Betting money puts the odds of constitutional collapse in the United States at about one in 25. Anyone can wager three or four cents on Polymarket, Kalshi, or PredictIt that will pay out $1 if Donald Trump wins a third term in the 2028 election—an impossibility, according to the plain text of the Twenty-Second Amendment: “No person shall be elected to the office of the President more than twice.”
Dmitri Mehlhorn, a former Democratic strategist, thinks that the chance of political apocalypse is about 20 times higher—and that Americans need to start preparing now. He recently secured dual citizenship for his family on the Caribbean island of St. Kitts and is obsessively thinking through how people should respond if Trump tries to maintain power with the threat of force. He styles himself a doomsday philosopher of this worst-case scenario.
On a Tuesday last month, this effort brought him to a co-working space in Manhattan’s SoHo neighborhood to play a war game of sorts with about 15 finance professionals, nonprofit leaders, technology executives, and former Democratic-campaign advisers—and me.
“Let’s just use fiction to just imagine things so that we’re not all bogged down in prose,” Mehlhorn told the assembled players as the first glasses of wine were poured. The end of the enlightened order, it turns out, is best contemplated with libations and crudités. “Just imagine a world where certain assumptions are true.”
The assumptions were these: It is December 2025, and a term-limited U.S. president is rapidly consolidating control over the military and law enforcement, and pardoning criminal allies. To win, “President Buzz Windrip” and his team must increase their own wealth and maintain power or secure legal amnesty through 2030. They are competing against two other teams: one representing the business community, which seeks to protect and grow its capital and avoid prison, and one representing the defenders of the U.S. constitutional system and the rule of law. The game plays out over several 30-minute rounds, as players submit their actions so that AI agents—the game masters—can calculate the impact of each move and present new challenges. The premise of the game, like a round of Dungeons & Dragons, encourages players to check their personal politics and morality at the door—and to try to think more radically.
“You have a man who breaks the law, and that man is the strongman. Which bends, the man or the law? That’s the question,” Mehlhorn told me when we first met to discuss his project. “If the president has proven in his first term that he will ignore subpoenas and ignore congressional budget authorizations and pardon anybody who also does, then suddenly, there’s no power. What are the remaining checks? Every check is gone.”
[Read: Trump’s plan to subvert the midterms is already under way]
Mehlhorn’s ideas exist well outside the Democratic mainstream, which remains focused on the midterms and 2028 election—contests that he believes will have little meaning. He assumes that federal law enforcement will operate at Trump’s whims and that Democrat-run states need to build deterrence—threats of federal-tax boycotts, an expansive embrace of states’ rights, a new understanding of the importance of gun ownership, to name a few. His defenders see him as a colorful provocateur who is forcing uncomfortable conversations. His critics in Democratic politics consider his obsessions dangerous distractions, built on fanciful thinking, from the difficult work of winning elections and regaining power.
The game—held in a neighborhood better known for $5,000 handbags than fantastical End Times—is a test of Mehlhorn’s most outlandish ideas. The three teams prepared to split up and move into separate conference rooms to plot their strategy. “It can be zero or one or two winners,” Mehlhorn announced, “but it cannot be three winners.” He didn’t say what soon became clear: The zero-winner scenario means that the nation has collapsed into civil war.
If Mehlhorn is inclined to imagine a president discarding settled norms and restraints, perhaps it’s because he’s so often done so himself. He experienced Trump’s 2016 victory as a century-defining challenge to the American experiment. He responded by transforming himself from a venture capitalist into an iconoclastic Democratic strategist. His ticket was a friendship with Reid Hoffman, the billionaire co-founder of LinkedIn, whom Mehlhorn told me he knows from their time together at Stanford. As Hoffman’s donor-adviser, Mehlhorn helped direct more than $1 billion in anti-Trump spending from 2017 to 2024, particularly by rallying other wealthy liberals who had ties to Silicon Valley.
Their projects were often unconventional and frequently controversial. They backed prescient planning in 2020 for the possibility that Trump would resist the peaceful transfer of power, and two years later, they warned that “MAGA leaders intend to use 2022 midterm wins to install Trump in 2024 regardless of the vote.” They spoke openly of their efforts to dampen voter enthusiasm in Republican communities, the sort of vote suppression that other Democrats have long tried to combat.
Once it became clear that Trump would run for president again in 2024, Mehlhorn said that he and Hoffman helped fund the failed challenge to Trump’s candidacy under Section 3 of the Fourteenth Amendment, which prohibits federal officials who have “engaged in insurrection or rebellion” from holding office again. (The Colorado Supreme Court ruled in their favor, only to have the decision overturned by the U.S. Supreme Court.) They helped pay the legal bills of E. Jean Carroll, a writer who successfully won a civil judgment against Trump for sexual abuse and defamation. They supported the Republican candidacy of former South Carolina Governor Nikki Haley during the 2024 primaries, in an effort to weaken Trump. They were early backers of anti-Trump groups such as the Courier Newsroom and Republican Voters Against Trump, and of efforts aimed at turning out young urban voters through giveaways and parties. Frustrated with Democrats’ dependence on television advertising, they funded apolitical Facebook communities that were designed to attract audiences of less likely Democratic voters, who would find themselves bombarded with get-out-the-vote messages.
“The people who had been funding elections through 2016 were extremely conservative and cautious tactically,” Michael Podhorzer, a former strategist for the AFL-CIO who worked with Mehlhorn during the 2020 election cycle, told me. “What he brought from a Silicon Valley perspective was the idea, as he put it, that he would be happy if he funded 10 things and nine failed.”
Mehlhorn was also the rare political operative who did not mind making enemies—or opposing liberal causes, such as voting-rights reform in 2021, which he fought because he thought that it would distract from economic issues important to voters and lacked a path to becoming law. I spoke with one of Mehlhorn’s admiring colleagues, who requested anonymity to speak frankly and described him as “an asshole to a bunch of people who never have people tell them hard things.” Other Democratic strategists who know Mehlhorn were less admiring. Several declined to comment, because they did not share his eagerness to fight with allies. Others remain scarred by public and private feuds with him over strategy, experiences that Mehlhorn acknowledges as “bullying my own allies.”
In 2017, Mehlhorn secretly helped fund a deceptive social-media botnet campaign, which mimicked foreign-interference tactics and was designed to hurt a Republican Senate candidate. The backlash was swift. Hoffman later apologized for how his money had been used, saying that he had not been aware of the project. Mehlhorn said that he would no longer fund such efforts.
Then, in the summer of 2024, Mehlhorn crossed an even brighter line, suggesting in an email to a small network of politicos that the assassination attempt against Trump could have been a false-flag attack in the tradition of authoritarian regimes, such as the one run by Russian President Vladimir Putin. “I know I am prone to bias on this, but this is a classic Putin play and given the facts seems more plausible,” Mehlhorn wrote. “Look at the actual shot. Look at the staging. Look at how ready Trump is to rally.” He pressed recipients to investigate whether the shooting might have been staged so that Trump could benefit from the photos and the backlash.
When the conspiratorial email was published by Semafor, it ignited bipartisan outrage. Mehlhorn apologized. Within a week, Hoffman cut ties. Even as many Democrats distanced themselves, Mehlhorn quickly launched his own effort to support Kamala Harris’s bid for the presidency. But after Trump won, he announced that he was done with electoral politics. Mehlhorn didn’t think that Trump intended to allow free and fair elections again anyway. “This moment demands recalibration, resilience, and resolve,” he wrote to friends weeks after the election. “Even as authoritarian forces ascend, history shows that ingenuity and collaboration can prevail.”
[Read: Inside the Democratic rupture that undermined Kamala Harris’s presidential hopes]
He set about reinventing himself again, this time as a political thinker. He was, he said, no longer fighting the war but working “in the war college.” When we first met for this story, he gave me a list of 30 authors who have shaped his thinking, including the philosophers John Locke, Karl Popper, and Hannah Arendt; the astronomer Carl Sagan; and the retired federal judge J. Michael Luttig. But he felt that the world had already spun beyond the bounds of what traditional white papers could anticipate. So he decided to start working in fiction.
Mehlhorn has written a science-fiction book imagining America’s collapse into autocracy, and he programmed an AI to create a set of fictional Substack authors who write about parts of his political vision. Mehlhorn also founded a new group called the Atoll Society, named for the ring of islands that forms after an underwater volcanic eruption. In this metaphor, Mehlhorn casts the United States as we know it as the exploded volcano, a void. The surrounding islands—a national and an international “archipelago of light”—will have to form the new resistance.
In any role-playing game, the rules shape reality. Players are given win conditions that define their motives, and the game masters are given the ability to throw any obstacle in their way. As the cheese plates arrived in SoHo, the three teams found themselves in uncharted territory. (To get the full, immersive experience, I participated in the game as one of five players on the fictional president’s team. As a condition of my joining, I agreed not to identify or quote the other participants.)
The AI game masters told the players that an Oregon sheriff had started organizing local law enforcement in support of the Constitution and against the president’s consolidation of federal forces. The president’s team moved quickly to prosecute the sheriff for treason. The business community bought up national-media properties in the hope of increasing its leverage with the administration. The constitutional defenders began recruiting retired generals to counteract the president’s consolidation of power. Both the president’s team and the rule-of-law defenders launched back-channel negotiations with the capitalists.
Conditions deteriorated rapidly. The National Guard in blue states began to defect after being federalized by the president, loosening his hold on the military. Wary of the economic instability, the business community tried to relocate money outside the country, prompting the federal government to seize $600 billion. Then the president’s team launched a false-flag cyberattack on electricity infrastructure and pinned it on North Korea.
The exercise quickly descended into the realm of dystopian fiction, as the players wrestled with the same ideas that Mehlhorn has been developing. The constitutional defenders, operating through Democratic state leaders, found themselves leaning into the “states’ rights” powers of the Tenth Amendment. By the later rounds, the blue states had launched a tax boycott, worked to win over the military, embraced the Oregon sheriff as a champion of liberty, and abandoned the long-standing liberal aversion to guns.
“There are a lot of people with guns and military experience in this country. A lot of them. The Democratic Party does not have relationships with them at all,” Mehlhorn said. “You start to have alliances built around the Bill of Rights and start having a notion that the tax, the golden goose of all this wealth creation, is not just going to keep on paying if the federal Constitution is clearly in breach. That set of conversations can create a set of incentives where you actually deter the far-worse outcomes.”
Listening to Mehlhorn, I found it hard not to wonder how he would react if a similar exercise were conducted by Trump supporters—gaming out how to combat a dastardly Democratic power grab by rallying an armed coalition to stop it. Many in the president’s coalition are already warning about a wave of what the White House adviser Stephen Miller calls “domestic terrorism” to obstruct federal authority—a catchall term that Trump aides use sometimes to describe activity protected by the First Amendment, and sometimes to describe the physical obstruction of law-enforcement efforts. Meanwhile, Trump supporters, such as the lawyer Alan Dershowitz and the former Trump adviser Steve Bannon, have been busily preparing their own outlandish scenarios for the president’s indefinite rule. In important ways, the political debate has begun to shift away from what is happening to focus on the darkest visions of what might happen next.
Trump started selling Trump 2028 hats three months after taking office last year. Eight months in, he gathered military leadership to focus on the “war from within.” When Democratic lawmakers released a video reaffirming that the military code does not permit carrying out illegal orders, Trump accused them of seditious behavior “punishable by DEATH.”
Trump has been, for many of his most devoted supporters, delightfully ambiguous about whether he will actually try to run again in 2028. Trump said in March that “there are methods” for getting around the Constitution’s third-term prohibitions. When asked in August, he said that he would “probably not” seek another term but then added, “I’d like to.” He told reporters in October that the Constitution is “pretty clear” that he’s “not allowed to run,” only to have a White House spokesperson say last month that the nation would be “lucky” for such a thing to happen. His chief of staff, Susie Wiles, told Vanity Fair that he would not run again. “But he sure is having fun with it,” she added. “When you think of it, we shouldn’t even have an election,” Trump told Reuters about the 2026 midterms last week. (The White House spokesperson Abigail Jackson provided a statement attacking The Atlantic when she was asked about Trump’s plans.)
Dershowitz, a Harvard Law professor who has worked for both Trump and Jeffrey Epstein, told The Wall Street Journal last month that he had given Trump a draft of a book he wrote that includes suggestions for working around the Twenty-Second Amendment. In one such scenario, Dershowitz told the Journal, he suggests that Republican electors representing Trump, if he were to win, would decline to cast a vote in the Electoral College, which would pitch the decision to Congress, which would then appoint Trump. “They then select, and not elect, the president,” Dershowitz told the Journal, adding that he does not think Trump will pursue this path.
Mehlhorn believes that the most likely scenario is simpler: Trump would announce his candidacy for president late, leading the Republican Party to easily nominate him. The Supreme Court then would either split on the question of whether this is legal or rule against Trump with no effect. (J. D. Vance has repeatedly endorsed the idea of the president defying Supreme Court rulings.) As the votes are being counted, Mehlhorn imagines, Trump and federal law enforcement would declare rampant fraud in urban centers and move to disqualify residents of those districts’ votes. In January 2029, Vance would do what Mike Pence did not do in 2021: use his power to throw the election to Trump. If there is resistance, another a pro-Trump mob—including many who have already been pardoned for the January 6, 2021, attack—could provide backup.
“As president, who stops him?” Mehlhorn asked me as he described the scenario. “Who prevents that from happening? What person? What combination of people?”
[Read: Trump Exhaustion Syndrome]
Even if you accept the premise that Trump would take such radical steps to achieve a third term, though, there are several answers to this question. The first is that Trump will turn 82 before the 2028 election, making him older than Joe Biden was in 2024. (Mehlhorn agrees that health could be a factor.) The strongest internal check might be the American electorate, which would still have the ability to vote Trump’s allies out of office, splintering his coalition and potentially providing the votes for his impeachment and conviction. A recent Economist/YouGov poll found that 68 percent of the country, including 45 percent of Trump voters in 2024, did not want him to serve a third term. Trump’s recent pivot to try to reverse his falling approval numbers on the economy are an indication, if nothing else, of how much this year’s midterm elections will matter in determining whether he can hold together his coalition.
Then there are other parts of the executive branch, including military leaders, Trump’s inner circle in the West Wing, and the national party. Many people now view his consolidation of power as a necessary shift, but their support is based on the belief that these changes are happening within the constitutional framework. If Trump defies the Supreme Court, cancels urban votes en masse, or outwardly defies the Twenty-Second Amendment’s plain text, even some of his top advisers might defect. Investigations into the January 6 attack have revealed that many Trump aides, including ones who continue to work for him, were appalled at what happened that day, and some tried to intervene.
Mehlhorn welcomes the debate, in part because he is hoping that he is wrong. And after spending time with him, I realized that he is interested less in the most likely outcome of any situation than in the most dangerous plausible outcome. Bill Kristol, an anti-Trump activist who previously edited the conservative Weekly Standard, estimates that the chance that Trump runs for a third term is about one in four—much lower than Mehlhorn’s prediction but higher than the betting markets’. He told me that the scenarios raised by Mehlhorn have value.
“I do think there is just a consistent underestimation of the problem or the challenge of how alarmed we should all be,” Kristol said. “I think someone like Dmitri is useful in leaning against the overconfidence.”
Within an hour of the game’s completion, the AI game masters produced a 10-page verdict on the winners and losers, summarizing what had happened and why. In the final round of the game, the business community turned on the president—by a vote of 3 to 2 in their conference room—using their media empire to expose his role in the false-flag attack on U.S. infrastructure. The president eventually fled on a military aircraft to Hungary. The Oregon sheriff was sworn in as the president at the California state capitol instead of in Washington, D.C., a sign of how fragile the nation had become.
As with the first time the game was played, in Washington, there were two winners: the business community and the defenders of the prior constitutional system. The country as a whole, however, did poorly, falling into a recession. “The Republic held. Barely,” the AI game engine concluded. “And only because enough people, across enough fractures, chose to hold it.”
Mehlhorn said he’d programmed conditions that would allow the president’s team to win two-thirds of the time. But at both events where he has staged the game so far, the president’s team has been the only loser. In his view, this may reflect a design flaw that he can address ahead of future gatherings, including one planned in Midtown Manhattan this month and events scheduled later this year in Paris and Berlin. Business interests in real life would likely not move as a single block. Maybe his assumptions are simply wrong, and American democracy is more robust than he fears.
But Mehlhorn also suggested that the president’s team in both game runs may have been too timid—that the elite groups of players he had gathered lacked imaginative cruelty. “Perhaps participants from rule-of-law cultures instinctively hesitated to deploy violence,” he wrote to the players in a follow-up email, “even in simulation.”
The post Democracy’s Doomsday Prepper appeared first on The Atlantic.
