnership' labels from Kalshi and Polymarket. The companies whose shifting odds were being cited as proof of a rigged count were, in some cases, paying for the posts making that argument.
- 2026-06-07-2026-06-08NPR reports the arrangement. After being asked about the sponsored posts, Kalshi says it told the influencers to take them down, and several are deleted. Popular Information reports the same pattern and, in a follow-up, that Polymarket ends its sponsorship of election misinformation.
- 2026-06-08The Associated Press calls the second runoff slot for Nithya Raman, with more than 90 percent of ballots counted. Bass and Raman advance to the general election on 3 November 2026. No evidence of manipulation is produced by anyone, and no legal challenge establishes any.
Contradicted. Keep the two layers apart. What happened is not in dispute: in the 2 June 2026 Los Angeles mayoral primary, Spencer Pratt led the early returns for the second runoff slot with about 30.4 percent to City Councilwoman Nithya Raman's 22.3 percent, and over the following days Raman passed him, with the Associated Press calling the second slot for her on 8 June. The claim this file rates is that the shift was engineered, that California deliberately extends counting until a Democrat pulls ahead. It is debunked. The count runs long because state law requires it to: mail ballots postmarked by election day are accepted for seven more days, signature problems trigger a mandated cure process, and counties are given up to 30 days to finish. Late-counted ballots have leaned differently from election-day ballots in California for years, which is a known pattern rather than a hidden one, and nobody produced evidence of manipulation. One documented detail is worth carrying: some of the loudest fraud posts were sponsored content, labelled paid partnerships with the prediction markets Kalshi and Polymarket, and were taken down once reporters asked about them.
Reviewed by The Conspiratory Editors · Last reviewed July 31, 2026 · How we rate
Common questions
Is 'They count until they win' true?
Contradicted. Keep the two layers apart. What happened is not in dispute: in the 2 June 2026 Los Angeles mayoral primary, Spencer Pratt led the early returns for the second runoff slot with about 30.4 percent to City Councilwoman Nithya Raman's 22.3 percent, and over the following days Raman passed him, with the Associated Press calling the second slot for her on 8 June. The claim this file rates is that the shift was engineered, that California deliberately extends counting until a Democrat pulls ahead. It is debunked. The count runs long because state law requires it to: mail ballots postmarked by election day are accepted for seven more days, signature problems trigger a mandated cure process, and counties are given up to 30 days to finish. Late-counted ballots have leaned differently from election-day ballots in California for years, which is a known pattern rather than a hidden one, and nobody produced evidence of manipulation. One documented detail is worth carrying: some of the loudest fraud posts were sponsored content, labelled paid partnerships with the prediction markets Kalshi and Polymarket, and were taken down once reporters asked about them.
What is 'They count until they win'?
When the Los Angeles mayoral primary count stretched past a week in June 2026 and a candidate who had led for the second runoff spot fell behind, a familiar explanation arrived: California must be counting until it gets the answer it wants. This case file separates the parts. The lead genuinely did change, the count genuinely was slow, a…
What does the evidence show?
They do, predictably, when different kinds of ballots are counted at different times. California accepts mail ballots postmarked by election day for a further seven days, and those late arrivals have leaned more Democratic than election-day ballots for several cycles, a pattern documented well before this race. A lead built on the first…
Why do people believe it?
The experience is genuinely disorienting. Watching a candidate lead for days and then lose feels like something being taken away, in a way that a single final number never would. The emotional shape of a slow reversal fits theft better than it fits arithmetic, even when the arithmetic is ordinary.
What is still unresolved?
Slow counting is a legitimate policy argument even though it is not evidence of fraud. Whether a seven-day mail-ballot window and a 30-day canvass are the right trade-off against public confidence in fast results is a real debate, and people who want faster counts are not making a conspiracy claim by saying so.
Sources
- 1.Kalshi and Polymarket crack down on paid influencers claiming election fraud, NPR (2026)
- 2.Official Canvass: Vote Counting Process, California Secretary of State
- 3.California's vote counting process by law: counting can take 30 days to ensure voters' rights and integrity of elections, California Secretary of State (2022)
- 4.Why timing, postmarks and signatures matter for California's vote-by-mail majority, CapRadio (2026)
- 5.Why does it take so long to count votes in California? Here's why the process may feel even slower this election, CBS Los Angeles (2026)
- 6.Polymarket sponsoring election conspiracies from far-right influencers, Popular Information (2026)
- 7.July 2026 Intelligence Project Dispatch: Trends and incidents of the hard right, Southern Poverty Law Center (2026)
- 8.Los Angeles Mayor Primary 2026 Live Results, NBC News (2026)
- 9.2026 Los Angeles mayoral election, Wikipedia (2026)
Embed this verdict on your site
Paste this snippet to show our sourced verdict as a small card, with a link back to the full case file. Free to use.
<iframe src="https://theconspiratory.com/embed/california-vote-count-rigged" title="The Conspiratory verdict" width="520" height="190" style="border:0;max-width:100%" loading="lazy"></iframe>Help us investigate
This is a living case file. If you spot an error or know evidence we missed, tell us, and weigh in on where you land.
Where do you land?
Cast your read on this one.
Comments
Add your take. Comments are read and approved by a human before they appear, so keep it on topic and civil. Please do not accuse named, living people of crimes.